path
stringlengths 15
77
| type
stringclasses 1
value | project
stringclasses 1
value | commit_hash
stringlengths 40
40
| commit_message
stringlengths 15
198
| ground_truth
stringlengths 26
155
| main_code
stringlengths 176
2.5k
| context
stringlengths 91
9.37k
|
|---|---|---|---|---|---|---|---|
tests.coeditor.test_code_change/test_change_scope
|
Modified
|
temp-1
|
88a80dd462a4273154c266f5ef7d055f0b4fbfaf
|
Fix new scope headers. Fix unit tests.
|
<0>:<add> assert_str_equal(inner_class_code, indent(inner_attr1_expect, " " * 4))
|
# module: tests.coeditor.test_code_change
def test_change_scope():
<s>code, indent(f1_expect, " " * 4))
- assert_str_equal(f1_code, f1_expect)
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.subscopes[ProjectPath("code1", "f2")].spans_code
assert_str_equal(f2_code, f2_expect)
attr1_expect = dedent(
"""\
attr1: int
"""
)
attr1_code = scope.subscopes[ProjectPath("code1", "A")].spans_code
assert_str_equal(attr1_code, indent(attr1_expect, " " * 4))
method1_expect = dedent(
"""\
@staticmethod
def method1():
return 1
"""
)
method1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.method1")]
+ .all_code
- .spans_code
)
assert_str_equal(method1_code, indent(method1_expect, " " * 4))
inner_attr1_expect = dedent(
"""\
+ class B:
+ inner_attr1: int
- inner_attr1: int
"""
)
+ inner_class_code = (
- inner_attr1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.B")]
+ .all_code
- .spans_code
)
- assert_str_equal(inner_attr1_code, indent(inner_attr1_expect, " " * 8))
<0>
|
===========above chunk 0===========
# module: tests.coeditor.test_code_change
def test_change_scope():
# offset: -1
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
- def f1():
+ global x
- global x
+ x *= 5
- x *= 5
+ return x
- return x
"""
)
f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
+ assert_str_equal(f1_code, indent(f1_expect, " " * 4))
- assert_str_equal(f1_code, f1</s>
===========unchanged ref 0===========
at: textwrap
indent(text: str, prefix: str, predicate: Optional[Callable[[str], bool]]=...) -> str
|
tests.coeditor.test_code_change/TestChangedSpan.test_comments_change
|
Modified
|
temp-1
|
88a80dd462a4273154c266f5ef7d055f0b4fbfaf
|
Fix new scope headers. Fix unit tests.
|
<0>:<add> (Modified, 1),
|
# module: tests.coeditor.test_code_change
class TestChangedSpan:
def test_comments_change(self):
# have to update code as well for the changes to count
code2 = dedent(
"""\
import os
x = 1
# belongs to f1
def f1():
"added doc string"
global x
x *= 5
return x + 1
# belongs to main
if __name__ == "__main__":
print(f1() + x + 1) # belongs to print
"""
)
scope2 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code2))
self.check_changed_spans(
get_changed_spans(Modified(self.scope1, scope2)),
(Modified, -1),
- (Modified, 3),
<0> (Modified, 1),
)
|
===========unchanged ref 0===========
at: tests.coeditor.test_code_change.TestChangedSpan
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
"""
)
scope1 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code1))
check_changed_spans(changed_spans: Sequence[ChangedSpan], *expects: tuple[type, int])
===========changed ref 0===========
# module: tests.coeditor.test_code_change
def test_change_scope():
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
- def f1():
+ global x
- global x
+ x *= 5
- x *= 5
+ return x
- return x
"""
)
f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
+ assert_str_equal(f1_code, indent(f1_expect, " " * 4))
- assert_str_equal(f1_code, f1_expect)
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.</s>
===========changed ref 1===========
# module: tests.coeditor.test_code_change
def test_change_scope():
# offset: 1
<s> = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.subscopes[ProjectPath("code1", "f2")].spans_code
assert_str_equal(f2_code, f2_expect)
attr1_expect = dedent(
"""\
attr1: int
"""
)
attr1_code = scope.subscopes[ProjectPath("code1", "A")].spans_code
assert_str_equal(attr1_code, indent(attr1_expect, " " * 4))
method1_expect = dedent(
"""\
@staticmethod
def method1():
return 1
"""
)
method1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.method1")]
+ .all_code
- .spans_code
)
assert_str_equal(method1_code, indent(method1_expect, " " * 4))
inner_attr1_expect = dedent(
"""\
+ class B:
+ inner_attr1: int
- inner_attr1: int
"""
)
+ inner_class_code = (
- inner_attr1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.B")]
+ .all_code
- .spans_code
)
+ assert_str_equal(inner_class_code, indent(inner_attr1_expect, " " * 4))
- assert_str_equal(inner_attr1_code, indent(inner_attr1_expect, " " * 8))
|
coeditor.code_change/ChangeScope.header_code
|
Modified
|
temp-1
|
88a80dd462a4273154c266f5ef7d055f0b4fbfaf
|
Fix new scope headers. Fix unit tests.
|
<0>:<add> return "".join(snippets)
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def header_code(self) -> str:
if isinstance(self.tree, ptree.Module):
return f"# module: {self.path.module}"
# get the first non-empty line of self.tree
+ tree = self.tree
+ to_visit = []
+ parent = not_none(tree.parent)
+ while parent.type in ("decorated", "async_funcdef"):
+ to_visit.insert(0, parent.children[0])
+ parent = not_none(parent.parent)
+ to_visit.extend(tree.children)
+
+ snippets = list[str]()
- first_line = "#" if isinstance(self.tree, ptree.Function) else ""
+ for i, c in enumerate(to_visit):
- for i, c in enumerate(self.tree.children):
+ if c.type == "suite":
+ break
snippet = cast(str, c.get_code())
if i == 0:
# remove leading newlines
snippet = snippet.lstrip("\n")
assert isinstance(snippet, str)
- if count_lines(snippet) == 1:
- first_line += snippet
- else:
- first_line += snippet.splitlines()[0]
- break
- return first_line
+ snippets.append(snippet)
<0>
|
===========unchanged ref 0===========
at: coeditor._utils
not_none(x: Optional[T1]) -> T1
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[ProjectPath, Self]
at: functools
cached_property(func: Callable[[Any], _T])
at: parso.python.tree
Module(children)
at: parso.tree.BaseNode.__init__
self.children = children
self.parent: Optional[BaseNode] = None
at: parso.tree.NodeOrLeaf
__slots__ = ('parent',)
type: str
parent: 'Optional[BaseNode]'
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ @cached_property
+ def all_code(self) -> str:
+ return f"{self.header_code}\n{self.spans_code}"
+
===========changed ref 1===========
# module: coeditor.code_change
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
- ScopeTree = ptree.Module | ptree.Function | ptree.Class
PyNode = ptree.PythonBaseNode | ptree.PythonNode
_tlogger = TimeLogger()
===========changed ref 2===========
# module: tests.coeditor.test_code_change
class TestChangedSpan:
def test_comments_change(self):
# have to update code as well for the changes to count
code2 = dedent(
"""\
import os
x = 1
# belongs to f1
def f1():
"added doc string"
global x
x *= 5
return x + 1
# belongs to main
if __name__ == "__main__":
print(f1() + x + 1) # belongs to print
"""
)
scope2 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code2))
self.check_changed_spans(
get_changed_spans(Modified(self.scope1, scope2)),
(Modified, -1),
+ (Modified, 1),
- (Modified, 3),
(Modified, 1),
)
===========changed ref 3===========
# module: tests.coeditor.test_code_change
def test_change_scope():
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
- def f1():
+ global x
- global x
+ x *= 5
- x *= 5
+ return x
- return x
"""
)
f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
+ assert_str_equal(f1_code, indent(f1_expect, " " * 4))
- assert_str_equal(f1_code, f1_expect)
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.</s>
===========changed ref 4===========
# module: tests.coeditor.test_code_change
def test_change_scope():
# offset: 1
<s> = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.subscopes[ProjectPath("code1", "f2")].spans_code
assert_str_equal(f2_code, f2_expect)
attr1_expect = dedent(
"""\
attr1: int
"""
)
attr1_code = scope.subscopes[ProjectPath("code1", "A")].spans_code
assert_str_equal(attr1_code, indent(attr1_expect, " " * 4))
method1_expect = dedent(
"""\
@staticmethod
def method1():
return 1
"""
)
method1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.method1")]
+ .all_code
- .spans_code
)
assert_str_equal(method1_code, indent(method1_expect, " " * 4))
inner_attr1_expect = dedent(
"""\
+ class B:
+ inner_attr1: int
- inner_attr1: int
"""
)
+ inner_class_code = (
- inner_attr1_code = (
scope.subscopes[ProjectPath("code1", "A")]
.subscopes[ProjectPath("code1", "A.B")]
+ .all_code
- .spans_code
)
+ assert_str_equal(inner_class_code, indent(inner_attr1_expect, " " * 4))
- assert_str_equal(inner_attr1_code, indent(inner_attr1_expect, " " * 8))
|
coeditor.code_change/ChangeScope.from_tree
|
Modified
|
temp-1
|
88a80dd462a4273154c266f5ef7d055f0b4fbfaf
|
Fix new scope headers. Fix unit tests.
|
<0>:<add> subscopes[spath] = subscope
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
<s>as_any(s)):
- current_stmts.append(s)
- else:
- if current_stmts:
- spans.append(StatementSpan(current_stmts))
- current_stmts = []
- if current_stmts:
- spans.append(StatementSpan(current_stmts))
+ current_stmts = []
+ content = (
+ tree.children
+ if isinstance(tree, ptree.Module)
+ else cast(ptree.PythonNode, tree.get_suite()).children
+ )
+ for s in content:
+ # we don't create inner scopes for function contents
+ if is_func or _is_scope_statement(as_any(s)):
+ current_stmts.append(s)
+ else:
+ if current_stmts:
+ spans.append(StatementSpan(current_stmts))
+ current_stmts = []
+ if current_stmts:
+ spans.append(StatementSpan(current_stmts))
+
+ if is_func:
+ # we don't create inner scopes for function contents
+ return scope
+ for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
- for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
+ stree: ptree.Function | ptree.Class
- stree: ptree.Function | ptree.Class
+ spath = path.append(cast(ptree.Name, stree.name).value)
- spath = path.append(cast(ptree.Name, stree.name).value)
+ subscope = ChangeScope.from_tree(spath, stree)
- subscope = ChangeScope.from_tree(spath, stree)
- subscopes[spath] = subscope
<0> return scope
|
===========above chunk 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
# offset: -1
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes)
+ assert isinstance(tree, ScopeTree)
+ is_func = isinstance(tree, ptree.Function)
- if isinstance(tree, ptree.Function):
- span = StatementSpan([_to_decorated(tree)])
- spans.append(span)
- else:
- assert isinstance(tree, (ptree.Module, ptree.Class))
- current_stmts = []
- content = (
- tree.children
- if isinstance(tree, ptree.Module)
- else cast(ptree.PythonNode, tree.get_suite()).children
- )
- for s in content:
- if _is_scope_statement(as_any(s)):
- current_stmts.append(s)
- else:
- if current_stmts</s>
===========unchanged ref 0===========
at: coeditor._utils
as_any(x) -> Any
at: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[ProjectPath, Self])
_is_scope_statement(stmt: PyNode) -> bool
StatementSpan(statements: Sequence[PyNode])
at: coeditor.code_change.ChangeScope.header_code
snippets = list[str]()
at: parso.python.tree
PythonNode()
Module(children)
Function(children)
at: parso.python.tree.Scope
__slots__ = ()
get_suite()
at: parso.tree.BaseNode.__init__
self.children = children
at: parso.tree.NodeOrLeaf
type: str
get_code(include_prefix=True)
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ @cached_property
+ def all_code(self) -> str:
+ return f"{self.header_code}\n{self.spans_code}"
+
===========changed ref 1===========
# module: coeditor.code_change
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
- ScopeTree = ptree.Module | ptree.Function | ptree.Class
PyNode = ptree.PythonBaseNode | ptree.PythonNode
_tlogger = TimeLogger()
===========changed ref 2===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def header_code(self) -> str:
if isinstance(self.tree, ptree.Module):
return f"# module: {self.path.module}"
# get the first non-empty line of self.tree
+ tree = self.tree
+ to_visit = []
+ parent = not_none(tree.parent)
+ while parent.type in ("decorated", "async_funcdef"):
+ to_visit.insert(0, parent.children[0])
+ parent = not_none(parent.parent)
+ to_visit.extend(tree.children)
+
+ snippets = list[str]()
- first_line = "#" if isinstance(self.tree, ptree.Function) else ""
+ for i, c in enumerate(to_visit):
- for i, c in enumerate(self.tree.children):
+ if c.type == "suite":
+ break
snippet = cast(str, c.get_code())
if i == 0:
# remove leading newlines
snippet = snippet.lstrip("\n")
assert isinstance(snippet, str)
- if count_lines(snippet) == 1:
- first_line += snippet
- else:
- first_line += snippet.splitlines()[0]
- break
- return first_line
+ snippets.append(snippet)
+ return "".join(snippets)
===========changed ref 3===========
# module: tests.coeditor.test_code_change
class TestChangedSpan:
def test_comments_change(self):
# have to update code as well for the changes to count
code2 = dedent(
"""\
import os
x = 1
# belongs to f1
def f1():
"added doc string"
global x
x *= 5
return x + 1
# belongs to main
if __name__ == "__main__":
print(f1() + x + 1) # belongs to print
"""
)
scope2 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code2))
self.check_changed_spans(
get_changed_spans(Modified(self.scope1, scope2)),
(Modified, -1),
+ (Modified, 1),
- (Modified, 3),
(Modified, 1),
)
===========changed ref 4===========
# module: tests.coeditor.test_code_change
def test_change_scope():
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
- def f1():
+ global x
- global x
+ x *= 5
- x *= 5
+ return x
- return x
"""
)
f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
+ assert_str_equal(f1_code, indent(f1_expect, " " * 4))
- assert_str_equal(f1_code, f1_expect)
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
- f2_code = scope.</s>
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder._encode_parent_scopes
|
Modified
|
temp-1
|
88a80dd462a4273154c266f5ef7d055f0b4fbfaf
|
Fix new scope headers. Fix unit tests.
|
<0>:<add> return scope_tks
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_parent_scopes(
self, scope_changes: Sequence[Change[ChangeScope]], offset: int
) -> TokenSeq:
scope_tks = join_list(
(self._encode_scope_change(c) for c in scope_changes), sep=Newline_id
)
if offset != 0:
ending = encode_basic(f"\n# offset: {offset}\n")
else:
ending = [Newline_id]
+ scope_tks = truncate_section(
+ scope_tks + ending, TruncateAt.Left, self.max_scope_tks
+ )
- return scope_tks + ending
<0>
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
_encode_scope_change(c: Change[ChangeScope]) -> TokenSeq
at: coeditor.encoding
Newline_id = get_tk_id("\n")
truncate_section(sec: TokenSeq, direction: TruncateAt.Value, limit: int, add_bos: bool=True, inplace: bool=False) -> TokenSeq
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
+ max_scope_tks: int = 128
- max_scope_tks: int = 50
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ @cached_property
+ def all_code(self) -> str:
+ return f"{self.header_code}\n{self.spans_code}"
+
===========changed ref 2===========
# module: coeditor.code_change
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
- ScopeTree = ptree.Module | ptree.Function | ptree.Class
PyNode = ptree.PythonBaseNode | ptree.PythonNode
_tlogger = TimeLogger()
===========changed ref 3===========
# module: tests.coeditor.test_code_change
class TestChangedSpan:
def test_comments_change(self):
# have to update code as well for the changes to count
code2 = dedent(
"""\
import os
x = 1
# belongs to f1
def f1():
"added doc string"
global x
x *= 5
return x + 1
# belongs to main
if __name__ == "__main__":
print(f1() + x + 1) # belongs to print
"""
)
scope2 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code2))
self.check_changed_spans(
get_changed_spans(Modified(self.scope1, scope2)),
(Modified, -1),
+ (Modified, 1),
- (Modified, 3),
(Modified, 1),
)
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def header_code(self) -> str:
if isinstance(self.tree, ptree.Module):
return f"# module: {self.path.module}"
# get the first non-empty line of self.tree
+ tree = self.tree
+ to_visit = []
+ parent = not_none(tree.parent)
+ while parent.type in ("decorated", "async_funcdef"):
+ to_visit.insert(0, parent.children[0])
+ parent = not_none(parent.parent)
+ to_visit.extend(tree.children)
+
+ snippets = list[str]()
- first_line = "#" if isinstance(self.tree, ptree.Function) else ""
+ for i, c in enumerate(to_visit):
- for i, c in enumerate(self.tree.children):
+ if c.type == "suite":
+ break
snippet = cast(str, c.get_code())
if i == 0:
# remove leading newlines
snippet = snippet.lstrip("\n")
assert isinstance(snippet, str)
- if count_lines(snippet) == 1:
- first_line += snippet
- else:
- first_line += snippet.splitlines()[0]
- break
- return first_line
+ snippets.append(snippet)
+ return "".join(snippets)
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes)
+ assert isinstance(tree, ScopeTree)
+ is_func = isinstance(tree, ptree.Function)
- if isinstance(tree, ptree.Function):
- span = StatementSpan([_to_decorated(tree)])
- spans.append(span)
- else:
- assert isinstance(tree, (ptree.Module, ptree.Class))
- current_stmts = []
- content = (
- tree.children
- if isinstance(tree, ptree.Module)
- else cast(ptree.PythonNode, tree.get_suite()).children
- )
- for s in content:
- if _is_scope_statement(as_any(s)):
- current_stmts.append(s)
- else:
- if current_stmts:
- spans.append(StatementSpan(current_stmts))
- current_stmts = []
- if current_stmts:
- spans.append(StatementSpan(current_stmts))
+ current_stmts = []
+ content = (
+ tree.children
+ if isinstance(tree, ptree.Module)
+ else cast(ptree.PythonNode, tree.get_suite()).children
+ )
+ for s in content:
+ # we don't create inner scopes for function contents
+ if is_func or _is_scope_statement(as_any(s)):
+ current_stmts.append(s)
+ else:
+ if current_stmts:
+ spans.append(StatementSpan(current_stmts))
+ current_stmts = []
+ if current_stmts:
+ spans.append(StatementSpan(current_stmts))
+
+ if is_func:
+ # we don't create inner scopes for function contents
+ return scope</s>
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
# offset: 1
<s>stmts))
+
+ if is_func:
+ # we don't create inner scopes for function contents
+ return scope
+ for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
- for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
+ stree: ptree.Function | ptree.Class
- stree: ptree.Function | ptree.Class
+ spath = path.append(cast(ptree.Name, stree.name).value)
- spath = path.append(cast(ptree.Name, stree.name).value)
+ subscope = ChangeScope.from_tree(spath, stree)
- subscope = ChangeScope.from_tree(spath, stree)
+ subscopes[spath] = subscope
- subscopes[spath] = subscope
return scope
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.encode_change
|
Modified
|
temp-1
|
02aa6e2654a126be45b9b14e746d5bf2ad329e65
|
Fix static analysis.
|
<0>:<add> result.append((ProjectPath(used.full_name, ""), str(used)))
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
<s> _get_relevant(span: ChangedSpan):
if isinstance(span.change, Added):
# nothing to analyze
return []
path = span.parent_scopes[-1].earlier().path
line_usages = mod2usages[path.module]
+ all_used = set[PyDefinition]()
- all_used = set[PyFullName]()
l_start, l_end = span.line_range
for l in range(l_start, l_end + 1):
+ for pydef in line_usages.line2usages.get(l, set()):
+ if (
+ pydef.module == path.module
+ and l_start <= pydef.start_pos[0] <= l_end
+ ):
+ # skip self references
+ print(f"Skip: {pydef}")
+ continue
+ all_used.add(pydef)
- all_used.update(line_usages.line2usages.get(l, tuple()))
result = list[tuple[ProjectPath, str]]()
for used in all_used:
- result.append((ProjectPath("?", used), used))
<0> return result
sorted_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
yield CtxCodeChangeProblem(
span,
relevant_changes=sorted_cspans.copy(),
relevant_unchanged=_get_relevant(span),
)
sorted_cspans.append(span)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -1
def _get_relevant(span: ChangedSpan):
if isinstance(span.change, Added):
# nothing to analyze
return</s>
===========unchanged ref 0===========
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[tuple[ProjectPath, str]])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
fast_goto(script: jedi.Script, tree_name: tree.Name, *, follow_imports=False, follow_builtin_imports=False, only_stubs=False, prefer_stubs=False) -> set[classes.Name]
at: coeditor.ctx_change_encoder.CtxCodeChangeProblemGenerator.__init__
self.analysis = analysis
at: coeditor.ctx_change_encoder.CtxCodeChangeProblemGenerator.pre_edit_analysis
jmod = modules[mod_path]
script = jedi.Script(path=project.path / mod_path, project=project)
at: coeditor.ctx_change_encoder.JediUsageAnalysis
follow_imports: bool = True
only_same_project_usages: bool = False
get_module_usages(self, script: jedi.Script, proj_root: Path, silent: bool=False)
get_module_usages(script: jedi.Script, proj_root: Path, silent: bool=False)
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script.__init__
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
===========unchanged ref 1===========
at: jedi.api.project
Project(path, *, environment_path=None, load_unsafe_extensions=False, sys_path=None, added_sys_path=(), smart_sys_path=True)
at: jedi.api.project.Project
_environment = None
at: jedi.api.project.Project.__init__
self._path = path
at: pathlib.Path
__slots__ = ()
exists() -> bool
at: spot.static_analysis
ModuleName = str
sort_modules_by_imports(imports: Mapping[ModuleName, set[ModuleName]]) -> list[ModuleName]
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Mapping = _alias(collections.abc.Mapping, 2)
Sequence = _alias(collections.abc.Sequence, 1)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class LineUsageAnalysis:
+ line2usages: Mapping[int, set[PyDefinition]]
- line2usages: dict[int, set[PyFullName]]
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
+ @dataclass(unsafe_hash=True)
+ class PyDefinition:
+ full_name: PyFullName
+ module: ModuleName
+ file: Path
+ start_pos: tuple[int, int]
+ end_pos: tuple[int, int]
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
+ @dataclass(unsafe_hash=True)
+ class PyDefinition:
+ @staticmethod
+ def from_signatures(
+ name: classes.BaseName, project: Path | None = None
+ ) -> Iterable["PyDefinition"]:
+ if name.in_builtin_module():
+ return
+ for sig in name.get_signatures():
+ if (
+ not sig.in_builtin_module()
+ and (full_name := sig.full_name)
+ and (file := sig.module_path)
+ and (project in file.parents)
+ and (module := sig.module_name)
+ and (start_pos := sig.get_definition_start_position())
+ and (end_pos := sig.get_definition_end_position())
+ ):
+ full_name = PyFullName(full_name)
+ yield PyDefinition(full_name, module, file, start_pos, end_pos)
+
|
coeditor.ctx_change_encoder/JediUsageAnalysis.get_module_usages
|
Modified
|
temp-1
|
02aa6e2654a126be45b9b14e746d5bf2ad329e65
|
Fix static analysis.
|
<0>:<add> errors[repr(e)] = errors.setdefault(str(e), 0) + 1
|
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalysis:
def get_module_usages(
self, script: jedi.Script, proj_root: Path, silent: bool = False
):
<s> all_names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
+ resolve_cache = dict[_ObjId, set[PyDefinition]]()
for name in tqdm(all_names, f"Analyzing {script.path}", disable=silent):
name: tree.Name
line = name.start_pos[0]
usages = line2usages.setdefault(line, set())
try:
defs = fast_goto(
script,
name,
follow_imports=self.follow_imports,
follow_builtin_imports=False,
)
for d in defs:
- if (
- d.module_path
- and d.full_name
- and (
- not self.only_same_project_usages
- or (proj_root in d.module_path.parents)
- )
- ):
- usages.add(PyFullName(d.full_name))
+ key = _ObjId(id(d))
+ if (defs := resolve_cache.get(key)) is None:
+ defs = set(PyDefinition.from_signatures(d, proj_root))
+ resolve_cache[key] = defs
+ usages.update(defs)
+
except (AttributeError, AssertionError) as e:
+ text = repr(e)
- text = str(e)
errors[text] = errors.setdefault(text, 0) + 1
except ValueError as e:
# if the message is "not enough values to unpack"
if "not enough values to unpack (expected 2" in str(e):
- errors[str(e)] = errors.setdefault(str(e), 0) + 1
<0> else:
raise
return LineUsageAnalysis(line2usages)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalysis:
def get_module_usages(
self, script: jedi.Script, proj_root: Path, silent: bool = False
):
# offset: -1
jmod: tree.Module = script._module_node
+ line2usages = dict[int, set[PyDefinition]]()
- line2usages = dict[int, set[PyFullName]]()
all_names = [
name for k, names in jmod.get_used_names()._dict.items() for name in names
]
all_names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
+ resolve</s>
===========unchanged ref 0===========
at: coeditor._utils
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor.ctx_change_encoder
PyDefinition(full_name: PyFullName, module: ModuleName, file: Path, start_pos: tuple[int, int], end_pos: tuple[int, int])
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
_encode_parent_scopes(scope_changes: Sequence[Change[ChangeScope]], offset: int) -> TokenSeq
_encode_change(change: Change[str]) -> TokenSeq
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder._group_encode_changed_refs
all_chunks = list[TokenSeq]()
file_tks = TokenSeq()
at: coeditor.encoding
Newline_id = get_tk_id("\n")
break_into_chunks(tks: TokenSeq, header_f: Callable[[int], TokenSeq], chunk_size: int, overlap: int, right_to_left: bool=False, add_bos: bool=True, max_return_chunks: int | None=None) -> list[TokenSeq]
===========unchanged ref 1===========
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script.__init__
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
at: parso.python.tree
Module(children)
at: pathlib
Path()
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
+ @dataclass(unsafe_hash=True)
+ class PyDefinition:
+ full_name: PyFullName
+ module: ModuleName
+ file: Path
+ start_pos: tuple[int, int]
+ end_pos: tuple[int, int]
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalysis:
follow_imports: bool = True
+ only_same_project_usages: bool = False
- only_same_project_usages: bool = True
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
@dataclass
class LineUsageAnalysis:
+ line2usages: Mapping[int, set[PyDefinition]]
- line2usages: dict[int, set[PyFullName]]
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
+ @dataclass(unsafe_hash=True)
+ class PyDefinition:
+ @staticmethod
+ def from_signatures(
+ name: classes.BaseName, project: Path | None = None
+ ) -> Iterable["PyDefinition"]:
+ if name.in_builtin_module():
+ return
+ for sig in name.get_signatures():
+ if (
+ not sig.in_builtin_module()
+ and (full_name := sig.full_name)
+ and (file := sig.module_path)
+ and (project in file.parents)
+ and (module := sig.module_name)
+ and (start_pos := sig.get_definition_start_position())
+ and (end_pos := sig.get_definition_end_position())
+ ):
+ full_name = PyFullName(full_name)
+ yield PyDefinition(full_name, module, file, start_pos, end_pos)
+
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
def _get_relevant(span: ChangedSpan):
if isinstance(span.change, Added):
# nothing to analyze
return []
path = span.parent_scopes[-1].earlier().path
line_usages = mod2usages[path.module]
+ all_used = set[PyDefinition]()
- all_used = set[PyFullName]()
l_start, l_end = span.line_range
for l in range(l_start, l_end + 1):
+ for pydef in line_usages.line2usages.get(l, set()):
+ if (
+ pydef.module == path.module
+ and l_start <= pydef.start_pos[0] <= l_end
+ ):
+ # skip self references
+ print(f"Skip: {pydef}")
+ continue
+ all_used.add(pydef)
- all_used.update(line_usages.line2usages.get(l, tuple()))
result = list[tuple[ProjectPath, str]]()
for used in all_used:
+ result.append((ProjectPath(used.full_name, ""), str(used)))
- result.append((ProjectPath("?", used), used))
return result
sorted_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
yield CtxCodeChangeProblem(
span,
relevant_changes=sorted_cspans.copy(),
relevant_unchanged=</s>
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> yield CtxCodeChangeProblem(
span,
relevant_changes=sorted_cspans.copy(),
relevant_unchanged=_get_relevant(span),
)
sorted_cspans.append(span)
|
coeditor.code_change/_edits_from_commit_history
|
Modified
|
temp-1
|
3fc7535603679d696adc71d9db9ffc11f0ba700e
|
Analyzing only changed lines to speed up.
|
<0>:<add> checkout_commit(commit_next.hash)
|
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
<s>_path = RelPath(path.relative_to(project))
match path_change:
case Added():
mod = parse_module(path)
new_path2module[rel_path] = mod
changed[mod.mname] = JModuleChange.from_modules(Added(mod))
case Deleted():
mod = new_path2module.pop(rel_path)
changed[mod.mname] = JModuleChange.from_modules(Deleted(mod))
case Modified(path1, path2):
assert path1 == path2
mod_old = new_path2module[rel_path]
new_path2module[rel_path] = mod_new = parse_module(path)
changed[mod_new.mname] = JModuleChange.from_modules(
Modified(mod_old, mod_new)
)
with _tlogger.timed("post_edit_analysis"):
post_analysis = change_encoder.post_edit_analysis(
proj,
new_path2module,
+ changed,
- path_changes,
)
+
+ # now go backwards in time to perform pre-edit analysis
+ checkout_commit(commit_now.hash)
+ with _tlogger.timed("pre_edit_analysis"):
+ pre_analysis = change_encoder.pre_edit_analysis(
+ proj,
+ path2module,
+ changed,
+ )
<0>
pchange = JProjectChange(changed, commit_next)
with _tlogger.timed("encode_change"):
encs = change_encoder.encode_change(pchange, pre_analysis, post_analysis)
yield from encs
commit_now = commit_next
path2module = new_path2module
|
===========above chunk 0===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -1
<s>) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.endswith("A"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if tag.endswith("M"):
path_changes.append(Modified(path, path))
elif len(segs) == 3:
tag, path1, path2 = segs
assert tag.startswith("R")
if not is_src(path1) or not is_src(path2):
continue
path_changes.append(Deleted(path1))
path_changes.append(Added(path2))
- with _tlogger.timed("pre_edit_analysis"):
- pre_analysis = change_encoder.pre_edit_analysis(
- proj,
- path2module,
- path_changes,
- )
+ checkout_commit(commit_next.hash)
- # check out commit_next
- with _tlogger.timed("checkout"):
- subprocess.run(
- ["git", "checkout", commit_next.hash],
- cwd=project,
- capture_output=True,
- check=True,
- )
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
new_path2module = path2module.copy()
changed = dict[ModuleName, JModuleChange]()
for path_change in path_changes:
path = project / path_change.earlier()
rel_path = RelPath(path.relative_to(project))
match path_change:
case Added():
mod =</s>
===========above chunk 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -2
<s>logger.timed("checkout"):
- subprocess.run(
- ["git", "checkout", "-f", commit_now.hash],
- cwd=project,
- capture_output=True,
- check=True,
- )
+ checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
RelPath(f): parse_module(project / f) for f in get_python_files(project)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len(segs) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.ends</s>
===========above chunk 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -3
<s>)
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = copy.deepcopy(s._module_node) # needed due to reusing
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
+ def checkout_commit(commit_hash: str, force: bool = False):
+ with _tlogger.timed("checkout"):
+ subprocess.run(
+ ["git", "checkout", "-f", commit_hash],
+ cwd=project,
+ capture_output=True,
+ check=True,
+ )
- # turn this off so we don't have to deep copy the Modules
- # jedi.settings.fast_parser = False
- # Update: Have to use deep copy for now due to a bug in jedi: https://github.com/davidhalter/jedi/issues/1888
# checkout to the first commit
commit_now = history[-1]
- with</s>
===========above chunk 3===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -4
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project</s>
|
coeditor.ctx_change_encoder/PyDefinition.from_signatures
|
Modified
|
temp-1
|
3fc7535603679d696adc71d9db9ffc11f0ba700e
|
Analyzing only changed lines to speed up.
|
<0>:<add> yield PyDefinition(full_name, import_module, start_pos, end_pos)
|
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
@staticmethod
- def from_signatures(
- name: classes.BaseName, project: Path | None = None
- ) -> Iterable["PyDefinition"]:
+ def from_signatures(name: classes.BaseName) -> Iterable["PyDefinition"]:
+ cast(classes.Name, name).is_definition()
+ if (
+ not name.in_builtin_module()
- if name.in_builtin_module():
- return
- for sig in name.get_signatures():
- if (
- not sig.in_builtin_module()
+ and (full_name := name.full_name)
- and (full_name := sig.full_name)
- and (file := sig.module_path)
- and (project in file.parents)
+ and (import_module := name.module_name)
- and (module := sig.module_name)
+ and (start_pos := name.get_definition_start_position())
- and (start_pos := sig.get_definition_start_position())
+ and (end_pos := name.get_definition_end_position())
- and (end_pos := sig.get_definition_end_position())
+ ):
- ):
+ full_name = PyFullName(full_name)
- full_name = PyFullName(full_name)
- yield PyDefinition(full_name, module, file, start_pos, end_pos)
<0>
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
PyFullName = NewType("PyPathStr", str)
PyDefinition(full_name: PyFullName, import_module: ModuleName, start_pos: tuple[int, int], end_pos: tuple[int, int])
at: coeditor.ctx_change_encoder.PyDefinition
full_name: PyFullName
import_module: ModuleName
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: jedi.api.classes
BaseName(inference_state, name)
Name(inference_state, definition)
at: jedi.api.classes.BaseName
_mapping = {
'posixpath': 'os.path',
'riscospath': 'os.path',
'ntpath': 'os.path',
'os2emxpath': 'os.path',
'macpath': 'os.path',
'genericpath': 'os.path',
'posix': 'os',
'_io': 'io',
'_functools': 'functools',
'_collections': 'collections',
'_socket': 'socket',
'_sqlite3': 'sqlite3',
}
_tuple_mapping = dict((tuple(k.split('.')), v) for (k, v) in {
'argparse._ActionsContainer': 'argparse.ArgumentParser',
}.items())
in_builtin_module()
get_definition_start_position()
get_definition_end_position()
at: jedi.api.classes.Name
is_definition()
===========unchanged ref 1===========
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
Iterable = _alias(collections.abc.Iterable, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
+ """Note that the module and positions can be referring to either the import
+ statement or the actual definition."""
+
full_name: PyFullName
+ import_module: ModuleName
- module: ModuleName
- file: Path
start_pos: tuple[int, int]
end_pos: tuple[int, int]
===========changed ref 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project=proj)
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = copy.deepcopy(s._module_node) # needed due to reusing
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
+ def checkout_commit(commit_hash: str, force: bool = False):
+ with _tlogger.timed("checkout"):
+ subprocess.run(
+ ["git", "checkout", "-f", commit_hash],
+ cwd=project,
+ capture_output=True,
+ check=True,
+ )
- # turn this off so we don't have to deep copy the Modules
- # jedi.settings.fast_parser = False
- # Update: Have to use deep copy for now due to a bug in jedi: https://github.com/davidhalter/jedi/issues/1888
# checkout to the first commit
commit_now = history[-1]
- </s>
===========changed ref 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: 1
<s>/jedi/issues/1888
# checkout to the first commit
commit_now = history[-1]
- with _tlogger.timed("checkout"):
- subprocess.run(
- ["git", "checkout", "-f", commit_now.hash],
- cwd=project,
- capture_output=True,
- check=True,
- )
+ checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
RelPath(f): parse_module(project / f) for f in get_python_files(project)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len</s>
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.pre_edit_analysis
|
Modified
|
temp-1
|
3fc7535603679d696adc71d9db9ffc11f0ba700e
|
Analyzing only changed lines to speed up.
|
<0>:<add> result[mname] = line_usages
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def pre_edit_analysis(
self,
project: jedi.Project,
modules: Mapping[RelPath, JModule],
+ changes: Mapping[ModuleName, JModuleChange],
- file_changes: Sequence[Change[str]],
) -> Mapping[ModuleName, LineUsageAnalysis]:
<s>path)
result = dict[ModuleName, LineUsageAnalysis]()
+
+ src_map = {m.mname: f for f, m in modules.items()}
+ for mname, mchange in changes.items():
- for change in file_changes:
+ if not isinstance(mchange.module_change, Modified):
- if not isinstance(change, Modified):
continue
+
+ lines_to_analyze = set[int]()
+ for span in mchange.changed.values():
+ if span.change is Added:
+ continue
+ start, end = span.line_range
+ lines_to_analyze.update(range(start, end + 1))
+
+ mod_path = src_map[mname]
+ assert (
+ src_file := project.path / mod_path
+ ).exists(), f"src file missing: {src_file}"
- mod_path = RelPath(Path(change.before))
- jmod = modules[mod_path]
- assert (project.path / mod_path).exists()
+ script = jedi.Script(path=src_file, project=project)
- script = jedi.Script(path=project.path / mod_path, project=project)
+ line_usages = self.analysis.get_line_usages(
- line_usages = self.analysis.get_module_usages(
+ script, project.path, lines_to_analyze, silent=True
- script, project.path, silent=True
)
- result[jmod.mname] = line_usages
<0> return result
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def pre_edit_analysis(
self,
project: jedi.Project,
modules: Mapping[RelPath, JModule],
+ changes: Mapping[ModuleName, JModuleChange],
- file_changes: Sequence[Change[str]],
) -> Mapping[ModuleName, LineUsageAnalysis]:
# offset: -1
"Return the definition usages of each line."
# proot = Path(project._path)
result = dict[ModuleName, LineUsageAnalysis]()
+
+ src_map = {m.mname:</s>
===========unchanged ref 0===========
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
+ """Note that the module and positions can be referring to either the import
+ statement or the actual definition."""
+
full_name: PyFullName
+ import_module: ModuleName
- module: ModuleName
- file: Path
start_pos: tuple[int, int]
end_pos: tuple[int, int]
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
@staticmethod
- def from_signatures(
- name: classes.BaseName, project: Path | None = None
- ) -> Iterable["PyDefinition"]:
+ def from_signatures(name: classes.BaseName) -> Iterable["PyDefinition"]:
+ cast(classes.Name, name).is_definition()
+ if (
+ not name.in_builtin_module()
- if name.in_builtin_module():
- return
- for sig in name.get_signatures():
- if (
- not sig.in_builtin_module()
+ and (full_name := name.full_name)
- and (full_name := sig.full_name)
- and (file := sig.module_path)
- and (project in file.parents)
+ and (import_module := name.module_name)
- and (module := sig.module_name)
+ and (start_pos := name.get_definition_start_position())
- and (start_pos := sig.get_definition_start_position())
+ and (end_pos := name.get_definition_end_position())
- and (end_pos := sig.get_definition_end_position())
+ ):
- ):
+ full_name = PyFullName(full_name)
- full_name = PyFullName(full_name)
+ yield PyDefinition(full_name, import_module, start_pos, end_pos)
- yield PyDefinition(full_name, module, file, start_pos, end_pos)
===========changed ref 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project=proj)
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = copy.deepcopy(s._module_node) # needed due to reusing
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
+ def checkout_commit(commit_hash: str, force: bool = False):
+ with _tlogger.timed("checkout"):
+ subprocess.run(
+ ["git", "checkout", "-f", commit_hash],
+ cwd=project,
+ capture_output=True,
+ check=True,
+ )
- # turn this off so we don't have to deep copy the Modules
- # jedi.settings.fast_parser = False
- # Update: Have to use deep copy for now due to a bug in jedi: https://github.com/davidhalter/jedi/issues/1888
# checkout to the first commit
commit_now = history[-1]
- </s>
===========changed ref 3===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: 1
<s>/jedi/issues/1888
# checkout to the first commit
commit_now = history[-1]
- with _tlogger.timed("checkout"):
- subprocess.run(
- ["git", "checkout", "-f", commit_now.hash],
- cwd=project,
- capture_output=True,
- check=True,
- )
+ checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
RelPath(f): parse_module(project / f) for f in get_python_files(project)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len</s>
|
coeditor.code_change/ChangeScope.all_code
|
Modified
|
temp-1
|
b2f78bf15287dca9f1312cdd7720e6e9175fdeef
|
Fix line ranges. Improve analysis and add tests.
|
<0>:<add> return self.header_code + self.spans_code
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def all_code(self) -> str:
- return f"{self.header_code}\n{self.spans_code}"
<0>
|
===========changed ref 0===========
# module: coeditor.code_change
+ def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
+ if not can_be_empty and start >= end:
+ raise ValueError(f"Bad line range: {start=}, {end=}")
+ return _LineRange((start, end))
+
===========changed ref 1===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 2===========
# module: coeditor.code_change
+ def _strip_empty_lines(s: str):
+ s1 = s.lstrip("\n")
+ s2 = s1.rstrip("\n")
+ e_lines_left = len(s) - len(s1)
+ e_lines_right = len(s1) - len(s2)
+ return s2, e_lines_left, e_lines_right
+
===========changed ref 3===========
+ # module: tests.coeditor.testcases.usages
+
+
===========changed ref 4===========
+ # module: tests.coeditor.testcases.defs
+
+
===========changed ref 5===========
+ # module: tests.coeditor.testcases.usages
+ def get_named_changes(*args):
+ raise NotImplementedError
+
===========changed ref 6===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def iter_imports(self, tree):
+ raise NotImplementedError
+
===========changed ref 7===========
+ # module: tests.coeditor.testcases.usages
+ get_modified_spans = as_any(None)
+
===========changed ref 8===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @cached_property
+ def spans_code(self) -> str:
+ return "\n".join(s.code for s in self.spans)
+
===========changed ref 9===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def _to_scope(self) -> ChangeScope:
+ return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
+
===========changed ref 10===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ "A light wrapper around a jedi module."
+ mname: ModuleName
+ tree: ptree.Module
+
===========changed ref 11===========
+ # module: tests.coeditor.testcases.defs
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
+ ChangedSpan = NewType("ChangedSpan", str)
+
===========changed ref 12===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ """
+ A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
+ - functions and classes that are inside a parent function are considered hidden.
+ """
+
+ path: ProjectPath
+ tree: ScopeTree
+ spans: Sequence
+ subscopes: Mapping[ProjectPath, Self]
+
===========changed ref 13===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @staticmethod
+ def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
+ spans = []
+ subscopes = dict()
+ scope = ChangeScope(path, tree, spans, subscopes)
+ assert isinstance(tree, ScopeTree)
+ is_func = isinstance(tree, ptree.Function)
+
+ current_stmts = []
+ content = (
+ tree.children
+ if isinstance(tree, ptree.Module)
+ else cast(ptree.PythonNode, tree.get_suite()).children
+ )
+ raise NotImplementedError
+
===========changed ref 14===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ @cached_property
+ def imported_names(self):
+ names = set[ptree.Name]()
+ for stmt in self.iter_imports(self.tree):
+ if isinstance(stmt, ptree.ImportFrom):
+ for n in stmt.get_from_names():
+ assert isinstance(n, ptree.Name)
+ names.add(n)
+ elif isinstance(stmt, ptree.ImportName):
+ for n in stmt.get_defined_names():
+ assert isinstance(n, ptree.Name)
+ names.add(n)
+ return names
+
===========changed ref 15===========
+ # module: tests.coeditor.testcases.usages
+ def recurse(scope_change: Change[ChangeScope], parent_changes) -> Iterable[ChangedSpan]:
+ parent_changes = (*parent_changes, scope_change)
+ match scope_change:
+ case Modified(old_scope, new_scope):
+ # compute statement differences
+ yield from get_modified_spans(old_scope, new_scope, parent_changes)
+ for sub_change in get_named_changes(
+ old_scope.subscopes, new_scope.subscopes
+ ).values():
+ yield from recurse(sub_change, parent_changes)
+ case Added(scope) | Deleted(scope):
+ for span in scope.spans:
+ code_change = scope_change.new_value(span.code)
+ yield ChangedSpan(
+ code_change,
+ parent_changes,
+ span.line_range,
+ )
+ for s in scope.subscopes.values():
+ s_change = scope_change.new_value(s)
+ yield from recurse(s_change, parent_changes)
+
|
coeditor.code_change/_edits_from_commit_history
|
Modified
|
temp-1
|
b2f78bf15287dca9f1312cdd7720e6e9175fdeef
|
Fix line ranges. Improve analysis and add tests.
|
<0>:<add> pchange = JProjectChange(changed, modules_mod, commit_next)
|
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
<s>change:
case Added():
mod = parse_module(path)
new_path2module[rel_path] = mod
changed[mod.mname] = JModuleChange.from_modules(Added(mod))
case Deleted():
mod = new_path2module.pop(rel_path)
changed[mod.mname] = JModuleChange.from_modules(Deleted(mod))
case Modified(path1, path2):
assert path1 == path2
mod_old = new_path2module[rel_path]
new_path2module[rel_path] = mod_new = parse_module(path)
changed[mod_new.mname] = JModuleChange.from_modules(
Modified(mod_old, mod_new)
)
with _tlogger.timed("post_edit_analysis"):
post_analysis = change_encoder.post_edit_analysis(
proj,
new_path2module,
changed,
)
# now go backwards in time to perform pre-edit analysis
checkout_commit(commit_now.hash)
with _tlogger.timed("pre_edit_analysis"):
pre_analysis = change_encoder.pre_edit_analysis(
proj,
path2module,
changed,
)
checkout_commit(commit_next.hash)
+ modules_mod = Modified(path2module.values(), new_path2module.values())
- pchange = JProjectChange(changed, commit_next)
<0>
with _tlogger.timed("encode_change"):
encs = change_encoder.encode_change(pchange, pre_analysis, post_analysis)
yield from encs
commit_now = commit_next
path2module = new_path2module
|
===========above chunk 0===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -1
<s>0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len(segs) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.endswith("A"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if tag.endswith("M"):
path_changes.append(Modified(path, path))
elif len(segs) == 3:
tag, path1, path2 = segs
assert tag.startswith("R")
if not is_src(path1) or not is_src(path2):
continue
path_changes.append(Deleted(path1))
path_changes.append(Added(path2))
checkout_commit(commit_next.hash)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
new_path2module = path2module.copy()
changed = dict[ModuleName, JModuleChange]()
for path_change in path_changes:
path = project / path_change.earlier()
rel_path = RelPath(path.relative_to(project))
match path_change:
case Added():
mod = parse_module(path)
new_path2module[rel_path] =</s>
===========above chunk 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -2
<s> # m = parso.parse(path.read_text())
return JModule(mname, m)
def checkout_commit(commit_hash: str, force: bool = False):
with _tlogger.timed("checkout"):
subprocess.run(
["git", "checkout", "-f", commit_hash],
cwd=project,
capture_output=True,
check=True,
)
# checkout to the first commit
commit_now = history[-1]
checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
+ RelPath(f): parse_module(project / f)
+ for f in tqdm(
+ get_python_files(project), desc="building initial project", disable=silent
+ )
- RelPath(f): parse_module(project / f) for f in get_python_files(project)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(</s>
===========above chunk 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_encoder: ProjectChangeProcessor[TEnc],
ignore_dirs: set[str],
silent: bool,
) -> Iterable[TEnc]:
# offset: -3
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project=proj)
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = copy.deepcopy(s._module_node) # needed due to reusing
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))</s>
===========unchanged ref 0===========
at: coeditor._utils.TimeLogger
times: dict[str, list[float]] = field(default_factory=dict)
timed(name: str)
at: coeditor.code_change
_tlogger = TimeLogger()
JModule(mname: ModuleName, tree: ptree.Module)
JModuleChange(module_change: Change[JModule], changed: Mapping[ProjectPath, ChangedSpan])
get_python_files(project: Path)
DefaultIgnoreDirs = {".venv", ".mypy_cache", ".git", "venv", "build"}
TEnc = TypeVar("TEnc", covariant=True)
ProjectChangeProcessor()
NoProcessing()
at: coeditor.code_change.JModuleChange
module_change: Change[JModule]
changed: Mapping[ProjectPath, ChangedSpan]
from_modules(module_change: Change[JModule])
at: coeditor.common
RelPath = NewType("RelPath", Path)
run_command(args: Sequence[str], cwd: str | Path) -> str
print_err(*, sep: Optional[str]=..., end: Optional[str]=..., flush: bool=...) -> None
at: copy
deepcopy(x: _T, memo: Optional[Dict[int, Any]]=..., _nil: Any=...) -> _T
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script
_get_module_context()
at: jedi.api.Script.__init__
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.__init__
|
Modified
|
temp-1
|
b2f78bf15287dca9f1312cdd7720e6e9175fdeef
|
Fix line ranges. Improve analysis and add tests.
|
<0>:<add> analysis = JediUsageAnalyzer()
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
+ def __init__(self, analysis: "JediUsageAnalyzer | None"):
- def __init__(self, analysis: "JediUsageAnalysis | None"):
if analysis is None:
- analysis = JediUsageAnalysis()
<0> self.analysis = analysis
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[tuple[ProjectPath, str]])
JediUsageAnalyzer(follow_imports: bool=True)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
- jedi.cache.clear_time_caches = lambda: None
-
===========changed ref 1===========
+ # module: tests.coeditor.test_analysis
+
+
===========changed ref 2===========
+ # module: tests.coeditor.testcases.usages
+
+
===========changed ref 3===========
+ # module: tests.coeditor.testcases.defs
+
+
===========changed ref 4===========
+ # module: tests.coeditor.testcases.usages
+ def get_named_changes(*args):
+ raise NotImplementedError
+
===========changed ref 5===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def iter_imports(self, tree):
+ raise NotImplementedError
+
===========changed ref 6===========
+ # module: tests.coeditor.testcases.usages
+ get_modified_spans = as_any(None)
+
===========changed ref 7===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @cached_property
+ def spans_code(self) -> str:
+ return "\n".join(s.code for s in self.spans)
+
===========changed ref 8===========
+ # module: tests.coeditor.test_analysis
+ testcase_root = proj_root() / "tests" / "coeditor" / "testcases"
+
===========changed ref 9===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def _to_scope(self) -> ChangeScope:
+ return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
+
===========changed ref 10===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ "A light wrapper around a jedi module."
+ mname: ModuleName
+ tree: ptree.Module
+
===========changed ref 11===========
+ # module: tests.coeditor.testcases.defs
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
+ ChangedSpan = NewType("ChangedSpan", str)
+
===========changed ref 12===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def all_code(self) -> str:
+ return self.header_code + self.spans_code
- return f"{self.header_code}\n{self.spans_code}"
===========changed ref 13===========
# module: coeditor.code_change
@dataclass
class JProjectChange:
changed: Mapping[ModuleName, JModuleChange]
+ all_modules: Modified[Collection[JModule]]
commit_info: "CommitInfo | None"
===========changed ref 14===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
+ @property
+ def header_line_range(self) -> _LineRange:
+ parent_scope = self.parent_scopes[-1].earlier()
+ hrange = parent_scope.header_line_range
+ return hrange
+
===========changed ref 15===========
# module: coeditor.code_change
+ def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
+ if not can_be_empty and start >= end:
+ raise ValueError(f"Bad line range: {start=}, {end=}")
+ return _LineRange((start, end))
+
===========changed ref 16===========
+ # module: tests.coeditor.test_analysis
+ def assert_has_usages(defs: Collection[PyDefinition], *full_names: str):
+ nameset = list(d.full_name for d in defs)
+ for name in full_names:
+ if PyFullName(name) not in nameset:
+ raise AssertionError(f"{name} not in {nameset}")
+
===========changed ref 17===========
+ # module: tests.coeditor.test_analysis
+ def assert_no_usages(defs: Collection[PyDefinition], *full_names: str):
+ nameset = list(d.full_name for d in defs)
+ for name in full_names:
+ if PyFullName(name) in nameset:
+ raise AssertionError(f"{name} should not be in {nameset}")
+
===========changed ref 18===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 19===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
"Represents the changes made to a statement span."
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
- line_range: tuple[int, int]
- old_statements: Sequence[PyNode]
+ line_range: _LineRange
===========changed ref 20===========
# module: coeditor.code_change
+ def _strip_empty_lines(s: str):
+ s1 = s.lstrip("\n")
+ s2 = s1.rstrip("\n")
+ e_lines_left = len(s) - len(s1)
+ e_lines_right = len(s1) - len(s2)
+ return s2, e_lines_left, e_lines_right
+
===========changed ref 21===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ """
+ A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
+ - functions and classes that are inside a parent function are considered hidden.
+ """
+
+ path: ProjectPath
+ tree: ScopeTree
+ spans: Sequence
+ subscopes: Mapping[ProjectPath, Self]
+
===========changed ref 22===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @staticmethod
+ def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
+ spans = []
+ subscopes = dict()
+ scope = ChangeScope(path, tree, spans, subscopes)
+ assert isinstance(tree, ScopeTree)
+ is_func = isinstance(tree, ptree.Function)
+
+ current_stmts = []
+ content = (
+ tree.children
+ if isinstance(tree, ptree.Module)
+ else cast(ptree.PythonNode, tree.get_suite()).children
+ )
+ raise NotImplementedError
+
===========changed ref 23===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ @cached_property
+ def imported_names(self):
+ names = set[ptree.Name]()
+ for stmt in self.iter_imports(self.tree):
+ if isinstance(stmt, ptree.ImportFrom):
+ for n in stmt.get_from_names():
+ assert isinstance(n, ptree.Name)
+ names.add(n)
+ elif isinstance(stmt, ptree.ImportName):
+ for n in stmt.get_defined_names():
+ assert isinstance(n, ptree.Name)
+ names.add(n)
+ return names
+
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.pre_edit_analysis
|
Modified
|
temp-1
|
b2f78bf15287dca9f1312cdd7720e6e9175fdeef
|
Fix line ranges. Improve analysis and add tests.
|
<0>:<add> lines_to_analyze.update(range(*span.header_line_range))
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def pre_edit_analysis(
self,
project: jedi.Project,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> Mapping[ModuleName, LineUsageAnalysis]:
"Return the definition usages of each line."
# proot = Path(project._path)
result = dict[ModuleName, LineUsageAnalysis]()
src_map = {m.mname: f for f, m in modules.items()}
for mname, mchange in changes.items():
if not isinstance(mchange.module_change, Modified):
continue
lines_to_analyze = set[int]()
for span in mchange.changed.values():
if span.change is Added:
continue
- start, end = span.line_range
+ lines_to_analyze.update(range(*span.line_range))
- lines_to_analyze.update(range(start, end + 1))
<0>
mod_path = src_map[mname]
assert (
src_file := project.path / mod_path
).exists(), f"src file missing: {src_file}"
script = jedi.Script(path=src_file, project=project)
line_usages = self.analysis.get_line_usages(
script, project.path, lines_to_analyze, silent=True
)
result[mname] = line_usages
return result
|
===========unchanged ref 0===========
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
at: coeditor.ctx_change_encoder.CtxCodeChangeProblemGenerator.__init__
analysis = JediUsageAnalyzer()
self.analysis = analysis
at: coeditor.ctx_change_encoder.JediUsageAnalyzer
follow_imports: bool = True
get_line_usages(script: jedi.Script, proj_root: Path, lines_to_analyze: Collection[int], silent: bool=False)
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.project
Project(path, *, environment_path=None, load_unsafe_extensions=False, sys_path=None, added_sys_path=(), smart_sys_path=True)
at: jedi.api.project.Project
_environment = None
at: pathlib.Path
__slots__ = ()
exists() -> bool
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
+ def __init__(self, analysis: "JediUsageAnalyzer | None"):
- def __init__(self, analysis: "JediUsageAnalysis | None"):
if analysis is None:
+ analysis = JediUsageAnalyzer()
- analysis = JediUsageAnalysis()
self.analysis = analysis
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
- jedi.cache.clear_time_caches = lambda: None
-
===========changed ref 2===========
+ # module: tests.coeditor.test_analysis
+
+
===========changed ref 3===========
+ # module: tests.coeditor.testcases.usages
+
+
===========changed ref 4===========
+ # module: tests.coeditor.testcases.defs
+
+
===========changed ref 5===========
+ # module: tests.coeditor.testcases.usages
+ def get_named_changes(*args):
+ raise NotImplementedError
+
===========changed ref 6===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def iter_imports(self, tree):
+ raise NotImplementedError
+
===========changed ref 7===========
+ # module: tests.coeditor.testcases.usages
+ get_modified_spans = as_any(None)
+
===========changed ref 8===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @cached_property
+ def spans_code(self) -> str:
+ return "\n".join(s.code for s in self.spans)
+
===========changed ref 9===========
+ # module: tests.coeditor.test_analysis
+ testcase_root = proj_root() / "tests" / "coeditor" / "testcases"
+
===========changed ref 10===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ def _to_scope(self) -> ChangeScope:
+ return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
+
===========changed ref 11===========
+ # module: tests.coeditor.testcases.usages
+ @dataclass
+ class JModule:
+ "A light wrapper around a jedi module."
+ mname: ModuleName
+ tree: ptree.Module
+
===========changed ref 12===========
+ # module: tests.coeditor.testcases.defs
+ ScopeTree = ptree.Function | ptree.Class | ptree.Module
+ ChangedSpan = NewType("ChangedSpan", str)
+
===========changed ref 13===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@cached_property
def all_code(self) -> str:
+ return self.header_code + self.spans_code
- return f"{self.header_code}\n{self.spans_code}"
===========changed ref 14===========
# module: coeditor.code_change
@dataclass
class JProjectChange:
changed: Mapping[ModuleName, JModuleChange]
+ all_modules: Modified[Collection[JModule]]
commit_info: "CommitInfo | None"
===========changed ref 15===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
+ @property
+ def header_line_range(self) -> _LineRange:
+ parent_scope = self.parent_scopes[-1].earlier()
+ hrange = parent_scope.header_line_range
+ return hrange
+
===========changed ref 16===========
# module: coeditor.code_change
+ def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
+ if not can_be_empty and start >= end:
+ raise ValueError(f"Bad line range: {start=}, {end=}")
+ return _LineRange((start, end))
+
===========changed ref 17===========
+ # module: tests.coeditor.test_analysis
+ def assert_has_usages(defs: Collection[PyDefinition], *full_names: str):
+ nameset = list(d.full_name for d in defs)
+ for name in full_names:
+ if PyFullName(name) not in nameset:
+ raise AssertionError(f"{name} not in {nameset}")
+
===========changed ref 18===========
+ # module: tests.coeditor.test_analysis
+ def assert_no_usages(defs: Collection[PyDefinition], *full_names: str):
+ nameset = list(d.full_name for d in defs)
+ for name in full_names:
+ if PyFullName(name) in nameset:
+ raise AssertionError(f"{name} should not be in {nameset}")
+
===========changed ref 19===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 20===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
"Represents the changes made to a statement span."
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
- line_range: tuple[int, int]
- old_statements: Sequence[PyNode]
+ line_range: _LineRange
===========changed ref 21===========
# module: coeditor.code_change
+ def _strip_empty_lines(s: str):
+ s1 = s.lstrip("\n")
+ s2 = s1.rstrip("\n")
+ e_lines_left = len(s) - len(s1)
+ e_lines_right = len(s1) - len(s2)
+ return s2, e_lines_left, e_lines_right
+
===========changed ref 22===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ """
+ A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
+ - functions and classes that are inside a parent function are considered hidden.
+ """
+
+ path: ProjectPath
+ tree: ScopeTree
+ spans: Sequence
+ subscopes: Mapping[ProjectPath, Self]
+
===========changed ref 23===========
+ # module: tests.coeditor.testcases.defs
+ @dataclass
+ class ChangeScope:
+ @staticmethod
+ def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
+ spans = []
+ subscopes = dict()
+ scope = ChangeScope(path, tree, spans, subscopes)
+ assert isinstance(tree, ScopeTree)
+ is_func = isinstance(tree, ptree.Function)
+
+ current_stmts = []
+ content = (
+ tree.children
+ if isinstance(tree, ptree.Module)
+ else cast(ptree.PythonNode, tree.get_suite()).children
+ )
+ raise NotImplementedError
+
|
coeditor.encoding/change_to_line_diffs
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> diffs = compute_line_diffs(splitlines(before), splitlines(after))
|
# module: coeditor.encoding
def change_to_line_diffs(change: Change[str]) -> list[str]:
"Encode a change as a token sequence."
match change:
case Modified(before, after):
+ if change.unchanged:
+ diffs = []
+ else:
- diffs = compute_line_diffs(splitlines(before), splitlines(after))
<0> # rearrange_diffs_(diffs)
if not diffs:
# as a special case, `unified_diff` would return an empty when there is no change.
diffs = [" " + l for l in splitlines(before)]
case Added(after):
diffs = ["+" + l for l in splitlines(after)]
case Deleted(before):
diffs = ["-" + l for l in splitlines(before)]
case _:
raise ValueError(f"Invalid change type: {change}.")
return diffs
|
===========unchanged ref 0===========
at: coeditor._utils
compute_line_diffs(before: Sequence[str], after: Sequence[str], keep_explain_lines: bool=False)
at: coeditor.common
splitlines(text: str) -> list[str]
|
tests.coeditor.test_code_change/test_change_scope
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> inner_class_code = scope.subscopes["A"].subscopes["B"].all_code
|
# module: tests.coeditor.test_code_change
def test_change_scope():
<s>
assert_str_equal(f1_code, indent(f1_expect, " " * 4))
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes["f2"].all_code
- f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
assert_str_equal(f2_code, f2_expect)
attr1_expect = dedent(
"""\
attr1: int
"""
)
+ attr1_code = scope.subscopes["A"].spans_code
- attr1_code = scope.subscopes[ProjectPath("code1", "A")].spans_code
assert_str_equal(attr1_code, indent(attr1_expect, " " * 4))
method1_expect = dedent(
"""\
@staticmethod
def method1():
return 1
"""
)
- method1_code = (
- scope.subscopes[ProjectPath("code1", "A")]
- .subscopes[ProjectPath("code1", "A.method1")]
- .all_code
- )
+ method1_code = scope.subscopes["A"].subscopes["method1"].all_code
assert_str_equal(method1_code, indent(method1_expect, " " * 4))
inner_attr1_expect = dedent(
"""\
class B:
inner_attr1: int
"""
)
- inner_class_code = (
- scope.subscopes[ProjectPath("code1", "A")]
- .subscopes[ProjectPath("code1", "A.B")]
- .all_code
- )
<0> assert_str_equal(inner_class_code, indent(inner_attr1_expect, " " * 4))
|
===========above chunk 0===========
# module: tests.coeditor.test_code_change
def test_change_scope():
# offset: -1
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
global x
x *= 5
return x
"""
)
+ f1_code = scope.subscopes["f1"].spans_code
- f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
assert_str_equal(f1_code, indent(f1_expect, " " * 4))
f2</s>
===========unchanged ref 0===========
at: tests.coeditor.test_code_change.TestChangedSpan
scope1 = ChangeScope.from_tree(ProjectPath("code1", ""), code_to_module(code1))
at: textwrap
indent(text: str, prefix: str, predicate: Optional[Callable[[str], bool]]=...) -> str
===========changed ref 0===========
# module: coeditor.encoding
def change_to_line_diffs(change: Change[str]) -> list[str]:
"Encode a change as a token sequence."
match change:
case Modified(before, after):
+ if change.unchanged:
+ diffs = []
+ else:
+ diffs = compute_line_diffs(splitlines(before), splitlines(after))
- diffs = compute_line_diffs(splitlines(before), splitlines(after))
# rearrange_diffs_(diffs)
if not diffs:
# as a special case, `unified_diff` would return an empty when there is no change.
diffs = [" " + l for l in splitlines(before)]
case Added(after):
diffs = ["+" + l for l in splitlines(after)]
case Deleted(before):
diffs = ["-" + l for l in splitlines(before)]
case _:
raise ValueError(f"Invalid change type: {change}.")
return diffs
|
coeditor.code_change/ChangeScope.from_tree
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> subscopes[name] = subscope
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
+ scope = ChangeScope(path, tree, spans, subscopes, None)
- scope = ChangeScope(path, tree, spans, subscopes)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(current_stmts, scope))
- spans.append(StatementSpan(current_stmts))
current_stmts = []
if current_stmts:
+ spans.append(StatementSpan(current_stmts, scope))
- spans.append(StatementSpan(current_stmts))
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
+ name = cast(ptree.Name, stree.name).value
- spath = path.append(cast(ptree.Name, stree.name).value)
+ spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
+ subscope.parent_scope = scope
- subscopes[spath] = subscope
<0> return scope
|
===========unchanged ref 0===========
at: coeditor._utils
not_none(x: Optional[T1]) -> T1
assert_eq(x: T1, y: T1, message: Callable[[], str]=lambda: "") -> None
at: coeditor.code_change
_LineRange = NewType("LineRange", tuple[int, int])
_line_range(start: int, end: int, can_be_empty: bool=False) -> _LineRange
_strip_empty_lines(s: str)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
at: coeditor.code_change.ChangeScope.__post_init__
tree = self.tree
at: coeditor.code_change.StatementSpan.__post_init__
self.line_range: _LineRange = _line_range(start, end)
at: coeditor.common
count_lines(text: str) -> int
at: parso.tree
NodeOrLeaf()
at: parso.tree.BaseNode.__init__
self.children = children
self.parent: Optional[BaseNode] = None
at: parso.tree.NodeOrLeaf
__slots__ = ('parent',)
type: str
parent: 'Optional[BaseNode]'
get_code(include_prefix=True)
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search_span(self, line: int) -> "StatementSpan | None":
+ for span in self.spans:
+ if span.line_range[0] <= line < span.line_range[1]:
+ return span
+ return None
+
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search(self, path: ElemPath, line: int) -> Self | "StatementSpan":
+ scope = self._search_scope(path)
+ if scope.header_line_range[0] <= line < scope.header_line_range[1]:
+ return scope
+ span = scope._search_span(line)
+ return span or scope
+
===========changed ref 2===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search_scope(self, path: ElemPath) -> Self:
+ """Find the scope that can potentially contain the given path. Follow the
+ path segments until no more subscopes are found."""
+ segs = path.split(".")
+ scope = self
+ for s in segs:
+ if s in scope.subscopes:
+ scope = scope.subscopes[s]
+ else:
+ break
+ return scope
+
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
"""
A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
- functions and classes that are inside a parent function are considered hidden.
"""
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
+ subscopes: Mapping[str, Self]
- subscopes: Mapping[ProjectPath, Self]
+ parent_scope: "ChangeScope | None"
===========changed ref 4===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
+ @staticmethod
+ def from_unchanged(v: T1) -> "Modified[T1]":
+ return Modified(v, v, unchanged=True)
+
===========changed ref 5===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
before: E1
after: E1
+ unchanged: bool = False
===========changed ref 6===========
# module: coeditor.encoding
def change_to_line_diffs(change: Change[str]) -> list[str]:
"Encode a change as a token sequence."
match change:
case Modified(before, after):
+ if change.unchanged:
+ diffs = []
+ else:
+ diffs = compute_line_diffs(splitlines(before), splitlines(after))
- diffs = compute_line_diffs(splitlines(before), splitlines(after))
# rearrange_diffs_(diffs)
if not diffs:
# as a special case, `unified_diff` would return an empty when there is no change.
diffs = [" " + l for l in splitlines(before)]
case Added(after):
diffs = ["+" + l for l in splitlines(after)]
case Deleted(before):
diffs = ["-" + l for l in splitlines(before)]
case _:
raise ValueError(f"Invalid change type: {change}.")
return diffs
===========changed ref 7===========
# module: tests.coeditor.test_code_change
def test_change_scope():
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
for i, code in enumerate(global_spans):
assert_str_equal(scope.spans[i].code, code)
f1_expect = dedent(
"""\
global x
x *= 5
return x
"""
)
+ f1_code = scope.subscopes["f1"].spans_code
- f1_code = scope.subscopes[ProjectPath("code1", "f1")].spans_code
assert_str_equal(f1_code, indent(f1_expect, " " * 4))
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
+ f2_code = scope.subscopes["f2"].all_code
- f2_code = scope.subscopes[ProjectPath("code1", "f2")].all_code
assert_str_equal(f2_code, f2</s>
|
coeditor.code_change/JModuleChange.from_modules
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> module_change.map(lambda m: m.as_scope), tuple()
|
# module: coeditor.code_change
@dataclass
class JModuleChange:
@staticmethod
def from_modules(module_change: Change[JModule]):
"Compute the change spans from two versions of the same module."
with _tlogger.timed("JModuleChange.from_modules"):
changed = dict[ProjectPath, ChangedSpan]()
for cspan in get_changed_spans(
- module_change.map(lambda m: m._to_scope()), tuple()
<0> ):
path = cspan.parent_scopes[-1].earlier().path
changed[path] = cspan
return JModuleChange(module_change, changed)
|
===========unchanged ref 0===========
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: _LineRange
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: functools
cached_property(func: Callable[[Any], _T])
at: parso.python.tree
Module(children)
at: spot.static_analysis
ModuleName = str
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class JModule:
- def _to_scope(self) -> ChangeScope:
- return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
-
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def ancestors(self) -> list[Self]:
+ scope = self
+ result = [scope]
+ while scope := scope.parent_scope:
+ result.append(scope)
+ result.reverse()
+ return result
+
===========changed ref 2===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search_span(self, line: int) -> "StatementSpan | None":
+ for span in self.spans:
+ if span.line_range[0] <= line < span.line_range[1]:
+ return span
+ return None
+
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
"""
A statement span is a set of lines inside the same change scope. It is the basic unit of code changes handled by our model.
- For a modified function, the span is the function itself.
- For a modified module, the spans are the regions between the functions and classes plus
the spans recursively generated.
- For a modified class, the spans are the regions between methods plus
the spans recursively generated.
"""
statements: Sequence[PyNode]
+ scope: ChangeScope
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search(self, path: ElemPath, line: int) -> Self | "StatementSpan":
+ scope = self._search_scope(path)
+ if scope.header_line_range[0] <= line < scope.header_line_range[1]:
+ return scope
+ span = scope._search_span(line)
+ return span or scope
+
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
+ def _search_scope(self, path: ElemPath) -> Self:
+ """Find the scope that can potentially contain the given path. Follow the
+ path segments until no more subscopes are found."""
+ segs = path.split(".")
+ scope = self
+ for s in segs:
+ if s in scope.subscopes:
+ scope = scope.subscopes[s]
+ else:
+ break
+ return scope
+
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
"""
A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
- functions and classes that are inside a parent function are considered hidden.
"""
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
+ subscopes: Mapping[str, Self]
- subscopes: Mapping[ProjectPath, Self]
+ parent_scope: "ChangeScope | None"
===========changed ref 7===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
+ scope = ChangeScope(path, tree, spans, subscopes, None)
- scope = ChangeScope(path, tree, spans, subscopes)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(current_stmts, scope))
- spans.append(StatementSpan(current_stmts))
current_stmts = []
if current_stmts:
+ spans.append(StatementSpan(current_stmts, scope))
- spans.append(StatementSpan(current_stmts))
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
+ name = cast(ptree.Name, stree.name).value
- spath = path.append(cast(ptree.Name, stree.name).value)
+ spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
+ subscope.parent_scope = scope
+ subscopes[name] = subscope
- subscopes[spath] = subscope
return scope
===========changed ref 8===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
+ @staticmethod
+ def from_unchanged(v: T1) -> "Modified[T1]":
+ return Modified(v, v, unchanged=True)
+
===========changed ref 9===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
before: E1
after: E1
+ unchanged: bool = False
===========changed ref 10===========
# module: coeditor.encoding
def change_to_line_diffs(change: Change[str]) -> list[str]:
"Encode a change as a token sequence."
match change:
case Modified(before, after):
+ if change.unchanged:
+ diffs = []
+ else:
+ diffs = compute_line_diffs(splitlines(before), splitlines(after))
- diffs = compute_line_diffs(splitlines(before), splitlines(after))
# rearrange_diffs_(diffs)
if not diffs:
# as a special case, `unified_diff` would return an empty when there is no change.
diffs = [" " + l for l in splitlines(before)]
case Added(after):
diffs = ["+" + l for l in splitlines(after)]
case Deleted(before):
diffs = ["-" + l for l in splitlines(before)]
case _:
raise ValueError(f"Invalid change type: {change}.")
return diffs
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.encode_change
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> result.append(cspan)
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
<s>("."))
if path is None:
continue
+ jmod = before_mod_map[path.module]
+ scope = jmod.as_scope
+ elem = scope._search_scope(path.path)
+ match elem:
+ case ChangeScope(path=path, tree=ptree.Function()):
+ ancestors = elem.ancestors()
+ body_code = " " * len(ancestors) + "..."
+ h_end = elem.header_line_range[1]
+ cspan = ChangedSpan(
+ Modified.from_unchanged(body_code),
+ [Modified.from_unchanged(s) for s in ancestors],
+ _line_range(h_end, h_end + 1),
+ )
+ result.append(cspan)
+ case _:
+ cspan = ChangedSpan(
+ Modified.from_unchanged(f"path={path}"),
+ [Modified.from_unchanged(s) for s in elem.ancestors()],
+ _line_range(0, 1),
+ )
- result.append((ProjectPath(used.full_name, ""), str(used)))
<0> return result
sorted_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
yield CtxCodeChangeProblem(
span,
relevant_changes=sorted_cspans.copy(),
relevant_unchanged=_get_relevant(span),
)
sorted_cspans.append(span)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def encode_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -1
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
def _get_relevant(span: ChangedSpan):
if isinstance(span.change, Added):
# nothing to analyze
return []
path = span.parent_scopes[-1].earlier().path
line_usages = mod2usages[path.module]
all_used = set[PyDefinition]()
all_lines = set(range(*span.line_range))
all_lines.update(range(*span.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
+ pydef.full_name.startswith(path.module)
- pydef.full_name.startswith(pydef.import_module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
all_used.add(pydef)
+ result = list[ChangedSpan]()
- result = list[tuple[ProjectPath, str]]()
for used in all_used:
path = mod_hier.resolve_path(used.full_name.split("."))
if path is None:
continue
+ jmod = before_mod_map[path.module]
</s>
===========unchanged ref 0===========
at: coeditor.code_change
_line_range(start: int, end: int, can_be_empty: bool=False) -> _LineRange
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[ProjectPath, Self])
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: _LineRange)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[ProjectPath, Self]
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: _LineRange
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan])
PyDefinition(full_name: PyFullName, import_module: ModuleName, start_pos: tuple[int, int], end_pos: tuple[int, int])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
at: coeditor.ctx_change_encoder.CtxCodeChangeProblemGenerator.post_edit_analysis
module_order = sort_modules_by_imports(module_deps)
at: coeditor.ctx_change_encoder.LineUsageAnalysis
line2usages: Mapping[int, set[PyDefinition]]
at: coeditor.ctx_change_encoder.PyDefinition
full_name: PyFullName
import_module: ModuleName
start_pos: tuple[int, int]
end_pos: tuple[int, int]
at: parso.python.tree
Function(children)
at: spot.static_analysis
ModuleName = str
ModuleHierarchy()
at: spot.static_analysis.ModuleHierarchy
resolve_path(segs: list[str]) -> ProjectPath | None
===========unchanged ref 1===========
from_modules(modules: Iterable[str]) -> "ModuleHierarchy"
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Mapping = _alias(collections.abc.Mapping, 2)
Sequence = _alias(collections.abc.Sequence, 1)
at: typing.Mapping
get(key: _KT) -> Optional[_VT_co]
get(key: _KT, default: Union[_VT_co, _T]) -> Union[_VT_co, _T]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
# jedi.cache.clear_time_caches = lambda: None
@dataclass
class CtxCodeChangeProblem:
span: ChangedSpan
# most relevant to least relevant
relevant_changes: list[ChangedSpan]
# most relevant to least relevant
+ relevant_unchanged: list[ChangedSpan]
- relevant_unchanged: list[tuple[ProjectPath, str]]
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
"""
A change scope is a python module, non-hidden function, or a non-hidden class, or a python module.
- functions and classes that are inside a parent function are considered hidden.
"""
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
+ subscopes: Mapping[str, Self]
- subscopes: Mapping[ProjectPath, Self]
+ parent_scope: "ChangeScope | None"
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
+ @staticmethod
+ def from_name(name: classes.BaseName) -> Iterable["PyDefinition"]:
+ cast(classes.Name, name).is_definition()
+ if (
+ not name.in_builtin_module()
+ and (full_name := name.full_name)
+ and (import_module := name.module_name)
+ and (start_pos := name.get_definition_start_position())
+ and (end_pos := name.get_definition_end_position())
+ ):
+ full_name = PyFullName(full_name)
+ yield PyDefinition(full_name, import_module, start_pos, end_pos)
+
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
- @staticmethod
- def from_signatures(name: classes.BaseName) -> Iterable["PyDefinition"]:
- cast(classes.Name, name).is_definition()
- if (
- not name.in_builtin_module()
- and (full_name := name.full_name)
- and (import_module := name.module_name)
- and (start_pos := name.get_definition_start_position())
- and (end_pos := name.get_definition_end_position())
- ):
- full_name = PyFullName(full_name)
- yield PyDefinition(full_name, import_module, start_pos, end_pos)
-
===========changed ref 4===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
+ @staticmethod
+ def from_unchanged(v: T1) -> "Modified[T1]":
+ return Modified(v, v, unchanged=True)
+
===========changed ref 5===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
before: E1
after: E1
+ unchanged: bool = False
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class JModule:
+ @cached_property
+ def as_scope(self) -> ChangeScope:
+ return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
+
===========changed ref 7===========
# module: coeditor.code_change
@dataclass
class JModule:
- def _to_scope(self) -> ChangeScope:
- return ChangeScope.from_tree(ProjectPath(self.mname, ""), self.tree)
-
|
coeditor.ctx_change_encoder/JediUsageAnalyzer.get_line_usages
|
Modified
|
temp-1
|
d0fe36e93da8ca33161e524783335898bd47ebbc
|
- More analysis bug fix. - Encode unchanged using CSpan.
|
<0>:<add> usages.update(PyDefinition.from_name(d))
|
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
jmod: tree.Module = script._module_node
line2usages = dict[int, set[PyDefinition]]()
all_names = [
name for k, names in jmod.get_used_names()._dict.items() for name in names
]
all_names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
resolve_cache = dict[_ObjId, set[PyDefinition]]()
for name in tqdm(all_names, f"Analyzing {script.path}", disable=silent):
name: tree.Name
line = name.start_pos[0]
if line not in lines_to_analyze:
continue
usages = line2usages.setdefault(line, set())
try:
defs = fast_goto(
script,
name,
follow_imports=self.follow_imports,
follow_builtin_imports=False,
)
for d in defs:
- usages.update(PyDefinition.from_signatures(d))
<0>
except (AttributeError, AssertionError) as e:
text = repr(e)
errors[text] = errors.setdefault(text, 0) + 1
except ValueError as e:
# if the message is "not enough values to unpack"
if "not enough values to unpack (expected 2" in str(e):
errors[repr(e)] = errors.setdefault(str(e), 0) + 1
else:
raise
return LineUsageAnalysis(line2usages)
|
===========unchanged ref 0===========
at: coeditor._utils
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor.ctx_change_encoder
PyDefinition(full_name: PyFullName, import_module: ModuleName, start_pos: tuple[int, int], end_pos: tuple[int, int])
_ObjId = NewType("_ObjId", int)
fast_goto(script: jedi.Script, tree_name: tree.Name, *, follow_imports=False, follow_builtin_imports=False, only_stubs=False, prefer_stubs=False) -> set[classes.Name]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
max_chunks_per_ref: int = 4
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder._group_encode_changed_refs
all_chunks = list[TokenSeq]()
mod_chunks = break_into_chunks(
file_tks,
lambda i: self._encode_parent_scopes(mod_change, i),
self.max_ref_tks,
overlap=self.ref_chunk_overlap,
max_return_chunks=self.max_chunks_per_ref,
)
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script.__init__
self.path = path.absolute() if path else None
===========unchanged ref 1===========
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
at: parso.python.tree
Name()
Module(children)
at: parso.python.tree.Module
__slots__ = ('_used_names',)
type = 'file_input'
get_used_names()
at: parso.python.tree.UsedNamesMapping.__init__
self._dict = dct
at: parso.tree.Leaf.__init__
self.start_pos = start_pos
at: pathlib
Path()
at: tqdm.std
tqdm(iterable=None, desc=None, total=None, leave=True, file=None, ncols=None, mininterval=0.1, maxinterval=10.0, miniters=None, ascii=None, disable=False, unit='it', unit_scale=False, dynamic_ncols=False, smoothing=0.3, bar_format=None, initial=0, position=None, postfix=None, unit_divisor=1000, write_bytes=False, lock_args=None, nrows=None, colour=None, delay=0, gui=False, **kwargs)
at: typing
Collection = _alias(collections.abc.Collection, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
+ def _group_encode_unchanged_refs(
+ self, elems: Sequence[ChangedSpan]
+ ) -> Sequence[TokenSeq]:
+ return self._group_encode_changed_refs(elems)
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
- def _encode_unchanged_ref(
- self, path: ProjectPath, content: str
- ) -> Iterable[TokenSeq]:
- if (key := (path, content)) in self._value_cache:
- return self._value_cache[key]
- main_tks = encode_basic(f"#{str(path)}\n{content}")
- ref_chunks = (truncate_section(main_tks, TruncateAt.Right, self.max_ref_tks),)
- self._value_cache[key] = ref_chunks
- return ref_chunks
-
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_parent_scopes(
self, scope_changes: Sequence[Change[ChangeScope]], offset: int
) -> TokenSeq:
scope_tks = join_list((self._encode_scope_change(c) for c in scope_changes))
if offset != 0:
+ scope_tks.extend(encode_basic(f"# offset: {offset}\n"))
- ending = encode_basic(f"\n# offset: {offset}\n")
- else:
- ending = [Newline_id]
- scope_tks = truncate_section(
+ scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
- scope_tks + ending, TruncateAt.Left, self.max_scope_tks
- )
return scope_tks
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
# jedi.cache.clear_time_caches = lambda: None
@dataclass
class CtxCodeChangeProblem:
span: ChangedSpan
# most relevant to least relevant
relevant_changes: list[ChangedSpan]
# most relevant to least relevant
+ relevant_unchanged: list[ChangedSpan]
- relevant_unchanged: list[tuple[ProjectPath, str]]
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
+ @staticmethod
+ def from_name(name: classes.BaseName) -> Iterable["PyDefinition"]:
+ cast(classes.Name, name).is_definition()
+ if (
+ not name.in_builtin_module()
+ and (full_name := name.full_name)
+ and (import_module := name.module_name)
+ and (start_pos := name.get_definition_start_position())
+ and (end_pos := name.get_definition_end_position())
+ ):
+ full_name = PyFullName(full_name)
+ yield PyDefinition(full_name, import_module, start_pos, end_pos)
+
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
- @staticmethod
- def from_signatures(name: classes.BaseName) -> Iterable["PyDefinition"]:
- cast(classes.Name, name).is_definition()
- if (
- not name.in_builtin_module()
- and (full_name := name.full_name)
- and (import_module := name.module_name)
- and (start_pos := name.get_definition_start_position())
- and (end_pos := name.get_definition_end_position())
- ):
- full_name = PyFullName(full_name)
- yield PyDefinition(full_name, import_module, start_pos, end_pos)
-
===========changed ref 6===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
+ @staticmethod
+ def from_unchanged(v: T1) -> "Modified[T1]":
+ return Modified(v, v, unchanged=True)
+
===========changed ref 7===========
# module: coeditor.history
@dataclass
class Modified(_ChangeBase[E1]):
before: E1
after: E1
+ unchanged: bool = False
|
coeditor.code_change/ChangeScope.__post_init__
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> self.header_line_range: LineRange = header_line_range
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
<s> = f"# module: {self.path.module}"
+ header_line_range = line_range(0, 0, can_be_empty=True)
- header_line_range = _line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
header_code, _, e_right = _strip_empty_lines(header_code)
h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
+ header_line_range = line_range(h_start, h_end)
- header_line_range = _line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
- self.header_line_range: _LineRange = header_line_range
<0>
|
===========above chunk 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: -1
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
+ header_line_range = line_range(0, 0,</s>
===========unchanged ref 0===========
at: coeditor._utils
not_none(x: Optional[T1]) -> T1
assert_eq(x: T1, y: T1, message: Callable[[], str]=lambda: "") -> None
at: coeditor.code_change
LineRange = NewType("LineRange", tuple[int, int])
line_range(start: int, end: int, can_be_empty: bool=False) -> LineRange
_strip_empty_lines(s: str)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
at: coeditor.code_change.StatementSpan.__post_init__
self.line_range: LineRange = line_range(start, end)
at: coeditor.common
count_lines(text: str) -> int
at: parso.python.tree
Module(children)
at: parso.tree
NodeOrLeaf()
at: parso.tree.BaseNode.__init__
self.children = children
self.parent: Optional[BaseNode] = None
at: parso.tree.NodeOrLeaf
__slots__ = ('parent',)
type: str
parent: 'Optional[BaseNode]'
get_code(include_prefix=True)
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 1===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 2===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ LineRange = NewType("LineRange", tuple[int, int])
- _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 3===========
# module: tests.coeditor.test_edits
class TestChangeIdentities:
cases = {
"empty": Modified("", ""),
"generation": Modified("", "123"),
"no change": Modified(
dedent(
"""\
def f1():
x = 1
"""
),
dedent(
"""\
def f1():
x = 1
"""
),
),
+ "unchanged=True": Modified.from_unchanged(
+ dedent(
+ """\
+ def f1():
+ x = 1
+ """
+ ),
+ ),
+ # this test case cannot pass for some reason. Tokenizer bug?
+ # "leading_whitespace": Modified.from_unchanged(" ..."),
"replace last": Modified(
dedent(
"""\
def f1():
x = 1"""
),
dedent(
"""\
def f1():
x = 2
return x * 2"""
),
),
"no special tokens": Modified(
dedent(
"""\
def f1():
x = 1
y = 2
z = x + y
return z
def f2():
f1()"""
),
dedent(
"""\
# new comment
def f_new():
x = 1
if x > 0:
y = 2 * x
y *= 2
z = x + y
return z
def f2():
f1()
return f_new() + a
new_var = 0
"""
),
),
"with special tokens": Modified(
dedent(
"""\
def f1():
x = "<add>"
y = "<del>\tx"
return x + y
"""
),
dedent(
"""\
# new comment 1
# new comment 2
def f1():
if newcond:
x = "<add</s>
===========changed ref 4===========
# module: tests.coeditor.test_edits
class TestChangeIdentities:
# offset: 1
<s>
# new comment 1
# new comment 2
def f1():
if newcond:
x = "<add>"
new_var = 5
y = "<del>"
return x + new_var + y
"""
),
),
"super long": Modified(
"\n".join(f"x = {i}" for i in range(0, 200)),
"\n".join(f"x = {2* (i // 2)}" for i in range(0, 200)),
),
}
|
coeditor.code_change/ChangeScope.from_tree
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> spans.append(StatementSpan(len(spans), current_stmts, scope))
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
current_stmts = []
if current_stmts:
- spans.append(StatementSpan(current_stmts, scope))
<0>
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
|
===========unchanged ref 0===========
at: coeditor._utils
as_any(x) -> Any
at: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
_is_scope_statement(stmt: PyNode) -> bool
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
at: coeditor.code_change.ChangeScope
parent_scope: "ChangeScope | None"
at: parso.python.tree
PythonNode()
Name()
Module(children)
Class(children)
Function(children)
at: parso.python.tree.Class
type = 'classdef'
__slots__ = ()
at: parso.python.tree.ClassOrFunc
__slots__ = ()
at: parso.python.tree.Function
type = 'funcdef'
at: parso.python.tree.Scope
__slots__ = ()
_search_in_scope(*names)
get_suite()
at: parso.tree.BaseNode.__init__
self.children = children
at: parso.tree.Leaf.__init__
self.value = value
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: spot.static_analysis.ProjectPath
append(path: ElemPath) -> "ProjectPath"
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 1===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 2===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ LineRange = NewType("LineRange", tuple[int, int])
- _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
+ header_line_range = line_range(0, 0, can_be_empty=True)
- header_line_range = _line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
header_code, _, e_right = _strip_empty_lines(header_code)
h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
+ header_line_range = line_range(h_start, h_end)
- header_line_range = _line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"</s>
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: 1
<s>spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
+ self.header_line_range: LineRange = header_line_range
- self.header_line_range: _LineRange = header_line_range
===========changed ref 5===========
# module: tests.coeditor.test_edits
class TestChangeIdentities:
cases = {
"empty": Modified("", ""),
"generation": Modified("", "123"),
"no change": Modified(
dedent(
"""\
def f1():
x = 1
"""
),
dedent(
"""\
def f1():
x = 1
"""
),
),
+ "unchanged=True": Modified.from_unchanged(
+ dedent(
+ """\
+ def f1():
+ x = 1
+ """
+ ),
+ ),
+ # this test case cannot pass for some reason. Tokenizer bug?
+ # "leading_whitespace": Modified.from_unchanged(" ..."),
"replace last": Modified(
dedent(
"""\
def f1():
x = 1"""
),
dedent(
"""\
def f1():
x = 2
return x * 2"""
),
),
"no special tokens": Modified(
dedent(
"""\
def f1():
x = 1
y = 2
z = x + y
return z
def f2():
f1()"""
),
dedent(
"""\
# new comment
def f_new():
x = 1
if x > 0:
y = 2 * x
y *= 2
z = x + y
return z
def f2():
f1()
return f_new() + a
new_var = 0
"""
),
),
"with special tokens": Modified(
dedent(
"""\
def f1():
x = "<add>"
y = "<del>\tx"
return x + y
"""
),
dedent(
"""\
# new comment 1
# new comment 2
def f1():
if newcond:
x = "<add</s>
|
coeditor.code_change/StatementSpan.__post_init__
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> self.line_range: LineRange = line_range(start, end)
|
# module: coeditor.code_change
@dataclass
class StatementSpan:
def __post_init__(self):
assert self.statements
+ # remove leading newlines
+ n_leading_newlines = 0
+ stmts = self.statements
+ for s in stmts:
+ if s.type == ptree.Newline.type:
+ n_leading_newlines += 1
+ else:
+ break
+ if n_leading_newlines:
+ self.statements = stmts[n_leading_newlines:]
+
origin_code = "".join(s.get_code() for s in self.statements)
code, _, e_right = _strip_empty_lines(origin_code)
start = self.statements[0].start_pos[0]
end = self.statements[-1].end_pos[0] + 1 - e_right
self.code: str = code + "\n"
try:
- self.line_range: _LineRange = _line_range(start, end)
<0> except ValueError:
print_err(f"{origin_code=}, {e_right=}, {start=}, {end=}")
raise
|
===========unchanged ref 0===========
at: coeditor.code_change.StatementSpan
nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
at: parso.python.tree
Newline()
at: parso.python.tree.Newline
__slots__ = ()
type = 'newline'
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
"""
A statement span is a set of lines inside the same change scope. It is the basic unit of code changes handled by our model.
- For a modified function, the span is the function itself.
- For a modified module, the spans are the regions between the functions and classes plus
the spans recursively generated.
- For a modified class, the spans are the regions between methods plus
the spans recursively generated.
"""
+ nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
===========changed ref 1===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 2===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 3===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ LineRange = NewType("LineRange", tuple[int, int])
- _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
current_stmts = []
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
+ header_line_range = line_range(0, 0, can_be_empty=True)
- header_line_range = _line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
header_code, _, e_right = _strip_empty_lines(header_code)
h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
+ header_line_range = line_range(h_start, h_end)
- header_line_range = _line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"</s>
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: 1
<s>spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
+ self.header_line_range: LineRange = header_line_range
- self.header_line_range: _LineRange = header_line_range
|
coeditor.ctx_change_encoder/PyDefinition.from_name
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> raise ValueError(f"Inconsistent module: {full_name=}, {import_module=}")
|
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
@staticmethod
def from_name(name: classes.BaseName) -> Iterable["PyDefinition"]:
- cast(classes.Name, name).is_definition()
if (
not name.in_builtin_module()
and (full_name := name.full_name)
and (import_module := name.module_name)
and (start_pos := name.get_definition_start_position())
and (end_pos := name.get_definition_end_position())
):
full_name = PyFullName(full_name)
+ if not full_name.startswith(import_module):
<0> yield PyDefinition(full_name, import_module, start_pos, end_pos)
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
PyFullName = NewType("PyFullName", str)
at: coeditor.ctx_change_encoder.PyDefinition
full_name: PyFullName
import_module: ModuleName
start_pos: tuple[int, int]
end_pos: tuple[int, int]
at: jedi.api.classes
BaseName(inference_state, name)
at: jedi.api.classes.BaseName
_mapping = {
'posixpath': 'os.path',
'riscospath': 'os.path',
'ntpath': 'os.path',
'os2emxpath': 'os.path',
'macpath': 'os.path',
'genericpath': 'os.path',
'posix': 'os',
'_io': 'io',
'_functools': 'functools',
'_collections': 'collections',
'_socket': 'socket',
'_sqlite3': 'sqlite3',
}
_tuple_mapping = dict((tuple(k.split('.')), v) for (k, v) in {
'argparse._ActionsContainer': 'argparse.ArgumentParser',
}.items())
in_builtin_module()
get_definition_start_position()
get_definition_end_position()
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
+ PyFullName = NewType("PyFullName", str)
- PyFullName = NewType("PyPathStr", str)
===========changed ref 1===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 2===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
"Represents the changes made to a statement span."
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
+ line_range: LineRange
- line_range: _LineRange
===========changed ref 4===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ LineRange = NewType("LineRange", tuple[int, int])
- _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
"""
A statement span is a set of lines inside the same change scope. It is the basic unit of code changes handled by our model.
- For a modified function, the span is the function itself.
- For a modified module, the spans are the regions between the functions and classes plus
the spans recursively generated.
- For a modified class, the spans are the regions between methods plus
the spans recursively generated.
"""
+ nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
def __post_init__(self):
assert self.statements
+ # remove leading newlines
+ n_leading_newlines = 0
+ stmts = self.statements
+ for s in stmts:
+ if s.type == ptree.Newline.type:
+ n_leading_newlines += 1
+ else:
+ break
+ if n_leading_newlines:
+ self.statements = stmts[n_leading_newlines:]
+
origin_code = "".join(s.get_code() for s in self.statements)
code, _, e_right = _strip_empty_lines(origin_code)
start = self.statements[0].start_pos[0]
end = self.statements[-1].end_pos[0] + 1 - e_right
self.code: str = code + "\n"
try:
+ self.line_range: LineRange = line_range(start, end)
- self.line_range: _LineRange = _line_range(start, end)
except ValueError:
print_err(f"{origin_code=}, {e_right=}, {start=}, {end=}")
raise
===========changed ref 7===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
current_stmts = []
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder.encode_problem
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> named_references.append((f"changed ref {i}", chunk))
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
span = problem.span
named_references = list[tuple[str, TokenSeq]]()
# compute the references that are relevant to this span
- # FIXME
+ relevant_chunks = self._group_encode_changed_refs(problem.relevant_changes)
- # relevant_chunks = self._group_encode_changed_refs(problem.relevant_changes)
+ for i, chunk in enumerate(relevant_chunks):
- # for i, chunk in enumerate(relevant_chunks):
- # named_references.append((f"changed ref {i}", chunk))
<0> relevant_chunks = self._group_encode_unchanged_refs(problem.relevant_unchanged)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"unchanged ref {i}", chunk))
diffs = change_to_line_diffs(span.change)
original, delta = line_diffs_to_original_delta(diffs)
origin_lines = split_list(encode_basic(original), Newline_id)
tk_delta = delta.to_tk_delta()
chunk_id = 0
chunk_start_l = 0
scope_tks = self._encode_parent_scopes(span.parent_scopes, 0)
chunk_input = TokenSeq()
input_limit = self.max_query_tks - len(scope_tks)
chunk_lines = 0
chunk_output = TokenSeq()
prev_change_tks = TokenSeq()
def get_problem(chunk_input, chunk_output):
# try move some prev_change_tks into the input
above_tks = prev_change_tks
below_tks = join_list(origin_lines[l:], Newline_id)
chunk_input, above_tks, below_tks = self._inline_some_</s>
|
===========below chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: 1
<s>l:], Newline_id)
chunk_input, above_tks, below_tks = self._inline_some_context(
chunk_input, above_tks, below_tks, input_limit
)
# limit the input size if it's too long (can happen for later chunks)
chunk_input = truncate_section(chunk_input, TruncateAt.Right, input_limit)
chunk_output = truncate_output_tks(chunk_input, chunk_output)
chunk_output = truncate_section(
chunk_output, TruncateAt.Right, self.max_output_tks, add_bos=False
)
above_chunks = break_into_chunks(
above_tks,
lambda i: self._encode_parent_scopes(span.parent_scopes, -1 - i),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
right_to_left=True,
)
if finished:
below_chunks = []
else:
below_chunks = break_into_chunks(
below_tks,
lambda i: self._encode_parent_scopes(span.parent_scopes, i + 1),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
)
above_chunks = [
(f"above chunk {i}", chunk) for i, chunk in enumerate(above_chunks)
]
below_chunks = [
(f"below chunk {i}", chunk) for i, chunk in enumerate(below_chunks)
]
return TkCtxCodeChangeProblem(
scope_tks + chunk_input,
chunk_output,
path=span.parent_scopes[-1].earlier().path,
change_type=</s>
===========below chunk 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: 2
<s>,
chunk_output,
path=span.parent_scopes[-1].earlier().path,
change_type=span.change.map(lambda _: None),
named_references=above_chunks + below_chunks + named_references,
)
for l in range(len(tk_delta.deltas) + 1):
finished = l == len(tk_delta.deltas)
input_growth = len(origin_lines[l]) + 2 if l < len(origin_lines) else 1
if (
finished
or chunk_lines >= self.max_lines_to_edit
or len(chunk_input) + input_growth > input_limit
):
if has_change(chunk_output):
yield get_problem(chunk_input, chunk_output)
if finished:
break
chunk_main_input = join_list(origin_lines[chunk_start_l:l], Newline_id)
chunk_main_delta = tk_delta.for_input_range((chunk_start_l, l))
chunk_main_change = chunk_main_delta.to_change_tks(chunk_main_input)
prev_change_tks.extend(chunk_main_change)
prev_change_tks.append(Newline_id)
chunk_id += 1
chunk_input = TokenSeq()
chunk_lines = 0
chunk_output = TokenSeq()
chunk_start_l = l
chunk_input.append(get_extra_id(chunk_lines))
if l < len(origin_lines):
chunk_input.extend(origin_lines[l])
chunk_input.append(</s>
===========below chunk 2===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: 3
<s>id)
line_change = join_list(tk_delta.deltas[l], Newline_id)
chunk_output.append(get_extra_id(chunk_lines))
chunk_output.extend(line_change)
if line_change and line_change[-1] != Del_id:
chunk_output.append(Newline_id)
chunk_lines += 1
===========unchanged ref 0===========
at: cachetools
FIFOCache(maxsize, getsizeof=None)
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: _LineRange
at: coeditor.common
TokenSeq = list[Token]
split_list(lst: list[T1], sep: T1) -> list[list[T1]]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan])
TkCtxCodeChangeProblem(input_tks: TokenSeq, output_tks: TokenSeq, path: ProjectPath, change_type: Change[None], named_references: Sequence[tuple[str, TokenSeq]])
at: coeditor.ctx_change_encoder.CtxCodeChangeProblem
span: ChangedSpan
relevant_changes: list[ChangedSpan]
relevant_unchanged: list[ChangedSpan]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
_encode_parent_scopes(scope_changes: Sequence[Change[ChangeScope]], offset: int) -> TokenSeq
_inline_some_context(input: TokenSeq, above_ctx: TokenSeq, below_ctx: TokenSeq, size_limit: int) -> tuple[TokenSeq, TokenSeq, TokenSeq]
_group_encode_unchanged_refs(elems: Sequence[ChangedSpan]) -> Sequence[TokenSeq]
_group_encode_changed_refs(changes: Sequence[ChangedSpan]) -> Sequence[TokenSeq]
_group_encode_changed_refs(self, changes: Sequence[ChangedSpan]) -> Sequence[TokenSeq]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder.encode_problem
finished = l == len(tk_delta.deltas)
|
coeditor.ctx_change_encoder/JediUsageAnalyzer.get_line_usages
|
Modified
|
temp-1
|
d58fffa6f57746cb134bd98b5de3a8e668c4311d
|
Finish unchanged ref encoding.
|
<0>:<add> follow_imports=True,
|
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
jmod: tree.Module = script._module_node
line2usages = dict[int, set[PyDefinition]]()
all_names = [
name for k, names in jmod.get_used_names()._dict.items() for name in names
]
all_names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
- resolve_cache = dict[_ObjId, set[PyDefinition]]()
for name in tqdm(all_names, f"Analyzing {script.path}", disable=silent):
name: tree.Name
line = name.start_pos[0]
if line not in lines_to_analyze:
continue
usages = line2usages.setdefault(line, set())
try:
defs = fast_goto(
script,
name,
- follow_imports=self.follow_imports,
<0> follow_builtin_imports=False,
)
for d in defs:
usages.update(PyDefinition.from_name(d))
except (AttributeError, AssertionError) as e:
text = repr(e)
errors[text] = errors.setdefault(text, 0) + 1
except ValueError as e:
# if the message is "not enough values to unpack"
if "not enough values to unpack (expected 2" in str(e):
errors[repr(e)] = errors.setdefault(str(e), 0) + 1
else:
raise
return LineUsageAnalysis(line2usages)
|
===========unchanged ref 0===========
at: coeditor._utils
groupby(iterable: Iterable[T1], keyfunc: Callable[[T1], T2]) -> dict[T2, list[T1]]
at: coeditor.code_change
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: _LineRange)
at: coeditor.code_change.ChangedSpan
parent_scopes: Sequence[Change[ChangeScope]]
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
max_ref_tks: int = 512
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
_encode_parent_scopes(scope_changes: Sequence[Change[ChangeScope]], offset: int) -> TokenSeq
_encode_change(change: Change[str]) -> TokenSeq
at: coeditor.encoding
Newline_id = get_tk_id("\n")
break_into_chunks(tks: TokenSeq, header_f: Callable[[int], TokenSeq], chunk_size: int, overlap: int, right_to_left: bool=False, add_bos: bool=True, max_return_chunks: int | None=None) -> list[TokenSeq]
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ChangedSpan:
"Represents the changes made to a statement span."
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
+ line_range: LineRange
- line_range: _LineRange
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
- follow_imports: bool = True
-
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
+ PyFullName = NewType("PyFullName", str)
- PyFullName = NewType("PyPathStr", str)
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
@staticmethod
def from_name(name: classes.BaseName) -> Iterable["PyDefinition"]:
- cast(classes.Name, name).is_definition()
if (
not name.in_builtin_module()
and (full_name := name.full_name)
and (import_module := name.module_name)
and (start_pos := name.get_definition_start_position())
and (end_pos := name.get_definition_end_position())
):
full_name = PyFullName(full_name)
+ if not full_name.startswith(import_module):
+ raise ValueError(f"Inconsistent module: {full_name=}, {import_module=}")
yield PyDefinition(full_name, import_module, start_pos, end_pos)
===========changed ref 4===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 5===========
# module: coeditor.code_change
- def _line_range(start: int, end: int, can_be_empty: bool = False) -> _LineRange:
- if not can_be_empty and start >= end:
- raise ValueError(f"Bad line range: {start=}, {end=}")
- return _LineRange((start, end))
-
===========changed ref 6===========
# module: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
PyNode = ptree.PythonBaseNode | ptree.PythonNode
+ LineRange = NewType("LineRange", tuple[int, int])
- _LineRange = NewType("LineRange", tuple[int, int])
_tlogger = TimeLogger()
===========changed ref 7===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
"""
A statement span is a set of lines inside the same change scope. It is the basic unit of code changes handled by our model.
- For a modified function, the span is the function itself.
- For a modified module, the spans are the regions between the functions and classes plus
the spans recursively generated.
- For a modified class, the spans are the regions between methods plus
the spans recursively generated.
"""
+ nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
===========changed ref 8===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
def __post_init__(self):
assert self.statements
+ # remove leading newlines
+ n_leading_newlines = 0
+ stmts = self.statements
+ for s in stmts:
+ if s.type == ptree.Newline.type:
+ n_leading_newlines += 1
+ else:
+ break
+ if n_leading_newlines:
+ self.statements = stmts[n_leading_newlines:]
+
origin_code = "".join(s.get_code() for s in self.statements)
code, _, e_right = _strip_empty_lines(origin_code)
start = self.statements[0].start_pos[0]
end = self.statements[-1].end_pos[0] + 1 - e_right
self.code: str = code + "\n"
try:
+ self.line_range: LineRange = line_range(start, end)
- self.line_range: _LineRange = _line_range(start, end)
except ValueError:
print_err(f"{origin_code=}, {e_right=}, {start=}, {end=}")
raise
===========changed ref 9===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
current_stmts = []
if current_stmts:
+ spans.append(StatementSpan(len(spans), current_stmts, scope))
- spans.append(StatementSpan(current_stmts, scope))
if is_func:
# we don't create inner scopes for function contents
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
|
coeditor.code_change/get_python_files
|
Modified
|
temp-1
|
c5644c5542948a30b4d5a99c74d50454fb62315a
|
Speed up parsing with on-demand deepcopying.
|
<0>:<add> files.append(to_rel_path(Path(f.path).relative_to(project)))
|
# module: coeditor.code_change
+ def get_python_files(project: RelPath) -> list[RelPath]:
- def get_python_files(project: Path):
+ files = list[RelPath]()
- files = list[Path]()
for f in recurse_find_python_files(FolderIO(str(project))):
f: FileIO
- files.append(Path(f.path).relative_to(project))
<0> return files
|
===========unchanged ref 0===========
at: coeditor.common
RelPath = NewType("RelPath", Path)
to_rel_path(path: os.PathLike | str) -> RelPath
at: jedi.file_io
FolderIO(path)
FileIO(path: Union[os.PathLike, str])
at: jedi.inference.references
recurse_find_python_files(folder_io, except_paths=())
at: parso.file_io.FileIO.__init__
self.path = path
at: pathlib
Path()
at: pathlib.PurePath
__slots__ = (
'_drv', '_root', '_parts',
'_str', '_hash', '_pparts', '_cached_cparts',
)
drive = property(attrgetter('_drv'),
doc="""The drive prefix (letter or UNC path), if any.""")
root = property(attrgetter('_root'),
doc="""The root of the path, if any.""")
relative_to(*other: Union[str, _PathLike]) -> _P
===========changed ref 0===========
# module: coeditor.common
+ def to_rel_path(path: Path) -> RelPath:
+ if path.is_absolute():
+ raise ValueError(f"Expected a relative path, got: {path}")
+ return RelPath(path)
+
===========changed ref 1===========
# module: coeditor.common
+ def to_abs_path(path: Path) -> AbsPath:
+ return AbsPath(path.resolve())
+
|
coeditor.code_change/edits_from_commit_history
|
Modified
|
temp-1
|
c5644c5542948a30b4d5a99c74d50454fb62315a
|
Speed up parsing with on-demand deepcopying.
|
<0>:<add> tempdir, history, change_processor, edit_encoder, ignore_dirs, silent
|
<s>_from_commit_history(
project_dir: Path,
history: Sequence[CommitInfo],
tempdir: Path,
+ change_processor: ProjectChangeProcessor[TProb] = NoProcessing(),
- change_encoder: ProjectChangeProcessor[TEnc] = NoProcessing(),
+ edit_encoder: Callable[[TProb], Iterable[T1]] = lambda x: [x],
ignore_dirs=DefaultIgnoreDirs,
silent: bool = False,
+ ) -> Sequence[T1]:
- ) -> Iterable[TEnc]:
"""Incrementally compute the edits to a project from the git history.
Note that this will change the file states in the project directory, so
you should make a copy of the project before calling this function.
-
- Note that this returns an iterator, and the file state cleaning up will
- only happen when the iterator is exhausted.
"""
tempdir = tempdir.resolve()
if tempdir.exists():
raise FileExistsError(f"Workdir '{tempdir}' already exists.")
tempdir.mkdir(parents=True, exist_ok=False)
use_fast_parser = jedi.settings.fast_parser
try:
run_command(
["cp", "-r", str(project_dir / ".git"), str(tempdir)],
cwd=project_dir.parent,
)
+ return _edits_from_commit_history(
- yield from _edits_from_commit_history(
- tempdir, history, change_encoder, ignore_dirs, silent
<0> )
finally:
run_command(["rm", "-rf", str(tempdir)], cwd=tempdir.parent)
jedi.settings.fast_parser = use_fast_parser
|
===========unchanged ref 0===========
at: coeditor.code_change
DefaultIgnoreDirs = {".venv", ".mypy_cache", ".git", "venv", "build"}
TProb = TypeVar("TProb", covariant=True)
ProjectChangeProcessor()
NoProcessing()
_edits_from_commit_history(project: Path, history: Sequence[CommitInfo], change_processor: ProjectChangeProcessor[TProb], edit_encoder: Callable[[TProb], Iterable[T1]], ignore_dirs: set[str], silent: bool) -> Sequence[T1]
at: coeditor.common
T1 = TypeVar("T1")
run_command(args: Sequence[str], cwd: str | Path) -> str
at: jedi.settings
fast_parser = True
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
resolve(strict: bool=...) -> _P
mkdir(mode: int=..., parents: bool=..., exist_ok: bool=...) -> None
exists() -> bool
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Callable = _CallableType(collections.abc.Callable, 2)
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.code_change
class NoProcessing(ProjectChangeProcessor[JProjectChange]):
- def encode_change(
- self,
- pchange: JProjectChange,
- pre_analysis,
- post_analysis,
- ) -> Iterable[JProjectChange]:
- yield pchange
-
===========changed ref 1===========
# module: coeditor.code_change
class NoProcessing(ProjectChangeProcessor[JProjectChange]):
- def encode_change(
- self,
- pchange: JProjectChange,
- pre_analysis,
- post_analysis,
- ) -> Iterable[JProjectChange]:
- yield pchange
-
===========changed ref 2===========
# module: coeditor.code_change
+ class ProjectChangeProcessor(Generic[TProb]):
- class ProjectChangeProcessor(Generic[TEnc]):
- def encode_change(
- self, pchange: "JProjectChange", pre_analysis: Any, post_analysis: Any
- ) -> Iterable[TEnc]:
- ...
-
===========changed ref 3===========
# module: coeditor.code_change
+ class ProjectChangeProcessor(Generic[TProb]):
- class ProjectChangeProcessor(Generic[TEnc]):
- def encode_change(
- self, pchange: "JProjectChange", pre_analysis: Any, post_analysis: Any
- ) -> Iterable[TEnc]:
- ...
-
===========changed ref 4===========
# module: coeditor.code_change
+ TProb = TypeVar("TProb", covariant=True)
- TEnc = TypeVar("TEnc", covariant=True)
===========changed ref 5===========
# module: coeditor.code_change
+ def get_python_files(project: RelPath) -> list[RelPath]:
- def get_python_files(project: Path):
+ files = list[RelPath]()
- files = list[Path]()
for f in recurse_find_python_files(FolderIO(str(project))):
f: FileIO
+ files.append(to_rel_path(Path(f.path).relative_to(project)))
- files.append(Path(f.path).relative_to(project))
return files
===========changed ref 6===========
# module: coeditor.common
+ def to_abs_path(path: Path) -> AbsPath:
+ return AbsPath(path.resolve())
+
===========changed ref 7===========
# module: coeditor.common
+ def to_rel_path(path: Path) -> RelPath:
+ if path.is_absolute():
+ raise ValueError(f"Expected a relative path, got: {path}")
+ return RelPath(path)
+
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.pre_edit_analysis
|
Modified
|
temp-1
|
65a052117c316099d5beea02c38dd4ed3cd1b187
|
Speed up analyses using script caching.
|
<0>:<add> script = pstate.scripts[mod_path]
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def pre_edit_analysis(
self,
+ pstate: ProjectState,
- project: jedi.Project,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> Mapping[ModuleName, LineUsageAnalysis]:
"Return the definition usages of each line."
+ project = pstate.project
- # proot = Path(project._path)
result = dict[ModuleName, LineUsageAnalysis]()
src_map = {m.mname: f for f, m in modules.items()}
for mname, mchange in changes.items():
if not isinstance(mchange.module_change, Modified):
continue
lines_to_analyze = set[int]()
for span in mchange.changed.values():
if span.change is Added:
continue
lines_to_analyze.update(range(*span.line_range))
lines_to_analyze.update(range(*span.header_line_range))
mod_path = src_map[mname]
- assert (
- src_file := project.path / mod_path
- ).exists(), f"src file missing: {src_file}"
- script = jedi.Script(path=src_file, project=project)
<0> line_usages = self.analysis.get_line_usages(
script, project.path, lines_to_analyze, silent=True
)
result[mname] = line_usages
return result
|
===========unchanged ref 0===========
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.JModule
mname: ModuleName
tree: ptree.Module
at: coeditor.code_change.JModuleChange
module_change: Change[JModule]
changed: Mapping[ProjectPath, ChangedSpan]
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
at: coeditor.ctx_change_encoder.CtxCodeChangeProblemGenerator.__init__
self.analysis = analysis
at: coeditor.ctx_change_encoder.JediUsageAnalyzer
get_line_usages(script: jedi.Script, proj_root: Path, lines_to_analyze: Collection[int], silent: bool=False)
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
values() -> ValuesView[_VT_co]
===========changed ref 0===========
# module: coeditor.code_change
+ @dataclass
+ class ProjectState:
+ project: jedi.Project
+ scripts: Mapping[RelPath, jedi.Script]
+
===========changed ref 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[T1]:
+ scripts = dict[RelPath, jedi.Script]()
+
+ def parse_module(path: Path):
- def parse_module(path: Path, deep_copy: bool = False):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project=proj)
+ scripts[to_rel_path(path.relative_to(proj._path))] = s
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = s._module_node
- if deep_copy:
- m = copy.deepcopy(m)
-
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
def checkout_commit(commit_hash: str, force: bool = False):
with _tlogger.timed("checkout"):
subprocess.run(
["git", "checkout", "-f", commit_hash],
cwd=project,
capture_output=True,
check=True,
)
# checkout to the first commit
commit</s>
===========changed ref 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[T1]:
# offset: 1
<s>,
capture_output=True,
check=True,
)
# checkout to the first commit
commit_now = history[-1]
checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
+ pstate = ProjectState(proj, scripts)
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
+ f: parse_module(project / f)
- f: parse_module(project / f, deep_copy=False)
for f in tqdm(
get_python_files(project), desc="building initial project", disable=silent
)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
results = list[T1]()
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line</s>
===========changed ref 3===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[T1]:
# offset: 2
<s>_files:
segs = line.split("\t")
if len(segs) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.endswith("A"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if tag.endswith("M"):
path_changes.append(Modified(path, path))
elif len(segs) == 3:
tag, path1, path2 = segs
assert tag.startswith("R")
if not is_src(path1) or not is_src(path2):
continue
path_changes.append(Deleted(path1))
path_changes.append(Added(path2))
# make deep copys of changed modules
to_copy = {
to_rel_path(Path(path_change.before))
for path_change in path_changes
if not isinstance(path_change, Added)
}
_deep_copy_subset_(path2module, to_copy)
checkout_commit(commit_next.hash)
- proj = jedi.Project(path=project, added_sys_path=[project / "src"])
-
new_path2module = path2module.copy()
changed = dict[ModuleName, JModuleChange]()
for path_change in path_changes:
path = project / path_change.earlier()
rel_path = to_rel_path(path.relative</s>
|
coeditor.code_change/ChangeScope.__post_init__
|
Modified
|
temp-1
|
74b3e2333e9961157b553323ce501210bcaaebda
|
Fix span ranges.
|
<0>:<add> h_start = not_none(visited[0].get_start_pos_of_prefix())[0] + e_left
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
header_line_range = line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
+ header_code, e_left, e_right = _strip_empty_lines(header_code)
- header_code, _, e_right = _strip_empty_lines(header_code)
- h_start = visited[0].start_pos[0]
<0> h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
header_line_range = line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
self.header_line_range: LineRange = header_line_range
|
===========unchanged ref 0===========
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
_search_span(line: int) -> "StatementSpan | None"
at: coeditor.code_change.ChangeScope.__post_init__
self.header_code: str = header_code + "\n"
self.header_line_range: LineRange = header_line_range
at: coeditor.code_change.StatementSpan.__post_init__
code, e_left, e_right = _strip_empty_lines(origin_code)
self.code: str = code + "\n"
at: functools
cached_property(func: Callable[[Any], _T])
at: spot.static_analysis
ElemPath = str
|
coeditor.code_change/ChangeScope.from_tree
|
Modified
|
temp-1
|
74b3e2333e9961157b553323ce501210bcaaebda
|
Fix span ranges.
|
<0>:<add> raise ValueError(f"Function with no spans: {path=}, {tree.get_code()=}")
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
<s>tree, ptree.Function)
+
+ def mk_span(stmts):
+ # remove leading newlines
+ n_leading_newlines = 0
+ for s in stmts:
+ if s.type == ptree.Newline.type:
+ n_leading_newlines += 1
+ else:
+ break
+ if n_leading_newlines:
+ stmts = stmts[n_leading_newlines:]
+ if stmts:
+ yield StatementSpan(len(spans), stmts, scope)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.extend(mk_span(current_stmts))
- spans.append(StatementSpan(len(spans), current_stmts, scope))
current_stmts = []
if current_stmts:
+ spans.extend(mk_span(current_stmts))
- spans.append(StatementSpan(len(spans), current_stmts, scope))
if is_func:
# we don't create inner scopes for function contents
+ if not spans:
<0> return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
|
===========above chunk 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
# offset: -1
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
+
+ def mk_span(stmts):
+ # remove leading newlines
+ n</s>
===========unchanged ref 0===========
at: coeditor._utils
as_any(x) -> Any
at: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
_is_scope_statement(stmt: PyNode) -> bool
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
at: parso.python.tree
PythonNode()
Newline()
Module(children)
Function(children)
at: parso.python.tree.Newline
__slots__ = ()
type = 'newline'
at: parso.python.tree.Scope
__slots__ = ()
get_suite()
at: parso.tree.BaseNode.__init__
self.children = children
at: parso.tree.NodeOrLeaf
__slots__ = ('parent',)
type: str
parent: 'Optional[BaseNode]'
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
header_line_range = line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
+ header_code, e_left, e_right = _strip_empty_lines(header_code)
- header_code, _, e_right = _strip_empty_lines(header_code)
+ h_start = not_none(visited[0].get_start_pos_of_prefix())[0] + e_left
- h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
header_line_range = line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code</s>
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: 1
<s>end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
self.header_line_range: LineRange = header_line_range
|
coeditor.code_change/StatementSpan.__post_init__
|
Modified
|
temp-1
|
74b3e2333e9961157b553323ce501210bcaaebda
|
Fix span ranges.
|
<0>:<add> print_err(s)
|
# module: coeditor.code_change
@dataclass
class StatementSpan:
def __post_init__(self):
assert self.statements
- # remove leading newlines
- n_leading_newlines = 0
- stmts = self.statements
- for s in stmts:
- if s.type == ptree.Newline.type:
- n_leading_newlines += 1
- else:
- break
- if n_leading_newlines:
- self.statements = stmts[n_leading_newlines:]
-
origin_code = "".join(s.get_code() for s in self.statements)
+ code, e_left, e_right = _strip_empty_lines(origin_code)
- code, _, e_right = _strip_empty_lines(origin_code)
+ start = not_none(self.statements[0].get_start_pos_of_prefix())[0] + e_left
- start = self.statements[0].start_pos[0]
end = self.statements[-1].end_pos[0] + 1 - e_right
self.code: str = code + "\n"
try:
self.line_range: LineRange = line_range(start, end)
except ValueError:
+ print_err(f"{e_right=}, {start=}, {end=}")
- print_err(f"{origin_code=}, {e_right=}, {start=}, {end=}")
+ print_err("Origin code:")
+ print_err(origin_code)
+ print_err("Stmts:")
+ for s in self.statements:
<0> raise
|
===========unchanged ref 0===========
at: coeditor._utils
not_none(x: Optional[T1]) -> T1
at: coeditor.code_change
PyNode = ptree.PythonBaseNode | ptree.PythonNode
_strip_empty_lines(s: str)
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: parso.tree.BaseNode
__slots__ = ('children',)
get_start_pos_of_prefix()
get_code(include_prefix=True)
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.code_change
+ _non_scope_stmt_types = {
+ "decorated",
+ "async_stmt",
+ ptree.Class.type,
+ ptree.Function.type,
+ }
- _non_scope_stmt_types = {"decorated", "async_stmt"}
===========changed ref 1===========
# module: coeditor.code_change
def _is_scope_statement(stmt: PyNode) -> bool:
- match stmt:
- case ptree.PythonNode(type=node_type) if node_type not in _non_scope_stmt_types:
- return stmt.children[0].type not in ptree._IMPORTS
- case ptree.Flow():
- return True
- case _:
+ """Will only return False for functions, classes, and import statments"""
+ if stmt.type in _non_scope_stmt_types:
+ return False
- return False
+ if stmt.type == "simple_stmt" and stmt.children[0].type in ptree._IMPORTS:
+ return False
+ return True
===========changed ref 2===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
+
+ def mk_span(stmts):
+ # remove leading newlines
+ n_leading_newlines = 0
+ for s in stmts:
+ if s.type == ptree.Newline.type:
+ n_leading_newlines += 1
+ else:
+ break
+ if n_leading_newlines:
+ stmts = stmts[n_leading_newlines:]
+ if stmts:
+ yield StatementSpan(len(spans), stmts, scope)
current_stmts = []
content = (
tree.children
if isinstance(tree, ptree.Module)
else cast(ptree.PythonNode, tree.get_suite()).children
)
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
+ spans.extend(mk_span(current_stmts))
- spans.append(StatementSpan(len(spans), current_stmts, scope))
current_stmts = []
if current_stmts:
+ spans.extend(mk_span(current_stmts))
- spans.append(StatementSpan(len(spans), current_stmts, scope))
if is_func:
# we don't create inner scopes for function contents
+ if not spans:
+ raise ValueError(f"Function with no spans: {path=}, {tree.get_code()=}")
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.</s>
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
# offset: 1
<s>_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
header_line_range = line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
+ header_code, e_left, e_right = _strip_empty_lines(header_code)
- header_code, _, e_right = _strip_empty_lines(header_code)
+ h_start = not_none(visited[0].get_start_pos_of_prefix())[0] + e_left
- h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
header_line_range = line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code</s>
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: 1
<s>end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
self.header_line_range: LineRange = header_line_range
|
coeditor.code_change/_edits_from_commit_history
|
Modified
|
temp-1
|
74b3e2333e9961157b553323ce501210bcaaebda
|
Fix span ranges.
|
<0>:<add> results = list[TEnc]()
|
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
<s>_sys_path=[project / "src"])
pstate = ProjectState(proj, scripts)
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
f: parse_module(project / f)
for f in tqdm(
get_python_files(project), desc="building initial project", disable=silent
)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
- results = list[T1]()
<0> for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len(segs) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.endswith("A"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if</s>
|
===========above chunk 0===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
# offset: -1
<s> name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = s._module_node
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
+ def checkout_commit(commit_hash: str):
- def checkout_commit(commit_hash: str, force: bool = False):
with _tlogger.timed("checkout"):
subprocess.run(
["git", "checkout", "-f", commit_hash],
cwd=project,
capture_output=True,
check=True,
)
+ # to ensure sure we are not accidentally overriding real code changes
+ if list(project.iterdir()) != [project / ".git"]:
+ raise FileExistsError(f"Directory '{project}' should contain only '.git'.")
+
# checkout to the first commit
commit_now = history[-1]
+ checkout_commit(commit_now.hash)
- checkout_commit(commit_now.hash, force=True)
proj = jedi.Project(path=project, added_sys_path=[project / "src"])
pstate = ProjectState(proj, scripts)
# now we can</s>
===========above chunk 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
# offset: -2
scripts = dict[RelPath, jedi.Script]()
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
s = jedi.Script(path=path, project=proj)
scripts[to_rel_path(path.relative_to(proj._path))] = s
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj</s>
===========below chunk 0===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
# offset: 1
<s>Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if tag.endswith("M"):
path_changes.append(Modified(path, path))
elif len(segs) == 3:
tag, path1, path2 = segs
assert tag.startswith("R")
if not is_src(path1) or not is_src(path2):
continue
path_changes.append(Deleted(path1))
path_changes.append(Added(path2))
# make deep copys of changed modules
to_copy = {
to_rel_path(Path(path_change.before))
for path_change in path_changes
if not isinstance(path_change, Added)
}
_deep_copy_subset_(path2module, to_copy)
checkout_commit(commit_next.hash)
new_path2module = path2module.copy()
changed = dict[ModuleName, JModuleChange]()
for path_change in path_changes:
path = project / path_change.earlier()
rel_path = to_rel_path(path.relative_to(project))
match path_change:
case Added():
mod = parse_module(path)
new_path2module[rel_path] = mod
changed[mod.mname] = JModuleChange.from_modules(Added(mod))
case Deleted():
mod = new_path2module.pop(rel_path)
changed[mod.mname] = J</s>
===========below chunk 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
# offset: 2
<s> case Deleted():
mod = new_path2module.pop(rel_path)
changed[mod.mname] = JModuleChange.from_modules(Deleted(mod))
case Modified(path1, path2):
assert path1 == path2
mod_old = new_path2module[rel_path]
new_path2module[rel_path] = mod_new = parse_module(path)
changed[mod_new.mname] = JModuleChange.from_modules(
Modified(mod_old, mod_new)
)
with _tlogger.timed("post_edit_analysis"):
post_analysis = change_processor.post_edit_analysis(
pstate,
new_path2module,
changed,
)
# now go backwards in time to perform pre-edit analysis
checkout_commit(commit_now.hash)
with _tlogger.timed("pre_edit_analysis"):
pre_analysis = change_processor.pre_edit_analysis(
pstate,
path2module,
changed,
)
checkout_commit(commit_next.hash)
modules_mod = Modified(path2module.values(), new_path2module.values())
pchange = JProjectChange(changed, modules_mod, commit_next)
with _tlogger.timed("process_change"):
processed = list(
change_processor.process_change(pchange, pre</s>
===========below chunk 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
+ edit_encoder: Callable[[TProb], Iterable[TEnc]],
- edit_encoder: Callable[[TProb], Iterable[T1]],
ignore_dirs: set[str],
silent: bool,
+ ) -> Sequence[TEnc]:
- ) -> Sequence[T1]:
# offset: 3
<s>, post_analysis)
)
with _tlogger.timed("change_encoder"):
for change in processed:
results.extend(edit_encoder(change))
commit_now = commit_next
path2module = new_path2module
return results
|
coeditor.ctx_change_encoder/PyDefinition.from_name
|
Modified
|
temp-1
|
74b3e2333e9961157b553323ce501210bcaaebda
|
Fix span ranges.
|
<0>:<add> yield PyDefinition(full_name, start_pos, end_pos)
|
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
@staticmethod
def from_name(name: classes.BaseName) -> Iterable["PyDefinition"]:
if (
not name.in_builtin_module()
and (full_name := name.full_name)
+ # and (import_module := name.module_name)
- and (import_module := name.module_name)
and (start_pos := name.get_definition_start_position())
and (end_pos := name.get_definition_end_position())
):
full_name = PyFullName(full_name)
+ # if not full_name.startswith(import_module):
- if not full_name.startswith(import_module):
+ # raise ValueError(f"Inconsistent module: {full_name=}, {import_module=}")
- raise ValueError(f"Inconsistent module: {full_name=}, {import_module=}")
- yield PyDefinition(full_name, import_module, start_pos, end_pos)
<0>
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
PyFullName = NewType("PyFullName", str)
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
at: coeditor.ctx_change_encoder.PyDefinition
full_name: PyFullName
start_pos: tuple[int, int]
end_pos: tuple[int, int]
at: jedi.api.classes
BaseName(inference_state, name)
at: jedi.api.classes.BaseName
_mapping = {
'posixpath': 'os.path',
'riscospath': 'os.path',
'ntpath': 'os.path',
'os2emxpath': 'os.path',
'macpath': 'os.path',
'genericpath': 'os.path',
'posix': 'os',
'_io': 'io',
'_functools': 'functools',
'_collections': 'collections',
'_socket': 'socket',
'_sqlite3': 'sqlite3',
}
_tuple_mapping = dict((tuple(k.split('.')), v) for (k, v) in {
'argparse._ActionsContainer': 'argparse.ArgumentParser',
}.items())
in_builtin_module()
get_definition_start_position()
get_definition_end_position()
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
"""Note that the module and positions can be referring to either the import
statement or the actual definition."""
full_name: PyFullName
- import_module: ModuleName
start_pos: tuple[int, int]
end_pos: tuple[int, int]
===========changed ref 1===========
# module: coeditor.code_change
TProb = TypeVar("TProb", covariant=True)
+ TEnc = TypeVar("TEnc", covariant=True)
===========changed ref 2===========
# module: coeditor.code_change
+ _non_scope_stmt_types = {
+ "decorated",
+ "async_stmt",
+ ptree.Class.type,
+ ptree.Function.type,
+ }
- _non_scope_stmt_types = {"decorated", "async_stmt"}
===========changed ref 3===========
# module: coeditor.code_change
def _is_scope_statement(stmt: PyNode) -> bool:
- match stmt:
- case ptree.PythonNode(type=node_type) if node_type not in _non_scope_stmt_types:
- return stmt.children[0].type not in ptree._IMPORTS
- case ptree.Flow():
- return True
- case _:
+ """Will only return False for functions, classes, and import statments"""
+ if stmt.type in _non_scope_stmt_types:
+ return False
- return False
+ if stmt.type == "simple_stmt" and stmt.children[0].type in ptree._IMPORTS:
+ return False
+ return True
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class StatementSpan:
def __post_init__(self):
assert self.statements
- # remove leading newlines
- n_leading_newlines = 0
- stmts = self.statements
- for s in stmts:
- if s.type == ptree.Newline.type:
- n_leading_newlines += 1
- else:
- break
- if n_leading_newlines:
- self.statements = stmts[n_leading_newlines:]
-
origin_code = "".join(s.get_code() for s in self.statements)
+ code, e_left, e_right = _strip_empty_lines(origin_code)
- code, _, e_right = _strip_empty_lines(origin_code)
+ start = not_none(self.statements[0].get_start_pos_of_prefix())[0] + e_left
- start = self.statements[0].start_pos[0]
end = self.statements[-1].end_pos[0] + 1 - e_right
self.code: str = code + "\n"
try:
self.line_range: LineRange = line_range(start, end)
except ValueError:
+ print_err(f"{e_right=}, {start=}, {end=}")
- print_err(f"{origin_code=}, {e_right=}, {start=}, {end=}")
+ print_err("Origin code:")
+ print_err(origin_code)
+ print_err("Stmts:")
+ for s in self.statements:
+ print_err(s)
raise
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# compute header
if isinstance(self.tree, ptree.Module):
header_code = f"# module: {self.path.module}"
header_line_range = line_range(0, 0, can_be_empty=True)
else:
h_start, h_end = 0, 0
tree = self.tree
to_visit = list[NodeOrLeaf]()
parent = not_none(tree.parent)
while parent.type in ("decorated", "async_funcdef"):
to_visit.insert(0, parent.children[0])
parent = not_none(parent.parent)
to_visit.extend(tree.children)
visited = list[NodeOrLeaf]()
for c in to_visit:
if c.type == "suite":
break
visited.append(c)
header_code = "".join(cast(str, c.get_code()) for c in visited)
+ header_code, e_left, e_right = _strip_empty_lines(header_code)
- header_code, _, e_right = _strip_empty_lines(header_code)
+ h_start = not_none(visited[0].get_start_pos_of_prefix())[0] + e_left
- h_start = visited[0].start_pos[0]
h_end = visited[-1].end_pos[0] + 1 - e_right
assert_eq(count_lines(header_code) == h_end - h_start)
header_line_range = line_range(h_start, h_end)
if self.spans and h_end > self.spans[0].line_range[0]:
raise ValueError(
f"Header covers the fisrt span: {self.path=}, {h_start=}, {h_end=} "
f"{self.spans[0].line_range=}"
)
self.header_code</s>
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
def __post_init__(self):
# offset: 1
<s>end=} "
f"{self.spans[0].line_range=}"
)
self.header_code: str = header_code + "\n"
self.header_line_range: LineRange = header_line_range
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.process_change
|
Modified
|
temp-1
|
bfd57146137ab2ed5c1a9cd99939b8c49797f0e4
|
Record src_info in edits.
|
<0>:<add> processed_cspans.append(span)
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
<s> for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
used_defs.add(pydef)
# only keep unique changed spans
seen = set[tuple[ModuleName, LineRange]]()
for cspan in other_changes:
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in used_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key)
return result
+ processed_cspans = list[ChangedSpan]()
- sorted_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
+ relevant_changes = processed_cspans.copy()
- relevant_changes = sorted_cspans.copy()
yield CtxCodeChangeProblem(
span,
relevant_changes=relevant_changes,
relevant_unchanged=get_relevant_unchanged(
span, relevant_changes
),
+ src_info={"commit": pchange.commit_info},
)
- sorted_cspans.append(span)
<0>
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -1
<s>")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
used_defs = set[PyDefinition]()
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -2
<s>) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "#</s>
===========above chunk 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -3
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used:</s>
===========unchanged ref 0===========
at: coeditor.code_change
LineRange = NewType("LineRange", tuple[int, int])
line_range(start: int, end: int, can_be_empty: bool=False) -> LineRange
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: LineRange)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
ancestors() -> list[Self]
at: coeditor.code_change.ChangeScope.__post_init__
self.header_line_range: LineRange = header_line_range
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.StatementSpan
nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
at: coeditor.code_change.StatementSpan.__post_init__
self.code: str = code + "\n"
self.line_range: LineRange = line_range(start, end)
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan], src_info: dict[str, Any])
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
|
coeditor.ctx_change_encoder/TkCtxCodeChangeProblem.meta_data_lines
|
Modified
|
temp-1
|
bfd57146137ab2ed5c1a9cd99939b8c49797f0e4
|
Record src_info in edits.
|
<0>:<add> f"src_info: {self.src_info}",
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeProblem(TokenizedEdit):
def meta_data_lines(self) -> list[str]:
return [
f"path: {self.path}",
f"n_references: {len(self.references)}",
f"total_reference_tks: {sum(len(ref) for ref in self.references)}",
<0> ]
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeProblem
input_tks: TokenSeq
output_tks: TokenSeq
path: ProjectPath
change_type: Change[None]
named_references: Sequence[tuple[str, TokenSeq]]
src_info: dict[str, Any]
at: coeditor.encoding.TokenizedEdit
input_tks: TokenSeq
output_tks: TokenSeq
main_tks: TokenSeq
path: ProjectPath
change_type: Change[None]
BAD_DELETE = encode_single_line("((bad delete))")
all_ctxs(self) -> dict[str, TokenSeq]
meta_data_lines(self) -> list[str]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeProblem(TokenizedEdit):
input_tks: TokenSeq
output_tks: TokenSeq
path: ProjectPath
change_type: Change[None]
# most relevant to least relevant
named_references: Sequence[tuple[str, TokenSeq]]
+ src_info: dict[str, Any]
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
# jedi.cache.clear_time_caches = lambda: None
@dataclass
class CtxCodeChangeProblem:
span: ChangedSpan
# most relevant to least relevant
relevant_changes: list[ChangedSpan]
# most relevant to least relevant
relevant_unchanged: list[ChangedSpan]
+ # some optional information about how the problem was generated
+ src_info: dict[str, Any]
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-</s>
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
used_defs = set[PyDefinition]</s>
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 2
<s> all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
used_defs.add(pydef)
# only keep unique changed spans
seen = set[tuple[ModuleName, LineRange]]()
for cspan in other_changes:
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in used_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key)
return result
+ processed_cspans = list[ChangedSpan]()
- sorted_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
+ relevant_changes = processed_cspans.copy()
- relevant_changes = sorted_cspans.copy()
yield CtxCodeChange</s>
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder.encode_problem
|
Modified
|
temp-1
|
bfd57146137ab2ed5c1a9cd99939b8c49797f0e4
|
Record src_info in edits.
|
<0>:<add> src_info=problem.src_info,
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
<s>
else:
below_chunks = break_into_chunks(
below_tks,
lambda i: self._encode_parent_scopes(span.parent_scopes, i + 1),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
)
above_chunks = [
(f"above chunk {i}", chunk) for i, chunk in enumerate(above_chunks)
]
below_chunks = [
(f"below chunk {i}", chunk) for i, chunk in enumerate(below_chunks)
]
return TkCtxCodeChangeProblem(
scope_tks + chunk_input,
chunk_output,
path=span.parent_scopes[-1].earlier().path,
change_type=span.change.map(lambda _: None),
named_references=above_chunks + below_chunks + named_references,
<0> )
for l in range(len(tk_delta.deltas) + 1):
finished = l == len(tk_delta.deltas)
input_growth = len(origin_lines[l]) + 2 if l < len(origin_lines) else 1
if (
finished
or chunk_lines >= self.max_lines_to_edit
or len(chunk_input) + input_growth > input_limit
):
if has_change(chunk_output):
yield get_problem(chunk_input, chunk_output)
if finished:
break
chunk_main_input = join_list(origin_lines[chunk_start_l:l], Newline_id)
chunk_main_delta = tk_delta.for_input_range((chunk_start_l, l))
chunk_main_change = chunk_main_delta.to_change_tks(chunk_main</s>
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: -1
<s>to_tk_delta()
chunk_id = 0
chunk_start_l = 0
scope_tks = self._encode_parent_scopes(span.parent_scopes, 0)
chunk_input = TokenSeq()
input_limit = self.max_query_tks - len(scope_tks)
chunk_lines = 0
chunk_output = TokenSeq()
prev_change_tks = TokenSeq()
def get_problem(chunk_input, chunk_output):
# try move some prev_change_tks into the input
above_tks = prev_change_tks
below_tks = join_list(origin_lines[l:], Newline_id)
chunk_input, above_tks, below_tks = self._inline_some_context(
chunk_input, above_tks, below_tks, input_limit
)
# limit the input size if it's too long (can happen for later chunks)
chunk_input = truncate_section(chunk_input, TruncateAt.Right, input_limit)
chunk_output = truncate_output_tks(chunk_input, chunk_output)
chunk_output = truncate_section(
chunk_output, TruncateAt.Right, self.max_output_tks, add_bos=False
)
above_chunks = break_into_chunks(
above_tks,
lambda i: self._encode_parent_scopes(span.parent_scopes, -1 - i),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
right_to_left=True,
)
if finished:
below_chunks = []
else:
below_chunks = break_into_chunks(
below_tks,
lambda i: self._</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: -2
span = problem.span
named_references = list[tuple[str, TokenSeq]]()
# compute the references that are relevant to this span
relevant_chunks = self._group_encode_changed_refs(problem.relevant_changes)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"changed ref {i}", chunk))
relevant_chunks = self._group_encode_unchanged_refs(problem.relevant_unchanged)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"unchanged ref {i}", chunk))
diffs = change_to_line_diffs(span.change)
original, delta = line_diffs_to_original_delta(diffs)
origin_lines = split_list(encode_basic(original), Newline_id)
tk_delta = delta.to_tk_delta()
chunk_id = 0
chunk_start_l = 0
scope_tks = self</s>
===========below chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def encode_problem(
self,
problem: CtxCodeChangeProblem,
) -> Iterable[TkCtxCodeChangeProblem]:
# offset: 1
<s>start_l, l))
chunk_main_change = chunk_main_delta.to_change_tks(chunk_main_input)
prev_change_tks.extend(chunk_main_change)
prev_change_tks.append(Newline_id)
chunk_id += 1
chunk_input = TokenSeq()
chunk_lines = 0
chunk_output = TokenSeq()
chunk_start_l = l
chunk_input.append(get_extra_id(chunk_lines))
if l < len(origin_lines):
chunk_input.extend(origin_lines[l])
chunk_input.append(Newline_id)
line_change = join_list(tk_delta.deltas[l], Newline_id)
chunk_output.append(get_extra_id(chunk_lines))
chunk_output.extend(line_change)
if line_change and line_change[-1] != Del_id:
chunk_output.append(Newline_id)
chunk_lines += 1
===========unchanged ref 0===========
at: cachetools
FIFOCache(maxsize, getsizeof=None)
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
at: coeditor.common
TokenSeq = list[Token]
split_list(lst: list[T1], sep: T1) -> list[list[T1]]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan], src_info: dict[str, Any])
TkCtxCodeChangeProblem(input_tks: TokenSeq, output_tks: TokenSeq, path: ProjectPath, change_type: Change[None], named_references: Sequence[tuple[str, TokenSeq]], src_info: dict[str, Any])
_ObjId = NewType("_ObjId", int)
at: coeditor.ctx_change_encoder.CtxCodeChangeProblem
span: ChangedSpan
relevant_changes: list[ChangedSpan]
relevant_unchanged: list[ChangedSpan]
src_info: dict[str, Any]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
_encode_parent_scopes(scope_changes: Sequence[Change[ChangeScope]], offset: int) -> TokenSeq
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.process_change
|
Modified
|
temp-1
|
265b47b3f37487794989decf11d99183abf08228
|
Add parent scopes as relevant unchanged refs.
|
<0>:<add> relevant_changes = list(reversed(processed_cspans))
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
<s>startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
+ if pydef not in used_defs:
+ used_defs.add(pydef)
- used_defs.add(pydef)
+ sorted_defs.append(pydef)
+ # return unique cspans
- # only keep unique changed spans
seen = set[tuple[ModuleName, LineRange]]()
+ # consider other changes as seen
for cspan in other_changes:
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
+ for used in sorted_defs:
- for used in used_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key)
return result
processed_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
+ # latest changes are more relevant
- relevant_changes = processed_cspans.copy()
<0> yield CtxCodeChangeProblem(
span,
relevant_changes=relevant_changes,
relevant_unchanged=get_relevant_unchanged(
span, relevant_changes
),
src_info={"commit": pchange.commit_info},
)
processed_cspans.append(span)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -1
<s>(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
+ # parent defs are also considered as used
+ parent_defs = [
+ PyDefinition.from_scope(c.earlier()) for c in this_change.parent_scopes
+ ]
+ # immediate parents are more relevant
+ sorted_defs = list(reversed(parent_defs))
+ used_defs = set(sorted_defs)
- used_defs = set[PyDefinition]()
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
#</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -2
<s>as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.</s>
===========above chunk 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -3
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = j</s>
===========unchanged ref 0===========
at: coeditor.code_change
LineRange = NewType("LineRange", tuple[int, int])
line_range(start: int, end: int, can_be_empty: bool=False) -> LineRange
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: LineRange)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
ancestors() -> list[Self]
at: coeditor.code_change.ChangeScope.__post_init__
self.header_line_range: LineRange = header_line_range
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.StatementSpan
nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
at: coeditor.code_change.StatementSpan.__post_init__
self.code: str = code + "\n"
self.line_range: LineRange = line_range(start, end)
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan], src_info: dict[str, Any])
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder._group_encode_changed_refs
|
Modified
|
temp-1
|
265b47b3f37487794989decf11d99183abf08228
|
Add parent scopes as relevant unchanged refs.
|
<0>:<add> file_tks.append(Newline_id)
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _group_encode_changed_refs(
self, changes: Sequence[ChangedSpan]
) -> Sequence[TokenSeq]:
module2changes = groupby(changes, lambda c: c.path.module)
all_chunks = list[TokenSeq]()
for change_group in module2changes.values():
change_group.sort(key=lambda c: c.line_range[0])
file_tks = TokenSeq()
# we'll add module as the chunk header, so we start within the module
last_scope = change_group[0].parent_scopes[:1]
for c in change_group:
scope_diff = []
for i, s in enumerate(c.parent_scopes):
if (
i >= len(last_scope)
or s.earlier().path != last_scope[i].earlier().path
):
scope_diff.append(s)
if scope_diff:
header_tks = self._encode_parent_scopes(scope_diff, 0)
file_tks.extend(header_tks)
body_tks = self._encode_change(c.change)
file_tks.extend(body_tks)
file_tks.append(Newline_id)
<0> last_scope = c.parent_scopes
mod_change = change_group[0].parent_scopes[:1]
mod_chunks = break_into_chunks(
file_tks,
lambda i: self._encode_parent_scopes(mod_change, i),
self.max_ref_tks,
overlap=self.ref_chunk_overlap,
max_return_chunks=self.max_chunks_per_ref,
)
all_chunks.extend(mod_chunks)
return all_chunks
|
===========unchanged ref 0===========
at: coeditor._utils
groupby(iterable: Iterable[T1], keyfunc: Callable[[T1], T2]) -> dict[T2, list[T1]]
at: coeditor.code_change
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: LineRange)
at: coeditor.code_change.ChangedSpan
parent_scopes: Sequence[Change[ChangeScope]]
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder
_ObjId = NewType("_ObjId", int)
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder.__post_init__
self._id_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder._inline_some_context
truncated_above, truncated_below = truncate_sections(
extra_space,
(above_ctx, TruncateAt.Left),
(below_ctx, TruncateAt.Right),
add_bos=True,
)
truncated_above, truncated_below = truncate_sections(
extra_space,
(above_ctx, TruncateAt.Left),
(below_ctx, TruncateAt.Right),
add_bos=True,
)
===========unchanged ref 1===========
above_ctx = truncate_section(
above_ctx, TruncateAt.Right, above_left + self.ref_chunk_overlap
)
above_ctx = TokenSeq()
at: coeditor.encoding
change_to_tokens(change: Change[str]) -> TokenSeq
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass(unsafe_hash=True)
class PyDefinition:
+ @staticmethod
+ def from_scope(scope: ChangeScope) -> "PyDefinition":
+ path = scope.path
+ full_name = PyFullName(f"{path.module}.{path.path}")
+ start_pos = scope.header_line_range[0], 0
+ end_pos = scope.header_line_range[1], 0
+ return PyDefinition(full_name, start_pos, end_pos)
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-</s>
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
+ # parent defs are also considered as used</s>
|
coeditor.ctx_change_encoder/CtxCodeChangeProblemGenerator.process_change
|
Modified
|
temp-1
|
f9c5006c3d426a199c472d33cc31b910793b0357
|
- Fix header trailing newline encoding. - Fix change hiding.
|
<0>:<add> for cspan in (this_change, *other_changes):
|
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
<s>(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
if pydef not in used_defs:
used_defs.add(pydef)
sorted_defs.append(pydef)
# return unique cspans
seen = set[tuple[ModuleName, LineRange]]()
+ # we don't need to show the changed parts again
- # consider other changes as seen
- for cspan in other_changes:
<0> seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in sorted_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key)
return result
processed_cspans = list[ChangedSpan]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
if span.change.as_char() == Modified.as_char():
# latest changes are more relevant
relevant_changes = list(reversed(processed_cspans))
yield CtxCodeChangeProblem(
span,
relevant_changes=relevant_changes,</s>
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -1
<s> body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
# parent defs are also considered as used
parent_defs = [
PyDefinition.from_scope(c.earlier()) for c in this_change.parent_scopes
]
# immediate parents are more relevant
sorted_defs = list(reversed(parent_defs))
used_defs = set(sorted_defs)
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -2
<s>cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
</s>
===========above chunk 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: -3
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in c</s>
===========below chunk 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s>reversed(processed_cspans))
yield CtxCodeChangeProblem(
span,
relevant_changes=relevant_changes,
relevant_unchanged=get_relevant_unchanged(
span, relevant_changes
),
src_info={"commit": pchange.commit_info},
)
processed_cspans.append(span)
===========unchanged ref 0===========
at: coeditor.code_change
LineRange = NewType("LineRange", tuple[int, int])
line_range(start: int, end: int, can_be_empty: bool=False) -> LineRange
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: LineRange)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
ancestors() -> list[Self]
at: coeditor.code_change.ChangeScope.__post_init__
self.header_line_range: LineRange = header_line_range
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.StatementSpan
nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
at: coeditor.code_change.StatementSpan.__post_init__
self.code: str = code + "\n"
self.line_range: LineRange = line_range(start, end)
at: coeditor.ctx_change_encoder
CtxCodeChangeProblem(span: ChangedSpan, relevant_changes: list[ChangedSpan], relevant_unchanged: list[ChangedSpan], src_info: dict[str, Any])
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder._encode_scope_change
|
Modified
|
temp-1
|
f9c5006c3d426a199c472d33cc31b910793b0357
|
- Fix header trailing newline encoding. - Fix change hiding.
|
<0>:<add> hchange = c.map(lambda s: s.header_code.strip("\n"))
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq:
if (key := _ObjId(id(c))) in self._scope_cache:
return self._scope_cache[key]
- hchange = c.map(lambda s: s.header_code)
<0> tks = truncate_section(
change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
)
self._scope_cache[key] = tks
return tks
|
===========unchanged ref 0===========
at: coeditor.code_change
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder
_ObjId = NewType("_ObjId", int)
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
VERSION = "0.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_chunks_per_ref: int = 4
max_lines_per_function: int = 500
skip_unchanged_problems: bool = True
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder.__post_init__
self._scope_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
at: coeditor.encoding
change_to_tokens(change: Change[str]) -> TokenSeq
TruncateAt()
truncate_section(sec: TokenSeq, direction: TruncateAt.Value, limit: int, add_bos: bool=True, inplace: bool=False) -> TokenSeq
at: coeditor.encoding.TruncateAt
Value = int
Left = 0
Right = 1
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-</s>
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
# parent defs are also considered as used
</s>
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 2
<s>_defs = [
PyDefinition.from_scope(c.earlier()) for c in this_change.parent_scopes
]
# immediate parents are more relevant
sorted_defs = list(reversed(parent_defs))
used_defs = set(sorted_defs)
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
if pydef not in used_defs:
used_defs.add(pydef)
sorted_defs.append(pydef)
# return unique cspans
seen = set[tuple[ModuleName, LineRange]]()
+ # we don't need to show the changed parts again
- # consider other changes as seen
+ for cspan in (this_change, *other_changes):
- for cspan in other_changes:
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in sorted_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key</s>
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder._encode_parent_scopes
|
Modified
|
temp-1
|
f9c5006c3d426a199c472d33cc31b910793b0357
|
- Fix header trailing newline encoding. - Fix change hiding.
|
<0>:<add> scope_tks.append(Newline_id)
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_parent_scopes(
self, scope_changes: Sequence[Change[ChangeScope]], offset: int
) -> TokenSeq:
+ scope_tks = join_list(
+ (self._encode_scope_change(c) for c in scope_changes), Newline_id
- scope_tks = join_list((self._encode_scope_change(c) for c in scope_changes))
+ )
if offset != 0:
+ scope_tks.extend(encode_basic(f"\n# offset: {offset}\n"))
- scope_tks.extend(encode_basic(f"# offset: {offset}\n"))
+ else:
<0> scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
return scope_tks
|
===========unchanged ref 0===========
at: coeditor.code_change
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
at: coeditor.common
TokenSeq = list[Token]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder
_encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq
_encode_scope_change(c: Change[ChangeScope]) -> TokenSeq
at: coeditor.encoding
Newline_id = get_tk_id("\n")
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq:
if (key := _ObjId(id(c))) in self._scope_cache:
return self._scope_cache[key]
+ hchange = c.map(lambda s: s.header_code.strip("\n"))
- hchange = c.map(lambda s: s.header_code)
tks = truncate_section(
change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
)
self._scope_cache[key] = tks
return tks
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-</s>
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
# parent defs are also considered as used
</s>
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 2
<s>_defs = [
PyDefinition.from_scope(c.earlier()) for c in this_change.parent_scopes
]
# immediate parents are more relevant
sorted_defs = list(reversed(parent_defs))
used_defs = set(sorted_defs)
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
if pydef not in used_defs:
used_defs.add(pydef)
sorted_defs.append(pydef)
# return unique cspans
seen = set[tuple[ModuleName, LineRange]]()
+ # we don't need to show the changed parts again
- # consider other changes as seen
+ for cspan in (this_change, *other_changes):
- for cspan in other_changes:
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in sorted_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key</s>
|
coeditor.ctx_change_encoder/TkCtxCodeChangeEncoder._encode_change
|
Modified
|
temp-1
|
f9c5006c3d426a199c472d33cc31b910793b0357
|
- Fix header trailing newline encoding. - Fix change hiding.
|
<0>:<add> change = change.map(lambda s: s.strip("\n"))
|
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_change(self, change: Change[str]) -> TokenSeq:
if (key := _ObjId(id(change))) in self._id_cache:
return self._id_cache[key]
<0> change_tks = change_to_tokens(change)
self._id_cache[key] = change_tks
return change_tks
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder
_ObjId = NewType("_ObjId", int)
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder.__post_init__
self._id_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
at: coeditor.ctx_change_encoder.TkCtxCodeChangeEncoder._inline_some_context
truncated_above, truncated_below = truncate_sections(
extra_space,
(above_ctx, TruncateAt.Left),
(below_ctx, TruncateAt.Right),
add_bos=True,
)
truncated_above, truncated_below = truncate_sections(
extra_space,
(above_ctx, TruncateAt.Left),
(below_ctx, TruncateAt.Right),
add_bos=True,
)
above_ctx = truncate_section(
above_ctx, TruncateAt.Right, above_left + self.ref_chunk_overlap
)
above_ctx = TokenSeq()
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq:
if (key := _ObjId(id(c))) in self._scope_cache:
return self._scope_cache[key]
+ hchange = c.map(lambda s: s.header_code.strip("\n"))
- hchange = c.map(lambda s: s.header_code)
tks = truncate_section(
change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
)
self._scope_cache[key] = tks
return tks
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class TkCtxCodeChangeEncoder:
def _encode_parent_scopes(
self, scope_changes: Sequence[Change[ChangeScope]], offset: int
) -> TokenSeq:
+ scope_tks = join_list(
+ (self._encode_scope_change(c) for c in scope_changes), Newline_id
- scope_tks = join_list((self._encode_scope_change(c) for c in scope_changes))
+ )
if offset != 0:
+ scope_tks.extend(encode_basic(f"\n# offset: {offset}\n"))
- scope_tks.extend(encode_basic(f"# offset: {offset}\n"))
+ else:
+ scope_tks.append(Newline_id)
scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
return scope_tks
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(used.full_name.split("."))
cspans = list[ChangedSpan]()
if path is None:
cspan_cache[used] = cspans
return cspans
jmod = before_mod_map[path.module]
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()
match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-</s>
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class CtxCodeChangeProblemGenerator(ProjectChangeProcessor[CtxCodeChangeProblem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Iterable[CtxCodeChangeProblem]:
# offset: 1
<s> ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
# parent defs are also considered as used
</s>
|
coeditor.dataset/_process_commits
|
Modified
|
temp-1
|
f9e7d3bbf0646c97dce2c651fb83fba9c8cfcca8
|
Switch data preparation to new format.
|
<0>:<add> jedi.settings.cache_directory = old_cache
|
# module: coeditor.dataset
def _process_commits(
root: Path,
+ workdir: Path,
commits: Sequence[CommitInfo],
- training: bool,
+ encoder: C3EditEncoder,
- encoder: EditEncoder[T1],
- drop_comments: bool,
- ) -> list[T1]:
+ ) -> Sequence[TkC3Problem]:
+ # use process-specific parso cache
+ old_cache = jedi.settings.cache_directory
+ jedi.settings.cache_directory = workdir / "jedi_cache"
try:
+ # cannot return here since subprocess will be killed after returning
+ return edits_from_commit_history(
+ root,
+ commits,
+ tempdir=workdir / "code",
+ change_processor=encoder.change_processor,
+ edit_encoder=encoder.edit_tokenizer.tokenize_problem,
+ silent=True,
- edits = list(
- edits_from_commit_history(root, commits, drop_comments=drop_comments)
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
return []
- tk_edits = list()
- if isinstance(encoder, AnalysisBasedEditEncoder) or isinstance(
- encoder, QueryRefEditEncoder
- ):
- tk_edits.extend(encoder.encode_pedits(edits, training))
- else:
- for pe in edits:
- tk_edits.extend(encoder.encode_pedit(pe, training))
- return tk_edits
+ finally:
<0>
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
at: coeditor.dataset.C3EditEncoder
change_processor: ProjectChangeProcessor[C3Problem] = field(
default_factory=C3ProblemGenerator
)
edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
at: coeditor.dataset._process_commits
old_cache = jedi.settings.cache_directory
at: jedi.settings
cache_directory = os.path.expanduser(_cache_directory)
at: pathlib
Path()
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
+ @dataclass
+ class C3EditEncoder:
+ change_processor: ProjectChangeProcessor[C3Problem] = field(
+ default_factory=C3ProblemGenerator
+ )
+ edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
+
===========changed ref 1===========
# module: coeditor.encoding
@dataclass
class TkDelta:
def to_change_tks(self, input: TokenSeq) -> TokenSeq:
lines = split_list(input, Newline_id)
+ if len(lines) > len(self.deltas):
- assert len(lines) <= len(self.deltas)
+ print_err(f"{self.deltas}")
+ print_err(f"{input}")
+ raise ValueError(
+ f"Delta is longer than input: {len(lines)=} > {len(self.deltas)=}"
+ )
new_lines = list[TokenSeq]()
for line, delta in zip(lines, self.deltas):
deleted = False
if delta:
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
elif action[0] == Del_id:
deleted = True
if deleted:
new_lines.append([Del_id] + line)
else:
new_lines.append(line)
if len(self.deltas) == len(lines) + 1:
delta = self.deltas[-1]
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
return join_list(new_lines, Newline_id)
|
coeditor.code_change/NoProcessing.process_change
|
Modified
|
temp-1
|
f9e7d3bbf0646c97dce2c651fb83fba9c8cfcca8
|
Switch data preparation to new format.
|
<0>:<add> return [pchange]
|
# module: coeditor.code_change
class NoProcessing(ProjectChangeProcessor[JProjectChange]):
def process_change(
self,
pchange: JProjectChange,
pre_analysis,
post_analysis,
+ ) -> Sequence[JProjectChange]:
- ) -> Iterable[JProjectChange]:
- yield pchange
<0>
|
===========unchanged ref 0===========
at: coeditor.code_change
JProjectChange(changed: Mapping[ModuleName, JModuleChange], all_modules: Modified[Collection[JModule]], commit_info: "CommitInfo | None")
ProjectChangeProcessor()
at: coeditor.code_change.ProjectChangeProcessor
VERSION = "1.0"
process_change(pchange: "JProjectChange", pre_analysis: Any, post_analysis: Any) -> Sequence[TProb]
process_change(self, pchange: "JProjectChange", pre_analysis: Any, post_analysis: Any) -> Sequence[TProb]
===========changed ref 0===========
# module: coeditor.dataset
+ @dataclass
+ class C3EditEncoder:
+ change_processor: ProjectChangeProcessor[C3Problem] = field(
+ default_factory=C3ProblemGenerator
+ )
+ edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
+
===========changed ref 1===========
# module: coeditor.dataset
def datasets_from_repos(
repos_root: Path,
+ encoder: C3EditEncoder,
- encoder: EditEncoder[TEdit],
- drop_comments: bool,
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> dict[str, TokenizedEditDataset[TkC3Problem]]:
- ) -> dict[str, TokenizedEditDataset[TEdit]]:
splits = ["test", "valid", "train"]
projects = dict[str, list[Path]]()
split_is_training = dict[str, list[bool]]()
for split in splits:
if not (repos_root / split).exists():
warnings.warn(f"Split {split} not found at {repos_root / split}.")
continue
ps = [p for p in (repos_root / split).iterdir() if p.is_dir]
projects[split] = ps
training = split == "train"
split_is_training[split] = [training] * len(ps)
if not ps:
warnings.warn(f"No projects found in {split} split")
dataset = dataset_from_projects(
join_list(projects.values()),
encoder=encoder,
- drop_comments=drop_comments,
repo_training=join_list(split_is_training.values()),
max_history_per_repo=max_history_per_repo,
workers=workers,
)
return {k: dataset.subset(v) for k, v in projects.items()}
===========changed ref 2===========
# module: coeditor.dataset
+ def make_or_load_datasets(
+ dataset_name: str,
+ encoder: C3EditEncoder,
+ recreate_data: bool = False,
+ workers: int = DefaultWorkers,
+ ) -> dict[str, TokenizedEditDataset[TkC3Problem]]:
+ config_str = (
+ repr_modified_args(encoder.change_processor)
+ + "-"
+ + repr_modified_args(encoder.edit_tokenizer)
+ )
+ save_dir = get_dataset_dir(dataset_name) / config_str
+
+ if recreate_data or not save_dir.exists():
+ if dataset_name == "SPOT":
+ datasets = {
+ "test": dataset_from_projects(
+ [proj_root()], encoder, [False], workers=workers
+ )
+ }
+ else:
+ datasets = datasets_from_repos(
+ get_dataset_dir(dataset_name) / "repos",
+ encoder,
+ workers=workers,
+ )
+ with timed_action("Saving datasets to disk"):
+ save_datasets(datasets, save_dir)
+ print("Tokenized dataset saved to:", save_dir)
+ print("Dataset stats:")
+ for group, dataset in datasets.items():
+ print("=" * 20, group, "=" * 20)
+ pretty_print_dict(dataset.overall_stats())
+ else:
+ with timed_action("Loading datasets from disk"):
+ datasets = load_datasets(save_dir)
+
+ return datasets
+
===========changed ref 3===========
# module: coeditor.encoding
@dataclass
class TkDelta:
def to_change_tks(self, input: TokenSeq) -> TokenSeq:
lines = split_list(input, Newline_id)
+ if len(lines) > len(self.deltas):
- assert len(lines) <= len(self.deltas)
+ print_err(f"{self.deltas}")
+ print_err(f"{input}")
+ raise ValueError(
+ f"Delta is longer than input: {len(lines)=} > {len(self.deltas)=}"
+ )
new_lines = list[TokenSeq]()
for line, delta in zip(lines, self.deltas):
deleted = False
if delta:
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
elif action[0] == Del_id:
deleted = True
if deleted:
new_lines.append([Del_id] + line)
else:
new_lines.append(line)
if len(self.deltas) == len(lines) + 1:
delta = self.deltas[-1]
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
return join_list(new_lines, Newline_id)
===========changed ref 4===========
# module: coeditor.dataset
def _process_commits(
root: Path,
+ workdir: Path,
commits: Sequence[CommitInfo],
- training: bool,
+ encoder: C3EditEncoder,
- encoder: EditEncoder[T1],
- drop_comments: bool,
- ) -> list[T1]:
+ ) -> Sequence[TkC3Problem]:
+ # use process-specific parso cache
+ old_cache = jedi.settings.cache_directory
+ jedi.settings.cache_directory = workdir / "jedi_cache"
try:
+ # cannot return here since subprocess will be killed after returning
+ return edits_from_commit_history(
+ root,
+ commits,
+ tempdir=workdir / "code",
+ change_processor=encoder.change_processor,
+ edit_encoder=encoder.edit_tokenizer.tokenize_problem,
+ silent=True,
- edits = list(
- edits_from_commit_history(root, commits, drop_comments=drop_comments)
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
return []
- tk_edits = list()
- if isinstance(encoder, AnalysisBasedEditEncoder) or isinstance(
- encoder, QueryRefEditEncoder
- ):
- tk_edits.extend(encoder.encode_pedits(edits, training))
- else:
- for pe in edits:
- tk_edits.extend(encoder.encode_pedit(pe, training))
- return tk_edits
+ finally:
+ jedi.settings.cache_directory = old_cache
|
coeditor.code_change/edits_from_commit_history
|
Modified
|
temp-1
|
f9e7d3bbf0646c97dce2c651fb83fba9c8cfcca8
|
Switch data preparation to new format.
|
<0>:<add> shutil.rmtree(tempdir)
|
# module: coeditor.code_change
def edits_from_commit_history(
project_dir: Path,
history: Sequence[CommitInfo],
tempdir: Path,
change_processor: ProjectChangeProcessor[TProb] = NoProcessing(),
edit_encoder: Callable[[TProb], Iterable[TEnc]] = lambda x: [x],
ignore_dirs=DefaultIgnoreDirs,
silent: bool = False,
) -> Sequence[TEnc]:
"""Incrementally compute the edits to a project from the git history.
Note that this will change the file states in the project directory, so
you should make a copy of the project before calling this function.
"""
tempdir = tempdir.resolve()
if tempdir.exists():
raise FileExistsError(f"Workdir '{tempdir}' already exists.")
+ use_fast_parser = jedi.settings.fast_parser
tempdir.mkdir(parents=True, exist_ok=False)
- use_fast_parser = jedi.settings.fast_parser
try:
run_command(
["cp", "-r", str(project_dir / ".git"), str(tempdir)],
cwd=project_dir.parent,
)
return _edits_from_commit_history(
tempdir, history, change_processor, edit_encoder, ignore_dirs, silent
)
finally:
- run_command(["rm", "-rf", str(tempdir)], cwd=tempdir.parent)
<0> jedi.settings.fast_parser = use_fast_parser
|
===========unchanged ref 0===========
at: coeditor.code_change
DefaultIgnoreDirs = {".venv", ".mypy_cache", ".git", "venv", "build"}
JProjectChange(changed: Mapping[ModuleName, JModuleChange], all_modules: Modified[Collection[JModule]], commit_info: "CommitInfo | None")
TProb = TypeVar("TProb", covariant=True)
TEnc = TypeVar("TEnc", covariant=True)
ProjectChangeProcessor()
NoProcessing()
_edits_from_commit_history(project: Path, history: Sequence[CommitInfo], change_processor: ProjectChangeProcessor[TProb], edit_encoder: Callable[[TProb], Iterable[TEnc]], ignore_dirs: set[str], silent: bool) -> Sequence[TEnc]
at: coeditor.common
run_command(args: Sequence[str], cwd: str | Path) -> str
at: jedi.settings
fast_parser = True
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
resolve(strict: bool=...) -> _P
mkdir(mode: int=..., parents: bool=..., exist_ok: bool=...) -> None
exists() -> bool
at: pathlib.PurePath
__slots__ = (
'_drv', '_root', '_parts',
'_str', '_hash', '_pparts', '_cached_cparts',
)
drive = property(attrgetter('_drv'),
doc="""The drive prefix (letter or UNC path), if any.""")
root = property(attrgetter('_root'),
doc="""The root of the path, if any.""")
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Callable = _CallableType(collections.abc.Callable, 2)
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.code_change
class NoProcessing(ProjectChangeProcessor[JProjectChange]):
def process_change(
self,
pchange: JProjectChange,
pre_analysis,
post_analysis,
+ ) -> Sequence[JProjectChange]:
- ) -> Iterable[JProjectChange]:
+ return [pchange]
- yield pchange
===========changed ref 1===========
# module: coeditor.dataset
+ @dataclass
+ class C3EditEncoder:
+ change_processor: ProjectChangeProcessor[C3Problem] = field(
+ default_factory=C3ProblemGenerator
+ )
+ edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
+
===========changed ref 2===========
# module: coeditor.dataset
def datasets_from_repos(
repos_root: Path,
+ encoder: C3EditEncoder,
- encoder: EditEncoder[TEdit],
- drop_comments: bool,
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> dict[str, TokenizedEditDataset[TkC3Problem]]:
- ) -> dict[str, TokenizedEditDataset[TEdit]]:
splits = ["test", "valid", "train"]
projects = dict[str, list[Path]]()
split_is_training = dict[str, list[bool]]()
for split in splits:
if not (repos_root / split).exists():
warnings.warn(f"Split {split} not found at {repos_root / split}.")
continue
ps = [p for p in (repos_root / split).iterdir() if p.is_dir]
projects[split] = ps
training = split == "train"
split_is_training[split] = [training] * len(ps)
if not ps:
warnings.warn(f"No projects found in {split} split")
dataset = dataset_from_projects(
join_list(projects.values()),
encoder=encoder,
- drop_comments=drop_comments,
repo_training=join_list(split_is_training.values()),
max_history_per_repo=max_history_per_repo,
workers=workers,
)
return {k: dataset.subset(v) for k, v in projects.items()}
===========changed ref 3===========
# module: coeditor.dataset
+ def make_or_load_datasets(
+ dataset_name: str,
+ encoder: C3EditEncoder,
+ recreate_data: bool = False,
+ workers: int = DefaultWorkers,
+ ) -> dict[str, TokenizedEditDataset[TkC3Problem]]:
+ config_str = (
+ repr_modified_args(encoder.change_processor)
+ + "-"
+ + repr_modified_args(encoder.edit_tokenizer)
+ )
+ save_dir = get_dataset_dir(dataset_name) / config_str
+
+ if recreate_data or not save_dir.exists():
+ if dataset_name == "SPOT":
+ datasets = {
+ "test": dataset_from_projects(
+ [proj_root()], encoder, [False], workers=workers
+ )
+ }
+ else:
+ datasets = datasets_from_repos(
+ get_dataset_dir(dataset_name) / "repos",
+ encoder,
+ workers=workers,
+ )
+ with timed_action("Saving datasets to disk"):
+ save_datasets(datasets, save_dir)
+ print("Tokenized dataset saved to:", save_dir)
+ print("Dataset stats:")
+ for group, dataset in datasets.items():
+ print("=" * 20, group, "=" * 20)
+ pretty_print_dict(dataset.overall_stats())
+ else:
+ with timed_action("Loading datasets from disk"):
+ datasets = load_datasets(save_dir)
+
+ return datasets
+
===========changed ref 4===========
# module: coeditor.encoding
@dataclass
class TkDelta:
def to_change_tks(self, input: TokenSeq) -> TokenSeq:
lines = split_list(input, Newline_id)
+ if len(lines) > len(self.deltas):
- assert len(lines) <= len(self.deltas)
+ print_err(f"{self.deltas}")
+ print_err(f"{input}")
+ raise ValueError(
+ f"Delta is longer than input: {len(lines)=} > {len(self.deltas)=}"
+ )
new_lines = list[TokenSeq]()
for line, delta in zip(lines, self.deltas):
deleted = False
if delta:
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
elif action[0] == Del_id:
deleted = True
if deleted:
new_lines.append([Del_id] + line)
else:
new_lines.append(line)
if len(self.deltas) == len(lines) + 1:
delta = self.deltas[-1]
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
return join_list(new_lines, Newline_id)
|
coeditor.code_change/_edits_from_commit_history
|
Modified
|
temp-1
|
f9e7d3bbf0646c97dce2c651fb83fba9c8cfcca8
|
Switch data preparation to new format.
|
<0>:<add> pchange, pre_analysis, post_analysis
|
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[TEnc]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[TEnc]:
<s>)
changed[mod.mname] = JModuleChange.from_modules(Deleted(mod))
case Modified(path1, path2):
assert path1 == path2
mod_old = new_path2module[rel_path]
new_path2module[rel_path] = mod_new = parse_module(path)
changed[mod_new.mname] = JModuleChange.from_modules(
Modified(mod_old, mod_new)
)
with _tlogger.timed("post_edit_analysis"):
post_analysis = change_processor.post_edit_analysis(
pstate,
new_path2module,
changed,
)
# now go backwards in time to perform pre-edit analysis
checkout_commit(commit_now.hash)
with _tlogger.timed("pre_edit_analysis"):
pre_analysis = change_processor.pre_edit_analysis(
pstate,
path2module,
changed,
)
checkout_commit(commit_next.hash)
modules_mod = Modified(path2module.values(), new_path2module.values())
pchange = JProjectChange(changed, modules_mod, commit_next)
with _tlogger.timed("process_change"):
+ processed = change_processor.process_change(
- processed = list(
- change_processor.process_change(pchange, pre_analysis, post_analysis)
<0> )
with _tlogger.timed("change_encoder"):
for change in processed:
results.extend(edit_encoder(change))
commit_now = commit_next
path2module = new_path2module
return results
|
===========above chunk 0===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[TEnc]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[TEnc]:
# offset: -1
<s>"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes.append(Deleted(path))
if tag.endswith("M"):
path_changes.append(Modified(path, path))
elif len(segs) == 3:
tag, path1, path2 = segs
assert tag.startswith("R")
if not is_src(path1) or not is_src(path2):
continue
path_changes.append(Deleted(path1))
path_changes.append(Added(path2))
# make deep copys of changed modules
to_copy = {
to_rel_path(Path(path_change.before))
for path_change in path_changes
if not isinstance(path_change, Added)
}
_deep_copy_subset_(path2module, to_copy)
checkout_commit(commit_next.hash)
new_path2module = path2module.copy()
changed = dict[ModuleName, JModuleChange]()
for path_change in path_changes:
path = project / path_change.earlier()
rel_path = to_rel_path(path.relative_to(project))
match path_change:
case Added():
mod = parse_module(path)
new_path2module[rel_path] = mod
changed[mod.mname] = JModuleChange.from_modules(Added(mod))
case Deleted():
mod = new_path2module.pop(rel_path)
changed[mod.mname] = JModuleChange.from_modules(Deleted(mod))
case Modified(path1</s>
===========above chunk 1===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[TEnc]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[TEnc]:
# offset: -2
<s>project, added_sys_path=[project / "src"])
pstate = ProjectState(proj, scripts)
# now we can get the first project state, although this not needed for now
# but we'll use it later for pre-edit analysis
path2module = {
f: parse_module(project / f)
for f in tqdm(
get_python_files(project), desc="building initial project", disable=silent
)
}
def is_src(path_s: str) -> bool:
path = Path(path_s)
return path.suffix == ".py" and all(p not in ignore_dirs for p in path.parts)
future_commits = list(reversed(history[:-1]))
results = list[TEnc]()
for commit_next in tqdm(
future_commits, smoothing=0, desc="processing commits", disable=silent
):
# get changed files
changed_files = run_command(
["git", "diff", commit_now.hash, commit_next.hash, "--name-status"],
cwd=project,
).splitlines()
path_changes = list[Change[str]]()
for line in changed_files:
segs = line.split("\t")
if len(segs) == 2:
tag, path = segs
if not is_src(path):
continue
if tag.endswith("A"):
path_changes.append(Added(path))
elif tag.endswith("D"):
path_changes</s>
===========above chunk 2===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[TEnc]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[TEnc]:
# offset: -3
<s>.relative_to(proj._path))] = s
mcontext = s._get_module_context()
assert isinstance(mcontext, ModuleContext)
mname = cast(str, mcontext.py__name__())
if mname.startswith("src."):
e = ValueError(f"Bad module name: {mname}")
files = list(project.iterdir())
print_err(f"project: {proj}", file=sys.stderr)
print_err(f"files in root: {files}", file=sys.stderr)
raise e
m = s._module_node
assert isinstance(m, ptree.Module)
# mname = PythonProject.rel_path_to_module_name(path.relative_to(proj.path))
# m = parso.parse(path.read_text())
return JModule(mname, m)
def checkout_commit(commit_hash: str):
with _tlogger.timed("checkout"):
subprocess.run(
["git", "checkout", "-f", commit_hash],
cwd=project,
capture_output=True,
check=True,
)
# to ensure sure we are not accidentally overriding real code changes
if list(project.iterdir()) != [project / ".git"]:
raise FileExistsError(f"Directory '{project}' should contain only '.git'.")
# checkout to the first commit
commit_now = history[-1]
checkout_commit(commit_now.hash)
proj = jedi.Project(</s>
===========above chunk 3===========
# module: coeditor.code_change
def _edits_from_commit_history(
project: Path,
history: Sequence[CommitInfo],
change_processor: ProjectChangeProcessor[TProb],
edit_encoder: Callable[[TProb], Iterable[TEnc]],
ignore_dirs: set[str],
silent: bool,
) -> Sequence[TEnc]:
# offset: -4
scripts = dict[RelPath, jedi.Script]()
def parse_module(path: Path):
with _tlogger.timed("parse_module"):
+ assert path.is_absolute(), f"Path is not absolute: {path=}"
s = jedi.Script(path=path, project=proj)
scripts[to_rel_path</s>
|
scripts.coeditor.train_retrieval_model/train_model
|
Modified
|
temp-1
|
f9e7d3bbf0646c97dce2c651fb83fba9c8cfcca8
|
Switch data preparation to new format.
|
<0>:<add> datasets = make_or_load_datasets(dataset_name, encoder, recreate_data=recreate_data)
|
<s>model(
dataset_name="medium",
model_variant="-sig-analysis-post_usees",
+ encoder: C3EditEncoder = C3EditEncoder(),
- encoder: QueryRefEditEncoder = QueryRefEditEncoder(),
- drop_comments: bool = True,
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# model_variant = "-file"
model_name = f"coeditor-{dataset_name}"
model_name += model_variant
dec_args = DecodingArgs()
if train_args.quicktest:
model_name = "quicktest-" + model_name
if not eval_only:
check_save_dir(model_name)
- datasets = make_or_load_datasets(
- dataset_name, encoder, drop_comments=drop_comments, recreate_data=recreate_data
- )
<0>
config_dict = {
k: get_modified_args(v)
for k, v in {
"data_args": batch_args,
"train_args": train_args,
"dec_args": dec_args,
}.items()
}
project = "Coeditor" if not train_args.quicktest else "Coeditor-quicktest"
wandb.init(dir="..", project=project, name=model_name, config=config_dict)
if train_args.quicktest:
print("Using fewer data for quick test.")
n_quick_exs = 20
for name, dataset in datasets.items():
datasets[name] = TokenizedEditDataset.from_edits(
dataset.all_edits()[:n_quick_exs]
)
if not eval_only:
model = RetrievalEditorModel.from_code_t5(
"base", reuse_embed=True, reinit_weights=train_args.re</s>
|
===========below chunk 0===========
<s>_name="medium",
model_variant="-sig-analysis-post_usees",
+ encoder: C3EditEncoder = C3EditEncoder(),
- encoder: QueryRefEditEncoder = QueryRefEditEncoder(),
- drop_comments: bool = True,
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: 1
<s>Model.from_code_t5(
"base", reuse_embed=True, reinit_weights=train_args.reinit_weights
)
else:
model = RetrievalEditorModel.load(get_model_dir() / model_name)
if os.getenv("CUDA_VISIBLE_DEVICES") is None:
warnings.warn(
"CUDA_VISIBLE_DEVICES not set, using 0. Note that "
"the Huggingface Trainer will use all visible GPUs for training."
)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
if not eval_only:
with timed_action("Warm-up Training"):
warmup_bargs = copy.deepcopy(batch_args)
warmup_bargs.max_total_ref_tks //= 4
warmup_bargs.min_queires *= 4
warmup_bargs.max_queries *= 2
warmup_targs = copy.deepcopy(train_args)
warmup_targs.learning_rate *= 4
warmup_targs.max_train_epochs = 1
all_edits = datasets["train"].all_edits()
warmup_edits = random_subset(all_edits, len(all_edits) // 4)
model.train_on_data(
model_name,
TokenizedEditDataset.from_edits(warmup_edits),
datasets["valid"],
warmup_targs,
batch_args=warmup_b</s>
===========below chunk 1===========
<s>_name="medium",
model_variant="-sig-analysis-post_usees",
+ encoder: C3EditEncoder = C3EditEncoder(),
- encoder: QueryRefEditEncoder = QueryRefEditEncoder(),
- drop_comments: bool = True,
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: 2
<s>up_edits),
datasets["valid"],
warmup_targs,
batch_args=warmup_bargs,
eval_batch_args=test_batch_args,
)
with timed_action("Fine-tune Training"):
model.train_on_data(
model_name,
datasets["train"],
datasets["valid"],
train_args,
batch_args=batch_args,
eval_batch_args=test_batch_args,
)
model.to("cuda")
with timed_action("Loss Evaluation"):
eval_result = model.eval_loss_on_data(datasets["test"], test_batch_args)
eval_dict = {f"test/{k}": v.average() for k, v in eval_result.items()}
wandb.log(eval_dict)
max_saved_samples = 300
with timed_action("Accuracy Evaluation"):
dec_result = model.predict_on_data(datasets["test"], test_batch_args, dec_args)
pickle_dump(get_model_dir() / model_name / "dec_result.pkl", dec_result)
exact_acc, exact_correct_map = dec_result.exact_match_accuracy()
wandb.log({"test/exact-acc": exact_acc.average()})
out_dir = get_</s>
===========below chunk 2===========
<s>_name="medium",
model_variant="-sig-analysis-post_usees",
+ encoder: C3EditEncoder = C3EditEncoder(),
- encoder: QueryRefEditEncoder = QueryRefEditEncoder(),
- drop_comments: bool = True,
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: 3
<s>dir() / model_name / "exact_match_samples"
dec_result.save_examples_to_dir(
out_dir, random_subset(exact_correct_map, max_saved_samples)
)
print("Exact-match samples saved to:", out_dir)
return model
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
timed_action(name: str, silent: bool=False)
pickle_dump(file: Path, obj: Any)
get_modified_args(instance, flatten: bool=False) -> dict[str, Any] | None
at: coeditor.common
get_model_dir(trained=True) -> Path
random_subset(all: Mapping[T1, T2], n: int, rng: random.Random | int | None=None) -> dict[T1, T2]
random_subset(all: Sequence[T1], n: int, rng: random.Random | int | None=None) -> list[T1]
at: coeditor.dataset
TokenizedEditDataset(_edits: list[TEdit])
at: coeditor.dataset.TokenizedEditDataset
_edits: list[TEdit]
from_edits(edits: Iterable[TEdit]) -> "TokenizedEditDataset[TEdit]"
at: coeditor.model
DecodingArgs(max_output_tks: int=512, do_sample: bool=False, top_p: float=0.9, num_beams: Optional[int]=1, length_penalty: float=0.0, marginalize_samples: int=1)
at: copy
deepcopy(x: _T, memo: Optional[Dict[int, Any]]=..., _nil: Any=...) -> _T
at: os
environ = _createenviron()
getenv(key: str, default: _T) -> Union[str, _T]
getenv(key: str) -> Optional[str]
at: train_model
check_save_dir(model_name: str) -> None
at: wandb
init = wandb_sdk.init
===========unchanged ref 1===========
log = _preinit.PreInitCallable("wandb.log", wandb_sdk.wandb_run.Run.log)
===========changed ref 0===========
# module: coeditor.code_change
class NoProcessing(ProjectChangeProcessor[JProjectChange]):
def process_change(
self,
pchange: JProjectChange,
pre_analysis,
post_analysis,
+ ) -> Sequence[JProjectChange]:
- ) -> Iterable[JProjectChange]:
+ return [pchange]
- yield pchange
===========changed ref 1===========
# module: coeditor.dataset
+ @dataclass
+ class C3EditEncoder:
+ change_processor: ProjectChangeProcessor[C3Problem] = field(
+ default_factory=C3ProblemGenerator
+ )
+ edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
+
|
coeditor.dataset/dataset_from_projects
|
Modified
|
temp-1
|
c2672194b1a2abfcf00b08e9e73e6851407c3d95
|
Fix jedi cache error. Print processor errors.
|
<0>:<add> pretty_print_dict(error_counts)
|
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
<s>, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
+ presults = pmap(
- tk_edits = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits = dict[Path, list[TkC3Problem]]()
+
+ error_counts = dict[str, int]()
+ for root, pr in zip(roots, presults):
- for root, edits in zip(roots, tk_edits):
+ project2edits.setdefault(root, []).extend(pr.edits)
- project2edits.setdefault(root, []).extend(edits)
+ for k, v in pr.processor_errors.items():
+ error_counts[k] = error_counts.get(k, 0) + v
+
+ print("Processor Errors:")
<0>
return TokenizedEditDataset(project2edits)
|
===========above chunk 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: -1
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
</s>
===========unchanged ref 0===========
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
pmap(f: Callable[..., T1], iter3: Iterable[Any], iter4: Iterable[Any], iter5: Iterable[Any], iter6: Iterable[Any], /, *iterables: Iterable[Any], desc: str | None=None, key_args: Mapping[str, Any] | None=None, max_workers: int | None=None, chunksize: int | None=None, tqdm_args: Mapping[str, Any] | None=None) -> list[T1]
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
_ProcessingResult(edits: Sequence[TkC3Problem], processor_errors: dict[str, int])
_process_commits(root: Path, workdir: Path, commits: Sequence[CommitInfo], encoder: C3EditEncoder) -> _ProcessingResult
at: coeditor.dataset.C3EditEncoder
change_processor: ProjectChangeProcessor[C3Problem] = field(
default_factory=C3ProblemGenerator
)
at: coeditor.dataset._process_commits
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
edits = []
at: math
ceil(x: SupportsFloat, /) -> int
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
exists() -> bool
at: shutil
rmtree(path: Union[bytes, StrPath], ignore_errors: bool=..., onerror: Optional[Callable[[Any, Any, Any], Any]]=...) -> None
at: tempfile
gettempdir() -> str
===========unchanged ref 1===========
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
+ @dataclass
+ class _ProcessingResult:
+ edits: Sequence[TkC3Problem]
+ processor_errors: dict[str, int]
+
===========changed ref 1===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
+ ) -> _ProcessingResult:
- ) -> Sequence[TkC3Problem]:
# use process-specific parso cache
- old_cache = jedi.settings.cache_directory
- jedi.settings.cache_directory = workdir / "jedi_cache"
+ _fix_jedi_cache(workdir)
try:
# cannot return here since subprocess will be killed after returning
+ edits = edits_from_commit_history(
- return edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
+ edits = []
+ return _ProcessingResult(
+ edits,
+ encoder.change_processor.get_errors(),
+ )
- return []
- finally:
- jedi.settings.cache_directory = old_cache
|
coeditor.ctx_change_encoder/C3ProblemGenerator.__init__
|
Modified
|
temp-1
|
c2672194b1a2abfcf00b08e9e73e6851407c3d95
|
Fix jedi cache error. Print processor errors.
|
<0>:<add> self.analyzer = analyzer
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __init__(self, analyzer: "JediUsageAnalyzer | None" = None):
- def __init__(self, analysis: "JediUsageAnalyzer | None" = None):
+ if analyzer is None:
- if analysis is None:
+ analyzer = JediUsageAnalyzer()
- analysis = JediUsageAnalyzer()
- self.analysis = analysis
<0>
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
JediUsageAnalyzer()
at: coeditor.ctx_change_encoder.C3ProblemGenerator
VERSION = "1.0"
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def clear_errors(self):
+ return None
+
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def get_errors(self) -> dict[str, int]:
+ return dict()
+
===========changed ref 2===========
# module: coeditor.dataset
+ @dataclass
+ class _ProcessingResult:
+ edits: Sequence[TkC3Problem]
+ processor_errors: dict[str, int]
+
===========changed ref 3===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
+ ) -> _ProcessingResult:
- ) -> Sequence[TkC3Problem]:
# use process-specific parso cache
- old_cache = jedi.settings.cache_directory
- jedi.settings.cache_directory = workdir / "jedi_cache"
+ _fix_jedi_cache(workdir)
try:
# cannot return here since subprocess will be killed after returning
+ edits = edits_from_commit_history(
- return edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
+ edits = []
+ return _ProcessingResult(
+ edits,
+ encoder.change_processor.get_errors(),
+ )
- return []
- finally:
- jedi.settings.cache_directory = old_cache
===========changed ref 4===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
+ presults = pmap(
- tk_edits = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder</s>
===========changed ref 5===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: 1
<s>process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits = dict[Path, list[TkC3Problem]]()
+
+ error_counts = dict[str, int]()
+ for root, pr in zip(roots, presults):
- for root, edits in zip(roots, tk_edits):
+ project2edits.setdefault(root, []).extend(pr.edits)
- project2edits.setdefault(root, []).extend(edits)
+ for k, v in pr.processor_errors.items():
+ error_counts[k] = error_counts.get(k, 0) + v
+
+ print("Processor Errors:")
+ pretty_print_dict(error_counts)
return TokenizedEditDataset(project2edits)
|
coeditor.ctx_change_encoder/C3ProblemGenerator.pre_edit_analysis
|
Modified
|
temp-1
|
c2672194b1a2abfcf00b08e9e73e6851407c3d95
|
Fix jedi cache error. Print processor errors.
|
<0>:<add> line_usages = self.analyzer.get_line_usages(
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def pre_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> Mapping[ModuleName, LineUsageAnalysis]:
"Return the definition usages of each line."
project = pstate.project
result = dict[ModuleName, LineUsageAnalysis]()
src_map = {m.mname: f for f, m in modules.items()}
for mname, mchange in changes.items():
if not isinstance(mchange.module_change, Modified):
continue
lines_to_analyze = set[int]()
for span in mchange.changed.values():
if span.change is Added:
continue
lines_to_analyze.update(range(*span.line_range))
lines_to_analyze.update(range(*span.header_line_range))
mod_path = src_map[mname]
script = pstate.scripts[mod_path]
- line_usages = self.analysis.get_line_usages(
<0> script, project.path, lines_to_analyze, silent=True
)
result[mname] = line_usages
return result
|
===========unchanged ref 0===========
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.JModule
mname: ModuleName
tree: ptree.Module
at: coeditor.code_change.JModuleChange
module_change: Change[JModule]
changed: Mapping[ProjectPath, ChangedSpan]
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
at: coeditor.ctx_change_encoder.C3ProblemGenerator.__init__
analyzer = JediUsageAnalyzer()
self.analyzer = analyzer
at: coeditor.ctx_change_encoder.JediUsageAnalyzer.__post_init__
self.error_counts = dict[str, int]()
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
values() -> ValuesView[_VT_co]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def clear_errors(self):
+ return self.analyzer.error_counts.clear()
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def get_errors(self) -> dict[str, int]:
+ return self.analyzer.error_counts
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __init__(self, analyzer: "JediUsageAnalyzer | None" = None):
- def __init__(self, analysis: "JediUsageAnalyzer | None" = None):
+ if analyzer is None:
- if analysis is None:
+ analyzer = JediUsageAnalyzer()
- analysis = JediUsageAnalyzer()
+ self.analyzer = analyzer
- self.analysis = analysis
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def clear_errors(self):
+ return None
+
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def get_errors(self) -> dict[str, int]:
+ return dict()
+
===========changed ref 5===========
# module: coeditor.dataset
+ @dataclass
+ class _ProcessingResult:
+ edits: Sequence[TkC3Problem]
+ processor_errors: dict[str, int]
+
===========changed ref 6===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
+ ) -> _ProcessingResult:
- ) -> Sequence[TkC3Problem]:
# use process-specific parso cache
- old_cache = jedi.settings.cache_directory
- jedi.settings.cache_directory = workdir / "jedi_cache"
+ _fix_jedi_cache(workdir)
try:
# cannot return here since subprocess will be killed after returning
+ edits = edits_from_commit_history(
- return edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
+ edits = []
+ return _ProcessingResult(
+ edits,
+ encoder.change_processor.get_errors(),
+ )
- return []
- finally:
- jedi.settings.cache_directory = old_cache
===========changed ref 7===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
+ presults = pmap(
- tk_edits = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder</s>
===========changed ref 8===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: 1
<s>process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits = dict[Path, list[TkC3Problem]]()
+
+ error_counts = dict[str, int]()
+ for root, pr in zip(roots, presults):
- for root, edits in zip(roots, tk_edits):
+ project2edits.setdefault(root, []).extend(pr.edits)
- project2edits.setdefault(root, []).extend(edits)
+ for k, v in pr.processor_errors.items():
+ error_counts[k] = error_counts.get(k, 0) + v
+
+ print("Processor Errors:")
+ pretty_print_dict(error_counts)
return TokenizedEditDataset(project2edits)
|
coeditor.ctx_change_encoder/C3ProblemGenerator.post_edit_analysis
|
Modified
|
temp-1
|
c2672194b1a2abfcf00b08e9e73e6851407c3d95
|
Fix jedi cache error. Print processor errors.
|
<0>:<add> for source in _fast_goto(
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def post_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> list[ModuleName]:
"Return the topological order among the modules."
# sort modules topologically
module_deps = dict[ModuleName, set[ModuleName]]()
for rel_path, module in modules.items():
names = {n for n in module.imported_names}
script = pstate.scripts[rel_path]
deps = module_deps.setdefault(module.mname, set())
for n in names:
- for source in fast_goto(
<0> script, n, follow_imports=True, follow_builtin_imports=False
):
deps.add(source.module_name)
module_order = sort_modules_by_imports(module_deps)
return module_order
|
===========unchanged ref 0===========
at: coeditor.code_change.JModule
mname: ModuleName
at: coeditor.common
RelPath = NewType("RelPath", Path)
at: coeditor.ctx_change_encoder.C3ProblemGenerator.__init__
analyzer = JediUsageAnalyzer()
self.analyzer = analyzer
at: coeditor.ctx_change_encoder.C3ProblemGenerator.pre_edit_analysis
project = pstate.project
lines_to_analyze = set[int]()
script = pstate.scripts[mod_path]
at: coeditor.ctx_change_encoder.JediUsageAnalyzer
get_line_usages(self, script: jedi.Script, proj_root: Path, lines_to_analyze: Collection[int], silent: bool=False)
get_line_usages(script: jedi.Script, proj_root: Path, lines_to_analyze: Collection[int], silent: bool=False)
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def clear_errors(self):
+ return self.analyzer.error_counts.clear()
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def get_errors(self) -> dict[str, int]:
+ return self.analyzer.error_counts
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __init__(self, analyzer: "JediUsageAnalyzer | None" = None):
- def __init__(self, analysis: "JediUsageAnalyzer | None" = None):
+ if analyzer is None:
- if analysis is None:
+ analyzer = JediUsageAnalyzer()
- analysis = JediUsageAnalyzer()
+ self.analyzer = analyzer
- self.analysis = analysis
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def pre_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> Mapping[ModuleName, LineUsageAnalysis]:
"Return the definition usages of each line."
project = pstate.project
result = dict[ModuleName, LineUsageAnalysis]()
src_map = {m.mname: f for f, m in modules.items()}
for mname, mchange in changes.items():
if not isinstance(mchange.module_change, Modified):
continue
lines_to_analyze = set[int]()
for span in mchange.changed.values():
if span.change is Added:
continue
lines_to_analyze.update(range(*span.line_range))
lines_to_analyze.update(range(*span.header_line_range))
mod_path = src_map[mname]
script = pstate.scripts[mod_path]
+ line_usages = self.analyzer.get_line_usages(
- line_usages = self.analysis.get_line_usages(
script, project.path, lines_to_analyze, silent=True
)
result[mname] = line_usages
return result
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def clear_errors(self):
+ return None
+
===========changed ref 5===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def get_errors(self) -> dict[str, int]:
+ return dict()
+
===========changed ref 6===========
# module: coeditor.dataset
+ @dataclass
+ class _ProcessingResult:
+ edits: Sequence[TkC3Problem]
+ processor_errors: dict[str, int]
+
===========changed ref 7===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
+ ) -> _ProcessingResult:
- ) -> Sequence[TkC3Problem]:
# use process-specific parso cache
- old_cache = jedi.settings.cache_directory
- jedi.settings.cache_directory = workdir / "jedi_cache"
+ _fix_jedi_cache(workdir)
try:
# cannot return here since subprocess will be killed after returning
+ edits = edits_from_commit_history(
- return edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
+ edits = []
+ return _ProcessingResult(
+ edits,
+ encoder.change_processor.get_errors(),
+ )
- return []
- finally:
- jedi.settings.cache_directory = old_cache
===========changed ref 8===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
+ presults = pmap(
- tk_edits = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder</s>
|
coeditor.ctx_change_encoder/JediUsageAnalyzer.get_line_usages
|
Modified
|
temp-1
|
c2672194b1a2abfcf00b08e9e73e6851407c3d95
|
Fix jedi cache error. Print processor errors.
|
<0>:<add> errors[err_text] = errors.get(err_text, 0) + 1
|
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
<s>names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
unexpected = dict[str, int]()
for name in tqdm(all_names, f"Analyzing {script.path}", disable=silent):
name: tree.Name
line = name.start_pos[0]
if line not in lines_to_analyze:
continue
usages = line2usages.setdefault(line, set())
try:
+ defs = _fast_goto(
- defs = fast_goto(
script,
name,
follow_imports=True,
follow_builtin_imports=False,
)
for d in defs:
usages.update(PyDefinition.from_name(d))
+ except KeyError:
+ # for debugging
+ raise
except Exception as e:
- # if the message is "not enough values to unpack"
err_text = repr(e)
is_known = any(err in err_text for err in _KnownJediErrors)
- errors[err_text] = errors.setdefault(err_text, 0) + 1
<0> if not is_known:
unexpected[err_text] = unexpected.get(err_text, 0) + 1
if unexpected:
project_name = proj_root.name
if script.path:
file_path = script.path.relative_to(proj_root)
else:
file_path = "<unknown>"
for err, count in unexpected.items():
logging.warn(
f"Unexpected error when analyzing '{project_name}/{file_path}' ({count=}): {err}"
)
return LineUsageAnalysis(line2usages)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
# offset: -1
jmod: tree.Module = script._module_node
line2usages = dict[int, set[PyDefinition]]()
all_names = [
name for k, names in jmod.get_used_names()._dict.items() for name in names
]
all_names.sort(key=lambda x: x.start_pos)
errors = self.error_counts
unexpected = dict[</s>
===========unchanged ref 0===========
at: coeditor._utils
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor.ctx_change_encoder
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
_KnownJediErrors = {
"not enough values to unpack (expected 2",
"'Newline' object has no attribute 'children'",
"trailer_op is actually <AssertStmt:",
"There's a scope that was not managed: <Module",
"maximum recursion depth exceeded",
}
_fast_goto(script: jedi.Script, tree_name: tree.Name, *, follow_imports=False, follow_builtin_imports=False, only_stubs=False, prefer_stubs=False) -> set[classes.Name]
at: coeditor.ctx_change_encoder.PyDefinition
full_name: PyFullName
start_pos: tuple[int, int]
end_pos: tuple[int, int]
from_name(name: classes.BaseName) -> Iterable["PyDefinition"]
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script.__init__
self.path = path.absolute() if path else None
===========unchanged ref 1===========
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
at: parso.python.tree
Name()
Module(children)
at: parso.python.tree.Module
__slots__ = ('_used_names',)
type = 'file_input'
get_used_names()
at: parso.python.tree.UsedNamesMapping.__init__
self._dict = dct
at: parso.tree.Leaf.__init__
self.start_pos = start_pos
at: pathlib
Path()
at: pathlib.PurePath
__slots__ = (
'_drv', '_root', '_parts',
'_str', '_hash', '_pparts', '_cached_cparts',
)
drive = property(attrgetter('_drv'),
doc="""The drive prefix (letter or UNC path), if any.""")
root = property(attrgetter('_root'),
doc="""The root of the path, if any.""")
at: tqdm.std
tqdm(iterable=None, desc=None, total=None, leave=True, file=None, ncols=None, mininterval=0.1, maxinterval=10.0, miniters=None, ascii=None, disable=False, unit='it', unit_scale=False, dynamic_ncols=False, smoothing=0.3, bar_format=None, initial=0, position=None, postfix=None, unit_divisor=1000, write_bytes=False, lock_args=None, nrows=None, colour=None, delay=0, gui=False, **kwargs)
at: typing
Collection = _alias(collections.abc.Collection, 1)
===========unchanged ref 2===========
at: typing.Mapping
get(key: _KT) -> Optional[_VT_co]
get(key: _KT, default: Union[_VT_co, _T]) -> Union[_VT_co, _T]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def clear_errors(self):
+ return self.analyzer.error_counts.clear()
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def get_errors(self) -> dict[str, int]:
+ return self.analyzer.error_counts
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __init__(self, analyzer: "JediUsageAnalyzer | None" = None):
- def __init__(self, analysis: "JediUsageAnalyzer | None" = None):
+ if analyzer is None:
- if analysis is None:
+ analyzer = JediUsageAnalyzer()
- analysis = JediUsageAnalyzer()
+ self.analyzer = analyzer
- self.analysis = analysis
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def post_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> list[ModuleName]:
"Return the topological order among the modules."
# sort modules topologically
module_deps = dict[ModuleName, set[ModuleName]]()
for rel_path, module in modules.items():
names = {n for n in module.imported_names}
script = pstate.scripts[rel_path]
deps = module_deps.setdefault(module.mname, set())
for n in names:
+ for source in _fast_goto(
- for source in fast_goto(
script, n, follow_imports=True, follow_builtin_imports=False
):
deps.add(source.module_name)
module_order = sort_modules_by_imports(module_deps)
return module_order
|
coeditor.encoding/TkDelta.to_change_tks
|
Modified
|
temp-1
|
ba5ac8335a7bc3fa47400469334a769b1be553e4
|
Fix TkDelta for empty inputs.
|
<0>:<add> f"Input is longer than delta: {len(lines)=} > {len(self.deltas)=}"
|
# module: coeditor.encoding
@dataclass
class TkDelta:
def to_change_tks(self, input: TokenSeq) -> TokenSeq:
+ lines = split_list(input, Newline_id) if input else []
- lines = split_list(input, Newline_id)
if len(lines) > len(self.deltas):
+ print_err(f"{self.deltas=}")
- print_err(f"{self.deltas}")
+ print_err(f"{input=}")
- print_err(f"{input}")
raise ValueError(
- f"Delta is longer than input: {len(lines)=} > {len(self.deltas)=}"
<0> )
new_lines = list[TokenSeq]()
for line, delta in zip(lines, self.deltas):
deleted = False
if delta:
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
elif action[0] == Del_id:
deleted = True
if deleted:
new_lines.append([Del_id] + line)
else:
new_lines.append(line)
if len(self.deltas) == len(lines) + 1:
delta = self.deltas[-1]
for action in delta:
if action[0] == Add_id:
new_lines.append(action)
return join_list(new_lines, Newline_id)
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
split_list(lst: list[T1], sep: T1) -> list[list[T1]]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
print_err(*, sep: Optional[str]=..., end: Optional[str]=..., flush: bool=...) -> None
at: coeditor.encoding
Add_id = get_tk_id(Add)
Del_id = get_tk_id(Del)
Newline_id = get_tk_id("\n")
at: coeditor.encoding.TkDelta
deltas: list[tuple[TokenSeq, ...]]
|
spot.static_analysis/ProjectPath.pop
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> p1 = ".".join(split_dots(self.path)[:-1])
|
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
- p1 = ".".join(self.path.split(".")[:-1])
<0> return ProjectPath(self.module, p1)
|
===========unchanged ref 0===========
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
at: spot.utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
===========changed ref 0===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
|
spot.static_analysis/PythonFunction.is_test_func
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> (file := split_dots(self.path.module)[-1]).startswith("test_")
|
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
- (file := self.path.module.split(".")[-1]).startswith("test_")
<0> or file.endswith("_test")
)
|
===========unchanged ref 0===========
at: functools
cached_property(func: Callable[[Any], _T])
at: spot.static_analysis.ProjectPath
module: ModuleName
at: spot.static_analysis.PythonFunction
name: str
path: ProjectPath
parent_class: ProjectPath | None
tree: cst.FunctionDef
at: spot.utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
===========changed ref 0===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
===========changed ref 1===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
|
spot.static_analysis/to_abs_import_path
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> result_segs = list(mod_segs[:-dots])
|
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
- result_segs = mod_segs[:-dots]
<0> rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
|
===========unchanged ref 0===========
at: spot.static_analysis
ModuleName = str
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
at: spot.utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
at: typing
Generator = _alias(collections.abc.Generator, 3)
===========changed ref 0===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
===========changed ref 1===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 2===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
|
spot.static_analysis/path_up
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> segs = split_dots(path)
|
# module: spot.static_analysis
@cache
def path_up(path: str) -> str:
- segs = split_import_path(path)
<0> return ".".join(segs[:-1])
|
===========unchanged ref 0===========
at: functools
cache(user_function: Callable[..., _T], /) -> _lru_cache_wrapper[_T]
===========changed ref 0===========
# module: spot.static_analysis
- @cache
- def split_import_path(path: str):
- return path.split(".")
-
===========changed ref 1===========
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
+ result_segs = list(mod_segs[:-dots])
- result_segs = mod_segs[:-dots]
rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
===========changed ref 2===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 3===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
===========changed ref 4===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
|
spot.static_analysis/ModuleHierarchy.from_modules
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> root.add_module(split_dots(m))
|
# module: spot.static_analysis
class ModuleHierarchy:
@staticmethod
def from_modules(modules: Iterable[str]) -> "ModuleHierarchy":
root = ModuleHierarchy()
for m in modules:
- root.add_module(split_import_path(m))
<0> return root
|
===========unchanged ref 0===========
at: spot.static_analysis
ModuleName = str
at: spot.static_analysis.ModuleHierarchy.from_modules
root = ModuleHierarchy()
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
===========changed ref 0===========
# module: spot.static_analysis
- @cache
- def split_import_path(path: str):
- return path.split(".")
-
===========changed ref 1===========
# module: spot.static_analysis
@cache
def path_up(path: str) -> str:
+ segs = split_dots(path)
- segs = split_import_path(path)
return ".".join(segs[:-1])
===========changed ref 2===========
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
+ result_segs = list(mod_segs[:-dots])
- result_segs = mod_segs[:-dots]
rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
===========changed ref 3===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 4===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
===========changed ref 5===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
|
spot.static_analysis/UsageAnalysis.find_class
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> segs = split_dots(abs_p)
|
# module: spot.static_analysis
class UsageAnalysis:
def find_class(self, mname: ModuleName, qname: QualifiedName) -> PythonClass | None:
cls_path = None
match qname.source:
case QualifiedNameSource.IMPORT:
for abs_p in to_abs_import_path(
mname, qname.name, self.add_implicit_rel_imports
):
- segs = split_import_path(abs_p)
<0> if len(segs) >= 2:
cls_path = self.ns_hier.resolve_path(segs)
break
case QualifiedNameSource.LOCAL:
cls_path = ProjectPath(mname, qname.name)
if cls_path in self.path2class:
return self.path2class[cls_path]
return None
|
===========unchanged ref 0===========
at: spot.static_analysis
ModuleName = str
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: spot.static_analysis.ModuleHierarchy
resolve_path(segs: Sequence[str]) -> ProjectPath | None
at: spot.static_analysis.UsageAnalysis
all_usages: list[ProjectUsage]
path2elem: dict[ProjectPath, PythonElem]
user2used: dict[ProjectPath, list[ProjectUsage]]
used2user: dict[ProjectPath, list[ProjectUsage]]
TLogger = TimeLogger()
at: spot.static_analysis.UsageAnalysis.__init__
self.add_implicit_rel_imports = add_implicit_rel_imports
self.ns_hier = ModuleHierarchy.from_modules(project.modules.keys())
self.path2class = {
cls.path: cls
for mod in project.modules.values()
for cls in mod.all_classes()
}
at: spot.utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
===========changed ref 0===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
===========changed ref 1===========
# module: spot.static_analysis
class ModuleHierarchy:
@staticmethod
def from_modules(modules: Iterable[str]) -> "ModuleHierarchy":
root = ModuleHierarchy()
for m in modules:
+ root.add_module(split_dots(m))
- root.add_module(split_import_path(m))
return root
===========changed ref 2===========
# module: spot.static_analysis
- @cache
- def split_import_path(path: str):
- return path.split(".")
-
===========changed ref 3===========
# module: spot.static_analysis
@cache
def path_up(path: str) -> str:
+ segs = split_dots(path)
- segs = split_import_path(path)
return ".".join(segs[:-1])
===========changed ref 4===========
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
+ result_segs = list(mod_segs[:-dots])
- result_segs = mod_segs[:-dots]
rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
===========changed ref 5===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 6===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
|
spot.static_analysis/UsageAnalysis.generate_usages
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> segs = split_dots(abs_p)
|
# module: spot.static_analysis
class UsageAnalysis:
def generate_usages(
self,
mname: ModuleName,
caller: ProjectPath,
qname: QualifiedName,
parent_call: cst.Call | None,
) -> Iterable[ProjectUsage]:
<s>fixture>", m]:
yield from gen_fixture_usages(m)
break
case [*prefix, cls, _, "<locals>", "self", m]:
segs = [*prefix, cls, m]
callee = ProjectPath(mname, ".".join(segs))
if callee in self.path2class:
if self.record_type_usages:
yield usage(callee, is_certain=True)
yield from gen_constructor_usages(self.path2class[callee])
elif callee in self.path2elem:
yield usage(self.path2elem[callee].path, is_certain=True)
elif len(segs) >= 2 and segs[-2] != "<locals>":
# method fuzzy match case 3
yield from gen_class_usages(segs[-1])
segs.pop()
continue
break
match qname.source:
case QualifiedNameSource.IMPORT:
for abs_p in to_abs_import_path(
mname, qname.name, self.add_implicit_rel_imports
):
- segs = split_import_path(abs_p)
<0> callee = self.ns_hier.resolve_path(segs)
if callee is None:
continue
if callee in self.path2class:
if self.record_type_usages:
yield usage(callee, is_certain=True)
yield from gen_constructor_usages(self.path2class[callee])
break
elif callee in self.path2elem:
yield usage(self.path2elem[callee].path, is_certain=True)
break
case QualifiedNameSource.LOCAL:
yield from resolve_local_usages(qname.name)
|
===========above chunk 0===========
# module: spot.static_analysis
class UsageAnalysis:
def generate_usages(
self,
mname: ModuleName,
caller: ProjectPath,
qname: QualifiedName,
parent_call: cst.Call | None,
) -> Iterable[ProjectUsage]:
# offset: -1
<s> return
for vf in visible_fixtures(fix_name):
if vf in candidates:
yield usage(vf, is_certain=True)
return
def gen_constructor_usages(cls: PythonClass):
if not parent_call:
return
cpath = cls.path
used_elems = list[tuple[ProjectPath, bool]]()
cons_path = cpath.append("__init__")
if cons_path in self.path2elem:
used_elems.append((cons_path, True))
else:
for v in cls.attributes.values():
used_elems.append((v.path, True))
# maybe also used members from parent class
for el in self.cls2members[cpath].values():
if isinstance(el, PythonVariable):
used_elems.append((el.path, False))
for u, certain in used_elems:
yield usage(self.path2elem[u].path, is_certain=certain)
def resolve_local_usages(name: str):
segs = name.split(".")
# try resolve all usages in the access chain until a certain usage is found
# For example, if we have `a.b.c.d` and we can resolve `a.b` to a certain element,
# but not `a.b.c`, we will then also generate class usages for `.c` and `.d`.
while segs:
match segs:
case ["<attr>", m]:
# method fuzzy match case 1
yield from gen_class_usages(m)
break
case ["<fixture>", m]:
yield from gen_fixture_usages(m)
break
case [*prefix,</s>
===========above chunk 1===========
# module: spot.static_analysis
class UsageAnalysis:
def generate_usages(
self,
mname: ModuleName,
caller: ProjectPath,
qname: QualifiedName,
parent_call: cst.Call | None,
) -> Iterable[ProjectUsage]:
# offset: -2
def usage(used: ProjectPath, is_certain: bool):
return ProjectUsage(caller, used, is_certain, parent_call)
def gen_class_usages(member_name: str):
if member_name.startswith("__") and member_name.endswith("__"):
# skip common methods like __init__
return
for e in self.name2class_member.get(member_name, []):
yield usage(e.path, is_certain=False)
def visible_fixtures(fix_name: str) -> Generator[ProjectPath, None, None]:
psegs = caller.path.split(".")
# try finding in the local module
while psegs:
psegs.pop()
psegs.append(fix_name)
yield ProjectPath(caller.module, ".".join(psegs))
psegs.pop()
# try finding in conftest files
msegs = mname.split(".")
while msegs:
msegs.pop()
msegs.append("conftest")
yield ProjectPath(".".join(msegs), fix_name)
msegs.pop()
def gen_fixture_usages(fix_name: str):
candidates = self.name2fixtures.get(fix_name, None)
if not candidates:
return
for vf in visible_fixtures(fix_name):
if vf in candidates:
yield usage(vf,</s>
===========unchanged ref 0===========
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
PythonVariable(name: str, path: ProjectPath, parent_class: ProjectPath | None, tree: cst.Name | cst.Attribute, assignments: list[
cst.Assign | cst.AnnAssign
], wrapped_assignments: list[
cst.Assign | cst.AnnAssign
])
PythonClass(name: str, path: ProjectPath, attributes: dict[str, PythonVariable], methods: dict[str, PythonFunction], inner_classes: dict[str, "PythonClass"], tree: cst.ClassDef, parent_class: ProjectPath | None)
to_abs_import_path(current_mod: ModuleName, path: str, allow_implicit: bool=True) -> Generator[ModuleName, None, None]
ProjectUsage(user: ProjectPath, used: ProjectPath, is_certain: bool, callsite: cst.Call | None)
at: spot.static_analysis.ModuleHierarchy
resolve_path(segs: Sequence[str]) -> ProjectPath | None
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
append(path: ElemPath) -> "ProjectPath"
from_str(s: str) -> "ProjectPath"
at: spot.static_analysis.PythonClass
name: str
path: ProjectPath
attributes: dict[str, PythonVariable]
methods: dict[str, PythonFunction]
inner_classes: dict[str, "PythonClass"]
tree: cst.ClassDef
parent_class: ProjectPath | None
at: spot.static_analysis.PythonFunction
path: ProjectPath
at: spot.static_analysis.PythonVariable
name: str
path: ProjectPath
parent_class: ProjectPath | None
tree: cst.Name | cst.Attribute
===========unchanged ref 1===========
assignments: list[
cst.Assign | cst.AnnAssign
] # only record assignments outside of functions
wrapped_assignments: list[
cst.Assign | cst.AnnAssign
] # assignments inside functions
at: spot.static_analysis.UsageAnalysis.__init__
self.add_implicit_rel_imports = add_implicit_rel_imports
self.record_type_usages = record_type_usages
self.ns_hier = ModuleHierarchy.from_modules(project.modules.keys())
self.path2elem = {v.path: v for v in project.all_elems()}
self.path2class = {
cls.path: cls
for mod in project.modules.values()
for cls in mod.all_classes()
}
self.cls2members = cls2members = dict[ProjectPath, dict[str, PythonElem]]()
self.name2class_member = groupby(
[self.path2elem[p] for p in all_class_members], lambda e: e.name
)
self.name2fixtures = {
n: {f.path for f in fs} for n, fs in name2fixtures.items()
}
at: spot.utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Generator = _alias(collections.abc.Generator, 3)
at: typing.Mapping
get(key: _KT) -> Optional[_VT_co]
get(key: _KT, default: Union[_VT_co, _T]) -> Union[_VT_co, _T]
|
coeditor.history/_select_ast_calls
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> segs = split_dots(path.path)
|
# module: coeditor.history
def _select_ast_calls(
node: ast.AST, path: ProjectPath
) -> Generator[ast.Call, None, None]:
"""Return all call nodes with the mathcing function name in the AST."""
- segs = path.path.split(".")
<0> if segs[-1] == "__init__":
f_name = segs[-2]
else:
f_name = segs[-1]
for n in ast.walk(node):
if isinstance(n, ast.Call) and isinstance(n.func, ast.Name):
if n.func.id == f_name:
yield n
|
===========unchanged ref 0===========
at: _ast
AST(*args: Any, **kwargs: Any)
Call(*args: Any, **kwargs: Any)
Name(*args: Any, **kwargs: Any)
at: ast
walk(node: AST) -> Iterator[AST]
at: coeditor._utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
at: typing
Generator = _alias(collections.abc.Generator, 3)
===========changed ref 0===========
# module: spot.static_analysis
- @cache
- def split_import_path(path: str):
- return path.split(".")
-
===========changed ref 1===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
===========changed ref 2===========
# module: spot.static_analysis
@cache
def path_up(path: str) -> str:
+ segs = split_dots(path)
- segs = split_import_path(path)
return ".".join(segs[:-1])
===========changed ref 3===========
# module: spot.static_analysis
class ModuleHierarchy:
@staticmethod
def from_modules(modules: Iterable[str]) -> "ModuleHierarchy":
root = ModuleHierarchy()
for m in modules:
+ root.add_module(split_dots(m))
- root.add_module(split_import_path(m))
return root
===========changed ref 4===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
===========changed ref 5===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 6===========
# module: spot.static_analysis
class UsageAnalysis:
def find_class(self, mname: ModuleName, qname: QualifiedName) -> PythonClass | None:
cls_path = None
match qname.source:
case QualifiedNameSource.IMPORT:
for abs_p in to_abs_import_path(
mname, qname.name, self.add_implicit_rel_imports
):
+ segs = split_dots(abs_p)
- segs = split_import_path(abs_p)
if len(segs) >= 2:
cls_path = self.ns_hier.resolve_path(segs)
break
case QualifiedNameSource.LOCAL:
cls_path = ProjectPath(mname, qname.name)
if cls_path in self.path2class:
return self.path2class[cls_path]
return None
===========changed ref 7===========
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
+ result_segs = list(mod_segs[:-dots])
- result_segs = mod_segs[:-dots]
rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
===========changed ref 8===========
# module: spot.static_analysis
class UsageAnalysis:
def generate_usages(
self,
mname: ModuleName,
caller: ProjectPath,
qname: QualifiedName,
parent_call: cst.Call | None,
) -> Iterable[ProjectUsage]:
def usage(used: ProjectPath, is_certain: bool):
return ProjectUsage(caller, used, is_certain, parent_call)
def gen_class_usages(member_name: str):
if member_name.startswith("__") and member_name.endswith("__"):
# skip common methods like __init__
return
for e in self.name2class_member.get(member_name, []):
yield usage(e.path, is_certain=False)
def visible_fixtures(fix_name: str) -> Generator[ProjectPath, None, None]:
psegs = caller.path.split(".")
# try finding in the local module
while psegs:
psegs.pop()
psegs.append(fix_name)
yield ProjectPath(caller.module, ".".join(psegs))
psegs.pop()
# try finding in conftest files
msegs = mname.split(".")
while msegs:
msegs.pop()
msegs.append("conftest")
yield ProjectPath(".".join(msegs), fix_name)
msegs.pop()
def gen_fixture_usages(fix_name: str):
candidates = self.name2fixtures.get(fix_name, None)
if not candidates:
return
for vf in visible_fixtures(fix_name):
if vf in candidates:
yield usage(vf, is_certain=True)
return
def gen_constructor_usages(cls: PythonClass):
if not parent_call:
return
cpath = cls.path
used_elems = list[tuple[ProjectPath, bool]]()
cons_path = cpath.</s>
|
coeditor.code_change/ChangeScope._search_scope
|
Modified
|
temp-1
|
2282c6c0bcb9fea8d0b3c73335181e13605c6c00
|
Optimize `split_dots`.
|
<0>:<add> segs = split_dots(path)
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
def _search_scope(self, path: ElemPath) -> Self:
"""Find the scope that can potentially contain the given path. Follow the
path segments until no more subscopes are found."""
- segs = path.split(".")
<0> scope = self
for s in segs:
if s in scope.subscopes:
scope = scope.subscopes[s]
else:
break
return scope
|
===========unchanged ref 0===========
at: coeditor._utils
_lru_cache_wrapper(*args: Hashable, **kwargs: Hashable) -> _T
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
at: spot.static_analysis
ElemPath = str
===========changed ref 0===========
# module: spot.static_analysis
- @cache
- def split_import_path(path: str):
- return path.split(".")
-
===========changed ref 1===========
# module: spot.utils
+ @cache
+ def split_dots(path: str) -> tuple[str, ...]:
+ return tuple(path.split("."))
+
===========changed ref 2===========
# module: spot.static_analysis
@cache
def path_up(path: str) -> str:
+ segs = split_dots(path)
- segs = split_import_path(path)
return ".".join(segs[:-1])
===========changed ref 3===========
# module: spot.static_analysis
class ModuleHierarchy:
@staticmethod
def from_modules(modules: Iterable[str]) -> "ModuleHierarchy":
root = ModuleHierarchy()
for m in modules:
+ root.add_module(split_dots(m))
- root.add_module(split_import_path(m))
return root
===========changed ref 4===========
# module: spot.static_analysis
class ProjectPath(NamedTuple):
def pop(self) -> "ProjectPath":
+ p1 = ".".join(split_dots(self.path)[:-1])
- p1 = ".".join(self.path.split(".")[:-1])
return ProjectPath(self.module, p1)
===========changed ref 5===========
# module: coeditor.history
def _select_ast_calls(
node: ast.AST, path: ProjectPath
) -> Generator[ast.Call, None, None]:
"""Return all call nodes with the mathcing function name in the AST."""
+ segs = split_dots(path.path)
- segs = path.path.split(".")
if segs[-1] == "__init__":
f_name = segs[-2]
else:
f_name = segs[-1]
for n in ast.walk(node):
if isinstance(n, ast.Call) and isinstance(n.func, ast.Name):
if n.func.id == f_name:
yield n
===========changed ref 6===========
# module: spot.static_analysis
@dataclass
class PythonFunction:
@cached_property
def is_test_func(self) -> bool:
# follow the pytest rules (but ignore the method requirements):
# https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#conventions-for-python-test-discovery
return self.name.startswith("test") and (
+ (file := split_dots(self.path.module)[-1]).startswith("test_")
- (file := self.path.module.split(".")[-1]).startswith("test_")
or file.endswith("_test")
)
===========changed ref 7===========
# module: spot.static_analysis
class UsageAnalysis:
def find_class(self, mname: ModuleName, qname: QualifiedName) -> PythonClass | None:
cls_path = None
match qname.source:
case QualifiedNameSource.IMPORT:
for abs_p in to_abs_import_path(
mname, qname.name, self.add_implicit_rel_imports
):
+ segs = split_dots(abs_p)
- segs = split_import_path(abs_p)
if len(segs) >= 2:
cls_path = self.ns_hier.resolve_path(segs)
break
case QualifiedNameSource.LOCAL:
cls_path = ProjectPath(mname, qname.name)
if cls_path in self.path2class:
return self.path2class[cls_path]
return None
===========changed ref 8===========
# module: spot.static_analysis
def to_abs_import_path(
current_mod: ModuleName,
path: str,
allow_implicit: bool = True,
) -> Generator[ModuleName, None, None]:
"""Given the current module and an import path, return the list of modules
(in absolute path) that import could potentially be referring to."""
dots = 0
while dots < len(path) and path[dots] == ".":
dots += 1
if dots == 0:
yield path
if allow_implicit:
yield path_join(path_up(current_mod), path)
return
+ mod_segs = split_dots(current_mod)
- mod_segs = split_import_path(current_mod)
assert len(mod_segs) >= dots, "Cannot go up more levels."
+ result_segs = list(mod_segs[:-dots])
- result_segs = mod_segs[:-dots]
rest = path[dots:]
if rest:
result_segs.append(rest)
yield ".".join(result_segs)
===========changed ref 9===========
# module: spot.static_analysis
class UsageAnalysis:
def generate_usages(
self,
mname: ModuleName,
caller: ProjectPath,
qname: QualifiedName,
parent_call: cst.Call | None,
) -> Iterable[ProjectUsage]:
def usage(used: ProjectPath, is_certain: bool):
return ProjectUsage(caller, used, is_certain, parent_call)
def gen_class_usages(member_name: str):
if member_name.startswith("__") and member_name.endswith("__"):
# skip common methods like __init__
return
for e in self.name2class_member.get(member_name, []):
yield usage(e.path, is_certain=False)
def visible_fixtures(fix_name: str) -> Generator[ProjectPath, None, None]:
psegs = caller.path.split(".")
# try finding in the local module
while psegs:
psegs.pop()
psegs.append(fix_name)
yield ProjectPath(caller.module, ".".join(psegs))
psegs.pop()
# try finding in conftest files
msegs = mname.split(".")
while msegs:
msegs.pop()
msegs.append("conftest")
yield ProjectPath(".".join(msegs), fix_name)
msegs.pop()
def gen_fixture_usages(fix_name: str):
candidates = self.name2fixtures.get(fix_name, None)
if not candidates:
return
for vf in visible_fixtures(fix_name):
if vf in candidates:
yield usage(vf, is_certain=True)
return
def gen_constructor_usages(cls: PythonClass):
if not parent_call:
return
cpath = cls.path
used_elems = list[tuple[ProjectPath, bool]]()
cons_path = cpath.</s>
|
spot.utils/get_pushover_config
|
Modified
|
temp-1
|
6d7e1f96fce615c099ceff3e3fc430818d70a407
|
Switch to frozen classes.
|
<0>:<add> warnings.warn(
|
# module: spot.utils
def get_pushover_config() -> dict[str, str] | None:
config_file = proj_root() / "config/pushover.json"
if config_file.exists():
match json.loads(config_file.read_text()):
case {"user": user, "token": token}:
return {"user": user, "token": token}
if not _pushover_warned[0]:
- logging.warning(
<0> f"No pushover config file found at {config_file}. Not able to push message."
)
_pushover_warned[0] = True
return None
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: json
loads(s: Union[str, bytes], *, cls: Optional[Type[JSONDecoder]]=..., object_hook: Optional[Callable[[Dict[Any, Any]], Any]]=..., parse_float: Optional[Callable[[str], Any]]=..., parse_int: Optional[Callable[[str], Any]]=..., parse_constant: Optional[Callable[[str], Any]]=..., object_pairs_hook: Optional[Callable[[List[Tuple[Any, Any]]], Any]]=..., **kwds: Any) -> Any
at: pathlib.Path
__slots__ = ()
read_text(encoding: Optional[str]=..., errors: Optional[str]=...) -> str
exists() -> bool
at: spot.utils
proj_root() -> Path
_pushover_warned = [False]
|
spot.utils/PickleCache.remove
|
Modified
|
temp-1
|
6d7e1f96fce615c099ceff3e3fc430818d70a407
|
Switch to frozen classes.
|
<0>:<add> warnings.warn(f"[PickleCache] File not found: '{path}'")
|
# module: spot.utils
class PickleCache:
def remove(self, rel_path: Path | str):
path = self.cache_dir / rel_path
if path.exists():
path.unlink()
else:
- logging.warning(f"[PickleCache] File not found: '{path}'")
<0>
|
===========unchanged ref 0===========
at: pathlib
Path()
at: pathlib.Path
unlink(missing_ok: bool=...) -> None
exists() -> bool
at: spot.utils.PickleCache.__init__
self.cache_dir = cache_dir
===========changed ref 0===========
# module: spot.utils
def get_pushover_config() -> dict[str, str] | None:
config_file = proj_root() / "config/pushover.json"
if config_file.exists():
match json.loads(config_file.read_text()):
case {"user": user, "token": token}:
return {"user": user, "token": token}
if not _pushover_warned[0]:
+ warnings.warn(
- logging.warning(
f"No pushover config file found at {config_file}. Not able to push message."
)
_pushover_warned[0] = True
return None
|
spot.utils/PickleCache.clear
|
Modified
|
temp-1
|
6d7e1f96fce615c099ceff3e3fc430818d70a407
|
Switch to frozen classes.
|
<0>:<add> warnings.warn(f"No cache found at: {self.cache_dir}, skip clearing.")
|
# module: spot.utils
class PickleCache:
def clear(self):
if self.cache_dir.exists():
logging.info(f"Clearing cache: at: {self.cache_dir}")
shutil.rmtree(self.cache_dir)
else:
- logging.warning(f"No cache found at: {self.cache_dir}, skip clearing.")
<0>
|
===========unchanged ref 0===========
at: logging
info(msg: Any, *args: Any, exc_info: _ExcInfoType=..., stack_info: bool=..., extra: Optional[Dict[str, Any]]=..., **kwargs: Any) -> None
at: pathlib.Path
exists() -> bool
at: shutil
rmtree(path: Union[bytes, StrPath], ignore_errors: bool=..., onerror: Optional[Callable[[Any, Any, Any], Any]]=...) -> None
at: spot.utils.PickleCache.__init__
self.cache_dir = cache_dir
===========changed ref 0===========
# module: spot.utils
class PickleCache:
def remove(self, rel_path: Path | str):
path = self.cache_dir / rel_path
if path.exists():
path.unlink()
else:
+ warnings.warn(f"[PickleCache] File not found: '{path}'")
- logging.warning(f"[PickleCache] File not found: '{path}'")
===========changed ref 1===========
# module: spot.utils
def get_pushover_config() -> dict[str, str] | None:
config_file = proj_root() / "config/pushover.json"
if config_file.exists():
match json.loads(config_file.read_text()):
case {"user": user, "token": token}:
return {"user": user, "token": token}
if not _pushover_warned[0]:
+ warnings.warn(
- logging.warning(
f"No pushover config file found at {config_file}. Not able to push message."
)
_pushover_warned[0] = True
return None
|
coeditor.retrieval_model/RetrievalEditorModel._reorder_cache
|
Modified
|
temp-1
|
6d7e1f96fce615c099ceff3e3fc430818d70a407
|
Switch to frozen classes.
|
<0>:<add> warnings.warn(
|
# module: coeditor.retrieval_model
class RetrievalEditorModel(T5PreTrainedModel):
def _reorder_cache(self, past, beam_idx):
if past is None:
- logging.warning(
<0> "You might want to consider setting `use_cache=True` to speed up decoding"
)
return past
reordered_decoder_past = ()
for layer_past_states in past:
# get the correct batch idx from layer past batch dim
# batch dim of `past` is at 2nd position
reordered_layer_past_states = ()
for layer_past_state in layer_past_states:
# need to set correct `past` for each of the four key / value states
reordered_layer_past_states = reordered_layer_past_states + (
layer_past_state.index_select(
0, beam_idx.to(layer_past_state.device)
),
)
# assert reordered_layer_past_states[0].shape == layer_past_states[0].shape
assert len(reordered_layer_past_states) == len(layer_past_states)
reordered_decoder_past = reordered_decoder_past + (
reordered_layer_past_states,
)
return reordered_decoder_past
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: transformers.generation.utils.GenerationMixin
_reorder_cache(self, past_key_values, beam_idx)
===========changed ref 0===========
# module: coeditor.encoding
+ @dataclass(frozen=True)
- @dataclass
class TkDelta:
"""The Tokenized version of :class:`StrDelta`."""
+ deltas: Sequence[tuple[TokenSeq, ...]]
- deltas: list[tuple[TokenSeq, ...]]
===========changed ref 1===========
# module: coeditor.encoding
+ @dataclass(frozen=True)
- @dataclass
class StrDelta:
"""Stores the line deltas for each line. A line delta is a list of added lines
(starting with a '+') followed by optionally a `-` line
(for deleting the current line)."""
+ deltas: Sequence[tuple[str, ...]]
- deltas: list[tuple[str, ...]]
===========changed ref 2===========
# module: spot.utils
class PickleCache:
def remove(self, rel_path: Path | str):
path = self.cache_dir / rel_path
if path.exists():
path.unlink()
else:
+ warnings.warn(f"[PickleCache] File not found: '{path}'")
- logging.warning(f"[PickleCache] File not found: '{path}'")
===========changed ref 3===========
# module: spot.utils
class PickleCache:
def clear(self):
if self.cache_dir.exists():
logging.info(f"Clearing cache: at: {self.cache_dir}")
shutil.rmtree(self.cache_dir)
else:
+ warnings.warn(f"No cache found at: {self.cache_dir}, skip clearing.")
- logging.warning(f"No cache found at: {self.cache_dir}, skip clearing.")
===========changed ref 4===========
# module: spot.utils
def get_pushover_config() -> dict[str, str] | None:
config_file = proj_root() / "config/pushover.json"
if config_file.exists():
match json.loads(config_file.read_text()):
case {"user": user, "token": token}:
return {"user": user, "token": token}
if not _pushover_warned[0]:
+ warnings.warn(
- logging.warning(
f"No pushover config file found at {config_file}. Not able to push message."
)
_pushover_warned[0] = True
return None
|
coeditor.common/normalize_code_by_ast
|
Modified
|
temp-1
|
c7532373c41c6d26f05afc40e4d1377108596556
|
Remove module docstrings when normalizing code.
|
<0>:<add> node = as_any(self.generic_visit(node))
|
# module: coeditor.common
def normalize_code_by_ast(
code: str, sort_keyargs: bool = True, remove_doc_string: bool = True
) -> str:
"""Normalize the code by parsing and unparsing it using the AST module.
If parsing fails, return the original code."""
class KeyargSorter(ast.NodeTransformer):
def visit_Call(self, node: ast.Call):
if node.keywords:
node.keywords.sort(key=lambda x: x.arg or "None")
return node
class DocStringremover(ast.NodeTransformer):
def visit_FunctionDef(self, node: ast.FunctionDef):
+ return self._visit_def(node)
+
+ def visit_Module(self, node: ast.Module) -> Any:
+ return self._visit_def(node)
- match node.body:
- case [ast.Expr(value=ast.Constant(value=str())), *body]:
- node.body = body
- return node
def visit_ClassDef(self, node: ast.ClassDef):
+ return self._visit_def(node)
+
+ def _visit_def(self, node):
- node = cast(ast.ClassDef, self.generic_visit(node))
<0> match node.body:
case [ast.Expr(value=ast.Constant(value=str())), *body]:
node.body = body
return node
try:
tree = ast.parse(dedent(code))
if remove_doc_string:
tree = DocStringremover().visit(tree)
if sort_keyargs:
tree = KeyargSorter().visit(tree)
return ast.unparse(tree)
except SyntaxError:
return code
|
===========unchanged ref 0===========
at: _ast
Module(*args: Any, **kwargs: Any)
FunctionDef(*args: Any, **kwargs: Any)
ClassDef(*args: Any, **kwargs: Any)
Expr(*args: Any, **kwargs: Any)
Call(*args: Any, **kwargs: Any)
Constant(*args: Any, **kwargs: Any)
at: _ast.Call
func: expr
args: typing.List[expr]
keywords: typing.List[keyword]
at: ast
parse(source: Union[str, bytes], filename: Union[str, bytes]=..., mode: Literal["exec"]=..., *, type_comments: bool=..., feature_version: Union[None, int, _typing.Tuple[int, int]]=...) -> Module
parse(source: Union[str, bytes], filename: Union[str, bytes]=..., mode: str=..., *, type_comments: bool=..., feature_version: Union[None, int, _typing.Tuple[int, int]]=...) -> AST
NodeTransformer()
at: ast.NodeTransformer
generic_visit(node: AST) -> AST
at: ast.NodeVisitor
visit(node: AST) -> Any
visit_Module(self, node: Module) -> Any
visit_FunctionDef(self, node: FunctionDef) -> Any
visit_ClassDef(self, node: ClassDef) -> Any
visit_Call(self, node: Call) -> Any
at: spot.utils
as_any(x) -> Any
at: textwrap
dedent(text: str) -> str
|
coeditor.code_change/_deep_copy_subset_
|
Modified
|
temp-1
|
68df1a188774d0da509adfb2b256212eec6e146e
|
Fix dataset creation.
|
<0>:<add> keys = {k for k in keys if k in dict}
|
# module: coeditor.code_change
def _deep_copy_subset_(dict: dict[T1, T2], keys: Collection[T1]) -> dict[T1, T2]:
"This is more efficient than deepcopying each value individually if they share common data."
<0> to_copy = {k: dict[k] for k in keys}
copies = copy.deepcopy(to_copy)
for k in keys:
dict[k] = copies[k]
return dict
|
===========unchanged ref 0===========
at: coeditor.common
T1 = TypeVar("T1")
T2 = TypeVar("T2")
at: copy
deepcopy(x: _T, memo: Optional[Dict[int, Any]]=..., _nil: Any=...) -> _T
at: typing
Collection = _alias(collections.abc.Collection, 1)
===========changed ref 0===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 1===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 2===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 3===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 4===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- VERSION = "1.0"
-
===========changed ref 5===========
# module: coeditor.dataset
@dataclass
class _ProcessingResult:
edits: Sequence[TkC3Problem]
+ stats: dict[str, dict | Any]
- processor_errors: dict[str, int]
===========changed ref 6===========
# module: spot.utils
@dataclass
class TimeLogger:
+ @staticmethod
+ def times_to_dataframe(times: dict[str, list[float]]):
+ names = list(times.keys())
+ total_times = [sum(ts) for ts in times.values()]
+ counts = [len(ts) for ts in times.values()]
+ avg_times = [sum(ts) / len(ts) for ts in times.values()]
+
+ df = pd.DataFrame(
+ {
+ "name": names,
+ "count": counts,
+ "avg_time": avg_times,
+ "total_time": total_times,
+ }
+ )
+ df.sort_values(by="total_time", ascending=False, inplace=True)
+ return df
+
===========changed ref 7===========
# module: spot.utils
@dataclass
class TimeLogger:
def as_dataframe(self):
- names = list(self.times.keys())
- total_times = [sum(ts) for ts in self.times.values()]
- counts = [len(ts) for ts in self.times.values()]
- avg_times = [sum(ts) / len(ts) for ts in self.times.values()]
+ return self.times_to_dataframe(self.times)
- df = pd.DataFrame(
- {
- "name": names,
- "count": counts,
- "avg_time": avg_times,
- "total_time": total_times,
- }
- )
- df.sort_values(by="total_time", ascending=False, inplace=True)
- return df
-
===========changed ref 8===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
+ coeditor.code_change._tlogger.clear()
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
edits = []
+ stats = dict()
+ encoder.change_processor.append_stats(stats)
+ rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
+ return _ProcessingResult(edits, stats)
- return _ProcessingResult(
- edits,
- encoder.change_processor.get_errors(),
- )
===========changed ref 9===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max</s>
|
coeditor.ctx_change_encoder/C3ProblemGenerator.post_edit_analysis
|
Modified
|
temp-1
|
68df1a188774d0da509adfb2b256212eec6e146e
|
Fix dataset creation.
|
<0>:<add> for source in srcs:
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def post_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> list[ModuleName]:
"Return the topological order among the modules."
# sort modules topologically
module_deps = dict[ModuleName, set[ModuleName]]()
for rel_path, module in modules.items():
names = {n for n in module.imported_names}
script = pstate.scripts[rel_path]
deps = module_deps.setdefault(module.mname, set())
for n in names:
+ try:
+ srcs = _fast_goto(
- for source in _fast_goto(
+ script, n, follow_imports=True, follow_builtin_imports=False
- script, n, follow_imports=True, follow_builtin_imports=False
+ )
- ):
+ except Exception as e:
+ if "There's a scope that was not managed:" in str(e):
+ self.analyzer.add_error(str(e))
+ continue
+ else:
+ raise
<0> deps.add(source.module_name)
module_order = sort_modules_by_imports(module_deps)
return module_order
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
_fast_goto(script: jedi.Script, tree_name: tree.Name, *, follow_imports=False, follow_builtin_imports=False, only_stubs=False, prefer_stubs=False) -> set[classes.Name]
at: coeditor.ctx_change_encoder.C3ProblemGenerator.__init__
self.analyzer = analyzer
at: coeditor.ctx_change_encoder.JediUsageAnalyzer
_KnownJediErrors = {
"not enough values to unpack (expected 2",
"'Newline' object has no attribute 'children'",
"trailer_op is actually ",
"There's a scope that was not managed: <Module",
"maximum recursion depth exceeded",
"'NoneType' object has no attribute 'type'",
}
add_error(err_text: str)
add_error(self, err_text: str)
at: spot.static_analysis
ModuleName = str
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
at: typing.Mapping
items() -> AbstractSet[Tuple[_KT, _VT_co]]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
- def clear_errors(self):
- return self.analyzer.error_counts.clear()
-
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
- def get_errors(self) -> dict[str, int]:
- return self.analyzer.error_counts
-
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def clear_stats(self) -> None:
+ return self.analyzer.error_counts.clear()
+
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __repr__(self) -> str:
+ return repr_modified_args(self)
+
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ VERSION = "1.1"
- VERSION = "1.0"
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def append_stats(self, stats: dict[str, Any]) -> None:
+ rec_add_dict_to(stats, {"analyzer_errors": self.analyzer.error_counts})
+
===========changed ref 6===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 7===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 8===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 9===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 10===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- VERSION = "1.0"
-
===========changed ref 11===========
# module: coeditor.dataset
@dataclass
class _ProcessingResult:
edits: Sequence[TkC3Problem]
+ stats: dict[str, dict | Any]
- processor_errors: dict[str, int]
===========changed ref 12===========
# module: coeditor.code_change
def _deep_copy_subset_(dict: dict[T1, T2], keys: Collection[T1]) -> dict[T1, T2]:
"This is more efficient than deepcopying each value individually if they share common data."
+ keys = {k for k in keys if k in dict}
to_copy = {k: dict[k] for k in keys}
copies = copy.deepcopy(to_copy)
for k in keys:
dict[k] = copies[k]
return dict
===========changed ref 13===========
# module: spot.utils
@dataclass
class TimeLogger:
+ @staticmethod
+ def times_to_dataframe(times: dict[str, list[float]]):
+ names = list(times.keys())
+ total_times = [sum(ts) for ts in times.values()]
+ counts = [len(ts) for ts in times.values()]
+ avg_times = [sum(ts) / len(ts) for ts in times.values()]
+
+ df = pd.DataFrame(
+ {
+ "name": names,
+ "count": counts,
+ "avg_time": avg_times,
+ "total_time": total_times,
+ }
+ )
+ df.sort_values(by="total_time", ascending=False, inplace=True)
+ return df
+
===========changed ref 14===========
# module: spot.utils
@dataclass
class TimeLogger:
def as_dataframe(self):
- names = list(self.times.keys())
- total_times = [sum(ts) for ts in self.times.values()]
- counts = [len(ts) for ts in self.times.values()]
- avg_times = [sum(ts) / len(ts) for ts in self.times.values()]
+ return self.times_to_dataframe(self.times)
- df = pd.DataFrame(
- {
- "name": names,
- "count": counts,
- "avg_time": avg_times,
- "total_time": total_times,
- }
- )
- df.sort_values(by="total_time", ascending=False, inplace=True)
- return df
-
===========changed ref 15===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
+ coeditor.code_change._tlogger.clear()
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
)
except UnicodeDecodeError as e:
# this might happen in rare cases
warnings.warn(f"Unable to process project: {root}\nError: {e}")
edits = []
+ stats = dict()
+ encoder.change_processor.append_stats(stats)
+ rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
+ return _ProcessingResult(edits, stats)
- return _ProcessingResult(
- edits,
- encoder.change_processor.get_errors(),
- )
|
coeditor.ctx_change_encoder/JediUsageAnalyzer.get_line_usages
|
Modified
|
temp-1
|
68df1a188774d0da509adfb2b256212eec6e146e
|
Fix dataset creation.
|
<0>:<add> self.add_error(err_text)
|
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
<s> unexpected = dict[str, int]()
for name in tqdm(all_names, f"Analyzing {script.path}", disable=silent):
name: tree.Name
line = name.start_pos[0]
if line not in lines_to_analyze:
continue
usages = line2usages.setdefault(line, set())
try:
defs = _fast_goto(
script,
name,
follow_imports=True,
follow_builtin_imports=False,
)
for d in defs:
usages.update(PyDefinition.from_name(d))
- except KeyError:
- # for debugging
- raise
except Exception as e:
err_text = repr(e)
- is_known = any(err in err_text for err in _KnownJediErrors)
- errors[err_text] = errors.get(err_text, 0) + 1
- if not is_known:
- unexpected[err_text] = unexpected.get(err_text, 0) + 1
- if unexpected:
- project_name = proj_root.name
- if script.path:
- file_path = script.path.relative_to(proj_root)
- else:
- file_path = "<unknown>"
- for err, count in unexpected.items():
- logging.warn(
- f"Unexpected error when analyzing '{project_name}/{file_path}' ({count=}): {err}"
- )
+ str_limit = 40
+ if len(err_text) > str_limit:
+ err_text = err_text[:str_limit] + "..."
<0> return LineUsageAnalysis(line2usages)
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class JediUsageAnalyzer:
def get_line_usages(
self,
script: jedi.Script,
proj_root: Path,
lines_to_analyze: Collection[int],
silent: bool = False,
):
# offset: -1
jmod: tree.Module = script._module_node
line2usages = dict[int, set[PyDefinition]]()
all_names = [
name for k, names in jmod.get_used_names()._dict.items() for name in names
]
all_names.sort(key=lambda x: x.start_pos)
- errors = self.error_counts
- unexpected = dict[str, int]()
for name in tqdm(all_names, f"Analyzing {script.path}"</s>
===========unchanged ref 0===========
at: coeditor._utils
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor.ctx_change_encoder
PyDefinition(full_name: PyFullName, start_pos: tuple[int, int], end_pos: tuple[int, int])
LineUsageAnalysis(line2usages: Mapping[int, set[PyDefinition]])
_fast_goto(script: jedi.Script, tree_name: tree.Name, *, follow_imports=False, follow_builtin_imports=False, only_stubs=False, prefer_stubs=False) -> set[classes.Name]
at: coeditor.ctx_change_encoder.PyDefinition
from_name(name: classes.BaseName) -> Iterable["PyDefinition"]
at: jedi.api
Script(code=None, *, path=None, environment=None, project=None)
at: jedi.api.Script.__init__
self.path = path.absolute() if path else None
self._module_node, code = self._inference_state.parse_and_get_code(
code=code,
path=self.path,
use_latest_grammar=path and path.suffix == '.pyi',
cache=False, # No disk cache, because the current script often changes.
diff_cache=settings.fast_parser,
cache_path=settings.cache_directory,
)
at: parso.python.tree
Name()
Module(children)
at: parso.python.tree.Module
__slots__ = ('_used_names',)
type = 'file_input'
get_used_names()
at: parso.python.tree.UsedNamesMapping.__init__
self._dict = dct
at: parso.tree.Leaf.__init__
self.start_pos = start_pos
at: pathlib
Path()
===========unchanged ref 1===========
at: tqdm.std
tqdm(iterable=None, desc=None, total=None, leave=True, file=None, ncols=None, mininterval=0.1, maxinterval=10.0, miniters=None, ascii=None, disable=False, unit='it', unit_scale=False, dynamic_ncols=False, smoothing=0.3, bar_format=None, initial=0, position=None, postfix=None, unit_divisor=1000, write_bytes=False, lock_args=None, nrows=None, colour=None, delay=0, gui=False, **kwargs)
at: typing
Collection = _alias(collections.abc.Collection, 1)
at: typing.Mapping
get(key: _KT) -> Optional[_VT_co]
get(key: _KT, default: Union[_VT_co, _T]) -> Union[_VT_co, _T]
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
_ObjId = NewType("_ObjId", int)
-
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
- def clear_errors(self):
- return self.analyzer.error_counts.clear()
-
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
- def get_errors(self) -> dict[str, int]:
- return self.analyzer.error_counts
-
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def clear_stats(self) -> None:
+ return self.analyzer.error_counts.clear()
+
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def __repr__(self) -> str:
+ return repr_modified_args(self)
+
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ VERSION = "1.1"
- VERSION = "1.0"
===========changed ref 6===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ def append_stats(self, stats: dict[str, Any]) -> None:
+ rec_add_dict_to(stats, {"analyzer_errors": self.analyzer.error_counts})
+
===========changed ref 7===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def post_edit_analysis(
self,
pstate: ProjectState,
modules: Mapping[RelPath, JModule],
changes: Mapping[ModuleName, JModuleChange],
) -> list[ModuleName]:
"Return the topological order among the modules."
# sort modules topologically
module_deps = dict[ModuleName, set[ModuleName]]()
for rel_path, module in modules.items():
names = {n for n in module.imported_names}
script = pstate.scripts[rel_path]
deps = module_deps.setdefault(module.mname, set())
for n in names:
+ try:
+ srcs = _fast_goto(
- for source in _fast_goto(
+ script, n, follow_imports=True, follow_builtin_imports=False
- script, n, follow_imports=True, follow_builtin_imports=False
+ )
- ):
+ except Exception as e:
+ if "There's a scope that was not managed:" in str(e):
+ self.analyzer.add_error(str(e))
+ continue
+ else:
+ raise
+ for source in srcs:
deps.add(source.module_name)
module_order = sort_modules_by_imports(module_deps)
return module_order
===========changed ref 8===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 9===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def clear_errors(self):
- return None
-
===========changed ref 10===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 11===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- def get_errors(self) -> dict[str, int]:
- return dict()
-
===========changed ref 12===========
# module: coeditor.code_change
@dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
- VERSION = "1.0"
-
===========changed ref 13===========
# module: coeditor.dataset
@dataclass
class _ProcessingResult:
edits: Sequence[TkC3Problem]
+ stats: dict[str, dict | Any]
- processor_errors: dict[str, int]
|
spot.utils/compute_line_diffs
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> return compute_line_diffs_fast(before, after)
|
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
- compute_line_diffs_fast(before, after)
<0> differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
|
===========unchanged ref 0===========
at: difflib
Differ(linejunk: Optional[_JunkCallback]=..., charjunk: Optional[_JunkCallback]=...)
at: difflib.Differ
compare(a: Sequence[_StrType], b: Sequence[_StrType]) -> Iterator[_StrType]
at: spot.utils
compute_line_diffs_fast(before: Sequence[str], after: Sequence[str])
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
|
coeditor.common/normalize_code_by_ast
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> except (SyntaxError, ValueError):
|
# module: coeditor.common
def normalize_code_by_ast(
code: str, sort_keyargs: bool = True, remove_doc_string: bool = True
) -> str:
"""Normalize the code by parsing and unparsing it using the AST module.
If parsing fails, return the original code."""
class KeyargSorter(ast.NodeTransformer):
def visit_Call(self, node: ast.Call):
if node.keywords:
node.keywords.sort(key=lambda x: x.arg or "None")
return node
class DocStringremover(ast.NodeTransformer):
def visit_FunctionDef(self, node: ast.FunctionDef):
return self._visit_def(node)
def visit_Module(self, node: ast.Module) -> Any:
return self._visit_def(node)
def visit_ClassDef(self, node: ast.ClassDef):
return self._visit_def(node)
def _visit_def(self, node):
node = as_any(self.generic_visit(node))
match node.body:
case [ast.Expr(value=ast.Constant(value=str())), *body]:
node.body = body
return node
try:
tree = ast.parse(dedent(code))
if remove_doc_string:
tree = DocStringremover().visit(tree)
if sort_keyargs:
tree = KeyargSorter().visit(tree)
return ast.unparse(tree)
- except SyntaxError:
<0> return code
|
===========unchanged ref 0===========
at: _ast
Module(*args: Any, **kwargs: Any)
FunctionDef(*args: Any, **kwargs: Any)
ClassDef(*args: Any, **kwargs: Any)
Expr(*args: Any, **kwargs: Any)
Call(*args: Any, **kwargs: Any)
Constant(*args: Any, **kwargs: Any)
at: _ast.Call
func: expr
args: typing.List[expr]
keywords: typing.List[keyword]
at: ast
parse(source: Union[str, bytes], filename: Union[str, bytes]=..., mode: Literal["exec"]=..., *, type_comments: bool=..., feature_version: Union[None, int, _typing.Tuple[int, int]]=...) -> Module
parse(source: Union[str, bytes], filename: Union[str, bytes]=..., mode: str=..., *, type_comments: bool=..., feature_version: Union[None, int, _typing.Tuple[int, int]]=...) -> AST
NodeTransformer()
unparse(ast_obj: AST) -> str
at: ast.NodeTransformer
generic_visit(node: AST) -> AST
at: ast.NodeVisitor
visit(node: AST) -> Any
visit_Module(self, node: Module) -> Any
visit_FunctionDef(self, node: FunctionDef) -> Any
visit_ClassDef(self, node: ClassDef) -> Any
visit_Call(self, node: Call) -> Any
at: spot.utils
as_any(x) -> Any
at: textwrap
dedent(text: str) -> str
===========changed ref 0===========
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
+ return compute_line_diffs_fast(before, after)
- compute_line_diffs_fast(before, after)
differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
|
coeditor.dataset/_process_commits
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> warnings.warn(f"Failed to process project: {root}\nError: {e}")
|
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
+ is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ encoder.change_processor.clear_stats()
+ encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
+ time_limit=time_limit_per_commit * (len(commits) + 10),
)
+ except Exception as e:
+ if isinstance(e, KeyboardInterrupt):
+ raise
- except UnicodeDecodeError as e:
- # this might happen in rare cases
- warnings.warn(f"Unable to process project: {root}\nError: {e}")
<0> edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
|
===========unchanged ref 0===========
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
_ProcessingResult(edits: Sequence[TkC3Problem], stats: dict[str, dict | Any])
time_limit_per_commit = 5.0
at: coeditor.dataset.C3EditEncoder
change_processor: ProjectChangeProcessor[C3Problem] = field(
default_factory=C3ProblemGenerator
)
edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
at: pathlib
Path()
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
+ return compute_line_diffs_fast(before, after)
- compute_line_diffs_fast(before, after)
differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
===========changed ref 1===========
# module: coeditor.common
def normalize_code_by_ast(
code: str, sort_keyargs: bool = True, remove_doc_string: bool = True
) -> str:
"""Normalize the code by parsing and unparsing it using the AST module.
If parsing fails, return the original code."""
class KeyargSorter(ast.NodeTransformer):
def visit_Call(self, node: ast.Call):
if node.keywords:
node.keywords.sort(key=lambda x: x.arg or "None")
return node
class DocStringremover(ast.NodeTransformer):
def visit_FunctionDef(self, node: ast.FunctionDef):
return self._visit_def(node)
def visit_Module(self, node: ast.Module) -> Any:
return self._visit_def(node)
def visit_ClassDef(self, node: ast.ClassDef):
return self._visit_def(node)
def _visit_def(self, node):
node = as_any(self.generic_visit(node))
match node.body:
case [ast.Expr(value=ast.Constant(value=str())), *body]:
node.body = body
return node
try:
tree = ast.parse(dedent(code))
if remove_doc_string:
tree = DocStringremover().visit(tree)
if sort_keyargs:
tree = KeyargSorter().visit(tree)
return ast.unparse(tree)
+ except (SyntaxError, ValueError):
- except SyntaxError:
return code
|
coeditor.dataset/dataset_from_projects
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> for k in sorted(errors.keys(), key=lambda k: errors[k], reverse=True):
|
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
<s> presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
+ chunk_training,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits = dict[Path, list[TkC3Problem]]()
try:
stats = dict[str, Any]()
for root, pr in zip(roots, presults):
project2edits.setdefault(root, []).extend(pr.edits)
rec_add_dict_to(stats, pr.stats)
if "tlogger" in stats:
df = TimeLogger.times_to_dataframe(stats.pop("tlogger"))
print("Time stats:")
display(df)
if "analyzer_errors" in list(stats.keys()):
errors: dict = stats.pop("analyzer_errors")
for k in list(errors.keys()):
if JediUsageAnalyzer.is_known_error(k):
errors.pop(k)
if errors:
print("Analyzer errors:")
- for k in sorted(errors.keys(), reverse=True):
<0> print(f"{k}:\t{errors[k]}")
if stats:
print("Other Stats:")
pretty_print_dict(stats)
except Exception as e:
if not isinstance(e, KeyboardInterrupt):
print("Error while printing stats:", e)
return TokenizedEditDataset(project2edits)
|
===========above chunk 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: -1
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 4))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,</s>
===========unchanged ref 0===========
at: IPython.core.display_functions
display(*, include=None, exclude=None, metadata=None, transient=None, display_id=None, raw=False, clear=False, source=_sentinel, **kwargs)
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
pmap(f: Callable[..., T1], iter3: Iterable[Any], iter4: Iterable[Any], iter5: Iterable[Any], iter6: Iterable[Any], /, *iterables: Iterable[Any], desc: str | None=None, key_args: Mapping[str, Any] | None=None, max_workers: int | None=None, chunksize: int | None=None, tqdm_args: Mapping[str, Any] | None=None) -> list[T1]
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor._utils.TimeLogger
times: dict[str, list[float]] = field(default_factory=dict)
times_to_dataframe(times: dict[str, list[float]])
at: coeditor.common
rec_add_dict_to(target: dict[str, Any], value: Mapping[str, Any], value_merger: Callable[[Any, Any], Any]=lambda x, y: x + y)
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
_ProcessingResult(edits: Sequence[TkC3Problem], stats: dict[str, dict | Any])
_process_commits(root: Path, workdir: Path, commits: Sequence[CommitInfo], is_training: bool, encoder: C3EditEncoder) -> _ProcessingResult
at: coeditor.dataset.C3EditEncoder
change_processor: ProjectChangeProcessor[C3Problem] = field(
default_factory=C3ProblemGenerator
)
===========unchanged ref 1===========
at: coeditor.dataset._ProcessingResult
edits: Sequence[TkC3Problem]
stats: dict[str, dict | Any]
at: coeditor.dataset._process_commits
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
edits = []
at: math
ceil(x: SupportsFloat, /) -> int
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
exists() -> bool
at: shutil
rmtree(path: Union[bytes, StrPath], ignore_errors: bool=..., onerror: Optional[Callable[[Any, Any, Any], Any]]=...) -> None
at: tempfile
gettempdir() -> str
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
at: typing.MutableMapping
pop(key: _KT, default: Union[_VT, _T]=...) -> Union[_VT, _T]
pop(key: _KT) -> _VT
===========changed ref 0===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
+ is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ encoder.change_processor.clear_stats()
+ encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
+ time_limit=time_limit_per_commit * (len(commits) + 10),
)
+ except Exception as e:
+ if isinstance(e, KeyboardInterrupt):
+ raise
- except UnicodeDecodeError as e:
- # this might happen in rare cases
+ warnings.warn(f"Failed to process project: {root}\nError: {e}")
- warnings.warn(f"Unable to process project: {root}\nError: {e}")
edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
===========changed ref 1===========
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
+ return compute_line_diffs_fast(before, after)
- compute_line_diffs_fast(before, after)
differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
|
coeditor.code_change/edits_from_commit_history
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> time_limit=time_limit,
|
# module: coeditor.code_change
def edits_from_commit_history(
project_dir: Path,
history: Sequence[CommitInfo],
tempdir: Path,
change_processor: ProjectChangeProcessor[TProb] = NoProcessing(),
edit_encoder: Callable[[TProb], Iterable[TEnc]] = lambda x: [x],
ignore_dirs=DefaultIgnoreDirs,
silent: bool = False,
+ time_limit: float | None = None,
) -> Sequence[TEnc]:
"""Incrementally compute the edits to a project from the git history.
Note that this will change the file states in the project directory, so
you should make a copy of the project before calling this function.
"""
tempdir = tempdir.resolve()
if tempdir.exists():
raise FileExistsError(f"Workdir '{tempdir}' already exists.")
use_fast_parser = jedi.settings.fast_parser
tempdir.mkdir(parents=True, exist_ok=False)
try:
run_command(
["cp", "-r", str(project_dir / ".git"), str(tempdir)],
cwd=project_dir.parent,
)
return _edits_from_commit_history(
+ tempdir,
+ history,
+ change_processor,
+ edit_encoder,
+ ignore_dirs,
+ silent,
- tempdir, history, change_processor, edit_encoder, ignore_dirs, silent
<0> )
finally:
shutil.rmtree(tempdir)
jedi.settings.fast_parser = use_fast_parser
|
===========unchanged ref 0===========
at: coeditor.code_change
DefaultIgnoreDirs = {".venv", ".mypy_cache", ".git", "venv", "build"}
JProjectChange(changed: Mapping[ModuleName, JModuleChange], all_modules: Modified[Collection[JModule]], commit_info: "CommitInfo | None")
TProb = TypeVar("TProb", covariant=True)
TEnc = TypeVar("TEnc", covariant=True)
ProjectChangeProcessor()
NoProcessing()
at: coeditor.common
run_command(args: Sequence[str], cwd: str | Path) -> str
at: jedi.settings
fast_parser = True
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
resolve(strict: bool=...) -> _P
mkdir(mode: int=..., parents: bool=..., exist_ok: bool=...) -> None
exists() -> bool
at: pathlib.PurePath
__slots__ = (
'_drv', '_root', '_parts',
'_str', '_hash', '_pparts', '_cached_cparts',
)
drive = property(attrgetter('_drv'),
doc="""The drive prefix (letter or UNC path), if any.""")
root = property(attrgetter('_root'),
doc="""The root of the path, if any.""")
at: typing
Iterable = _alias(collections.abc.Iterable, 1)
Callable = _CallableType(collections.abc.Callable, 2)
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.code_change
- @dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def set_training(self, is_training: bool) -> None:
+ return None
+
===========changed ref 1===========
# module: coeditor.code_change
@dataclass(frozen=True)
class ChangedSpan:
+ def _is_func_body(self) -> bool:
+ return self.parent_scopes[-1].earlier().tree.type == ptree.Function.type
+
===========changed ref 2===========
# module: coeditor.dataset
+ time_limit_per_commit = 5.0
===========changed ref 3===========
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
+ return compute_line_diffs_fast(before, after)
- compute_line_diffs_fast(before, after)
differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
===========changed ref 4===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
+ is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ encoder.change_processor.clear_stats()
+ encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
+ time_limit=time_limit_per_commit * (len(commits) + 10),
)
+ except Exception as e:
+ if isinstance(e, KeyboardInterrupt):
+ raise
- except UnicodeDecodeError as e:
- # this might happen in rare cases
+ warnings.warn(f"Failed to process project: {root}\nError: {e}")
- warnings.warn(f"Unable to process project: {root}\nError: {e}")
edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
===========changed ref 5===========
# module: coeditor.common
def normalize_code_by_ast(
code: str, sort_keyargs: bool = True, remove_doc_string: bool = True
) -> str:
"""Normalize the code by parsing and unparsing it using the AST module.
If parsing fails, return the original code."""
class KeyargSorter(ast.NodeTransformer):
def visit_Call(self, node: ast.Call):
if node.keywords:
node.keywords.sort(key=lambda x: x.arg or "None")
return node
class DocStringremover(ast.NodeTransformer):
def visit_FunctionDef(self, node: ast.FunctionDef):
return self._visit_def(node)
def visit_Module(self, node: ast.Module) -> Any:
return self._visit_def(node)
def visit_ClassDef(self, node: ast.ClassDef):
return self._visit_def(node)
def _visit_def(self, node):
node = as_any(self.generic_visit(node))
match node.body:
case [ast.Expr(value=ast.Constant(value=str())), *body]:
node.body = body
return node
try:
tree = ast.parse(dedent(code))
if remove_doc_string:
tree = DocStringremover().visit(tree)
if sort_keyargs:
tree = KeyargSorter().visit(tree)
return ast.unparse(tree)
+ except (SyntaxError, ValueError):
- except SyntaxError:
return code
|
coeditor.ctx_change_encoder/C3ProblemGenerator.__init__
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> self._is_training: bool = False
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def __init__(self, analyzer: "JediUsageAnalyzer | None" = None):
if analyzer is None:
analyzer = JediUsageAnalyzer()
+
self.analyzer = analyzer
+ # whether to only generate problems for editing functions
<0>
|
===========unchanged ref 0===========
at: coeditor.ctx_change_encoder
JediUsageAnalyzer()
at: coeditor.ctx_change_encoder.C3ProblemGenerator
VERSION = "1.1"
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ VERSION = "1.2"
- VERSION = "1.1"
===========changed ref 1===========
# module: coeditor.code_change
- @dataclass
class ProjectChangeProcessor(Generic[TProb], ABC):
+ def set_training(self, is_training: bool) -> None:
+ return None
+
===========changed ref 2===========
# module: coeditor.code_change
+ _Second = float
===========changed ref 3===========
# module: coeditor.dataset
+ time_limit_per_commit = 5.0
===========changed ref 4===========
# module: coeditor.code_change
@dataclass(frozen=True)
class ChangedSpan:
+ def _is_func_body(self) -> bool:
+ return self.parent_scopes[-1].earlier().tree.type == ptree.Function.type
+
===========changed ref 5===========
# module: spot.utils
def compute_line_diffs(
before: Sequence[str], after: Sequence[str], keep_explain_lines: bool = False
):
SizeLimit = 8000
if (
sum(len(x) for x in before) > SizeLimit
or sum(len(x) for x in after) > SizeLimit
):
+ return compute_line_diffs_fast(before, after)
- compute_line_diffs_fast(before, after)
differ = difflib.Differ()
result = []
for line in differ.compare(before, after):
assert len(line) >= 2
tag = line[0]
if keep_explain_lines or tag != "?":
result.append(tag + line[2:])
return result
===========changed ref 6===========
# module: coeditor.code_change
def edits_from_commit_history(
project_dir: Path,
history: Sequence[CommitInfo],
tempdir: Path,
change_processor: ProjectChangeProcessor[TProb] = NoProcessing(),
edit_encoder: Callable[[TProb], Iterable[TEnc]] = lambda x: [x],
ignore_dirs=DefaultIgnoreDirs,
silent: bool = False,
+ time_limit: float | None = None,
) -> Sequence[TEnc]:
"""Incrementally compute the edits to a project from the git history.
Note that this will change the file states in the project directory, so
you should make a copy of the project before calling this function.
"""
tempdir = tempdir.resolve()
if tempdir.exists():
raise FileExistsError(f"Workdir '{tempdir}' already exists.")
use_fast_parser = jedi.settings.fast_parser
tempdir.mkdir(parents=True, exist_ok=False)
try:
run_command(
["cp", "-r", str(project_dir / ".git"), str(tempdir)],
cwd=project_dir.parent,
)
return _edits_from_commit_history(
+ tempdir,
+ history,
+ change_processor,
+ edit_encoder,
+ ignore_dirs,
+ silent,
+ time_limit=time_limit,
- tempdir, history, change_processor, edit_encoder, ignore_dirs, silent
)
finally:
shutil.rmtree(tempdir)
jedi.settings.fast_parser = use_fast_parser
===========changed ref 7===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
+ is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ encoder.change_processor.clear_stats()
+ encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
+ time_limit=time_limit_per_commit * (len(commits) + 10),
)
+ except Exception as e:
+ if isinstance(e, KeyboardInterrupt):
+ raise
- except UnicodeDecodeError as e:
- # this might happen in rare cases
+ warnings.warn(f"Failed to process project: {root}\nError: {e}")
- warnings.warn(f"Unable to process project: {root}\nError: {e}")
edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
===========changed ref 8===========
# module: coeditor.common
def normalize_code_by_ast(
code: str, sort_keyargs: bool = True, remove_doc_string: bool = True
) -> str:
"""Normalize the code by parsing and unparsing it using the AST module.
If parsing fails, return the original code."""
class KeyargSorter(ast.NodeTransformer):
def visit_Call(self, node: ast.Call):
if node.keywords:
node.keywords.sort(key=lambda x: x.arg or "None")
return node
class DocStringremover(ast.NodeTransformer):
def visit_FunctionDef(self, node: ast.FunctionDef):
return self._visit_def(node)
def visit_Module(self, node: ast.Module) -> Any:
return self._visit_def(node)
def visit_ClassDef(self, node: ast.ClassDef):
return self._visit_def(node)
def _visit_def(self, node):
node = as_any(self.generic_visit(node))
match node.body:
case [ast.Expr(value=ast.Constant(value=str())), *body]:
node.body = body
return node
try:
tree = ast.parse(dedent(code))
if remove_doc_string:
tree = DocStringremover().visit(tree)
if sort_keyargs:
tree = KeyargSorter().visit(tree)
return ast.unparse(tree)
+ except (SyntaxError, ValueError):
- except SyntaxError:
return code
|
coeditor.ctx_change_encoder/C3ProblemGenerator.process_change
|
Modified
|
temp-1
|
ab374914700355dc76b7f09d9c8121b7a6855e67
|
- Only editing non-function spans during training. - Add time limit to data creation. - Fix fast diff. - Handle value error in `normalize_code_by_ast`. - Add max_total_ref_tks to C3ProblemTokenizer.
|
<0>:<add> if should_mk_problem:
|
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Sequence[C3Problem]:
<s>
sorted_defs.append(pydef)
# return unique cspans
seen = set[tuple[ModuleName, LineRange]]()
# we don't need to show the changed parts again
for cspan in (this_change, *other_changes):
seen.add((cspan.path.module, cspan.line_range))
result = list[ChangedSpan]()
for used in sorted_defs:
for cspan in get_def_spans(used):
key = (cspan.path.module, cspan.line_range)
if key not in seen:
result.append(cspan)
seen.add(key)
return result
processed_cspans = list[ChangedSpan]()
problems = list[C3Problem]()
for m in module_order:
if (mchange := pchange.changed.get(m)) is None:
continue
for span in mchange.changed.values():
+ should_mk_problem = (span.change.as_char() == Modified.as_char()) and (
- if span.change.as_char() == Modified.as_char():
+ # only consider function edits at test time
+ self._is_training
+ or span._is_func_body()
+ )
<0> # latest changes are more relevant
relevant_changes = list(reversed(processed_cspans))
prob = C3Problem(
span,
relevant_changes=relevant_changes,
relevant_unchanged=get_relevant_unchanged(
span, relevant_changes
),
src_info={"commit": pchange.commit_info},
)
problems.append(prob)
processed_cspans.append(span)
return problems
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Sequence[C3Problem]:
# offset: -1
<s>
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
stmt_span.line_range,
)
cspans.append(cspan)
cspan_cache[used] = cspans
return cspans
def get_relevant_unchanged(
this_change: ChangedSpan, other_changes: Sequence[ChangedSpan]
):
if isinstance(this_change.change, Added):
# nothing to analyze
return []
path = this_change.path
line_usages = mod2usages[path.module]
# parent defs are also considered as used
parent_defs = [
PyDefinition.from_scope(c.earlier()) for c in this_change.parent_scopes
]
# immediate parents are more relevant
sorted_defs = list(reversed(parent_defs))
used_defs = set(sorted_defs)
all_lines = set(range(*this_change.line_range))
all_lines.update(range(*this_change.header_line_range))
for l in all_lines:
for pydef in line_usages.line2usages.get(l, set()):
if (
pydef.full_name.startswith(path.module)
and pydef.start_pos[0] in all_lines
):
# skip self references
continue
if pydef not in used_defs:
used_defs.add(pydef)
sorted_defs.append(pydef)
# return unique cspans
seen = set[tuple[ModuleName, Line</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Sequence[C3Problem]:
# offset: -2
<s> match elem:
case ChangeScope(tree=ptree.Function()):
func_scopes.append(elem)
case ChangeScope(tree=ptree.Class()):
# add all attrs and methods
stmt_spans.extend(elem.spans)
func_scopes.extend(
s
for s in elem.subscopes.values()
if isinstance(s.tree, ptree.Function)
)
case StatementSpan():
stmt_spans.append(elem)
# add collapsed functions
for f_scope in func_scopes:
ancestors = f_scope.ancestors()
stmts = f_scope.spans[-1].statements
body_code = stmts[-1].get_code().strip("\n")
if len(stmts) > 1:
ellipsis = " " * (len(ancestors) - 1) + "# ...\n"
body_code = ellipsis + body_code
- h_end = f_scope.header_line_range[1]
cspan = ChangedSpan(
Modified.from_unchanged(body_code),
[Modified.from_unchanged(s) for s in ancestors],
+ f_scope.spans[-1].line_range,
- line_range(h_end, h_end + len(body_code)),
)
cspans.append(cspan)
# add statement spans
for stmt_span in stmt_spans:
ancestors = stmt_span.scope.ancestors()
body_code = stmt_span.code
cspan = ChangedSpan(
Modified.from_unchanged(body_</s>
===========above chunk 2===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
def process_change(
self,
pchange: JProjectChange,
mod2usages: Mapping[ModuleName, LineUsageAnalysis],
module_order: Sequence[ModuleName],
) -> Sequence[C3Problem]:
# offset: -3
before_mod_map = {m.mname: m for m in pchange.all_modules.before}
mod_hier = ModuleHierarchy.from_modules(before_mod_map)
cspan_cache = dict[PyDefinition, list[ChangedSpan]]()
def get_def_spans(used: PyDefinition) -> list[ChangedSpan]:
"Get the (pre-edit) spans for the given definition."
if used.full_name in cspan_cache:
return cspan_cache[used.full_name]
path = mod_hier.resolve_path(split_dots(used.full_name))
cspans = list[ChangedSpan]()
if path is None or (jmod := before_mod_map.get(path.module)) is None:
cspan_cache[used] = cspans
return cspans
scope = jmod.as_scope
elem = scope._search(path.path, used.start_pos[0])
func_scopes = list[ChangeScope]()
stmt_spans = list[StatementSpan]()</s>
===========unchanged ref 0===========
at: coeditor._utils
split_dots(path: str) -> tuple[str, ...]
at: coeditor.code_change
LineRange = NewType("LineRange", tuple[int, int])
line_range(start: int, end: int, can_be_empty: bool=False) -> LineRange
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
ChangedSpan(change: Change[str], parent_scopes: Sequence[Change[ChangeScope]], line_range: LineRange)
JProjectChange(changed: Mapping[ModuleName, JModuleChange], all_modules: Modified[Collection[JModule]], commit_info: "CommitInfo | None")
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
ancestors() -> list[Self]
at: coeditor.code_change.ChangeScope.__post_init__
self.header_line_range: LineRange = header_line_range
at: coeditor.code_change.ChangedSpan
change: Change[str]
parent_scopes: Sequence[Change[ChangeScope]]
line_range: LineRange
at: coeditor.code_change.JModuleChange
module_change: Change[JModule]
changed: Mapping[ProjectPath, ChangedSpan]
at: coeditor.code_change.JProjectChange
changed: Mapping[ModuleName, JModuleChange]
all_modules: Modified[Collection[JModule]]
commit_info: "CommitInfo | None"
at: coeditor.code_change.StatementSpan
nth_in_parent: int
statements: Sequence[PyNode]
scope: ChangeScope
|
coeditor.dataset/_process_commits
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> traceback.print_exception(e, limit=-6)
|
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
encoder.change_processor.clear_stats()
encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
except Exception as e:
if isinstance(e, KeyboardInterrupt):
raise
warnings.warn(f"Failed to process project: {root}\nError: {e}")
<0> edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
_ProcessingResult(edits: Sequence[TkC3Problem], stats: dict[str, dict | Any])
at: coeditor.dataset.C3EditEncoder
change_processor: ProjectChangeProcessor[C3Problem] = field(
default_factory=C3ProblemGenerator
)
edit_tokenizer: C3ProblemTokenizer = field(default_factory=C3ProblemTokenizer)
at: pathlib
Path()
at: traceback
print_exception(etype: Optional[Type[BaseException]], value: Optional[BaseException], tb: Optional[TracebackType], limit: Optional[int]=..., file: Optional[IO[str]]=..., chain: bool=...) -> None
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
+ time_limit_per_commit = 10.0
- time_limit_per_commit = 5.0
|
coeditor.dataset/dataset_from_projects
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> history_chunk_size = max(50, math.ceil(len(h) / 10))
|
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
<s>
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
- history_chunk_size = max(50, math.ceil(len(h) / 4))
<0> for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
chunk_training,
key_args={"encoder": encoder},
desc="Create tokenized edits",
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits</s>
|
===========above chunk 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: -1
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
</s>
===========below chunk 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
) -> "TokenizedEditDataset[TkC3Problem]":
# offset: 1
<s>exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
project2edits = dict[Path, list[TkC3Problem]]()
try:
stats = dict[str, Any]()
for root, pr in zip(roots, presults):
project2edits.setdefault(root, []).extend(pr.edits)
rec_add_dict_to(stats, pr.stats)
if "tlogger" in stats:
df = TimeLogger.times_to_dataframe(stats.pop("tlogger"))
print("Time stats:")
display(df)
if "analyzer_errors" in list(stats.keys()):
errors: dict = stats.pop("analyzer_errors")
for k in list(errors.keys()):
if JediUsageAnalyzer.is_known_error(k):
errors.pop(k)
if errors:
print("Analyzer errors:")
for k in sorted(errors.keys(), key=lambda k: errors[k], reverse=True):
print(f"{k}:\t{errors[k]}")
if stats:
print("Other Stats:")
pretty_print_dict(stats)
except Exception as e:
if not isinstance(e, KeyboardInterrupt):
print("Error while printing stats:", e)
return TokenizedEditDataset(project2edits)
===========unchanged ref 0===========
at: IPython.core.display_functions
display(*, include=None, exclude=None, metadata=None, transient=None, display_id=None, raw=False, clear=False, source=_sentinel, **kwargs)
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
pmap(f: Callable[..., T1], iter3: Iterable[Any], iter4: Iterable[Any], iter5: Iterable[Any], iter6: Iterable[Any], /, *iterables: Iterable[Any], desc: str | None=None, key_args: Mapping[str, Any] | None=None, max_workers: int | None=None, chunksize: int | None=None, tqdm_args: Mapping[str, Any] | None=None) -> list[T1]
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor._utils.TimeLogger
times: dict[str, list[float]] = field(default_factory=dict)
times_to_dataframe(times: dict[str, list[float]])
at: coeditor.common
rec_add_dict_to(target: dict[str, Any], value: Mapping[str, Any], value_merger: Callable[[Any, Any], Any]=lambda x, y: x + y)
at: coeditor.dataset
C3EditEncoder(change_processor: ProjectChangeProcessor[C3Problem]=field(
default_factory=C3ProblemGenerator
), edit_tokenizer: C3ProblemTokenizer=field(default_factory=C3ProblemTokenizer))
_ProcessingResult(edits: Sequence[TkC3Problem], stats: dict[str, dict | Any])
_process_commits(root: Path, workdir: Path, commits: Sequence[CommitInfo], is_training: bool, encoder: C3EditEncoder) -> _ProcessingResult
at: coeditor.dataset._ProcessingResult
edits: Sequence[TkC3Problem]
stats: dict[str, dict | Any]
===========unchanged ref 1===========
at: coeditor.dataset._process_commits
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
edits = []
stats = dict()
at: math
ceil(x: SupportsFloat, /) -> int
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
exists() -> bool
at: shutil
rmtree(path: Union[bytes, StrPath], ignore_errors: bool=..., onerror: Optional[Callable[[Any, Any, Any], Any]]=...) -> None
at: spot.utils
pretty_print_dict(d: dict, level: int=0, max_show_level: int=1000, float_precision: int=5)
at: tempfile
gettempdir() -> str
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
at: typing.MutableMapping
pop(key: _KT, default: Union[_VT, _T]=...) -> Union[_VT, _T]
pop(key: _KT) -> _VT
===========changed ref 0===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
is_training: bool,
encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
encoder.change_processor.clear_stats()
encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
change_processor=encoder.change_processor,
edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
except Exception as e:
if isinstance(e, KeyboardInterrupt):
raise
warnings.warn(f"Failed to process project: {root}\nError: {e}")
+ traceback.print_exception(e, limit=-6)
edits = []
stats = dict()
encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
===========changed ref 1===========
# module: coeditor.dataset
+ time_limit_per_commit = 10.0
- time_limit_per_commit = 5.0
|
coeditor.retrieval_model/_BatchSampler.estimate_n_batches
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> batches = tk_edits_to_batches(self.all_edits, self.batch_args, silent=True)
|
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
def estimate_n_batches(self) -> int:
- batches = query_edits_to_batches(self.all_edits, self.batch_args, silent=True)
<0> return len(batches)
|
===========unchanged ref 0===========
at: coeditor.retrieval_model.pad_token_seqs
max_len = max(len(ref) for ref in seqs)
pad_id = PAD_id
===========changed ref 0===========
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
+ all_edits: list[TkC3Problem]
- all_edits: list[BasicTkQueryEdit]
batch_args: BatchArgs
shuffle: bool
desc: str
tqdm_args: dict | None = None
===========changed ref 1===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
@classmethod
def eval_default(cls) -> Self:
return BatchArgs(
max_total_ref_tks=512 * 32,
max_queries=32,
- max_ref_dropout=0.0,
shuffle_extra_ids=False,
)
===========changed ref 2===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
max_output_tks: int = 256
max_query_tks: int = 512
min_queires: int = 1
max_queries: int = 8
max_ref_tks: int = 512
max_total_ref_tks: int = 512 * 16
- max_ref_dropout: float = 1.0
shuffle_extra_ids: bool = True
- use_only_modified: bool = True
===========changed ref 3===========
# module: coeditor.dataset
+ time_limit_per_commit = 10.0
- time_limit_per_commit = 5.0
===========changed ref 4===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
+ def process_edit(e: TkC3Problem):
+ labels = e.output_tks
+
+ labels = wrap_bos(labels)
+
+ if len(labels) > args.max_output_tks:
+ labels = labels[: args.max_output_tks]
+
+ input_ids = e.input_tks
+
+ if args.shuffle_extra_ids and random.random() < 0.5:
+ id_map = random_extra_id_map()
+ input_ids = [id_map.get(tk, tk) for tk in input_ids]
+ labels = [id_map.get(tk, tk) for tk in labels]
+
+ return input_ids, labels
+
+ cost_limit = args.cost_limit()
+ warned_batch_size = False
+
+ def edits_to_batches(
+ edits: Sequence[TkC3Problem],
+ ) -> Iterable[dict]:
+ def pack_batch(rows: list[dict]):
+ assert rows, "empty batch found"
+ input_ids = [x["input_tks"] for x in rows]
+ labels = [x["output_tks"] for x in rows]
+ refs = [x["references"] for x in rows]
+ id2ref = {id(ref): ref for row in refs for ref in row}
+ references = [id2ref[x] for x in id2ref]
+ id2order = {x: i for i, x in enumerate(id2ref)}
+ query_ref_list = [[id2order[id(ref)] for ref in row] for row in refs]
+ return {
+ "input_ids": input_ids,
+ "references": references,
+ "query_</s>
===========changed ref 5===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
# offset: 1
<s> <add> return {
+ "input_ids": input_ids,
+ "references": references,
+ "query_ref_list": query_ref_list,
+ "labels": labels,
+ }
+
+ # sample references for each query
+ current_batch = []
+ current_cost = 0
+ for edit in tqdm(edits, desc="edits_to_batches", disable=silent):
+ all_refs = [x[1] for x in edit.named_references]
+ ref_size_sum = 0
+ ref_selected = list[TokenSeq]()
+ for ref in all_refs:
+ if ref_size_sum + len(ref) <= args.max_total_ref_tks:
+ ref_selected.append(ref)
+ ref_size_sum += len(ref)
+ input_tks, output_tks = process_edit(edit)
+ cost = retrieval_cost_model(
+ ref_size=sum(len(x) for x in ref_selected),
+ query_size=len(input_tks),
+ output_size=len(output_tks),
+ )
+ row = {
+ "input_tks": input_tks,
+ "output_tks": output_tks,
+ "references": ref_selected,
+ }
+ nonlocal warned_batch_size
+ if cost > cost_limit and not warned_batch_size:
+ warned_batch_size = True
+ warnings.warn("Batch cost limit is too small.")
+ if (not current_batch) or (
+ cost +</s>
===========changed ref 6===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
# offset: 2
<s>cost <= cost_limit
+ and len(current_batch) < args.max_queries
+ ):
+ current_batch.append(row)
+ current_cost += cost
+ else:
+ yield pack_batch(current_batch)
+ current_batch = [row]
+ current_cost = cost
+ if current_batch:
+ yield pack_batch(current_batch)
+
+ batches = list[dict]()
+ bsizes = list[int]()
+ for batch in edits_to_batches(query_edits):
+ batches.append(batch)
+ bsizes.append(len(batch["input_ids"]))
+
+ batch_stats = {k: f"{v:.1f}" for k, v in scalar_stats(bsizes).items()}
+ if not silent:
+ cprint("blue", f"num batches: {len(batches)},", f"batch stats: {batch_stats}")
+
+ return batches
+
===========changed ref 7===========
# module: coeditor.retrieval_model
class RetrievalEditorModel(T5PreTrainedModel):
def eval_loss_on_data(
self, data: TokenizedEditDataset, batch_args: "BatchArgs"
) -> dict[str, WeightedSum]:
batch_args = copy.deepcopy(batch_args)
- batch_args.max_ref_dropout = 0.0
batch_args.shuffle_extra_ids = False
eval_loader = edits_to_dataloader(
data.all_edits(),
args=batch_args,
shuffle=False,
desc="Eval Epoch",
)
return self.eval_loss_on_loader(eval_loader)
|
coeditor.retrieval_model/_BatchSampler.__iter__
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> batches = tk_edits_to_batches(self.all_edits, self.batch_args)
|
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
def __iter__(self) -> Iterable[Mapping]:
if self.shuffle:
random.shuffle(self.all_edits)
- batches = query_edits_to_batches(self.all_edits, self.batch_args)
<0> if self.shuffle:
random.shuffle(batches)
tqdm_args = self.tqdm_args or {"smoothing": 0.0}
for b in tqdm(batches, desc=self.desc + f" {self.epochs}", **tqdm_args):
input_ids = pad_token_seqs(b["input_ids"])
labels = pad_token_seqs(b["labels"], pad_id=-100)
yield {
"input_ids": input_ids,
"references": b["references"],
"query_ref_list": b["query_ref_list"],
"labels": labels,
}
self.epochs += 1
|
===========unchanged ref 0===========
at: coeditor.retrieval_model.pad_token_seqs
rows = []
at: math
ceil(x: SupportsFloat, /) -> int
===========changed ref 0===========
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
def estimate_n_batches(self) -> int:
+ batches = tk_edits_to_batches(self.all_edits, self.batch_args, silent=True)
- batches = query_edits_to_batches(self.all_edits, self.batch_args, silent=True)
return len(batches)
===========changed ref 1===========
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
+ all_edits: list[TkC3Problem]
- all_edits: list[BasicTkQueryEdit]
batch_args: BatchArgs
shuffle: bool
desc: str
tqdm_args: dict | None = None
===========changed ref 2===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
@classmethod
def eval_default(cls) -> Self:
return BatchArgs(
max_total_ref_tks=512 * 32,
max_queries=32,
- max_ref_dropout=0.0,
shuffle_extra_ids=False,
)
===========changed ref 3===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
max_output_tks: int = 256
max_query_tks: int = 512
min_queires: int = 1
max_queries: int = 8
max_ref_tks: int = 512
max_total_ref_tks: int = 512 * 16
- max_ref_dropout: float = 1.0
shuffle_extra_ids: bool = True
- use_only_modified: bool = True
===========changed ref 4===========
# module: coeditor.dataset
+ time_limit_per_commit = 10.0
- time_limit_per_commit = 5.0
===========changed ref 5===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
+ def process_edit(e: TkC3Problem):
+ labels = e.output_tks
+
+ labels = wrap_bos(labels)
+
+ if len(labels) > args.max_output_tks:
+ labels = labels[: args.max_output_tks]
+
+ input_ids = e.input_tks
+
+ if args.shuffle_extra_ids and random.random() < 0.5:
+ id_map = random_extra_id_map()
+ input_ids = [id_map.get(tk, tk) for tk in input_ids]
+ labels = [id_map.get(tk, tk) for tk in labels]
+
+ return input_ids, labels
+
+ cost_limit = args.cost_limit()
+ warned_batch_size = False
+
+ def edits_to_batches(
+ edits: Sequence[TkC3Problem],
+ ) -> Iterable[dict]:
+ def pack_batch(rows: list[dict]):
+ assert rows, "empty batch found"
+ input_ids = [x["input_tks"] for x in rows]
+ labels = [x["output_tks"] for x in rows]
+ refs = [x["references"] for x in rows]
+ id2ref = {id(ref): ref for row in refs for ref in row}
+ references = [id2ref[x] for x in id2ref]
+ id2order = {x: i for i, x in enumerate(id2ref)}
+ query_ref_list = [[id2order[id(ref)] for ref in row] for row in refs]
+ return {
+ "input_ids": input_ids,
+ "references": references,
+ "query_</s>
===========changed ref 6===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
# offset: 1
<s> <add> return {
+ "input_ids": input_ids,
+ "references": references,
+ "query_ref_list": query_ref_list,
+ "labels": labels,
+ }
+
+ # sample references for each query
+ current_batch = []
+ current_cost = 0
+ for edit in tqdm(edits, desc="edits_to_batches", disable=silent):
+ all_refs = [x[1] for x in edit.named_references]
+ ref_size_sum = 0
+ ref_selected = list[TokenSeq]()
+ for ref in all_refs:
+ if ref_size_sum + len(ref) <= args.max_total_ref_tks:
+ ref_selected.append(ref)
+ ref_size_sum += len(ref)
+ input_tks, output_tks = process_edit(edit)
+ cost = retrieval_cost_model(
+ ref_size=sum(len(x) for x in ref_selected),
+ query_size=len(input_tks),
+ output_size=len(output_tks),
+ )
+ row = {
+ "input_tks": input_tks,
+ "output_tks": output_tks,
+ "references": ref_selected,
+ }
+ nonlocal warned_batch_size
+ if cost > cost_limit and not warned_batch_size:
+ warned_batch_size = True
+ warnings.warn("Batch cost limit is too small.")
+ if (not current_batch) or (
+ cost +</s>
===========changed ref 7===========
# module: coeditor.retrieval_model
+ def tk_edits_to_batches(
+ query_edits: Sequence[TkC3Problem],
+ args: BatchArgs,
+ silent: bool = False,
+ ) -> list[dict]:
# offset: 2
<s>cost <= cost_limit
+ and len(current_batch) < args.max_queries
+ ):
+ current_batch.append(row)
+ current_cost += cost
+ else:
+ yield pack_batch(current_batch)
+ current_batch = [row]
+ current_cost = cost
+ if current_batch:
+ yield pack_batch(current_batch)
+
+ batches = list[dict]()
+ bsizes = list[int]()
+ for batch in edits_to_batches(query_edits):
+ batches.append(batch)
+ bsizes.append(len(batch["input_ids"]))
+
+ batch_stats = {k: f"{v:.1f}" for k, v in scalar_stats(bsizes).items()}
+ if not silent:
+ cprint("blue", f"num batches: {len(batches)},", f"batch stats: {batch_stats}")
+
+ return batches
+
===========changed ref 8===========
# module: coeditor.retrieval_model
class RetrievalEditorModel(T5PreTrainedModel):
def eval_loss_on_data(
self, data: TokenizedEditDataset, batch_args: "BatchArgs"
) -> dict[str, WeightedSum]:
batch_args = copy.deepcopy(batch_args)
- batch_args.max_ref_dropout = 0.0
batch_args.shuffle_extra_ids = False
eval_loader = edits_to_dataloader(
data.all_edits(),
args=batch_args,
shuffle=False,
desc="Eval Epoch",
)
return self.eval_loss_on_loader(eval_loader)
|
coeditor.api/EditPredictionService.suggest_edit
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> batches = tk_edits_to_batches(qedits, self.batch_args)
|
# module: coeditor.api
@dataclass
class EditPredictionService:
def suggest_edit(
self,
file: Path,
line: int,
log_dir: Path | None = Path(".coeditor_logs"),
) -> ServiceResponse:
<s> elem_change = Modified(trans_elem, trans_elem)
with timed("encode edits"):
respect_lines = (
self.compute_offset(now_mod, now_elem, line, drop_comments=True) + 1
)
print(f"{respect_lines = }")
req = EditRequest(elem_change, respect_lines)
qedits = list(
self.encoder.encode_pedit(
pedit,
self.stub_cache,
queries=[req],
training=False,
)
)
if qedits[0].tk_pedit.module_stubs:
print("stub files:", qedits[0].tk_pedit.module_stubs.keys())
assert len(qedits) == 1
- batches = query_edits_to_batches(qedits, self.batch_args)
<0> assert len(batches) == 1
batch = batches[0]
with timed("run model"), torch.autocast("cuda"):
predictions = self.model.predict_on_batch(
batch, [req], self.dec_args, self.show_max_solutions
)
assert_eq(len(predictions), 1)
predictions = predictions[0]
assert predictions
if log_dir is not None:
log_dir.mkdir(exist_ok=True)
input_tks = batch["input_ids"][0]
references = batch["references"]
output_truth = batch["labels"][0]
print(f"Writing logs to: {log_dir}")
for i, pred in enumerate(predictions):
with (log_dir / f"solution-{i}.txt").open("w") as f:
pred_tks = pred.out_tks
</s>
|
===========above chunk 0===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def suggest_edit(
self,
file: Path,
line: int,
log_dir: Path | None = Path(".coeditor_logs"),
) -> ServiceResponse:
# offset: -1
<s>
cst.parse_module(now_code), mname, drop_comments=False
),
)
now_elem = get_elem_by_line(now_mod, line)
if now_elem is None:
raise ValueError(
f"No code element found at line {line} in file {file}."
)
if not isinstance(now_elem, PythonFunction):
raise ValueError(f"Only functions can be edited by the model.")
with timed("construct project edit"):
pedit = self.config.get_pedit(
project, file, self.prev_cache, self.now_cache
)
if mname not in pedit.changes:
assert mname in pedit.before.modules
assert mname in pedit.after.modules
pedit.changes[mname] = ModuleEdit.from_no_change(
pedit.before.modules[mname]
)
match [
c for c in pedit.all_elem_changes() if get_change_path(c) == now_elem.path
]:
case [Modified(PythonFunction(), PythonFunction()) as mf]:
elem_change = cast(Modified[PythonFunction], mf)
case [Added(PythonFunction()) as mf]:
elem_change = cast(Added[PythonFunction], mf)
case _:
if self.config.drop_comments:
trans_tree = remove_comments(now_elem.tree)
trans_elem = PythonFunction(
now_elem.name, now_elem.path, now_elem.parent_class, trans_tree
)
else:
trans_elem = now_elem
elem_change = Modified(trans_elem, trans_elem)
with timed("encode edits"):
respect_lines =</s>
===========above chunk 1===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def suggest_edit(
self,
file: Path,
line: int,
log_dir: Path | None = Path(".coeditor_logs"),
) -> ServiceResponse:
# offset: -2
"""Make the suggestion in-place at the given location."""
timed = self.tlogger.timed
project = self.project
if not file.is_absolute():
file = project / file
with timed("get target element"):
mname = PythonProject.rel_path_to_module_name(file.relative_to(project))
stamp = os.stat(file).st_mtime
now_code = file.read_text()
now_mod = self.now_cache.cached(
(mname, False),
stamp,
lambda: PythonModule.from_cst(
cst.parse_module(now_code), mname, drop_comments=False
),
)
</s>
===========below chunk 0===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def suggest_edit(
self,
file: Path,
line: int,
log_dir: Path | None = Path(".coeditor_logs"),
) -> ServiceResponse:
# offset: 1
<s> f"solution-{i}.txt").open("w") as f:
pred_tks = pred.out_tks
score = pred.score
print(f"{respect_lines = }", file=f)
print(f"{len(input_tks) = }", file=f)
print(f"{len(references) = }", file=f)
print(f"Solution score: {score:.3g}", file=f)
print(f"Marginalized samples:", pred.n_samples, file=f)
pred = RetrievalModelPrediction(
input_ids=input_tks,
output_ids=pred_tks,
labels=output_truth,
references=references,
)
pred_str = RetrievalDecodingResult.show_prediction(None, pred)
print(pred_str, file=f)
now_span = now_mod.location_map[now_elem.tree]
# old_elem_code = get_span(now_code, now_span)
old_elem_code = now_elem.code
respect_lines = (
self.compute_offset(now_mod, now_elem, line, drop_comments=False) + 1
)
suggestions = list[EditSuggestion]()
for pred in predictions:
suggested_change, preview = self.apply_edit_to_elem(
file,
now_mod,
now_elem,
line,
pred.out_tks,
)
suggestion = EditSuggestion(
score=pred.score,
change_preview=preview,
new_code=suggested_change.after,
)
suggestions.append(suggestion)
def as_tuple(x: CodePosition):
return (x.line</s>
===========below chunk 1===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def suggest_edit(
self,
file: Path,
line: int,
log_dir: Path | None = Path(".coeditor_logs"),
) -> ServiceResponse:
# offset: 2
<s> )
suggestions.append(suggestion)
def as_tuple(x: CodePosition):
return (x.line, x.column)
return ServiceResponse(
target_file=file.as_posix(),
edit_start=(now_span.start.line, 0),
edit_end=as_tuple(now_span.end),
old_code=old_elem_code,
suggestions=suggestions,
)
===========unchanged ref 0===========
at: coeditor._utils
assert_eq(x: T1, y: T1, message: Callable[[], str]=lambda: "") -> None
at: coeditor.api
EditSuggestion(score: float, change_preview: str, new_code: str)
ServiceResponse(target_file: str, edit_start: tuple[int, int], edit_end: tuple[int, int], old_code: str, suggestions: list[EditSuggestion])
get_elem_by_line(module: PythonModule, line: int) -> PythonElem | None
at: coeditor.api.ChangeDetectionConfig
untracked_as_additions: bool = True
ignore_dirs: Collection[str] = field(
default_factory=lambda: PythonProject.DefaultIgnoreDirs
)
drop_comments: DropComment = True
get_pedit(project_root: Path, target_file: Path, prev_cache: TimedCache[tuple[ModuleName, DropComment], PythonModule, str], now_cache: TimedCache[tuple[ModuleName, DropComment], PythonModule, float]) -> ProjectEdit
at: coeditor.api.EditPredictionService
compute_offset(now_mod: PythonModule, now_elem: PythonElem, line: int, drop_comments: bool)
apply_edit_to_elem(self, file: Path, now_mod: PythonModule, now_elem: PythonElem, cursor_line: int, out_tks: TokenSeq) -> tuple[Modified[str], str]
apply_edit_to_elem(file: Path, now_mod: PythonModule, now_elem: PythonElem, cursor_line: int, out_tks: TokenSeq) -> tuple[Modified[str], str]
at: coeditor.api.EditPredictionService.__init__
self.project = project
self.model = model
self.batch_args = batch_args
self.encoder = encoder
self.dec_args = dec_args
self.config = config
self.show_max_solutions = 3
|
coeditor.api/EditPredictionService.apply_edit_to_elem
|
Modified
|
temp-1
|
ec882028fee3333b952c7e20e8051e7a58402a6d
|
Switch to TkC3Problem in retrieval model code.
|
<0>:<add> new_change = new_change.map(lambda s: s.strip("\n"))
|
# module: coeditor.api
@dataclass
class EditPredictionService:
def apply_edit_to_elem(
self,
file: Path,
now_mod: PythonModule,
now_elem: PythonElem,
cursor_line: int,
out_tks: TokenSeq,
) -> tuple[Modified[str], str]:
<s>, drop_comments=True)
+ 1
)
new_out_tks = TokenSeq()
for k, seg in output_ids_as_seqs(out_tks).items():
rel_line = extra_id_to_number(k) + lines_no_comment
if rel_line not in line_map:
messages = []
messages.append(
f"predicted relative line {rel_line} (extra_id_{extra_id_to_number(k)}) is out of range."
)
messages.append(
f"{n_lines = }, {lines_no_comment = }, {lines_with_comment = }"
)
messages.append(f"{line_map = }")
if isinstance(code_change, Modified):
messages.append("Prev element:")
messages.append(add_line_numbers(code_change.before))
e = ValueError("\n".join(messages))
raise e
line = line_map[rel_line]
k1 = get_extra_id(line - lines_with_comment)
new_out_tks.append(k1)
new_out_tks.extend(seg)
out_tks = new_out_tks
change_tks = change_to_tokens(code_change)
new_change = apply_output_tks_to_change(change_tks, lines_with_comment, out_tks)
- new_change.before = new_change.before.strip("\n")
- new_change.after = new_change.after.strip("\n")
<0> preview = self.preview_changes(new_change, lines_with_comment)
return new_change, preview
|
===========above chunk 0===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def apply_edit_to_elem(
self,
file: Path,
now_mod: PythonModule,
now_elem: PythonElem,
cursor_line: int,
out_tks: TokenSeq,
) -> tuple[Modified[str], str]:
# offset: -1
<s>index_stamp(self.project, path_s),
lambda: PythonModule.from_cst(
cst.parse_module(file_content_from_commit(self.project, "", path_s)),
mname,
drop_comments=False,
),
)
now_code = now_elem.code
prev_elem = prev_mod.elems_dict.get(now_elem.path.path)
if prev_elem is None:
code_change = Added(now_code)
else:
code_change = Modified(prev_elem.code, now_code)
lines_with_comment = (
self.compute_offset(now_mod, now_elem, cursor_line, drop_comments=False) + 1
)
logging.info("Now respect lines:", lines_with_comment)
if self.config.drop_comments:
# map the changes to the original code locations with comments
remover = CommentRemover(prev_mod.location_map)
if prev_elem is not None:
elem1 = prev_elem.tree.visit(remover)
assert isinstance(elem1, cst.CSTNode)
line_map = remover.line_map(elem1)
else:
line_map = {0: 0, 1: 1}
n_lines = len(line_map)
line_map[n_lines] = line_map[n_lines - 1] + 1
lines_no_comment = (
self.compute_offset(now_mod, now_elem, cursor_line, drop_comments=True)
+ 1
)
new_out_tks = TokenSeq()
for k</s>
===========above chunk 1===========
# module: coeditor.api
@dataclass
class EditPredictionService:
def apply_edit_to_elem(
self,
file: Path,
now_mod: PythonModule,
now_elem: PythonElem,
cursor_line: int,
out_tks: TokenSeq,
) -> tuple[Modified[str], str]:
# offset: -2
mname = now_elem.path.module
path_s = file.relative_to(self.project).as_posix()
prev_mod = self.prev_cache.cached(
(path_s, False),
self.config.get_index_stamp(self.project, path_s),
lambda: PythonModule.from_cst(
cst.parse</s>
===========unchanged ref 0===========
at: coeditor.api.ChangeDetectionConfig
drop_comments: DropComment = True
get_index_stamp(project_root: Path, path_s: str)
at: coeditor.api.EditPredictionService
compute_offset(now_mod: PythonModule, now_elem: PythonElem, line: int, drop_comments: bool)
preview_changes(change: Modified[str], respect_lines: int) -> str
at: coeditor.api.EditPredictionService.__init__
self.project = project
self.config = config
self.prev_cache = TimedCache[
tuple[ModuleName, DropComment], PythonModule, str
]()
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.common.TimedCache
cached(key: T1, stamp: TStamp, f: Callable[[], T2]) -> T2
at: coeditor.encoding
get_extra_id(i: int) -> Token
extra_id_to_number(tk: Token) -> int
change_to_tokens(change: Change[str]) -> TokenSeq
at: logging
info(msg: Any, *args: Any, exc_info: _ExcInfoType=..., stack_info: bool=..., extra: Optional[Dict[str, Any]]=..., **kwargs: Any) -> None
at: pathlib
Path()
at: pathlib.PurePath
as_posix() -> str
relative_to(*other: Union[str, _PathLike]) -> _P
at: spot.data
output_ids_as_seqs(output_ids: Iterable[Token]) -> dict[Token, TokenSeq]
at: spot.static_analysis
PythonElem = PythonFunction | PythonVariable
===========unchanged ref 1===========
PythonModule(functions: list[PythonFunction], global_vars: list[PythonVariable], classes: list[PythonClass], name: ModuleName, imported_modules: set[ModuleName], defined_symbols: dict[str, ProjectPath], tree: cst.Module, location_map: dict[cst.CSTNode, CodeRange], elem2pos: dict[ElemPath, CodeRange], removed_comments: list[cst.CSTNode])
CommentRemover(src_map: Mapping[cst.CSTNode, CodeRange] | None=None)
at: spot.static_analysis.CommentRemover
line_map(post_node: cst.CSTNode) -> dict[int, int]
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
at: spot.static_analysis.PythonFunction
path: ProjectPath
at: spot.static_analysis.PythonModule
location_map: dict[cst.CSTNode, CodeRange]
from_cst(module: cst.Module, name: str, drop_comments: bool) -> "PythonModule"
at: spot.static_analysis.PythonVariable
path: ProjectPath
at: spot.utils
add_line_numbers(code: str)
at: typing.Mapping
get(key: _KT) -> Optional[_VT_co]
get(key: _KT, default: Union[_VT_co, _T]) -> Union[_VT_co, _T]
===========changed ref 0===========
# module: coeditor.dataset
+ time_limit_per_commit = 10.0
- time_limit_per_commit = 5.0
===========changed ref 1===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
@classmethod
def eval_default(cls) -> Self:
return BatchArgs(
max_total_ref_tks=512 * 32,
max_queries=32,
- max_ref_dropout=0.0,
shuffle_extra_ids=False,
)
===========changed ref 2===========
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
+ all_edits: list[TkC3Problem]
- all_edits: list[BasicTkQueryEdit]
batch_args: BatchArgs
shuffle: bool
desc: str
tqdm_args: dict | None = None
===========changed ref 3===========
# module: coeditor.retrieval_model
@dataclass
class _BatchSampler:
def estimate_n_batches(self) -> int:
+ batches = tk_edits_to_batches(self.all_edits, self.batch_args, silent=True)
- batches = query_edits_to_batches(self.all_edits, self.batch_args, silent=True)
return len(batches)
===========changed ref 4===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
max_output_tks: int = 256
max_query_tks: int = 512
min_queires: int = 1
max_queries: int = 8
max_ref_tks: int = 512
max_total_ref_tks: int = 512 * 16
- max_ref_dropout: float = 1.0
shuffle_extra_ids: bool = True
- use_only_modified: bool = True
|
coeditor.retrieval_model/BatchArgs.eval_default
|
Modified
|
temp-1
|
18bed1bae904b1d484977a957aa4477ec2128c00
|
Training parameter adjustment.
|
<0>:<add> max_total_ref_tks=512 * 24,
|
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
@classmethod
def eval_default(cls) -> Self:
return BatchArgs(
- max_total_ref_tks=512 * 32,
<0> max_queries=32,
shuffle_extra_ids=False,
)
|
===========unchanged ref 0===========
at: coeditor.retrieval_model
BatchArgs(max_output_tks: int=256, max_query_tks: int=512, min_queires: int=1, max_queries: int=8, max_ref_tks: int=512, max_total_ref_tks: int=512 * 12, shuffle_extra_ids: bool=True)
at: coeditor.retrieval_model.BatchArgs
max_output_tks: int = 256
max_query_tks: int = 512
min_queires: int = 1
max_queries: int = 8
max_ref_tks: int = 512
max_total_ref_tks: int = 512 * 12
shuffle_extra_ids: bool = True
===========changed ref 0===========
# module: coeditor.retrieval_model
@dataclass
class BatchArgs:
max_output_tks: int = 256
max_query_tks: int = 512
min_queires: int = 1
max_queries: int = 8
max_ref_tks: int = 512
+ max_total_ref_tks: int = 512 * 12
- max_total_ref_tks: int = 512 * 16
shuffle_extra_ids: bool = True
|
scripts.coeditor.train_retrieval_model/train_model
|
Modified
|
temp-1
|
18bed1bae904b1d484977a957aa4477ec2128c00
|
Training parameter adjustment.
|
<0>:<add> warmup_bargs.max_total_ref_tks //= 3
|
# module: scripts.coeditor.train_retrieval_model
def train_model(
dataset_name="medium",
model_variant="-sig-analysis-post_usees",
encoder: C3EditEncoder = C3EditEncoder(),
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
<s>, reinit_weights=train_args.reinit_weights
)
else:
model = RetrievalEditorModel.load(get_model_dir() / model_name)
if os.getenv("CUDA_VISIBLE_DEVICES") is None:
warnings.warn(
"CUDA_VISIBLE_DEVICES not set, using 0. Note that "
"the Huggingface Trainer will use all visible GPUs for training."
)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
if not eval_only:
with timed_action("Warm-up Training"):
warmup_bargs = copy.deepcopy(batch_args)
- warmup_bargs.max_total_ref_tks //= 4
<0> warmup_bargs.min_queires *= 4
warmup_bargs.max_queries *= 2
warmup_targs = copy.deepcopy(train_args)
warmup_targs.learning_rate *= 4
warmup_targs.max_train_epochs = 1
all_edits = datasets["train"].all_edits()
warmup_edits = random_subset(all_edits, len(all_edits) // 4)
model.train_on_data(
model_name,
TokenizedEditDataset.from_edits(warmup_edits),
datasets["valid"],
warmup_targs,
batch_args=warmup_bargs,
eval_batch_args=test_batch_args,
)
with timed_action("F</s>
|
===========above chunk 0===========
# module: scripts.coeditor.train_retrieval_model
def train_model(
dataset_name="medium",
model_variant="-sig-analysis-post_usees",
encoder: C3EditEncoder = C3EditEncoder(),
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: -1
# model_variant = "-file"
model_name = f"coeditor-{dataset_name}"
model_name += model_variant
dec_args = DecodingArgs()
if train_args.quicktest:
model_name = "quicktest-" + model_name
if not eval_only:
check_save_dir(model_name)
datasets = make_or_load_datasets(dataset_name, encoder, recreate_data=recreate_data)
config_dict = {
k: get_modified_args(v)
for k, v in {
"data_args": batch_args,
"train_args": train_args,
"dec_args": dec_args,
}.items()
}
project = "Coeditor" if not train_args.quicktest else "Coeditor-quicktest"
wandb.init(dir="..", project=project, name=model_name, config=config_dict)
if train_args.quicktest:
print("Using fewer data for quick test.")
n_quick_exs = 20
for name, dataset in datasets.items():
datasets[name] = TokenizedEditDataset.from_edits(
dataset.all_edits()[:n_quick_exs]
)
if not eval_only:
model = RetrievalEditorModel.from_code_t5(
"base", reuse_embed=True, reinit_weights=train_args.reinit_weights
)
else:
model = RetrievalEditorModel.</s>
===========below chunk 0===========
# module: scripts.coeditor.train_retrieval_model
def train_model(
dataset_name="medium",
model_variant="-sig-analysis-post_usees",
encoder: C3EditEncoder = C3EditEncoder(),
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: 1
<s>up_bargs,
eval_batch_args=test_batch_args,
)
with timed_action("Fine-tune Training"):
model.train_on_data(
model_name,
datasets["train"],
datasets["valid"],
train_args,
batch_args=batch_args,
eval_batch_args=test_batch_args,
)
model.to("cuda")
with timed_action("Loss Evaluation"):
eval_result = model.eval_loss_on_data(datasets["test"], test_batch_args)
eval_dict = {f"test/{k}": v.average() for k, v in eval_result.items()}
wandb.log(eval_dict)
max_saved_samples = 300
with timed_action("Accuracy Evaluation"):
dec_result = model.predict_on_data(datasets["test"], test_batch_args, dec_args)
pickle_dump(get_model_dir() / model_name / "dec_result.pkl", dec_result)
exact_acc, exact_correct_map = dec_result.exact_match_accuracy()
wandb.log({"test/exact-acc": exact_acc.average()})
out_dir = get_model_dir() / model_name / "exact_match_samples"
dec_result.save_examples_to_dir(
out_dir, random_subset(exact_correct_map, max_saved_samples)
)
print("Exact-match samples saved</s>
===========below chunk 1===========
# module: scripts.coeditor.train_retrieval_model
def train_model(
dataset_name="medium",
model_variant="-sig-analysis-post_usees",
encoder: C3EditEncoder = C3EditEncoder(),
batch_args=BatchArgs.train_default(),
test_batch_args=BatchArgs.eval_default(),
train_args=TrainingArgs(),
recreate_data: bool = False,
eval_only: bool = False,
):
# offset: 2
<s>dir, random_subset(exact_correct_map, max_saved_samples)
)
print("Exact-match samples saved to:", out_dir)
return model
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
timed_action(name: str, silent: bool=False)
pickle_dump(file: Path, obj: Any)
get_modified_args(instance, flatten: bool=False) -> dict[str, Any] | None
at: coeditor.common
get_model_dir(trained=True) -> Path
random_subset(all: Mapping[T1, T2], n: int, rng: random.Random | int | None=None) -> dict[T1, T2]
random_subset(all: Sequence[T1], n: int, rng: random.Random | int | None=None) -> list[T1]
at: coeditor.dataset
TokenizedEditDataset(_edits: list[TEdit])
at: coeditor.dataset.TokenizedEditDataset
_edits: list[TEdit]
from_edits(edits: Iterable[TEdit]) -> "TokenizedEditDataset[TEdit]"
at: coeditor.model
DecodingArgs(max_output_tks: int=512, do_sample: bool=False, top_p: float=0.9, num_beams: Optional[int]=1, length_penalty: float=0.0, marginalize_samples: int=1)
at: copy
deepcopy(x: _T, memo: Optional[Dict[int, Any]]=..., _nil: Any=...) -> _T
at: os
environ = _createenviron()
getenv(key: str, default: _T) -> Union[str, _T]
getenv(key: str) -> Optional[str]
at: train_model
check_save_dir(model_name: str) -> None
at: wandb
init = wandb_sdk.init
===========unchanged ref 1===========
log = _preinit.PreInitCallable("wandb.log", wandb_sdk.wandb_run.Run.log)
===========changed ref 0===========
+ # module: tests.coeditor.testcases.example
+ @dataclass
+ class Weight:
+ value: float
+
===========changed ref 1===========
+ # module: tests.coeditor.testcases.example
+ @dataclass
+ class Foo:
+ value: float
+
===========changed ref 2===========
+ # module: tests.coeditor.testcases.example
+ @dataclass
+ class Foo:
+ def bar(self):
+ return Weight(self.value)
+
===========changed ref 3===========
+ # module: tests.coeditor.testcases.example
+ @dataclass
+ class Weight:
+ def go(self, other):
+ return self.value + other.value
+
===========changed ref 4===========
+ # module: tests.coeditor.testcases.example
+ foo(Foo(1), Foo(2))
+
===========changed ref 5===========
+ # module: tests.coeditor.testcases.example
+ def foo(x, y):
+ return x.bar().go(y.bar())
+
|
coeditor.ctx_change_encoder/C3ProblemTokenizer.tokenize_problem
|
Modified
|
temp-1
|
f0a4cab073846109c9b3083f25a99bd2bde88b3a
|
Make C3Problem more suitable for serialization.
|
<0>:<add> path=span.headers[-1].path,
|
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
<s>chunk_overlap,
)
above_chunks = [
(f"above chunk {i}", chunk) for i, chunk in enumerate(above_chunks)
]
below_chunks = [
(f"below chunk {i}", chunk) for i, chunk in enumerate(below_chunks)
]
all_refs = above_chunks + below_chunks + named_references
size_sum = 0
kept_refs = list[tuple[str, TokenSeq]]()
for (name, chunk) in all_refs:
if size_sum + len(chunk) <= self.max_total_ref_tks:
size_sum += len(chunk)
kept_refs.append((name, chunk))
return TkC3Problem(
scope_tks + chunk_input,
chunk_output,
- path=span.parent_scopes[-1].earlier().path,
<0> change_type=span.change.map(lambda _: None),
named_references=kept_refs,
src_info=problem.src_info,
)
problems = list[TkC3Problem]()
for l in range(len(tk_delta.deltas) + 1):
finished = l == len(tk_delta.deltas)
input_growth = len(origin_lines[l]) + 2 if l < len(origin_lines) else 1
if (
finished
or chunk_lines >= self.max_lines_to_edit
or len(chunk_input) + input_growth > input_limit
):
if has_change(chunk_output):
problems.append(get_problem(chunk_input, chunk_output))
if len(problems) >= self.max_chunks_per_elem:
break
if finished:
break
chunk_main_input = join_list</s>
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: -1
<s> get_problem(chunk_input, chunk_output):
# try move some prev_change_tks into the input
above_tks = prev_change_tks
below_tks = join_list(origin_lines[l:], Newline_id)
chunk_input, above_tks, below_tks = self._inline_some_context(
chunk_input, above_tks, below_tks, input_limit
)
# limit the input size if it's too long (can happen for later chunks)
chunk_input = truncate_section(chunk_input, TruncateAt.Right, input_limit)
chunk_output = truncate_output_tks(chunk_input, chunk_output)
chunk_output = truncate_section(
chunk_output, TruncateAt.Right, self.max_output_tks, add_bos=False
)
above_chunks = break_into_chunks(
above_tks,
+ lambda i: self._encode_headers(span.headers, -1 - i),
- lambda i: self._encode_parent_scopes(span.parent_scopes, -1 - i),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
right_to_left=True,
)
if finished:
below_chunks = []
else:
below_chunks = break_into_chunks(
below_tks,
+ lambda i: self._encode_headers(span.headers, i + 1),
- lambda i: self._encode_parent_scopes(span.parent_scopes, i + 1),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
)
above_chunks = [
(f"above chunk {i}", chunk) for i,</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: -2
span = problem.span
named_references = list[tuple[str, TokenSeq]]()
# compute the references that are relevant to this span
relevant_chunks = self._group_encode_changed_refs(problem.relevant_changes)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"changed ref {i}", chunk))
relevant_chunks = self._group_encode_unchanged_refs(problem.relevant_unchanged)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"unchanged ref {i}", chunk))
diffs = change_to_line_diffs(span.change)
original, delta = line_diffs_to_original_delta(diffs)
origin_lines = split_list(encode_basic(original), Newline_id)
tk_delta = delta.to_tk_delta()
chunk_id = 0
chunk_start_l = 0
+ scope_tks = self._encode_headers(span.headers, 0)
- scope_tks = self._encode_parent_scopes(span.parent_scopes, 0)
chunk_input = TokenSeq()
input_limit = self.max_query_tks - len(scope_tks)
chunk_lines = 0
chunk_output = TokenSeq()
prev_change_tks = TokenSeq()
def get_problem(chunk_input, chunk_output):
# try move some prev_change_tks into the input
above</s>
===========below chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: 1
<s>chunks_per_elem:
break
if finished:
break
chunk_main_input = join_list(origin_lines[chunk_start_l:l], Newline_id)
chunk_main_delta = tk_delta.for_input_range((chunk_start_l, l))
chunk_main_change = chunk_main_delta.to_change_tks(chunk_main_input)
prev_change_tks.extend(chunk_main_change)
prev_change_tks.append(Newline_id)
chunk_id += 1
chunk_input = TokenSeq()
chunk_lines = 0
chunk_output = TokenSeq()
chunk_start_l = l
chunk_input.append(get_extra_id(chunk_lines))
if l < len(origin_lines):
chunk_input.extend(origin_lines[l])
chunk_input.append(Newline_id)
line_change = join_list(tk_delta.deltas[l], Newline_id)
chunk_output.append(get_extra_id(chunk_lines))
chunk_output.extend(line_change)
if line_change and line_change[-1] != Del_id:
chunk_output.append(Newline_id)
chunk_lines += 1
return problems
===========unchanged ref 0===========
at: cachetools.lru
LRUCache(maxsize: int, getsizeof: Optional[Callable[[_VT], int]]=...)
at: coeditor.common
TokenSeq = list[Token]
split_list(lst: list[T1], sep: T1) -> list[list[T1]]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder
C3Problem(span: ChangedCodeSpan, relevant_changes: Sequence[ChangedCodeSpan], relevant_unchanged: Sequence[ChangedCodeSpan], src_info: dict[str, Any])
TkC3Problem(input_tks: TokenSeq, output_tks: TokenSeq, path: ProjectPath, change_type: Change[None], named_references: Sequence[tuple[str, TokenSeq]], src_info: dict[str, Any])
at: coeditor.ctx_change_encoder.C3Problem
span: ChangedCodeSpan
relevant_changes: Sequence[ChangedCodeSpan]
relevant_unchanged: Sequence[ChangedCodeSpan]
src_info: dict[str, Any]
at: coeditor.ctx_change_encoder.C3ProblemTokenizer
_encode_headers(scope_changes: Sequence[ChangedHeader], offset: int) -> TokenSeq
_encode_headers(self, scope_changes: Sequence[ChangedHeader], offset: int) -> TokenSeq
_inline_some_context(self, input: TokenSeq, above_ctx: TokenSeq, below_ctx: TokenSeq, size_limit: int) -> tuple[TokenSeq, TokenSeq, TokenSeq]
_inline_some_context(input: TokenSeq, above_ctx: TokenSeq, below_ctx: TokenSeq, size_limit: int) -> tuple[TokenSeq, TokenSeq, TokenSeq]
_group_encode_unchanged_refs(elems: Sequence[ChangedCodeSpan]) -> Sequence[TokenSeq]
_group_encode_changed_refs(changes: Sequence[ChangedCodeSpan]) -> Sequence[TokenSeq]
|
coeditor.ctx_change_encoder/C3ProblemTokenizer._encode_change
|
Modified
|
temp-1
|
f0a4cab073846109c9b3083f25a99bd2bde88b3a
|
Make C3Problem more suitable for serialization.
|
<0>:<add> self._change_cache[change] = change_tks
|
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def _encode_change(self, change: Change[str]) -> TokenSeq:
- if (key := _ObjId(id(change))) in self._id_cache:
- return self._id_cache[key]
- change = change.map(lambda s: s.strip("\n"))
+ if (change_tks := self._change_cache.get(change)) is None:
+ change_tks = change_to_tokens(change)
- change_tks = change_to_tokens(change)
- self._id_cache[key] = change_tks
<0> return change_tks
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder
ChangedHeader(change: Change[str], type: str, line_range: LineRange, path: ProjectPath)
at: coeditor.ctx_change_encoder.C3ProblemTokenizer
max_scope_tks: int = 128
at: coeditor.ctx_change_encoder.C3ProblemTokenizer._encode_header_change
hchange = ch.change.map(lambda s: s.strip("\n"))
tks = truncate_section(
change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
)
at: coeditor.encoding
change_to_tokens(change: Change[str]) -> TokenSeq
TruncateAt()
at: coeditor.encoding.TruncateAt
Left = 0
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedHeader:
+ """Represents the changes made to a header.
+ This format does not store parent syntax nodes and is more suitable for serialization.
+ """
+
+ change: Change[str]
+ # below are pre-edit attributes
+ type: str
+ line_range: LineRange
+ path: ProjectPath
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
+ def _encode_header_change(self, ch: ChangedHeader) -> TokenSeq:
+ hchange = ch.change.map(lambda s: s.strip("\n"))
+ tks = truncate_section(
+ change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
+ )
+ return tks
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
- def _encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq:
- if (key := _ObjId(id(c))) in self._scope_cache:
- return self._scope_cache[key]
- hchange = c.map(lambda s: s.header_code.strip("\n"))
- tks = truncate_section(
- change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
- )
- self._scope_cache[key] = tks
- return tks
-
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
- def _encode_parent_scopes(
- self, scope_changes: Sequence[Change[ChangeScope]], offset: int
- ) -> TokenSeq:
- scope_tks = join_list(
- (self._encode_scope_change(c) for c in scope_changes), Newline_id
- )
- if offset != 0:
- scope_tks.extend(encode_basic(f"\n# offset: {offset}\n"))
- else:
- scope_tks.append(Newline_id)
- scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
- return scope_tks
-
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def __post_init__(self):
- self._id_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
+ self._change_cache = LRUCache[Change[str], TokenSeq](maxsize=5000)
- self._scope_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
- self._value_cache = FIFOCache[Any, Sequence[TokenSeq]](maxsize=1000)
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ VERSION = "2.0"
- VERSION = "1.2"
+ # change spans with more than this many lines will be ignored
+ max_span_lines: int = 500
===========changed ref 6===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedHeader:
+ def __repr__(self) -> str:
+ return (
+ f"ChangedHeader(path={self.path}, range={self.line_range}, "
+ f"change={self.change})"
+ )
+
===========changed ref 7===========
# module: coeditor.ctx_change_encoder
@dataclass(frozen=True)
class PyDefinition:
- @staticmethod
- def from_scope(scope: ChangeScope) -> "PyDefinition":
- path = scope.path
- full_name = PyFullName(f"{path.module}.{path.path}")
- start_pos = scope.header_line_range[0], 0
- end_pos = scope.header_line_range[1], 0
- return PyDefinition(full_name, start_pos, end_pos)
-
===========changed ref 8===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedCodeSpan:
+ """Represents the changes made to a span of code.
+ This format does not store parent syntax nodes and is more suitable for serialization.
+ """
+
+ headers: Sequence[ChangedHeader]
+ change: Change[str]
+ # below are pre-edit attributes
+ line_range: LineRange
+ module: ModuleName
+
===========changed ref 9===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
- @dataclass
class C3Problem:
"Contextual code change prediction problem."
+ span: ChangedCodeSpan
- span: ChangedSpan
# most relevant to least relevant
+ relevant_changes: Sequence[ChangedCodeSpan]
- relevant_changes: list[ChangedSpan]
# most relevant to least relevant
+ relevant_unchanged: Sequence[ChangedCodeSpan]
- relevant_unchanged: list[ChangedSpan]
# some optional information about how the problem was generated
src_info: dict[str, Any]
===========changed ref 10===========
# module: coeditor.history
@dataclass(frozen=True)
class Modified(_ChangeBase[E1]):
+ def __repr__(self):
+ if self.before == self.after:
+ return f"Modified(before=after={repr(self.before)})"
+ else:
+ return f"Modified(before={repr(self.before)}, after={repr(self.after)})"
+
===========changed ref 11===========
<s>.code_change
def edits_from_commit_history(
project_dir: Path,
history: Sequence[CommitInfo],
tempdir: Path,
change_processor: ProjectChangeProcessor[TProb] = NoProcessing(),
- edit_encoder: Callable[[TProb], Iterable[TEnc]] = lambda x: [x],
ignore_dirs=DefaultIgnoreDirs,
silent: bool = False,
time_limit: float | None = None,
+ ) -> Sequence[TProb]:
- ) -> Sequence[TEnc]:
"""Incrementally compute the edits to a project from the git history.
Note that this will change the file states in the project directory, so
you should make a copy of the project before calling this function.
"""
tempdir = tempdir.resolve()
if tempdir.exists():
raise FileExistsError(f"Workdir '{tempdir}' already exists.")
use_fast_parser = jedi.settings.fast_parser
tempdir.mkdir(parents=True, exist_ok=False)
try:
run_command(
["cp", "-r", str(project_dir / ".git"), str(tempdir)],
cwd=project_dir.parent,
)
return _edits_from_commit_history(
tempdir,
history,
change_processor,
- edit_encoder,
ignore_dirs,
silent,
time_limit=time_limit,
)
finally:
shutil.rmtree(tempdir)
jedi.settings.fast_parser = use_fast_parser
|
coeditor.ctx_change_encoder/C3ProblemTokenizer._group_encode_changed_refs
|
Modified
|
temp-1
|
f0a4cab073846109c9b3083f25a99bd2bde88b3a
|
Make C3Problem more suitable for serialization.
|
<0>:<add> lambda i: self._encode_headers(mod_change, i),
|
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def _group_encode_changed_refs(
+ self, changes: Sequence[ChangedCodeSpan]
- self, changes: Sequence[ChangedSpan]
) -> Sequence[TokenSeq]:
<s> header_diff = list[ChangedHeader]()
- scope_diff = []
+ for i, h in enumerate(c.headers):
- for i, s in enumerate(c.parent_scopes):
- if (
- i >= len(last_scope)
- or s.earlier().path != last_scope[i].earlier().path
- ):
+ if i >= len(last_scope) or h.path != last_scope[i].path:
+ header_diff.append(h)
- scope_diff.append(s)
+ if header_diff:
- if scope_diff:
+ header_tks = self._encode_headers(header_diff, 0)
- header_tks = self._encode_parent_scopes(scope_diff, 0)
file_tks.extend(header_tks)
+ body_tks = self._encode_change(c.change.map(lambda c: c.strip("\n")))
- body_tks = self._encode_change(c.change)
file_tks.extend(body_tks)
file_tks.append(Newline_id)
file_tks.append(Newline_id)
+ last_scope = c.headers
- last_scope = c.parent_scopes
+ mod_change = change_group[0].headers[:1]
- mod_change = change_group[0].parent_scopes[:1]
mod_chunks = break_into_chunks(
file_tks,
- lambda i: self._encode_parent_scopes(mod_change, i),
<0> self.max_ref_tks,
overlap=self.ref_chunk_overlap,
)
all_chunks.extend(mod_chunks)
return all_chunks
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def _group_encode_changed_refs(
+ self, changes: Sequence[ChangedCodeSpan]
- self, changes: Sequence[ChangedSpan]
) -> Sequence[TokenSeq]:
# offset: -1
+ module2changes = groupby(changes, lambda c: c.module)
- module2changes = groupby(changes, lambda c: c.path.module)
all_chunks = list[TokenSeq]()
for change_group in module2changes.values():
change_group.sort(key=lambda c: c.line_range[0])
file_tks = TokenSeq()
# we'll add module as the chunk header, so we start within the module
+ last_scope = change_group[0].headers[:1]
- last_scope = change_group[0].parent_scopes[:1]
for c in change_group:
+ header_diff = list[ChangedHeader]()
- scope_diff = []
+ for i, h in enumerate(c.</s>
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
at: coeditor.ctx_change_encoder.C3ProblemTokenizer
max_scope_tks: int = 128
ref_chunk_overlap: int = 32
at: coeditor.encoding
Newline_id = get_tk_id("\n")
TruncateAt()
truncate_section(sec: TokenSeq, direction: TruncateAt.Value, limit: int, add_bos: bool=True, inplace: bool=False) -> TokenSeq
truncate_sections(total_limit: int, *sections: tuple[TokenSeq, TruncateAt.Value], add_bos: bool, inplace: bool=False) -> tuple[TokenSeq, ...]
at: coeditor.encoding.TruncateAt
Left = 0
Right = 1
===========changed ref 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
+ def _encode_header_change(self, ch: ChangedHeader) -> TokenSeq:
+ hchange = ch.change.map(lambda s: s.strip("\n"))
+ tks = truncate_section(
+ change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
+ )
+ return tks
+
===========changed ref 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
+ def _encode_headers(
+ self, scope_changes: Sequence[ChangedHeader], offset: int
+ ) -> TokenSeq:
+ scope_tks = join_list(
+ (self._encode_header_change(c) for c in scope_changes), Newline_id
+ )
+ if offset != 0:
+ scope_tks.extend(encode_basic(f"\n# offset: {offset}\n"))
+ else:
+ scope_tks.append(Newline_id)
+ scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
+ return scope_tks
+
===========changed ref 2===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def _encode_change(self, change: Change[str]) -> TokenSeq:
- if (key := _ObjId(id(change))) in self._id_cache:
- return self._id_cache[key]
- change = change.map(lambda s: s.strip("\n"))
+ if (change_tks := self._change_cache.get(change)) is None:
+ change_tks = change_to_tokens(change)
- change_tks = change_to_tokens(change)
+ self._change_cache[change] = change_tks
- self._id_cache[key] = change_tks
return change_tks
===========changed ref 3===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
- def _encode_scope_change(self, c: Change[ChangeScope]) -> TokenSeq:
- if (key := _ObjId(id(c))) in self._scope_cache:
- return self._scope_cache[key]
- hchange = c.map(lambda s: s.header_code.strip("\n"))
- tks = truncate_section(
- change_to_tokens(hchange), TruncateAt.Left, self.max_scope_tks
- )
- self._scope_cache[key] = tks
- return tks
-
===========changed ref 4===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
- def _encode_parent_scopes(
- self, scope_changes: Sequence[Change[ChangeScope]], offset: int
- ) -> TokenSeq:
- scope_tks = join_list(
- (self._encode_scope_change(c) for c in scope_changes), Newline_id
- )
- if offset != 0:
- scope_tks.extend(encode_basic(f"\n# offset: {offset}\n"))
- else:
- scope_tks.append(Newline_id)
- scope_tks = truncate_section(scope_tks, TruncateAt.Left, self.max_scope_tks)
- return scope_tks
-
===========changed ref 5===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def __post_init__(self):
- self._id_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
+ self._change_cache = LRUCache[Change[str], TokenSeq](maxsize=5000)
- self._scope_cache = FIFOCache[_ObjId, TokenSeq](maxsize=1000)
- self._value_cache = FIFOCache[Any, Sequence[TokenSeq]](maxsize=1000)
===========changed ref 6===========
# module: coeditor.ctx_change_encoder
class C3ProblemGenerator(ProjectChangeProcessor[C3Problem]):
+ VERSION = "2.0"
- VERSION = "1.2"
+ # change spans with more than this many lines will be ignored
+ max_span_lines: int = 500
===========changed ref 7===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedHeader:
+ def __repr__(self) -> str:
+ return (
+ f"ChangedHeader(path={self.path}, range={self.line_range}, "
+ f"change={self.change})"
+ )
+
===========changed ref 8===========
# module: coeditor.ctx_change_encoder
@dataclass(frozen=True)
class PyDefinition:
- @staticmethod
- def from_scope(scope: ChangeScope) -> "PyDefinition":
- path = scope.path
- full_name = PyFullName(f"{path.module}.{path.path}")
- start_pos = scope.header_line_range[0], 0
- end_pos = scope.header_line_range[1], 0
- return PyDefinition(full_name, start_pos, end_pos)
-
===========changed ref 9===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedCodeSpan:
+ """Represents the changes made to a span of code.
+ This format does not store parent syntax nodes and is more suitable for serialization.
+ """
+
+ headers: Sequence[ChangedHeader]
+ change: Change[str]
+ # below are pre-edit attributes
+ line_range: LineRange
+ module: ModuleName
+
===========changed ref 10===========
# module: coeditor.ctx_change_encoder
+ @dataclass(frozen=True)
+ class ChangedHeader:
+ """Represents the changes made to a header.
+ This format does not store parent syntax nodes and is more suitable for serialization.
+ """
+
+ change: Change[str]
+ # below are pre-edit attributes
+ type: str
+ line_range: LineRange
+ path: ProjectPath
+
|
spot.utils/pmap
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> desc = "pmap: " + f.__name__
|
# module: spot.utils
def pmap(
f: Callable[..., T1],
*f_args: Any,
+ desc: str | None = None,
- desc: str = "parallel map",
key_args: Mapping[str, Any] | None = None,
max_workers: int | None = None,
chunksize: int | None = None,
tqdm_args: Mapping[str, Any] | None = None,
) -> list[T1]:
"""
Parallel map with progress displaying.
"""
n = len(f_args[0])
assert_eq(n, *(len(xs) for xs in f_args))
tqdm_args = dict(tqdm_args) if tqdm_args else {}
tqdm_args.setdefault("smoothing", 0.0)
+
+ if desc is None:
<0>
if key_args is None:
key_args = {}
if max_workers is None:
max_workers = DefaultWorkers
if max_workers <= 1:
outs = list[T1]()
for i in tqdm(range(n), desc=desc, **tqdm_args):
outs.append(f(*(a[i] for a in f_args), **key_args))
return outs
if chunksize is None:
chunksize = max(1, n // (50 * max_workers))
tag_f = _TaggedFunc(f, key_args)
arg_tuples = zip(range(n), *f_args)
with (
multiprocessing.Pool(max_workers) as pool,
tqdm(total=n, desc=desc, **tqdm_args) as pbar,
):
results = dict[int, T1]()
for i, r in pool.imap_unordered(tag_f, arg_tuples, chunksize=chunksize):
results[i] = r
pbar.update()
return [results[i] for i in range(n)]
|
===========unchanged ref 0===========
at: multiprocessing
Pool(processes: Optional[int]=..., initializer: Optional[Callable[..., Any]]=..., initargs: Iterable[Any]=..., maxtasksperchild: Optional[int]=...) -> pool.Pool
at: spot.utils
T1 = TypeVar("T1")
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
_TaggedFunc(f: Callable[..., T1], key_args: Mapping[str, Any])
assert_eq(x: T1, *xs: T1, extra_message: Callable[[], str]=lambda: "") -> None
at: tqdm.std
tqdm(iterable=None, desc=None, total=None, leave=True, file=None, ncols=None, mininterval=0.1, maxinterval=10.0, miniters=None, ascii=None, disable=False, unit='it', unit_scale=False, dynamic_ncols=False, smoothing=0.3, bar_format=None, initial=0, position=None, postfix=None, unit_divisor=1000, write_bytes=False, lock_args=None, nrows=None, colour=None, delay=0, gui=False, **kwargs)
at: typing
Callable = _CallableType(collections.abc.Callable, 2)
Mapping = _alias(collections.abc.Mapping, 2)
|
coeditor.dataset/TokenizedEditDataset.__repr__
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> return f"TokenizedEditDataset(n_edits={n_edits})"
|
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def __repr__(self) -> str:
- n_projects = len(self.project2edits)
- n_edits = sum(len(edits) for edits in self.project2edits.values())
+ n_edits = len(self.all_edits())
- return f"TokenizedEditDataset(n_projects={n_projects}, n_edits={n_edits})"
<0>
|
===========unchanged ref 0===========
at: coeditor.dataset.TokenizedEditDataset
_edits: list[TEdit]
all_edits() -> list[TEdit]
all_edits(self) -> list[TEdit]
===========changed ref 0===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
+ _edits: list[TEdit]
- project2edits: dict[Path, list[TEdit]]
===========changed ref 1===========
# module: spot.utils
def pmap(
f: Callable[..., T1],
*f_args: Any,
+ desc: str | None = None,
- desc: str = "parallel map",
key_args: Mapping[str, Any] | None = None,
max_workers: int | None = None,
chunksize: int | None = None,
tqdm_args: Mapping[str, Any] | None = None,
) -> list[T1]:
"""
Parallel map with progress displaying.
"""
n = len(f_args[0])
assert_eq(n, *(len(xs) for xs in f_args))
tqdm_args = dict(tqdm_args) if tqdm_args else {}
tqdm_args.setdefault("smoothing", 0.0)
+
+ if desc is None:
+ desc = "pmap: " + f.__name__
if key_args is None:
key_args = {}
if max_workers is None:
max_workers = DefaultWorkers
if max_workers <= 1:
outs = list[T1]()
for i in tqdm(range(n), desc=desc, **tqdm_args):
outs.append(f(*(a[i] for a in f_args), **key_args))
return outs
if chunksize is None:
chunksize = max(1, n // (50 * max_workers))
tag_f = _TaggedFunc(f, key_args)
arg_tuples = zip(range(n), *f_args)
with (
multiprocessing.Pool(max_workers) as pool,
tqdm(total=n, desc=desc, **tqdm_args) as pbar,
):
results = dict[int, T1]()
for i, r in pool.imap_unordered(tag_f, arg_tuples, chunksize=chunksize):
results[i] = r
pbar.update()
return [results[i] for i in range(n)]
|
coeditor.dataset/TokenizedEditDataset.all_edits
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> return self._edits
|
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def all_edits(self) -> list[TEdit]:
- return join_list(self.project2edits.values())
<0>
|
===========unchanged ref 0===========
at: dataclasses
field(*, init: bool=..., repr: bool=..., hash: Optional[bool]=..., compare: bool=..., metadata: Optional[Mapping[str, Any]]=...) -> Any
field(*, default_factory: Callable[[], _T], init: bool=..., repr: bool=..., hash: Optional[bool]=..., compare: bool=..., metadata: Optional[Mapping[str, Any]]=...) -> _T
field(*, default: _T, init: bool=..., repr: bool=..., hash: Optional[bool]=..., compare: bool=..., metadata: Optional[Mapping[str, Any]]=...) -> _T
===========changed ref 0===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def subset(self, repos: Iterable[Path]) -> "TokenizedEditDataset":
- return TokenizedEditDataset({repo: self.project2edits[repo] for repo in repos})
-
===========changed ref 1===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
+ _edits: list[TEdit]
- project2edits: dict[Path, list[TEdit]]
===========changed ref 2===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def map(self, f: Callable[[TEdit], TEdit]) -> "TokenizedEditDataset[TEdit]":
- repos = tqdm(self.project2edits.items(), desc="transforming dataset")
- return TokenizedEditDataset(
- {repo: [f(e) for e in edits] for repo, edits in repos}
- )
-
===========changed ref 3===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def __repr__(self) -> str:
- n_projects = len(self.project2edits)
- n_edits = sum(len(edits) for edits in self.project2edits.values())
+ n_edits = len(self.all_edits())
+ return f"TokenizedEditDataset(n_edits={n_edits})"
- return f"TokenizedEditDataset(n_projects={n_projects}, n_edits={n_edits})"
===========changed ref 4===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def overall_stats(self) -> dict:
all_edits = self.all_edits()
n_added = sum(isinstance(e.change_type, Added) for e in all_edits)
basic_stats = {
- "n_projects": len(self.project2edits),
"n_edits": len(all_edits),
"n_additions": n_added,
}
extra_stats = dict[str, list]()
for e in all_edits:
for k, v in e.stats().items():
if k in extra_stats:
extra_stats[k].append(v)
else:
extra_stats[k] = [v]
return basic_stats | {k: scalar_stats(v) for k, v in extra_stats.items()}
===========changed ref 5===========
# module: spot.utils
def pmap(
f: Callable[..., T1],
*f_args: Any,
+ desc: str | None = None,
- desc: str = "parallel map",
key_args: Mapping[str, Any] | None = None,
max_workers: int | None = None,
chunksize: int | None = None,
tqdm_args: Mapping[str, Any] | None = None,
) -> list[T1]:
"""
Parallel map with progress displaying.
"""
n = len(f_args[0])
assert_eq(n, *(len(xs) for xs in f_args))
tqdm_args = dict(tqdm_args) if tqdm_args else {}
tqdm_args.setdefault("smoothing", 0.0)
+
+ if desc is None:
+ desc = "pmap: " + f.__name__
if key_args is None:
key_args = {}
if max_workers is None:
max_workers = DefaultWorkers
if max_workers <= 1:
outs = list[T1]()
for i in tqdm(range(n), desc=desc, **tqdm_args):
outs.append(f(*(a[i] for a in f_args), **key_args))
return outs
if chunksize is None:
chunksize = max(1, n // (50 * max_workers))
tag_f = _TaggedFunc(f, key_args)
arg_tuples = zip(range(n), *f_args)
with (
multiprocessing.Pool(max_workers) as pool,
tqdm(total=n, desc=desc, **tqdm_args) as pbar,
):
results = dict[int, T1]()
for i, r in pool.imap_unordered(tag_f, arg_tuples, chunksize=chunksize):
results[i] = r
pbar.update()
return [results[i] for i in range(n)]
|
coeditor.dataset/TokenizedEditDataset.from_edits
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> return TokenizedEditDataset(list(edits))
|
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
@staticmethod
- def from_edits(
- edits: Iterable[TEdit], path=Path("all")
- ) -> "TokenizedEditDataset[TEdit]":
+ def from_edits(edits: Iterable[TEdit]) -> "TokenizedEditDataset[TEdit]":
- return TokenizedEditDataset({path: list(edits)})
<0>
|
===========unchanged ref 0===========
at: dataclasses
dataclass(_cls: Type[_T]) -> Type[_T]
dataclass(*, init: bool=..., repr: bool=..., eq: bool=..., order: bool=..., unsafe_hash: bool=..., frozen: bool=...) -> Callable[[Type[_T]], Type[_T]]
dataclass(_cls: None) -> Callable[[Type[_T]], Type[_T]]
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def all_edits(self) -> list[TEdit]:
+ return self._edits
- return join_list(self.project2edits.values())
===========changed ref 1===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def subset(self, repos: Iterable[Path]) -> "TokenizedEditDataset":
- return TokenizedEditDataset({repo: self.project2edits[repo] for repo in repos})
-
===========changed ref 2===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
+ _edits: list[TEdit]
- project2edits: dict[Path, list[TEdit]]
===========changed ref 3===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def map(self, f: Callable[[TEdit], TEdit]) -> "TokenizedEditDataset[TEdit]":
- repos = tqdm(self.project2edits.items(), desc="transforming dataset")
- return TokenizedEditDataset(
- {repo: [f(e) for e in edits] for repo, edits in repos}
- )
-
===========changed ref 4===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def __repr__(self) -> str:
- n_projects = len(self.project2edits)
- n_edits = sum(len(edits) for edits in self.project2edits.values())
+ n_edits = len(self.all_edits())
+ return f"TokenizedEditDataset(n_edits={n_edits})"
- return f"TokenizedEditDataset(n_projects={n_projects}, n_edits={n_edits})"
===========changed ref 5===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def overall_stats(self) -> dict:
all_edits = self.all_edits()
n_added = sum(isinstance(e.change_type, Added) for e in all_edits)
basic_stats = {
- "n_projects": len(self.project2edits),
"n_edits": len(all_edits),
"n_additions": n_added,
}
extra_stats = dict[str, list]()
for e in all_edits:
for k, v in e.stats().items():
if k in extra_stats:
extra_stats[k].append(v)
else:
extra_stats[k] = [v]
return basic_stats | {k: scalar_stats(v) for k, v in extra_stats.items()}
===========changed ref 6===========
# module: spot.utils
def pmap(
f: Callable[..., T1],
*f_args: Any,
+ desc: str | None = None,
- desc: str = "parallel map",
key_args: Mapping[str, Any] | None = None,
max_workers: int | None = None,
chunksize: int | None = None,
tqdm_args: Mapping[str, Any] | None = None,
) -> list[T1]:
"""
Parallel map with progress displaying.
"""
n = len(f_args[0])
assert_eq(n, *(len(xs) for xs in f_args))
tqdm_args = dict(tqdm_args) if tqdm_args else {}
tqdm_args.setdefault("smoothing", 0.0)
+
+ if desc is None:
+ desc = "pmap: " + f.__name__
if key_args is None:
key_args = {}
if max_workers is None:
max_workers = DefaultWorkers
if max_workers <= 1:
outs = list[T1]()
for i in tqdm(range(n), desc=desc, **tqdm_args):
outs.append(f(*(a[i] for a in f_args), **key_args))
return outs
if chunksize is None:
chunksize = max(1, n // (50 * max_workers))
tag_f = _TaggedFunc(f, key_args)
arg_tuples = zip(range(n), *f_args)
with (
multiprocessing.Pool(max_workers) as pool,
tqdm(total=n, desc=desc, **tqdm_args) as pbar,
):
results = dict[int, T1]()
for i, r in pool.imap_unordered(tag_f, arg_tuples, chunksize=chunksize):
results[i] = r
pbar.update()
return [results[i] for i in range(n)]
|
coeditor.dataset/_process_commits
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> change_processor.append_stats(stats)
|
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
is_training: bool,
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ change_processor.clear_stats()
- encoder.change_processor.clear_stats()
+ change_processor.set_training(is_training)
- encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
+ change_processor=change_processor,
- change_processor=encoder.change_processor,
- edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
except Exception as e:
if isinstance(e, KeyboardInterrupt):
raise
warnings.warn(f"Failed to process project: {root}\nError: {e}")
traceback.print_exception(e, limit=-6)
edits = []
stats = dict()
- encoder.change_processor.append_stats(stats)
<0> rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
at: coeditor.common
rec_add_dict_to(target: dict[str, Any], value: Mapping[str, Any], value_merger: Callable[[Any, Any], Any]=lambda x, y: x + y)
at: coeditor.dataset
_ProcessingResult(edits: Sequence[C3Problem], stats: dict[str, dict | Any])
time_limit_per_commit = 10.0
at: pathlib
Path()
at: traceback
print_exception(etype: Optional[Type[BaseException]], value: Optional[BaseException], tb: Optional[TracebackType], limit: Optional[int]=..., file: Optional[IO[str]]=..., chain: bool=...) -> None
at: typing
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
@dataclass
class _ProcessingResult:
+ edits: Sequence[C3Problem]
- edits: Sequence[TkC3Problem]
stats: dict[str, dict | Any]
===========changed ref 1===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
@staticmethod
- def from_edits(
- edits: Iterable[TEdit], path=Path("all")
- ) -> "TokenizedEditDataset[TEdit]":
+ def from_edits(edits: Iterable[TEdit]) -> "TokenizedEditDataset[TEdit]":
+ return TokenizedEditDataset(list(edits))
- return TokenizedEditDataset({path: list(edits)})
===========changed ref 2===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def all_edits(self) -> list[TEdit]:
+ return self._edits
- return join_list(self.project2edits.values())
===========changed ref 3===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def subset(self, repos: Iterable[Path]) -> "TokenizedEditDataset":
- return TokenizedEditDataset({repo: self.project2edits[repo] for repo in repos})
-
===========changed ref 4===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
+ _edits: list[TEdit]
- project2edits: dict[Path, list[TEdit]]
===========changed ref 5===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def map(self, f: Callable[[TEdit], TEdit]) -> "TokenizedEditDataset[TEdit]":
- repos = tqdm(self.project2edits.items(), desc="transforming dataset")
- return TokenizedEditDataset(
- {repo: [f(e) for e in edits] for repo, edits in repos}
- )
-
===========changed ref 6===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def __repr__(self) -> str:
- n_projects = len(self.project2edits)
- n_edits = sum(len(edits) for edits in self.project2edits.values())
+ n_edits = len(self.all_edits())
+ return f"TokenizedEditDataset(n_edits={n_edits})"
- return f"TokenizedEditDataset(n_projects={n_projects}, n_edits={n_edits})"
===========changed ref 7===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def overall_stats(self) -> dict:
all_edits = self.all_edits()
n_added = sum(isinstance(e.change_type, Added) for e in all_edits)
basic_stats = {
- "n_projects": len(self.project2edits),
"n_edits": len(all_edits),
"n_additions": n_added,
}
extra_stats = dict[str, list]()
for e in all_edits:
for k, v in e.stats().items():
if k in extra_stats:
extra_stats[k].append(v)
else:
extra_stats[k] = [v]
return basic_stats | {k: scalar_stats(v) for k, v in extra_stats.items()}
===========changed ref 8===========
# module: spot.utils
def pmap(
f: Callable[..., T1],
*f_args: Any,
+ desc: str | None = None,
- desc: str = "parallel map",
key_args: Mapping[str, Any] | None = None,
max_workers: int | None = None,
chunksize: int | None = None,
tqdm_args: Mapping[str, Any] | None = None,
) -> list[T1]:
"""
Parallel map with progress displaying.
"""
n = len(f_args[0])
assert_eq(n, *(len(xs) for xs in f_args))
tqdm_args = dict(tqdm_args) if tqdm_args else {}
tqdm_args.setdefault("smoothing", 0.0)
+
+ if desc is None:
+ desc = "pmap: " + f.__name__
if key_args is None:
key_args = {}
if max_workers is None:
max_workers = DefaultWorkers
if max_workers <= 1:
outs = list[T1]()
for i in tqdm(range(n), desc=desc, **tqdm_args):
outs.append(f(*(a[i] for a in f_args), **key_args))
return outs
if chunksize is None:
chunksize = max(1, n // (50 * max_workers))
tag_f = _TaggedFunc(f, key_args)
arg_tuples = zip(range(n), *f_args)
with (
multiprocessing.Pool(max_workers) as pool,
tqdm(total=n, desc=desc, **tqdm_args) as pbar,
):
results = dict[int, T1]()
for i, r in pool.imap_unordered(tag_f, arg_tuples, chunksize=chunksize):
results[i] = r
pbar.update()
return [results[i] for i in range(n)]
|
coeditor.dataset/dataset_from_projects
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> return project2edits
|
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
<s>,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
+ project2edits = dict[Path, list[C3Problem]]()
- project2edits = dict[Path, list[TkC3Problem]]()
try:
stats = dict[str, Any]()
for root, pr in zip(roots, presults):
project2edits.setdefault(root, []).extend(pr.edits)
rec_add_dict_to(stats, pr.stats)
if "tlogger" in stats:
df = TimeLogger.times_to_dataframe(stats.pop("tlogger"))
print("Time stats:")
display(df)
if "analyzer_errors" in list(stats.keys()):
errors: dict = stats.pop("analyzer_errors")
for k in list(errors.keys()):
if JediUsageAnalyzer.is_known_error(k):
errors.pop(k)
if errors:
print("Analyzer errors:")
for k in sorted(errors.keys(), key=lambda k: errors[k], reverse=True):
print(f"{k}:\t{errors[k]}")
if stats:
print("Other Stats:")
pretty_print_dict(stats)
except Exception as e:
if not isinstance(e, KeyboardInterrupt):
print("Error while printing stats:", e)
- return TokenizedEditDataset(project2edits)
<0>
|
===========above chunk 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
# offset: -1
<s> get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 10))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
chunk_training,
- key_args={"encoder": encoder},
- desc="Create tokenized edits",
+ key_args={"change_processor": change_processor},
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
</s>
===========above chunk 1===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
# offset: -2
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",</s>
===========unchanged ref 0===========
at: IPython.core.display_functions
display(*, include=None, exclude=None, metadata=None, transient=None, display_id=None, raw=False, clear=False, source=_sentinel, **kwargs)
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
pmap(f: Callable[..., T1], iter3: Iterable[Any], iter4: Iterable[Any], iter5: Iterable[Any], iter6: Iterable[Any], /, *iterables: Iterable[Any], desc: str | None=None, key_args: Mapping[str, Any] | None=None, max_workers: int | None=None, chunksize: int | None=None, tqdm_args: Mapping[str, Any] | None=None) -> list[T1]
TimeLogger(times: dict[str, list[float]]=field(default_factory=dict))
at: coeditor._utils.TimeLogger
times: dict[str, list[float]] = field(default_factory=dict)
times_to_dataframe(times: dict[str, list[float]])
at: coeditor.common
rec_add_dict_to(target: dict[str, Any], value: Mapping[str, Any], value_merger: Callable[[Any, Any], Any]=lambda x, y: x + y)
at: coeditor.dataset
_process_commits(root: Path, workdir: Path, commits: Sequence[CommitInfo], is_training: bool, change_processor: ProjectChangeProcessor[C3Problem]) -> _ProcessingResult
at: coeditor.dataset._ProcessingResult
edits: Sequence[C3Problem]
stats: dict[str, dict | Any]
===========unchanged ref 1===========
at: coeditor.dataset.dataset_from_projects
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
at: math
ceil(x: SupportsFloat, /) -> int
at: pathlib
Path()
at: pathlib.Path
__slots__ = ()
iterdir() -> Generator[_P, None, None]
exists() -> bool
at: shutil
rmtree(path: Union[bytes, StrPath], ignore_errors: bool=..., onerror: Optional[Callable[[Any, Any, Any], Any]]=...) -> None
at: spot.utils
pretty_print_dict(d: dict, level: int=0, max_show_level: int=1000, float_precision: int=5)
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
Sequence = _alias(collections.abc.Sequence, 1)
at: typing.MutableMapping
pop(key: _KT, default: Union[_VT, _T]=...) -> Union[_VT, _T]
pop(key: _KT) -> _VT
===========changed ref 0===========
# module: coeditor.dataset
def _process_commits(
root: Path,
workdir: Path,
commits: Sequence[CommitInfo],
is_training: bool,
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
) -> _ProcessingResult:
# use process-specific parso cache
_fix_jedi_cache(workdir)
coeditor.code_change._tlogger.clear()
+ change_processor.clear_stats()
- encoder.change_processor.clear_stats()
+ change_processor.set_training(is_training)
- encoder.change_processor.set_training(is_training)
try:
# cannot return here since subprocess will be killed after returning
edits = edits_from_commit_history(
root,
commits,
tempdir=workdir / "code",
+ change_processor=change_processor,
- change_processor=encoder.change_processor,
- edit_encoder=encoder.edit_tokenizer.tokenize_problem,
silent=True,
time_limit=time_limit_per_commit * (len(commits) + 10),
)
except Exception as e:
if isinstance(e, KeyboardInterrupt):
raise
warnings.warn(f"Failed to process project: {root}\nError: {e}")
traceback.print_exception(e, limit=-6)
edits = []
stats = dict()
+ change_processor.append_stats(stats)
- encoder.change_processor.append_stats(stats)
rec_add_dict_to(stats, {"tlogger": coeditor.code_change._tlogger.times})
return _ProcessingResult(edits, stats)
|
coeditor.dataset/datasets_from_repos
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> return {k: join_list(dataset[r] for r in repos) for k, repos in projects.items()}
|
# module: coeditor.dataset
def datasets_from_repos(
repos_root: Path,
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> Mapping[str, Sequence[C3Problem]]:
- ) -> dict[str, TokenizedEditDataset[TkC3Problem]]:
splits = ["test", "valid", "train"]
projects = dict[str, list[Path]]()
split_is_training = dict[str, list[bool]]()
for split in splits:
if not (repos_root / split).exists():
warnings.warn(f"Split {split} not found at {repos_root / split}.")
continue
ps = [p for p in (repos_root / split).iterdir() if p.is_dir]
projects[split] = ps
training = split == "train"
split_is_training[split] = [training] * len(ps)
if not ps:
warnings.warn(f"No projects found in {split} split")
dataset = dataset_from_projects(
join_list(projects.values()),
+ change_processor=change_processor,
- encoder=encoder,
repo_training=join_list(split_is_training.values()),
max_history_per_repo=max_history_per_repo,
workers=workers,
)
- return {k: dataset.subset(v) for k, v in projects.items()}
<0>
|
===========unchanged ref 0===========
at: _warnings
warn(message: str, category: Optional[Type[Warning]]=..., stacklevel: int=..., source: Optional[Any]=...) -> None
warn(message: Warning, category: Any=..., stacklevel: int=..., source: Optional[Any]=...) -> None
at: coeditor._utils
DefaultWorkers: int = multiprocessing.cpu_count() // 2
global DefaultWorkers
repr_modified_args(instance, flatten: bool=False) -> str
at: coeditor.common
get_dataset_dir(dataname: str) -> Path
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.dataset
dataset_from_projects(project_roots: Sequence[Path], change_processor: ProjectChangeProcessor[C3Problem], repo_training: Sequence[bool], max_history_per_repo: int=1000, workers: int=DefaultWorkers) -> "Mapping[Path, Sequence[C3Problem]]"
at: coeditor.dataset.datasets_from_repos
projects = dict[str, list[Path]]()
projects[split] = ps
ps = [p for p in (repos_root / split).iterdir() if p.is_dir]
training = split == "train"
at: pathlib.Path
exists() -> bool
at: typing
Mapping = _alias(collections.abc.Mapping, 2)
Sequence = _alias(collections.abc.Sequence, 1)
===========changed ref 0===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
"""
Create a TokenizedEditDataset from a list of project roots and a given encoder.
Args:
- max_history_per_repo (int, optional): When the repo history is longer than
this value, only the oldest portion is going to be used. Defaults to 1000.
"""
workdir = Path(tempfile.gettempdir()) / "dataset_from_projects"
histories = pmap(
get_commit_history,
project_roots,
max_workers=workers,
desc="Getting commit histories",
tqdm_args={"unit": "repo"},
)
# keep the oldest portion of the history
histories = [commits[-max_history_per_repo:] for commits in histories]
# break long commit sequences into chunks for parallelization
roots = list[Path]()
chunk_training = list[bool]()
chunked_histories = list[list[CommitInfo]]()
for root, h, train in zip(project_roots, histories, repo_training):
history_chunk_size = max(50, math.ceil(len(h) / 10))
for i in range(0, len(h), history_chunk_size):
roots.append(root)
chunk_training.append(train)
# note that we need 1 extra overlapping commit to get all diffs
chunked_histories.append(h[i : i + history_chunk_size + 1])
workdirs = [workdir / f"chunk-{i}" for i in range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,</s>
===========changed ref 1===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
# offset: 1
<s> range(len(roots))]
try:
presults = pmap(
_process_commits,
roots,
workdirs,
chunked_histories,
chunk_training,
- key_args={"encoder": encoder},
- desc="Create tokenized edits",
+ key_args={"change_processor": change_processor},
max_workers=workers,
tqdm_args={"unit": "chunk"},
)
finally:
if workdir.exists():
shutil.rmtree(workdir)
print("Workdir removed:", workdir)
+ project2edits = dict[Path, list[C3Problem]]()
- project2edits = dict[Path, list[TkC3Problem]]()
try:
stats = dict[str, Any]()
for root, pr in zip(roots, presults):
project2edits.setdefault(root, []).extend(pr.edits)
rec_add_dict_to(stats, pr.stats)
if "tlogger" in stats:
df = TimeLogger.times_to_dataframe(stats.pop("tlogger"))
print("Time stats:")
display(df)
if "analyzer_errors" in list(stats.keys()):
errors: dict = stats.pop("analyzer_errors")
for k in list(errors.keys()):
if JediUsageAnalyzer.is_known_error(k):
</s>
===========changed ref 2===========
# module: coeditor.dataset
def dataset_from_projects(
project_roots: Sequence[Path],
+ change_processor: ProjectChangeProcessor[C3Problem],
- encoder: C3EditEncoder,
repo_training: Sequence[bool],
max_history_per_repo: int = 1000,
workers: int = DefaultWorkers,
+ ) -> "Mapping[Path, Sequence[C3Problem]]":
- ) -> "TokenizedEditDataset[TkC3Problem]":
# offset: 2
<s>.pop(k)
if errors:
print("Analyzer errors:")
for k in sorted(errors.keys(), key=lambda k: errors[k], reverse=True):
print(f"{k}:\t{errors[k]}")
if stats:
print("Other Stats:")
pretty_print_dict(stats)
except Exception as e:
if not isinstance(e, KeyboardInterrupt):
print("Error while printing stats:", e)
+ return project2edits
- return TokenizedEditDataset(project2edits)
===========changed ref 3===========
# module: coeditor.dataset
@dataclass
class _ProcessingResult:
+ edits: Sequence[C3Problem]
- edits: Sequence[TkC3Problem]
stats: dict[str, dict | Any]
===========changed ref 4===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
@staticmethod
- def from_edits(
- edits: Iterable[TEdit], path=Path("all")
- ) -> "TokenizedEditDataset[TEdit]":
+ def from_edits(edits: Iterable[TEdit]) -> "TokenizedEditDataset[TEdit]":
+ return TokenizedEditDataset(list(edits))
- return TokenizedEditDataset({path: list(edits)})
===========changed ref 5===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
def all_edits(self) -> list[TEdit]:
+ return self._edits
- return join_list(self.project2edits.values())
===========changed ref 6===========
# module: coeditor.dataset
@dataclass
class TokenizedEditDataset(Generic[TEdit]):
- def subset(self, repos: Iterable[Path]) -> "TokenizedEditDataset":
- return TokenizedEditDataset({repo: self.project2edits[repo] for repo in repos})
-
|
coeditor.ctx_change_encoder/C3ProblemTokenizer.tokenize_problem
|
Modified
|
temp-1
|
2e823493b40c36c3f449ba3571e2fa26454e3493
|
Update datasets creation pipeline.
|
<0>:<add> named_references=all_refs,
|
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
<s>) for i, chunk in enumerate(above_chunks)
]
below_chunks = [
(f"below chunk {i}", chunk) for i, chunk in enumerate(below_chunks)
]
all_refs = above_chunks + below_chunks + named_references
- size_sum = 0
- kept_refs = list[tuple[str, TokenSeq]]()
- for (name, chunk) in all_refs:
- if size_sum + len(chunk) <= self.max_total_ref_tks:
- size_sum += len(chunk)
- kept_refs.append((name, chunk))
return TkC3Problem(
scope_tks + chunk_input,
chunk_output,
path=span.headers[-1].path,
change_type=span.change.map(lambda _: None),
- named_references=kept_refs,
<0> src_info=problem.src_info,
)
problems = list[TkC3Problem]()
for l in range(len(tk_delta.deltas) + 1):
finished = l == len(tk_delta.deltas)
input_growth = len(origin_lines[l]) + 2 if l < len(origin_lines) else 1
if (
finished
or chunk_lines >= self.max_lines_to_edit
or len(chunk_input) + input_growth > input_limit
):
if has_change(chunk_output):
problems.append(get_problem(chunk_input, chunk_output))
if len(problems) >= self.max_chunks_per_elem:
break
if finished:
break
chunk_main_input = join_list(origin_lines[chunk_start_l:l], Newline_id)
chunk_main_delta</s>
|
===========above chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: -1
<s>
chunk_output = TokenSeq()
prev_change_tks = TokenSeq()
def get_problem(chunk_input, chunk_output):
# try move some prev_change_tks into the input
above_tks = prev_change_tks
below_tks = join_list(origin_lines[l:], Newline_id)
chunk_input, above_tks, below_tks = self._inline_some_context(
chunk_input, above_tks, below_tks, input_limit
)
# limit the input size if it's too long (can happen for later chunks)
chunk_input = truncate_section(chunk_input, TruncateAt.Right, input_limit)
chunk_output = truncate_output_tks(chunk_input, chunk_output)
chunk_output = truncate_section(
chunk_output, TruncateAt.Right, self.max_output_tks, add_bos=False
)
above_chunks = break_into_chunks(
above_tks,
lambda i: self._encode_headers(span.headers, -1 - i),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
right_to_left=True,
)
if finished:
below_chunks = []
else:
below_chunks = break_into_chunks(
below_tks,
lambda i: self._encode_headers(span.headers, i + 1),
chunk_size=self.max_ref_tks,
overlap=self.ref_chunk_overlap,
)
above_chunks = [
(f"above chunk {i}", chunk) for i, chunk in enumerate(above_chunks)
]
below_chunks = [
(f"below chunk {</s>
===========above chunk 1===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: -2
span = problem.span
named_references = list[tuple[str, TokenSeq]]()
# compute the references that are relevant to this span
relevant_chunks = self._group_encode_changed_refs(problem.relevant_changes)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"changed ref {i}", chunk))
relevant_chunks = self._group_encode_unchanged_refs(problem.relevant_unchanged)
for i, chunk in enumerate(relevant_chunks):
named_references.append((f"unchanged ref {i}", chunk))
diffs = change_to_line_diffs(span.change)
original, delta = line_diffs_to_original_delta(diffs)
origin_lines = split_list(encode_basic(original), Newline_id)
tk_delta = delta.to_tk_delta()
chunk_id = 0
chunk_start_l = 0
scope_tks = self._encode_headers(span.headers, 0)
chunk_input = TokenSeq()
input_limit = self.max_query_tks - len(scope_tks)
chunk_lines = 0
chunk_output = TokenSeq()
prev_change_tks = TokenSeq()
def get_problem(chunk</s>
===========below chunk 0===========
# module: coeditor.ctx_change_encoder
@dataclass
class C3ProblemTokenizer:
def tokenize_problem(
self,
problem: C3Problem,
) -> Sequence[TkC3Problem]:
# offset: 1
<s>input = join_list(origin_lines[chunk_start_l:l], Newline_id)
chunk_main_delta = tk_delta.for_input_range((chunk_start_l, l))
chunk_main_change = chunk_main_delta.to_change_tks(chunk_main_input)
prev_change_tks.extend(chunk_main_change)
prev_change_tks.append(Newline_id)
chunk_id += 1
chunk_input = TokenSeq()
chunk_lines = 0
chunk_output = TokenSeq()
chunk_start_l = l
chunk_input.append(get_extra_id(chunk_lines))
if l < len(origin_lines):
chunk_input.extend(origin_lines[l])
chunk_input.append(Newline_id)
line_change = join_list(tk_delta.deltas[l], Newline_id)
chunk_output.append(get_extra_id(chunk_lines))
chunk_output.extend(line_change)
if line_change and line_change[-1] != Del_id:
chunk_output.append(Newline_id)
chunk_lines += 1
return problems
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
split_list(lst: list[T1], sep: T1) -> list[list[T1]]
join_list(segs: Iterable[Iterable[T1]], sep: T1 | None=None) -> list[T1]
at: coeditor.ctx_change_encoder
C3Problem(span: ChangedCodeSpan, relevant_changes: Sequence[ChangedCodeSpan], relevant_unchanged: Sequence[ChangedCodeSpan], src_info: dict[str, Any])
TkC3Problem(input_tks: TokenSeq, output_tks: TokenSeq, path: ProjectPath, change_type: Change[None], named_references: Sequence[tuple[str, TokenSeq]], src_info: dict[str, Any])
at: coeditor.ctx_change_encoder.C3Problem
span: ChangedCodeSpan
relevant_changes: Sequence[ChangedCodeSpan]
relevant_unchanged: Sequence[ChangedCodeSpan]
src_info: dict[str, Any]
at: coeditor.ctx_change_encoder.C3ProblemTokenizer
VERSION = "1.0"
max_ref_tks: int = 512
max_query_tks: int = 512
max_output_tks: int = 256
max_scope_tks: int = 128
max_lines_to_edit: int = 20
ref_chunk_overlap: int = 32
max_total_ref_tks: int = 512 * 64 # a very large threshold
max_chunks_per_elem: int = 4
skip_unchanged_problems: bool = True
_encode_headers(scope_changes: Sequence[ChangedHeader], offset: int) -> TokenSeq
_inline_some_context(input: TokenSeq, above_ctx: TokenSeq, below_ctx: TokenSeq, size_limit: int) -> tuple[TokenSeq, TokenSeq, TokenSeq]
_group_encode_unchanged_refs(elems: Sequence[ChangedCodeSpan]) -> Sequence[TokenSeq]
|
coeditor.code_change/ChangeScope.from_tree
|
Modified
|
temp-1
|
f106fd266a6e09194f46b8cad8930f3f0978a011
|
Fix missing spans.
|
<0>:<add> if isinstance(container, BaseNode):
|
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
spans = []
subscopes = dict()
scope = ChangeScope(path, tree, spans, subscopes, None)
assert isinstance(tree, ScopeTree)
is_func = isinstance(tree, ptree.Function)
def mk_span(stmts):
# remove leading newlines
n_leading_newlines = 0
for s in stmts:
if s.type == ptree.Newline.type:
n_leading_newlines += 1
else:
break
if n_leading_newlines:
stmts = stmts[n_leading_newlines:]
if stmts:
yield StatementSpan(len(spans), stmts, scope)
current_stmts = []
container = tree if isinstance(tree, ptree.Module) else tree.get_suite()
- if isinstance(container, Node):
<0> content = container.children
else:
content = []
for s in content:
# we don't create inner scopes for function contents
if is_func or _is_scope_statement(as_any(s)):
current_stmts.append(s)
else:
if current_stmts:
spans.extend(mk_span(current_stmts))
current_stmts = []
if current_stmts:
spans.extend(mk_span(current_stmts))
if is_func:
# we don't create inner scopes for function contents
if not spans:
raise ValueError(f"Function with no spans: {path=}, {tree.get_code()=}")
return scope
for stree in tree._search_in_scope(ptree.Function.type, ptree.Class.type):
stree: ptree.Function | ptree.Class
name = cast(ptree.Name, stree.name).value
spath = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.</s>
|
===========below chunk 0===========
# module: coeditor.code_change
@dataclass
class ChangeScope:
@staticmethod
def from_tree(path: ProjectPath, tree: ScopeTree) -> "ChangeScope":
# offset: 1
<s> = path.append(name)
subscope = ChangeScope.from_tree(spath, stree)
subscope.parent_scope = scope
subscopes[name] = subscope
return scope
===========unchanged ref 0===========
at: coeditor._utils
as_any(x) -> Any
at: coeditor.code_change
ScopeTree = ptree.Function | ptree.Class | ptree.Module
ChangeScope(path: ProjectPath, tree: ScopeTree, spans: Sequence["StatementSpan"], subscopes: Mapping[str, Self], parent_scope: "ChangeScope | None")
_is_scope_statement(stmt: PyNode) -> bool
StatementSpan(nth_in_parent: int, statements: Sequence[PyNode], scope: ChangeScope)
at: coeditor.code_change.ChangeScope
path: ProjectPath
tree: ScopeTree
spans: Sequence["StatementSpan"]
subscopes: Mapping[str, Self]
parent_scope: "ChangeScope | None"
at: parso.python.tree
Newline()
Name()
Module(children)
Class(children)
Function(children)
at: parso.python.tree.Class
type = 'classdef'
__slots__ = ()
at: parso.python.tree.ClassOrFunc
__slots__ = ()
at: parso.python.tree.Function
type = 'funcdef'
at: parso.python.tree.Newline
__slots__ = ()
type = 'newline'
at: parso.python.tree.Scope
__slots__ = ()
_search_in_scope(*names)
get_suite()
at: parso.tree
BaseNode(children: List[NodeOrLeaf])
at: parso.tree.BaseNode
__slots__ = ('children',)
get_code(include_prefix=True)
at: parso.tree.BaseNode.__init__
self.children = children
at: parso.tree.Leaf.__init__
self.value = value
at: parso.tree.NodeOrLeaf
__slots__ = ('parent',)
type: str
parent: 'Optional[BaseNode]'
===========unchanged ref 1===========
at: spot.static_analysis
ProjectPath(typename: str, fields: Iterable[Tuple[str, Any]]=..., **kwargs: Any)
at: spot.static_analysis.ProjectPath
module: ModuleName
path: ElemPath
append(path: ElemPath) -> "ProjectPath"
at: typing
cast(typ: Type[_T], val: Any) -> _T
cast(typ: str, val: Any) -> Any
cast(typ: object, val: Any) -> Any
===========changed ref 0===========
# module: tests.coeditor.test_edits
+ def test_extra_ids():
+ all_extra_ids = _Tokenizer.additional_special_tokens_ids
+
+ for x in all_extra_ids:
+ assert is_extra_id(x)
+ n = extra_id_to_number(x)
+ assert get_extra_id(n) == x
+
===========changed ref 1===========
# module: tests.coeditor.test_code_change
def test_change_scope():
code1 = dedent(
"""\
import os
x = 1
y = x + 1
def f1():
global x
x *= 5
return x
if __name__ == "__main__":
print(f1() + x)
@annotated
def f2():
return 1
@dataclass
class A:
attr1: int
@staticmethod
def method1():
return 1
class B:
inner_attr1: int
"""
)
mod_tree = code_to_module(code1)
scope = ChangeScope.from_tree(ProjectPath("code1", ""), mod_tree)
global_spans = [
dedent(
"""\
x = 1
y = x + 1
"""
),
dedent(
"""\
if __name__ == "__main__":
print(f1() + x)
"""
),
]
+ try:
+ for i, code in enumerate(global_spans):
- for i, code in enumerate(global_spans):
+ assert_str_equal(scope.spans[i].code, code)
- assert_str_equal(scope.spans[i].code, code)
+ except Exception:
+ print_err(f"{scope.spans=}")
+ raise
f1_expect = dedent(
"""\
global x
x *= 5
return x
"""
)
f1_code = scope.subscopes["f1"].spans_code
assert_str_equal(f1_code, indent(f1_expect, " " * 4))
f2_expect = dedent(
"""\
@annotated
def f2():
return 1
"""
)
f2_code = scope.subscopes["f2"].all_code
assert_str_</s>
===========changed ref 2===========
# module: tests.coeditor.test_code_change
def test_change_scope():
# offset: 1
<s> """
)
f2_code = scope.subscopes["f2"].all_code
assert_str_equal(f2_code, f2_expect)
attr1_expect = dedent(
"""\
attr1: int
"""
)
attr1_code = scope.subscopes["A"].spans_code
assert_str_equal(attr1_code, indent(attr1_expect, " " * 4))
method1_expect = dedent(
"""\
@staticmethod
def method1():
return 1
"""
)
method1_code = scope.subscopes["A"].subscopes["method1"].all_code
assert_str_equal(method1_code, indent(method1_expect, " " * 4))
inner_attr1_expect = dedent(
"""\
class B:
inner_attr1: int
"""
)
inner_class_code = scope.subscopes["A"].subscopes["B"].all_code
assert_str_equal(inner_class_code, indent(inner_attr1_expect, " " * 4))
|
coeditor.encoding/get_extra_id
|
Modified
|
temp-1
|
0a557eb3eef82b76ad61c98704f8bfdf960a3ca1
|
Improve encoder performance.
|
<0>:<add> return _min_extra_id + (N_Extra_Ids - 1 - i)
|
# module: coeditor.encoding
def get_extra_id(i: int) -> int:
assert 0 <= i < N_Extra_Ids
- return _Tokenizer.additional_special_tokens_ids[N_Extra_Ids - 1 - i]
<0>
|
===========unchanged ref 0===========
at: coeditor.encoding
N_Extra_Ids = 100
_min_extra_id = _Tokenizer.additional_special_tokens_ids[0]
|
coeditor.encoding/change_to_tokens
|
Modified
|
temp-1
|
0a557eb3eef82b76ad61c98704f8bfdf960a3ca1
|
Improve encoder performance.
|
<0>:<add> raise AssertionError(f"Not a change type: {change}")
|
# module: coeditor.encoding
def change_to_tokens(change: Change[str]) -> TokenSeq:
+ match change:
+ case Modified(before=before, after=after, unchanged=unchanged):
+ if unchanged or before == after:
+ return encode_basic(before)
+ else:
+ diffs = change_to_line_diffs(change)
- diffs = change_to_line_diffs(change)
+ return encode_diffs(diffs)
- return encode_diffs(diffs)
+ case Added() | Deleted():
+ lines = split_list(encode_basic(change.earlier()), Newline_id)
+ tk = Add_id if isinstance(change, Added) else Del_id
+ return join_list([tk] + line for line in lines)
+ case _:
<0>
|
===========unchanged ref 0===========
at: coeditor.common
TokenSeq = list[Token]
===========changed ref 0===========
# module: coeditor.encoding
def get_extra_id(i: int) -> int:
assert 0 <= i < N_Extra_Ids
+ return _min_extra_id + (N_Extra_Ids - 1 - i)
- return _Tokenizer.additional_special_tokens_ids[N_Extra_Ids - 1 - i]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.