data/lexdk/create.py

# /// script
# requires-python = ">=3.12"
# dependencies = [
#     "datasets",
#     "pandas",
# ]
# ///
"""download lexdk from alexandrainst/lexdk-open"""

from datetime import datetime
from pathlib import Path
from typing import cast

import pandas as pd
from datasets import Dataset, load_dataset

column_order = [
    "text",
    "source",
    "id",
    "added",
    "created",
    "license",
    "domain",
    "metadata",
]


def convert_sample(example: dict) -> dict:
    # from sample:
    # {
    #     "url": "https://denstoredanske.lex.dk/Kullmanns_M%C3%B8lle",
    #     "title": "Kullmanns Mølle",
    #     "clarification": "",
    #     "authors": ["https://brugere.lex.dk/6929"],
    #     "date": "2021-01-20T13:23:20+01:00",
    #     "license": "fri anvendelse",
    #     "text": "Kullmanns Mølle er en mølle i Gudhjem, opkaldt efter Matts Kullmann, der byggede møllen i 1893 til sin søn, Christian Kullmann, se Gudhjem Mølle.",
    # }
    date = datetime.fromisoformat(example["date"])
    text = f"{example['title']}\n\npubliceret: {date}\n{example['text']}"

    new_example = dict(
        text_new=text,
        id=example["url"],
        source="lexdk",
        domain="Conversation",
        license="cc-by-sa-4.0",
        added="2025-01-04",
        created=f"{date.date()}, {date.date()}",
        metadata={"source-pretty": "Lex.dk"},
    )

    return new_example


def main():
    ds = load_dataset("alexandrainst/lexdk-open", split="train")
    ds = cast(Dataset, ds)

    dates = [datetime.fromisoformat(date).date() for date in ds["date"]]
    print(str(min(dates)), ",", str(max(dates)))  # 2009-01-28, 2023-09-05

    assert len(set(ds["url"])) == len(ds)

    ds = ds.map(convert_sample, num_proc=4)
    ds = ds.select_columns(column_order[1:] + ["text_new"])
    ds = ds.rename_columns({"text_new": "text"})
    # ensure order
    ds = ds.select_columns(column_order)

    df = ds.to_pandas()
    df = cast(pd.DataFrame, df)
    dedup_df = df.drop_duplicates(keep="first", subset=["text"])
    print("N. duplicates: ", df.shape[0] - dedup_df.shape[0])  # 0

    ds = ds.select(dedup_df.index)
    assert len(set(ds["text"])) == len(ds)

    save_path = Path(__file__).parent / "lexdk.parquet"
    ds.to_parquet(save_path)


if __name__ == "__main__":
    main()