import json
import re
import random
from pathlib import Path
import matplotlib.pyplot as plt
import matplotlib.patches as patches
from PIL import Image


def extract_bboxes_from_response(response_text):
    """从模型回复中提取 bbox 坐标，支持两种格式"""
    bboxes = []
    
    # 格式1: 带 bbox_start/bbox_end 包裹
    pattern1 = r'<bbox_start><x(\d+)><y(\d+)><x(\d+)><y(\d+)><bbox_end>'
    matches1 = re.findall(pattern1, response_text)
    for match in matches1:
        x1, y1, x2, y2 = map(int, match)
        bboxes.append([x1, y1, x2, y2])
    
    # 格式2: 直接 <x><y><x><y>
    if not bboxes:
        pattern2 = r'<x(\d+)><y(\d+)><x(\d+)><y(\d+)>'
        matches2 = re.findall(pattern2, response_text)
        for match in matches2:
            x1, y1, x2, y2 = map(int, match)
            bboxes.append([x1, y1, x2, y2])
    
    return bboxes


def extract_bboxes_from_ground_truth(gt_text):
    """从 GT 中提取 bbox 坐标"""
    bboxes = []
    pattern = r'<x(\d+)><y(\d+)><x(\d+)><y(\d+)>'
    matches = re.findall(pattern, gt_text)
    for match in matches:
        x1, y1, x2, y2 = map(int, match)
        bboxes.append([x1, y1, x2, y2])
    return bboxes


def visualize_comparison(image_path, gt_bboxes, pred_bboxes, output_path, idx):
    """可视化 GT（红色）和预测（蓝色）的对比"""
    img = Image.open(image_path)
    fig, ax = plt.subplots(1, figsize=(12, 9))
    ax.imshow(img)
    
    # 画 GT 框（红色实线）
    for bbox in gt_bboxes:
        x1, y1, x2, y2 = bbox
        rect = patches.Rectangle(
            (x1, y1), x2 - x1, y2 - y1,
            linewidth=2, edgecolor='red', facecolor='none', alpha=0.8
        )
        ax.add_patch(rect)
    
    # 在 GT 框左上角加标签
    for i, bbox in enumerate(gt_bboxes):
        x1, y1, x2, y2 = bbox
        ax.text(x1, y1 - 5, f'GT_{i}', fontsize=8, color='red', weight='bold',
                bbox=dict(boxstyle='round,pad=0.2', facecolor='white', alpha=0.7, edgecolor='red'))
    
    # 画预测框（蓝色虚线）
    for bbox in pred_bboxes:
        x1, y1, x2, y2 = bbox
        rect = patches.Rectangle(
            (x1, y1), x2 - x1, y2 - y1,
            linewidth=2, edgecolor='blue', facecolor='none', linestyle='--', alpha=0.9
        )
        ax.add_patch(rect)
    
    # 在预测框右上角加标签
    for i, bbox in enumerate(pred_bboxes):
        x1, y1, x2, y2 = bbox
        ax.text(x2, y1 - 5, f'Pred_{i}', fontsize=8, color='blue', weight='bold',
                bbox=dict(boxstyle='round,pad=0.2', facecolor='white', alpha=0.7, edgecolor='blue'),
                horizontalalignment='right')
    
    # 添加图例
    from matplotlib.patches import Patch
    legend_elements = [
        Patch(facecolor='none', edgecolor='red', linewidth=2, label='GT Boxes'),
        Patch(facecolor='none', edgecolor='blue', linewidth=2, linestyle='--', label='Predicted Boxes')
    ]
    ax.legend(handles=legend_elements, loc='upper right', fontsize=10)
    
    # 添加统计信息
    ax.set_title(f"Sample {idx}: GT={len(gt_bboxes)}, Pred={len(pred_bboxes)}", fontsize=12, weight='bold')
    ax.axis('off')
    
    plt.tight_layout()
    plt.savefig(output_path, dpi=150, bbox_inches='tight')
    plt.close()


def main():
    # 读取推理结果
    result_file = Path("/home/disk2/hjl/ICL_QWEN/ICL_benchmark/inference_results/test_results_k2.json")
    output_dir = Path("/home/disk2/hjl/ICL_QWEN/ICL_benchmark/visualizations")
    output_dir.mkdir(parents=True, exist_ok=True)
    
    with open(result_file, 'r', encoding='utf-8') as f:
        results = json.load(f)
    
    print(f"加载了 {len(results)} 个样本")
    
    # 随机选择一些样本可视化
    num_samples = min(10, len(results))
    selected = random.sample(results, num_samples)
    
    for idx, result in enumerate(selected):
        image_path = result['image']
        model_response = result['model_response']
        ground_truth = result['ground_truth']
        
        # 提取 bboxes
        gt_bboxes = extract_bboxes_from_ground_truth(ground_truth)
        pred_bboxes = extract_bboxes_from_response(model_response)
        
        print(f"\n样本 {idx+1}: {Path(image_path).name}")
        print(f"  GT 框数: {len(gt_bboxes)}")
        print(f"  预测框数: {len(pred_bboxes)}")
        
        if not gt_bboxes:
            print(f"  警告: 未找到 GT bboxes")
        if not pred_bboxes:
            print(f"  警告: 未找到预测 bboxes")
        
        # 可视化
        output_path = output_dir / f"comparison_{idx+1}.png"
        try:
            visualize_comparison(image_path, gt_bboxes, pred_bboxes, output_path, idx+1)
            print(f"  保存: {output_path}")
        except Exception as e:
            print(f"  可视化失败: {e}")
    
    print(f"\n完成！可视化保存到: {output_dir}")


if __name__ == "__main__":
    random.seed(42)
    main()