根据图片中的起始位置的特殊内容将图片进行分组

根据图片中特殊的标识将图片进行分组

def group_images_by_project(png_img_paths):
    """
    按项目规则（项目: \d 开头 → 七、牵头单位 结束）对PNG图片进行分组
    参数:
        png_img_paths: 所有PNG图片路径列表（按原始顺序传入，保证拼接顺序正确）
    返回:
        project_image_groups: 字典，key=项目顺序编号（如"project_1"），value=该项目对应的图片路径列表
    """
    project_image_groups = {}
    current_project_images = []  # 当前项目的图片列表
    project_counter = 1  # 项目编号计数器
    is_in_project = False  # 标记是否在项目处理中
    
    for img_path in png_img_paths:
        # 1. 跳过非PNG文件
        if not img_path.lower().endswith('.png'):
            print(f"跳过非PNG文件：{img_path}")
            continue

        # 2. 单独提取当前图片的文本
        img_text = ""
        try:
            img = Image.open(img_path)
            img_text = extract_image_text(img)
            img.close()
        except Exception as e:
            print(f"处理图片 {img_path} 失败：{e}")
            # 即使图片读取失败，若正在处理项目中，仍将图片加入
            if is_in_project:
                current_project_images.append(img_path)
            continue

        # 3. 判断当前图片是否包含新项目开头
        project_start_list = batch_match_all_projects(img_text)
        
        if project_start_list:
            # 如果已在一个项目中，先保存当前项目
            if is_in_project and current_project_images:
                project_key = f"project_{project_counter}"
                project_image_groups[project_key] = current_project_images.copy()
                print(f"保存项目 {project_key}，图片数量：{len(current_project_images)}")
                project_counter += 1
            
            # 开始新项目
            is_in_project = True
            current_project_images = [img_path]
            project_name = project_start_list[0]  # 保留原始项目名称用于日志
            print(f"开始新项目 {project_counter}（原始名称：{project_name}），对应图片：{img_path}")
            
        # 4. 判断当前图片是否是项目结束标识
        elif "七、牵头单位" in img_text and is_in_project:
            # 将当前图片加入当前项目
            current_project_images.append(img_path)
            
            # 结束当前项目
            project_key = f"project_{project_counter}"
            project_image_groups[project_key] = current_project_images.copy()
            print(f"结束项目 {project_key}，图片数量：{len(current_project_images)}")
            
            # 重置状态
            is_in_project = False
            current_project_images = []
            project_counter += 1
            
        # 5. 既不是开头也不是结尾，若在项目中则加入
        elif is_in_project:
            current_project_images.append(img_path)
            # 可选：打印日志，跟踪图片加入情况
            # print(f"图片 {img_path} 加入项目 {project_counter}")

    # 6. 处理最后一个未正常结束的项目
    if is_in_project and current_project_images:
        project_key = f"project_{project_counter}"
        project_image_groups[project_key] = current_project_images
        print(f"处理未结束项目 {project_key}，图片数量：{len(current_project_images)}")

    return project_image_groups
## 提炼图片索引
def batch_match_all_projects(text):
    """批量提取文本中的项目标识（返回原始匹配列表）"""
    # 核心正则：兼容带/不带冒号、有/无空格的项目标识
    full_pattern = r'\b(项目\s*\d+\s*[:：]?)'
    all_full_projects = re.findall(full_pattern, text)
    return all_full_projects