Function create_frame_ocr_function

utils/advanced_ocr.py:541–605

创建适用于Frame的OCR函数，用于集成到VerifierOptions中 Args: processor: 高级OCR处理器实例 Returns: callable: 接受Frame参数的OCR函数

(processor: AdvancedOCRProcessor)

def create_frame_ocr_function(processor: AdvancedOCRProcessor) -> Callable:
    """
    创建适用于Frame的OCR函数，用于集成到VerifierOptions中
    
    Args:
        processor: 高级OCR处理器实例
        
    Returns:
        callable: 接受Frame参数的OCR函数
    """
    def frame_ocr(frame: Dict[str, Any]) -> Optional[str]:
        """
        从Frame中提取并识别图像文字
        
        Args:
            frame: 包含图像路径的Frame字典
            
        Returns:
            str: 识别的文字，失败时返回XML文本或None
        """
        # 获取图像路径
        image_path = frame.get("image")
        if not image_path or not os.path.exists(image_path):
            # 退化到改进的XML提取
            xml_text = frame.get("xml_text", "")
            if xml_text:
                xml_processed = extract_text_from_xml(xml_text)
                if xml_processed.cleaned:
                    frame['_xml_processed'] = xml_processed
                    logger.info(f"图像不可用，使用改进XML提取: {xml_processed.cleaned[:100]}...")
                    logger.debug(f"XML提取词语数: {len(xml_processed.words)}")
                    return f"{xml_processed.cleaned} {xml_processed.no_spaces} {' '.join(xml_processed.words)}"
                else:
                    logger.warning("图像不可用且XML提取失败")
                    return xml_text[:200] if xml_text else None
            return None
        
        # 使用OCR识别
        ocr_text, backup_text = processor.extract_text_from_image(image_path)
        xml_text = frame.get("xml_text", "")
        merged_parts: List[str] = []
        
        if ocr_text:
            # 处理OCR文本
            processed = processor.process_text(ocr_text)
            frame['_ocr_processed'] = processed
            merged_parts.extend([processed.cleaned, processed.no_spaces] + processed.words)
            logger.debug(f"识别图像 {os.path.basename(image_path)} -> 词语数: {len(processed.words)}")
            
        # 融合XML文本
        if xml_text:
            xml_processed = extract_text_from_xml(xml_text)
            if xml_processed.cleaned:
                frame['_xml_processed'] = xml_processed
                merged_parts.extend([xml_processed.cleaned, xml_processed.no_spaces] + xml_processed.words)
                logger.debug(f"融合XML文本 -> 词语数: {len(xml_processed.words)}")
                
        if merged_parts: