从Frame中提取文本列表，优先使用OCR识别 Args: frame: 包含图像路径的Frame字典 Returns: List[str]: 文本列表
(frame: Dict[str, Any])
| 616 | callable: 接受Frame参数，返回文本列表的函数 |
| 617 | """ |
| 618 | def frame_texts(frame: Dict[str, Any]) -> List[str]: |
| 619 | """ |
| 620 | 从Frame中提取文本列表，优先使用OCR识别 |
| 621 | |
| 622 | Args: |
| 623 | frame: 包含图像路径的Frame字典 |
| 624 | |
| 625 | Returns: |
| 626 | List[str]: 文本列表 |
| 627 | """ |
| 628 | # 获取图像路径 |
| 629 | image_path = frame.get("image") |
| 630 | if not image_path or not os.path.exists(image_path): |
| 631 | # 退化到改进的XML提取 |
| 632 | xml_text = frame.get("xml_text", "") |
| 633 | if xml_text: |
| 634 | xml_processed = extract_text_from_xml(xml_text) |
| 635 | if xml_processed.words: |
| 636 | frame['_xml_processed'] = xml_processed |
| 637 | # 返回多种格式的文本 |
| 638 | result_texts = xml_processed.words.copy() |
| 639 | if xml_processed.cleaned: |
| 640 | result_texts.append(xml_processed.cleaned) |
| 641 | if xml_processed.no_spaces: |
| 642 | result_texts.append(xml_processed.no_spaces) |
| 643 | logger.info(f"图像不可用，使用改进XML提取，得到 {len(result_texts)} 个文本片段") |
| 644 | return list(set(result_texts)) |
| 645 | else: |
| 646 | # 降级到传统方式 |
| 647 | texts = frame.get("xml_texts", []) |
| 648 | if not texts and xml_text: |
| 649 | texts = [xml_text] |
| 650 | logger.warning("XML解析失败，使用原始XML文本") |
| 651 | return texts |
| 652 | else: |
| 653 | texts = frame.get("xml_texts", []) |
| 654 | logger.warning("无图像也无XML文本") |
| 655 | return texts |
| 656 | |
| 657 | # 使用OCR获取文本 |
| 658 | ocr_text, backup_text = processor.extract_text_from_image(image_path) |
| 659 | words = [] |
| 660 | |
| 661 | if ocr_text: |
| 662 | processed = processor.process_text(ocr_text) |
| 663 | words.extend(processed.words) |
| 664 | if processed.cleaned: |
| 665 | words.append(processed.cleaned) |
| 666 | if processed.no_spaces: |
| 667 | words.append(processed.no_spaces) |
| 668 | |
| 669 | # 融合 XML |
| 670 | xml_text = frame.get("xml_text", "") |
| 671 | if xml_text: |
| 672 | xml_processed = extract_text_from_xml(xml_text) |
| 673 | if xml_processed.words: |
| 674 | frame['_xml_processed'] = xml_processed |
| 675 | words.extend(xml_processed.words) |
nothing calls this directly
no test coverage detected