caoxiaozhu f47611020a fix(data_process): 修复 Word 切分行号断档与预览标题缺失 ...

- 正文抽取下钻 SDT 内容控件，目录等内容不再整段丢失
- 切片投影匹配剥离序列化插入的列表自动编号，新增行锚点兜底与游标防回退
- fixed/semantic 切分路径定位失败时保留切片，不再静默丢弃
- Word 预览按段落大纲级别识别标题，未套标题样式的小节正常渲染
- 本地嵌入模型抽为共享单例，供语义分块与质量评分共用

2026-08-19 14:22:27 +08:00

..

parsers

fix(data_process): 修复 Word 切分行号断档与预览标题缺失

2026-08-19 14:22:27 +08:00

__init__.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

embedding.py

fix(data_process): 修复 Word 切分行号断档与预览标题缺失

2026-08-19 14:22:27 +08:00

format_detection.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

quality.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

structured_processing.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

text_utils.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

transforms.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00

types.py

refactor: 完整重构 data_process 模块并修复拆分遗留缺陷

2026-08-12 14:29:39 +08:00