Repository navigation
Three original synthetic PDFs, explicit row/header/value checks, and a bilingual viewer for six measured local extraction results.
- Source: b2c0380
- Includes a known failure and a positive control, JSON/HTML export, and reproducible local converter scripts.
- 11 checker/CLI tests, desktop/mobile browser validation, and a separate Ubuntu CI reproduction passed. CI reproduces known failures; it does not certify converter accuracy.
- The ZIP contains source, fixtures, saved outputs and the offline viewer. Python 3.10+ plus requirements.txt are needed for the checker. No PyPI package is published.
- These three cases are not a representative benchmark. Expected values are supplied by the author; this is not independent PDF verification or OCR.
Demo: https://lydiatools.github.io/pdf-table-traps/
ZIP SHA-256: 2511f4a35d019ecf4b1c979fd8144aaa35cc8e17b0b3644e8a2cf65c9b5f4784
中文：3 份原创合成 PDF、按行列关系核查关键值的命令行工具与中英文实测报告。ZIP 含源码、样本和离线查看页；运行检查器需 Python 与基础依赖。小样本不代表总体准确率，未知项不会自动通过。