Xiaomeng Zhao
2551b6bce6
Merge pull request #5398 from myhloli/dev
...
feat: implement surrogate pair restoration for improved Unicode handling in PDF text extraction
2026-08-14 15:44:03 +08:00
Xiaomeng Zhao
d30e7d94c1
Merge pull request #5358 from hwb96/fix/docx-table-special-chars-signature
...
Fix docx tables silently dropped when cells contain non-text special characters
2026-08-14 15:42:27 +08:00
myhloli
86ed6f065c
feat: implement surrogate pair restoration for improved Unicode handling in PDF text extraction
2026-08-14 15:33:51 +08:00
hanwenbo
dfd270c70d
fix: render special char elements in table signature for mammoth alignment
...
_xml_table_signature only joined w:t text, dropping non-text character
elements such as <w:noBreakHyphen/> and <w:sym/>, so the XML-side
signature (e.g. "NCICTCAE") diverged from mammoth's rendering
("NCI‑CTCAE"). Alignment failed, the isolated fallback parse crashed on
cells with numbered lists, and the whole table was silently lost.
Render w:t / w:noBreakHyphen(U+2011) / w:softHyphen(U+00AD) / w:sym in
document order, replicating mammoth's dingbats mapping (incl. the F0..
low-byte fallback) for sym, and exclude OMML m:t like mammoth does.
Log table parse failures at warning level instead of debug.
Add regression tests in tests/unittest/test_docx_special_char_tables.py.
2026-08-05 14:47:01 +08:00
Xiaomeng Zhao
6b959c9972
Merge pull request #5251 from myhloli/dev
...
Dev
2026-07-10 19:49:25 +08:00
myhloli
6826a9dec7
feat: add offset duplicate character detection for improved text rendering
2026-07-10 19:48:07 +08:00
myhloli
7d7f5d146d
feat: enhance PDF font analysis by adding Latin and CJK font usage detection
2026-07-10 19:20:26 +08:00
Xiaomeng Zhao
33143d027f
Merge pull request #5246 from myhloli/dev
...
feat: prevent circular inheritance in DOCX style processing by tracking visited styles
2026-07-08 15:44:59 +08:00
myhloli
595912d044
feat: prevent circular inheritance in DOCX style processing by tracking visited styles
2026-07-08 15:32:35 +08:00
Xiaomeng Zhao
f9b75623e8
Merge pull request #5232 from opendatalab/master
...
master->dev
2026-07-03 21:13:40 +08:00
myhloli
53fde6d298
Update version.py with new version
2026-07-03 13:06:35 +00:00
Xiaomeng Zhao
d28782c853
Merge pull request #5231 from myhloli/dev
...
3.4.2
2026-07-03 21:04:52 +08:00
Xiaomeng Zhao
645ad5b143
Merge pull request #5230 from myhloli/dev
...
feat: unify PDFium page object boundary retrieval for compatibility
2026-07-03 21:03:41 +08:00
myhloli
650a7e5949
feat: unify PDFium page object boundary retrieval for compatibility
2026-07-03 20:38:38 +08:00
Xiaomeng Zhao
f89c3c8544
Merge pull request #5222 from opendatalab/master
...
master->dev
2026-07-03 16:49:46 +08:00
myhloli
d14a6e3714
Update version.py with new version
2026-07-03 08:39:18 +00:00
Xiaomeng Zhao
a41edb9f07
Merge pull request #5220 from opendatalab/dev
...
3.4.1
2026-07-03 16:32:50 +08:00
Xiaomeng Zhao
4fe8794300
Merge pull request #5219 from myhloli/dev
...
feat: enhance header and footer classification logic with new boundary checks
2026-07-03 16:19:00 +08:00
myhloli
37b2b5fea1
feat: refactor label handling and improve box label assignment in PPDocLayoutV2
2026-07-03 16:17:41 +08:00
myhloli
202a0fc82a
feat: enhance header and footer classification logic with new boundary checks
2026-07-03 15:56:56 +08:00
Xiaomeng Zhao
24f67791ea
Merge pull request #5218 from myhloli/dev
...
feat: integrate support for pdftext 0.7 PageChars and legacy char handling
2026-07-03 15:13:47 +08:00
myhloli
f10efa56f1
feat: integrate support for pdftext 0.7 PageChars and legacy char handling
2026-07-03 14:46:15 +08:00
Xiaomeng Zhao
19073865bf
Merge pull request #5216 from myhloli/dev
...
feat: enhance character selection for span filling in PDF processing
2026-07-03 10:55:22 +08:00
myhloli
6980fb7357
feat: enhance character selection for span filling in PDF processing
2026-07-03 10:53:50 +08:00
Xiaomeng Zhao
792b5b88a2
Merge pull request #5153 from opendatalab/master
...
master->dev
2026-06-18 15:27:49 +08:00
myhloli
3e60291846
Update version.py with new version
2026-06-18 07:21:15 +00:00
Xiaomeng Zhao
562ba5849a
Merge pull request #5151 from opendatalab/release-3.4.0
...
Release 3.4.0
2026-06-18 15:17:18 +08:00
Xiaomeng Zhao
2c487da3be
Merge pull request #5150 from opendatalab/dev
...
Dev
2026-06-18 15:16:53 +08:00
Xiaomeng Zhao
be2a9a2d2b
Merge pull request #5149 from myhloli/dev
...
feat: improve model source documentation for accessibility checks and update link
2026-06-18 15:16:00 +08:00
Xiaomeng Zhao
d0c5d475a1
Merge pull request #5148 from opendatalab/release-3.4.0
...
Release 3.4.0
2026-06-18 15:15:08 +08:00
myhloli
d46c353afc
feat: improve model source documentation for accessibility checks and update link
2026-06-18 15:04:29 +08:00
Xiaomeng Zhao
71ee160093
Merge pull request #5147 from myhloli/dev
...
Update mineru package version and documentation for 3.4 release
2026-06-18 14:53:37 +08:00
myhloli
c5cc50e81e
feat: update changelog for version 3.4 release with OCR upgrades and model download optimizations
2026-06-18 14:51:23 +08:00
myhloli
f6ffdd0a4a
feat: update accuracy metrics in documentation
2026-06-18 14:42:49 +08:00
myhloli
e0e58fc0d8
feat: update mineru package version to 3.4.0 in Dockerfiles
2026-06-18 11:14:06 +08:00
Xiaomeng Zhao
3a4fcc78a8
Merge pull request #5144 from myhloli/dev
...
Refactor generation logic and enhance model configuration handling
2026-06-18 11:04:28 +08:00
myhloli
3c2e8f1ffe
feat: change log level to debug for local model path configuration
2026-06-18 10:49:18 +08:00
myhloli
fb512fdfe6
feat: add model configuration handling and improve download path normalization
2026-06-18 10:39:50 +08:00
myhloli
e2cec3408f
feat: add model source configuration and auto-detection logic
2026-06-18 03:41:21 +08:00
myhloli
53cbc5c89f
feat: enhance OCR model initialization and add caching for snapshot downloads
2026-06-18 02:36:27 +08:00
myhloli
09fb4a1769
feat: refactor generation logic and enhance DataLoader memory management
2026-06-17 20:31:23 +08:00
Xiaomeng Zhao
ceff8c69f5
Merge pull request #5139 from myhloli/dev
...
Optimize OCR batch processing and enhance detection accuracy
2026-06-17 20:15:59 +08:00
myhloli
8fecc190dc
feat: refactor orientation scoring tasks and enhance batch detection logic
2026-06-17 20:00:47 +08:00
myhloli
87440e9f8a
feat: add debug logging for OCR orientation recognition scores
2026-06-17 19:37:27 +08:00
myhloli
2b8ee2df65
feat: add blank cell result filtering for OCR to improve accuracy
2026-06-17 19:32:43 +08:00
myhloli
209ada00a1
feat: add new model path for paddle table classification
2026-06-17 18:37:42 +08:00
myhloli
62f09af4e8
feat: enhance table OCR detection with batch processing and new input handling
2026-06-17 18:16:23 +08:00
myhloli
57639ebd45
feat: optimize batch processing and inference mode for improved performance
2026-06-17 15:06:54 +08:00
Xiaomeng Zhao
a5b6c3fd46
Merge pull request #5134 from myhloli/dev
...
Integrate PP-OCRv6 model and enhance OCR language handling
2026-06-17 11:12:53 +08:00
myhloli
735db71ea6
feat: implement dynamic OCR inference precision handling for improved performance
2026-06-17 01:04:06 +08:00
Xiaomeng Zhao