GitLab

Release 1.3.7

refactor(table): replace ocr_engine with lang in table model prediction

- Fix lang parameter ineffectiveness during table parsing model initialization
- Resolve significant slowdown in OCR and table parsing speed in CPU mode
- Update changelog in README.md and README_zh-CN.md

- Fix lang parameter ineffectiveness during table parsing model initialization
- Resolve significant slowdown in OCR and table parsing speed in CPU mode
- Update changelog in README.md and README_zh-CN.md

- Automatically change to ch_lite model when using CPU for Chinese OCR
- This modification improves performance on CPU devices

- Remove OCR engine instantiation inside the loop
- Pass language directly to the table model instead of OCR engine
- Simplify code structure and improve readability

Release 1.3.6

feat: add support for JPEG images and update documentation 

- Add '.jpeg' to the list of supported image extensions in app.py and read_api.py
- Update projects READMEs to indicate that web_demo is deprecated

- Update the bug report template to direct users to search the MinerU Readme instead of Docs
- This change ensures users check the most relevant and up-to-date information source before reporting issues

docs: update issue templates and disable blank issues 

docs: update issue templates and disable blank issues

- Update bug report template with more detailed instructions and sections
- Add operating system version field to bug report
- Include support for MPS in device options
- Disable blank issues and provide alternative contact links
- Remove feature request template

update test case

master->dev

Release 1.3.5

refactor(office_to_pdf): simplify font checking and add logging

- Remove specific Chinese font list and detailed font checking
- Add logging warning if no Chinese fonts are detected
- Make font checking more robust and less platform-specific

- Temporarily disable Chinese font check for Windows systems
- This change allows bypassing the font check when the required fonts are not present

- Update magic-pdf installation to include specific version with full dependencies
- Add numpy, decorator, attrs, absl-py, cloudpickle, ml-dtypes, tornado, and einops as separate packages
- Specify numpy version to be less than 2

fix(test_table): update image path to use relative path

- Replace hardcoded image path with dynamic path generation
- Use os.path.join to create platform-independent file paths
- Improve code maintainability and portability across different environments

master->dev

Release 1.3.4

Dev

docs(README): update changelog for v1.3.4 release

- Update README.md and README_zh-CN.md with the latest changes
- Add new release notes for version 1.3.4
- Include improvements in OCR detection speed and page-level sorting

feat(pdf_parse): add footnote block handling in layout split

- Modify `ocr_detect_all_bboxes.py` to return footnote blocks
- Update `pdf_parse_union_core_v2.py` to handle footnote blocks in line sorting and layout splitting
- This change improves the accuracy of layout analysis by considering footnote blocks separately

test(table): update unit test to use RapidTable model

- Rename test file from test_tablemaster.py to test_rapidtable.py
- Replace TableMasterPaddleModel with RapidTableModel
- Update test case to use new model and adjust assertions accordingly
- Remove some outdated assertions and comments