GitLab

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- Remove some outdated assertions and comments

feat(model): add text region handling and improve overlap resolution

feat(model): add text region handling and improve overlap resolution 

- Add text region handling in get_res_list_from_layout_res function
- Implement remove_overlaps_min_blocks function to handle overlapping blocks
- Update OCR region handling to include text regions
- Improve overlap resolution for all regions in layout results

fix:Chinese Character Garbling in PPTX/DOCX Conversion by Adding Font Check and Installation

build(docker): add torch and torchvision installation

build(docker): add torch and torchvision installation

build(docker): add torch and torchvision installation

- Add pip install command for torch and torchvision
- Specify version2.3.1 for both packages
- Use Aliyun mirror for faster download

Dev

master->dev

Release 1.3.3

build(deps): downgrade dill to 0.3.8 for doclayout_yolo compatibility

build(deps): downgrade dill to 0.3.8 for doclayout_yolo compatibility

- Change dill dependency from >=0.3.9,<1 to >=0.3.8,<1
- Update dill version in both general and specific requirements

fix(magic_pdf): correct range for images in document analysis