GitLab

feat(ocr): add PP-OCRv5 models and update configurations

- Add new PP-OCRv5 detection and recognition models
- Update arch_config.yaml with new model architectures
- Modify models_config.yml to include PP-OCRv5 models for ch_lite configuration- Change dictionary file for ch_lite to ppocrv5_dict.txt

fix(ocr_mkcontent): improve image handling and footnote integration in markdown output

master->dev

Release 1.3.11

docs(changelog): remove pdfminer.six version pinning from release notes

docs(changelog): remove pdfminer.six version pinning from release notes

Release 1.3.11

fix(modeling): escape backslashes in LaTeX command descriptions

fix(requirements): update pdfminer.six version and restrict torch version upper limit

fix(setup): update python_requires to support Python 3.10 to 3.13

fix(Dockerfile): update modelscope installation command to use mirror

docs(installation): update Python version and CUDA installation instructions

docs(README): reorder installation commands for clarity

feat(modeling): add regex patterns for LaTeX symbol replacements

feat(modeling): add 'protect' command to removal patterns

Update app.py: Fix parameter parsing in /file_parse endpoint

I have updated the `/file_parse` endpoint in `app.py` to correctly handle boolean and string parameters when they are sent via `multipart/form-data` requests (commonly used for file uploads). Previously, these parameters were not being properly parsed because FastAPI expects them to be passed as query or JSON body parameters by default.

### Changes Made:
- Added `Form(...)` to all non-file parameters (`parse_method`, `is_json_md_dump`, `output_dir`, and return flags like `return_layout`, etc.).
- This ensures that FastAPI correctly reads these fields from form-data, allowing clients to send both files and structured configuration options in the same request.

### Why This Change Was Needed:
- When using `requests.post(..., data=data, files=files)`, the `data` dictionary is sent as form-encoded data.
- Without explicitly declaring these fields with `Form(...)`, FastAPI does not bind them correctly, leading to default values always being used (e.g., `False` for boolean flags).
- This change allows the API to accurately reflect the client's intent and enables features like `return_layout`, `return_images`, etc., to work as expected.

This update improves compatibility with HTTP clients that rely on standard form-based file upload mechanisms while preserving the existing behavior of the API.

feat(model): add logging for batch image processing

- Add logger info for each batch processed
- Include batch number and page count in log message

update version

Release 1.3.10

feat(model_utils): adjust table detection threshold and add features