GitLab

- Remove unnecessary threading.Lock in AtomModelSingleton
- Add threading.Lock to CustomPEKModel for OCR processing
- Simplify model initialization logic in AtomModelSingleton

perf(model): optimize model initialization

- Add condition to return existing model if already initialized
- Improve efficiency by avoiding redundant model creation

perf(model): add threading lock for OCR model initialization

- Introduce a lock to synchronize access to OCR model initialization- This change improves thread safety when multiple threads access the OCR model concurrently
- The lock ensures that the OCR model is initialized only once, even in multi-threaded scenarios

feat(gradio_app): implement dynamic concurrency limit based on VRAM

- Update VRAM checking logic in app.py and model_utils.py
- Add None and type checks for VRAM values
- Adjust concurrency limit calculation in app.py
- Modify clean_vram function to handle cases with no VRAM information

- Add get_concurrency_limit function to calculate concurrency limit based on VRAM
- Update clean_vram function and rename to get_vram for better clarity
- Apply concurrency limit to the to_markdown function in the Gradio app

master->dev

Release 0.10.5

Dev -> 0.10.5

fix(pre_proc): prevent errors when imageWriter is None

refactor(para): adjust line height multiplier for block splitting,fix(pre_proc): prevent errors when imageWriter is None

- Decrease the maximum width and height from 9000 to 4500 pixels
- This change aims to prevent excessive resource usage when rendering PDFs

- Updated cut_image.py to check for NoneType imageWriter
- Prevents AttributeError when imageWriter is not provided

refactor(para): adjust line height multiplier for block splitting

- Decrease the line height multiplier from 0.8 to 0.7 for both left and right sides
- This modification aims to improve the accuracy of paragraph splitting

Release 0.10.4

fix(mkcontent): optimize paragraph text merging and language detection

- Remove overlap between bboxes for block separation
- Sort bboxes by combined x and y coordinates for better layout handling
- Comment out previous overlap removal function

- Extract language detection to block level instead of line level
- Improve logic for handling Chinese, Japanese, and Korean languages
- Refactor code for better readability and performance
- Optimize handling of hyphenated words at line ends

- Introduce language detection to determine line spacing based on language context
- Implement different spacing rules for Chinese/Japanese/Korean and Western texts
- Adjust span content handling based on detected language and span type

master->dev

Release 0.10.3

refactor(ocr): Fix the error of paddleocr failing to initialize in a multi-threaded environment

refactor(pdf_parse): adjust character-axis alignment algorithm