Repository navigation
1.0.0
π¦ Version 1.0.0 β Stable Release of PdfPlucker
I'm excited to announce the first stable release of PdfPlucker! π
This version marks a major milestone, delivering full functionality, performance improvements, and a smoother user experience.
π What's New?
-
Full CUDA and CPU Support
Added complete support for GPU (CUDA) and CPU execution. Includes custom export options like:--force-ocr--markdown
-
Integrated Logging System
Terminal outputs are now replaced with structured log files, making debugging and monitoring much easier. -
New Python Function API
You can now use a Python-nativepdfplucker()function as an alternative to the CLI commands β perfect for integration in scripts and notebooks.
π οΈ What's Fixed & Improved?
-
Parallel Processing Stability
Fixed issues withProcessPoolExecutorto ensure reliable parallel processing. -
Batch Timeout Handling
Improved timeout logic that adapts better to different batch sizes. -
Performance Optimization
Introduced efficient garbage collection handling to reduce memory usage and improve runtime stability. -
Dependency Update
Addedtorch(CPU version) as a dependency to improve out-of-the-box compatibility.
β What's to come?
- Support for different export modes, with referenced and embedded images
- ETA feature for logging
- Low memory mode
Keep up with this project to see the new features to come!
π‘ Tip
If you're upgrading from a previous version, make sure to check out the new CLI flags and try out the function API inside your Jupyter notebooks or scripts!