aliansari22/pdfclear

PDFClear is an open source toolkit for working with PDF files entirely inside your web browser. Instead of uploading documents to a remote server the way many online PDF tools do, the application and all the programs it needs run on your own device, so your files never leave your computer during normal use. The toolkit covers most everyday PDF tasks. You can merge, split, reorder, rotate, and delete pages, compress files, and edit document metadata. You can add text, images, signatures, watermarks, and page numbers, fill in forms, and preview your changes before saving. It can convert JPG, PNG, plain text, HTML, and Markdown files into PDFs, turn PDF pages into images, pull text out of scanned documents using optical character recognition, and convert PDFs into Markdown. PDFClear also includes optional AI features that run locally in the browser: summarizing documents and searching them by meaning rather than exact keywords. These features load supporting model files once, after which you can disconnect from the internet and keep using them. The project explains clearly what network requests happen and why, since some supporting files, like OCR language packs and AI models, still need to be downloaded before first use. Under the hood it is built with React and TypeScript using the Vite build tool, and relies on established libraries such as PDF.js, pdf-lib, Tesseract.js for text recognition, and Transformers.js for the AI features. People who want to run their own copy can do so with Docker or by building the project directly from source with Node.js 20 or newer. The project treats security seriously, since it processes untrusted documents and files, and asks that vulnerabilities be reported privately rather than posted publicly. It is released under the Apache License 2.0 and currently has 3 stars on GitHub.