Classify and split mixed document packets with DocJev
DocJev shows how to extract complete page text from PDF, DOCX, or PPTX files, classify documents from natural-language rules, and find boundaries between adjacent documents.
Original by Jerry LiuClassificationGitHub README and engineSource reviewed
Before you dive in
What you’ll find in the original
Use local LiteParse extraction for complete page text and reserve optional cloud OCR for difficult inputs.
Define natural-language categories, then choose classification or packet splitting through the Python library, CLI, or local app.
Measure OCR and decision timing separately and inspect packet boundaries, especially adjacent documents of the same category.
Worth knowing
Requires Python 3.11+ and a TypeSafe API key; optional LlamaParse uses a cloud service. The included 40-document report is project evidence, not a broad independent benchmark.