What problem does it solve? Research and documentation pipelines need reliable full-text versions of PDF and Office files for indexing, quality checks, and note-taking tools like NotebookLM and Obsidian, but raw PDFs are often unreliable to index and lack usable text layers. ## Core Features & Use Cases - Single and Batch Conversion: Convert PDF, DOCX, PPTX, and Excel files to Markdown with uvx markitdown, including a batch script with cache skipping and automatic pdftotext fallback. - Corrupted PDF Handling: Detect damaged or textless PDFs, write explicit placeholder files instead of silently skipping, and route to arXiv/PMC alternatives, manual summaries, or OCR via Tesseract/marker-pdf. - Main Paper Detection: Filter out reference, template, and graphical PDFs in mixed directories so only the primary paper is converted. - Use Case: After downloading 40 reference PDFs into pdfs/, run the batch script to produce Markdown files in pdfs_md/ for NotebookLM indexing, with failures automatically falling back to pdftotext or placeholders. ## Quick Start Convert the PDF at 06-ref/pdfs/Smith1988.pdf to Markdown in the pdfs_md directory using markitdown, falling back to pdftotext if the output is too short.
