Docling

Docling converts documents, mainly PDF but also DOCX and images, into structured Markdown or JSON output while keeping headings, reading order and table structure. For scanned files it runs OCR through engines such as Tesseract, EasyOCR or RapidOCR, and you can run it from the command line on a single file or build it into a Python script. Processing is local, so it also suits sensitive documents. It is useful for turning specifications, standards and reports into clean text that can be searched, compared or fed to AI tools. Limitation: the installation is heavy and needs Python 3.10 or later, and on poor scans the result depends on the OCR engine chosen.