Make Scanned PDFs Searchable with OCR

A scanned PDF is not text at all, it's a photograph of a page embedded in a PDF wrapper. There is nothing to search, select, or copy because there is no text layer, only pixels. PDFHaul checks every page for an existing text layer before doing anything else, so pages that already have real text are never touched.

For pages that lack a text layer, PDFHaul rasterizes the page and runs it through an OCR engine that recognizes individual words and their positions. That recognized text is then written back into the PDF as an invisible layer, positioned precisely under the original scanned image at the same coordinates the words appear visually.

The result is a PDF that looks pixel-for-pixel identical to the original scan, but now has a real, invisible text layer underneath. Opening it in any PDF reader lets you search, select, and copy the text, even though visually nothing changed. Mixed documents, where some pages are scanned and others already have text, are handled in one pass: only the pages that actually need recognition are processed.

The OCR engine recognizes text in 15 languages in a single pass, so a document doesn't need to be pre-sorted by language before processing. Accuracy depends on scan quality: clean, reasonably straight scans at typical document resolution recognize well, while blurry, heavily skewed, or very low-resolution scans reduce accuracy, the same limitation any OCR system has.