Compare Two PDFs

How it works

A PDF is not a word-processing document with a dependable paragraph tree. Text is often stored as positioned glyph runs, and a producer may emit words in an order that is convenient for drawing rather than reading. This tool reconstructs lines from the coordinates in each page's text layer before comparing them. It preserves the text it extracted instead of silently rewriting a document.

The comparison is page-aware. Page 1 in the original is compared with page 1 in the changed copy, then page 2 with page 2, and so on. That choice is intentionally understandable. If a new cover page is inserted near the beginning, later pages may appear changed because their page numbers shifted; the tool reports that effect rather than pretending it has solved document alignment.

What the result means

Rows beginning with + were found in the changed PDF but not the original. Rows beginning with - were present in the original but absent from the changed version. Unchanged rows remain in the page result so a reviewer can read the surrounding passage instead of seeing a list of isolated words.

The output is a review aid, not a legal certification. Text extraction can lose reading order in columns, omit text drawn as an image, and interpret ligatures or unusual fonts differently. A PDF may also contain important information outside its text layer: images, annotations, form values, attachments, bookmarks, signatures, and metadata. Inspect those separately before approving a contract, invoice, or regulated record.

Privacy and limits

Both documents are uploaded, their text layers extracted here, and the files deleted — the uploads as soon as the extraction finishes, the results within 30 minutes. There is no account and nothing is stored. That is worth knowing before comparing drafts, agreements or internal reports: it is a smaller circle of trust than a third-party comparison service, and it is not zero. For a document where it has to be zero, diff <(pdftotext a.pdf -) <(pdftotext b.pdf -) does the same job on your own machine with free software. Very large documents or pages with enormous text layers may be refused by the size budget rather than left to time out.

For scanned files, use an OCR workflow first. OCR creates a new text layer by interpreting pixels, so the comparison can then find word changes, but a recognition mistake is not the same thing as a human edit. For visual changes, render the relevant pages and compare them as images. For a structural PDF inspection, use a PDF validator or a specialized document review workflow instead of relying on text alone.

When to use another tool

Use PDF to text when you need one readable extraction rather than a two-version comparison. Use the ordinary text diff when the content is already copied as text. Use a PDF redaction tool when confidential text must be permanently covered; comparing documents does not remove anything. Use the PDF compressor when the goal is file size, not change detection.