PDF Text Diff

If you are used to diffing files, this is that tool pointed at PDFs. Granularity is fixed at word level, which is the right unit for prose: a line-based diff flags an entire paragraph because one article changed inside it, while a character-based diff shreds a renamed term into a confetti of single-letter edits. Word alignment marks exactly the tokens that were added or removed and leaves the rest as context. The presentation follows diff conventions rather than document-review conventions — additions and removals in a single aligned stream, unchanged text as context, so you can read the shape of the change quickly. The reason a PDF cannot simply be fed to a normal diff tool is that PDFs are binary and their bytes change wholesale on every save; the text has to be extracted and reconstructed into reading order first, which is what happens here before any alignment is attempted.