Compare PDFs
Diff two PDFs by page count, page size and metadata. Runs entirely on your device.
Runs on your device. The file is never uploaded.
Compare PDFs reads two documents and prints a structural report: page count first, then the size and rotation of every page the two share, then four metadata fields. Each finding is one line naming both values. Only title, author, subject and producer are read, so a changed keyword list or creation date goes unreported.
Questions
Does it compare the words on the page?
No, and this is the main thing to know. It is a structural comparison only: page count, page size, page rotation and four metadata fields. Two documents with completely different text but the same shape will report no differences. For a content comparison, run PDF to text on both and diff the results.
What exactly does it check?
Page count first. Then, for every page position the two documents share, the width and height in points and the rotation angle. Then the title, author, subject and producer metadata. File sizes are reported at the end but are not counted as a difference, since two files can differ in bytes and be identical in structure.
Why does it say "drop exactly 2 PDFs to compare"?
Because a comparison needs a pair, and the tool refuses to guess which two of three you meant. Drop one PDF and you get the same message. The first file you drop is the one on the left of every reported line, so the order you drop them decides how the differences read.
What does the output look like?
A plain text report, with the two file names on the first line and one line per finding. Differences are written as one value versus the other, for example a page size line naming the page number and both dimensions. When nothing differs it says so, with the line "no structural differences found".
What if one document has more pages than the other?
The page count difference is reported, then the per-page checks run only over the pages both documents have. Pages past the end of the shorter file are not examined, since there is nothing to compare them against. Nothing is added or padded to make the counts match.
Do I get a file back?
No, only the report on the page. There is nothing to download, because nothing was created or modified: both of your PDFs are read and left exactly as they were. Copy the text out if you need to keep it. Everything runs in a Web Worker on your device with no upload.