Question 1

Where do the bookmarks come from?

Accepted Answer

From what the document already says. If it has a contents page, that is read directly — the author wrote those titles and page numbers, so they are not a guess. If it has none, headings are found by type size instead: the size most characters are set in is the body, and anything meaningfully larger is a heading, ranked into levels. A file that already has an outline offers that as a third starting point, so you can edit what is there rather than start over.

Question 2

The contents page says 6 but that chapter is on sheet 9. Does it get that wrong?

Accepted Answer

No, and this is the part that is easy to get wrong. A book's printed page 1 is usually the fourth or fifth sheet of the file, because the cover, the copyright page and the contents itself are not counted — so a parser that trusts the printed numbers puts every bookmark a few pages early, pointing at a real page, which is the kind of mistake that survives a quick check. Instead each contents entry is looked for in the body, and the shift that the most entries agree on is applied to all of them. The page says what that shift worked out to be.

Question 3

Are the bookmarks right without checking?

Accepted Answer

Check them. Reading a contents page is exact when the layout is regular and drifts when it is not — a title that wraps onto two lines, a row with no page number, a chapter whose title in the contents differs from the one on the page. That is why the list is editable rather than applied straight away: you can rename an entry, move it a level in or out, change its page or delete it before anything is written.

Question 4

Does it work on a scan?

Accepted Answer

Not on its own, because there is no text to read — neither a contents page nor a heading exists as far as the file is concerned, only pixels. Run OCR PDF first: it adds a text layer without touching the scan, and this tool then works on the result like any other document. The page says so rather than returning an empty list and leaving you to guess why.

Question 5

What does adding bookmarks do to the file?

Accepted Answer

The pages are untouched — we compared every character of extracted text before and after and they are identical. The outline itself is tiny: six entries on a thirteen-page book cost 471 bytes. The whole file is rewritten as it is saved, though, so the byte count will not match the original exactly; on that book it came out smaller, because the library that writes it compresses the document structure differently from the one that made it.

Question 6

Will the bookmarks survive if I split or merge the file afterwards?

Accepted Answer

Yes. Every tool here that rebuilds a document — merge, split, extract pages, remove pages, reorder — reads the outline and writes it back against the new page numbers, and merging groups each file's bookmarks under a heading of its own. A bookmark whose page did not make it into the output is dropped rather than repointed at whatever took its place. One thing is not carried: a restored bookmark opens at the top of its page rather than at the exact scroll position it originally named.

Add PDF bookmarks

FAQ