How to extract pages from a PDF

Entries are read left to right and each contributes its pages to the selection in the order written. A single number takes that page; a hyphenated pair takes the span inclusively, so 7-9 is three pages.

Because the list is read in order, the sequence you type is the sequence in the output. Entering 9, 2, 5 gives a document whose pages are in that order — extraction and reordering in one step, which is occasionally exactly what you want and occasionally a surprise. If you want the original order, write the entries ascending.

Repeating a page repeats it in the output, which is a legitimate way to duplicate a page. And the count in your selection is the page count you should see in the result, which is the quickest way to confirm you typed what you meant.

The two tools cut the same document in different directions. Extraction is subsetting: you take a chosen few pages out into one place and the remainder is simply not included. Splitting is partitioning: the whole document is divided into several output files that between them account for every page.

Ask what you want to end up holding. One document containing a selection is extraction. Several documents, each a slice of the original, is splitting. "Send me the appendix" is extraction; "turn this statement bundle into twelve monthly files" is splitting.

The overlap is real — extraction with the combine option off produces one file per extracted page, which is a partial split — so if the distinction blurs for your task, either tool will get you there.

Pages are copied losslessly, and a copied page brings the resources it references with it. If your two pages use an embedded font, that font subset comes along — the output is not charged a share of it, it needs the whole thing.

Where this really bites is images. A page displaying one scanned photograph carries that photograph in full, so a single extracted page from a scanned bundle can be several megabytes. The size of an extract is set by what is on those specific pages, not by how few of them there are.

If the extract needs to be small as well as short — sending it by email, say — compress it after extraction. That way the quality decision applies to the small document you are actually sending.