How to Extract PDF Pages to a New File

How to Extract Pages from a PDF into a Separate File

An eighty page report holds four pages you need, the ones with the calculation table. A contract holds one annex, a handbook holds one chapter. The request looks the same every time, but the tool has two output modes and one setting that quietly reshapes the page format.

Extract or remove: one job from two directions

The same result is available from two tools, and the choice between them is arithmetic rather than meaning.

extract-pages builds a new document out of the marked pages. remove-pages keeps the document as a document and throws the surplus out of it.

The rule is short: when you need few and would discard many, extract. When it is the other way round, remove. Marking four pages is quicker than marking seventy six.

There is a substantive difference too. Extraction always produces a new file, which is the right choice when what comes out is going to travel as a separate attachment. Removal keeps the source document whole, and it is chosen when the result should stay the same paper, only shorter.

Two output modes

The first is the default. The second returns an archive even when a single page is being extracted, so unpacking is always part of the deal.

The per page mode is worth choosing for a specific job rather than just in case: sending each signatory their own sheet, uploading pages one at a time into a system that only accepts single page files, or filing a batch of scanned applications into folders. In every other situation one document is more convenient to open, to attach and to check.

The file names in the archive keep the original page numbers

This detail is easy to miss, and it solves the main problem of page by page extraction: working out later what came from where.

The files inside the archive are named after the page number in the source document, not after their position in the selection. Extracting pages 2, 5 and 6 gives an archive holding `page_002.pdf`, `page_005.pdf` and `page_006.pdf`. No renumbering from one, no `part_1` and `part_2`.

The value of that is direct: the names never lose their link to the original. A month later, looking at `page_047.pdf` in a shared folder, you can still say which sheet of which document it was. The number is padded to three digits, so the files sort by name in the right order up to 999 pages.

How to extract the sheets you need

1. Open extract-pages and upload the document. 2. Mark the pages on the preview grid or type the numbers in. 3. Pick the output mode: one document, or an archive of separate files. 4. Leave the page size preservation on unless you are sure. 5. Run the job and download the result. 6. Open the result and check that you caught the right sheets: the number printed on a page and its position in the file rarely match.

The page order is always the document's own

Marks on the preview do not remember the order you clicked them, and the text field does not set an order. The selection is sorted ascending before it is submitted, so `5,2,4` means the same thing as `2,4,5`.

That is a division of labour rather than an oversight. Extraction picks pages; rearranging them belongs to organize-pdf, where the order is set by dragging thumbnails.

If the task is both to pick and to rearrange, do it in two steps: extract what you need first, then put it in order.

What turning off page size preservation does

The page size switch is on by default, and with it every page travels into the new file as it was, with its own width, height and orientation.

Switched off, it brings every extracted page to a single format. That format comes neither from a setting nor from the most common page: it comes from the first page in the selection. The rest are fitted into it whole, proportions kept and centred, so nothing is stretched and margins appear around it instead.

Tested on a document whose third page is landscape while the rest are portrait. Selecting pages 3, 1 and 2 with the switch off produced three landscape pages: the first in the selection set the format for all of them.

Hence the rule: when normalising a document to one size, watch which page comes first in the selection. Getting it wrong breaks no content, but it turns the whole result on its side.

The switch is worth leaving on almost always. Turning it off earns its keep when the result goes to a printer or into a system that rejects documents made of mixed formats.

What travels with the pages

Extracted pages are not redrawn: text stays text, links stay links and image quality does not change. Only what those pages refer to goes into the new file, so the weight comes out roughly proportional. On a 20 page, 10 KB document, extracting the first five gave 3 KB and extracting one page gave a little over a kilobyte.

The caveat applies to the per page mode. A font or a logo used across every page lands in every file of the archive as its own copy, so the sizes of all the parts added together can noticeably exceed the source.

Document properties are carried over from the source unchanged. The title and author on an extracted annex will be those of the large contract it came out of. If the file is going outside, that is worth fixing through set-pdf-metadata.

The outline is carried over with its numbers recalculated: entries pointing at extracted pages survive and point at their new positions, the rest are dropped. In the page by page mode every page travels as its own document, and there is no structure left in it.

Limits and refusals

Page numbers are written comma separated, ranges with a hyphen: `2,5,9-12`. The parsing is strict, and any inaccuracy stops the whole job rather than running it partially. A number past the end of the document, a reversed range like `10-3`, a zero at the start of the count: all of these are refusals with an explanation.

The overall selection is capped at twenty thousand pages. Ordinary work never reaches that; it exists against entries like `1-99999` repeated several times.

A document under an open password cannot be processed: lift the protection with unlock-pdf and come back. A password that only restricts printing or copying does not interfere.

One last word on picking the tool. If the document needs breaking into parts by size rather than picking individual sheets out of it, that is split-pdf. If what you extract then has to join something else, that is merge-pdf.