Extract pages
What Extract pages actually does
Lifting a selection out of a PDF creates something new, and the new thing deserves to arrive intact rather than as a stack of pictures. What usually comes back has the right imagery and nothing around it: a navigation panel with no entries, anything interactive rendered inert, and an identity belonging to whichever library performed the copy. This builds precisely the selection asked for, arranged precisely as asked, and brings its surroundings along. Arrangement is yours: the request reads left to right, so writing 5, 1-3 puts the fifth sheet ahead of the first three, and naming a sheet twice produces it twice. Navigation belonging to lifted sheets travels and is renumbered where those sheets landed, while navigation belonging to what stayed behind is discarded rather than misdirected. Typed-in values and the identity of the original, its name, author, subject and keywords, all make the trip.
How to use it
- Open a source document. This tab measures its length and counts its navigation entries and interactive boxes before writing anything at all.
- Say what to lift. Clicking builds a rising sequence; writing unlocks the full vocabulary, spans such as 3-7, falling spans such as 7-3, unbounded spans such as 4-, plus end, odd, even, all and reverse.
- Watch the running tally and the summary beneath it, which say how long the new document will be and what becomes of every kind of content the original was carrying.
- Build it, then collect the new document alongside a JSON note listing the originating sheets in the sequence they were written and the navigation and interactive counts that made it across.
Useful for
- Lift a single chapter out of a thick manual so it opens on its own, navigation panel and all.
- Take just the countersigned sheet from a returned agreement without losing what was typed into it.
- Assemble a reading sequence from a scan by writing sheets out of order, say 1, 5-7, 3.
Limits worth knowing
- A single unprotected source, no larger than 50 MiB, producing no more than 200 sheets. Protected or corrupt sources halt the job by name. Its terminal counterpart works at any length.
- Asking for a sheet beyond the end halts the job, and a request resolving to nothing at all halts it too, rather than producing a document with nothing inside.
- Where a sheet is written more than once, its navigation entry binds to the earliest copy. Cross references, attached files, printed sheet labels and accessibility structure do not travel.
- Cryptographic signing does not survive, because a selection is a different document from the whole that was signed.
Questions people ask
Can I extract pages in a different order from the original?
Yes. Requests read left to right, so 5, 1-3 puts the fifth sheet first and the opening three behind it. Falling spans such as 7-3 work, and so does the word reverse.
Do the bookmarks come with the extracted pages?
They do now. Until this release the lifted selection arrived with an empty navigation panel. Entries belonging to lifted sheets are re-aimed at wherever those sheets landed, and the accompanying note counts how many made it.
Can I take the same page twice?
Yes. Name a sheet repeatedly and it appears repeatedly. Its navigation entry attaches to the earliest copy written.
What is the difference between this and splitting a PDF?
Splitting carves one source into several outputs at once. This produces exactly one output holding exactly the selection named, arranged as named, which is what suits a single excerpt rather than a set of parts.
Does anything leave my machine?
Nothing does. Opening, previewing, lifting and saving all happen inside this tab using code delivered with the page.