Merge & split PDF

Why this is lossless and fast

A PDF is a tree of objects with a page list at the top. Each page points at its own content stream — the drawing commands — plus the fonts and images it needs. Merging two documents means building a new tree whose page list references the pages of both; splitting means building trees that reference a subset.

The content streams themselves are copied byte for byte and never decoded. Nothing is re-compressed, nothing is rasterised, no image is re-encoded. Text stays selectable because it is the same text objects; a scan keeps exactly the JPEG it arrived as. This is also why it is quick even on large files: the work is proportional to the number of objects, not to the number of pixels.

It is worth knowing what that implies about size. A merged file is roughly the sum of its inputs, because all the content is still there. Merging is not a way to make documents smaller, and any tool that shrinks a merged PDF noticeably has re-encoded something.

Page ranges, and why bad ones are refused

Ranges are written the way people write them: 1-3, 5, 8- gives three files, and the open-ended 8- means page eight to the end. Each range becomes its own document, so you get one file per range rather than one file with the pages you chose.

A range that runs past the end of the document is rejected rather than quietly trimmed. Silently clamping 1-99 to 1-20 would hand you a file that looks complete and is missing everything you asked for beyond page twenty — a mistake you would find much later, when the submission was already sent.

Encrypted files

A password-protected PDF is refused when you add it, and says so. That is deliberate: the alternative is discovering the problem partway through a merge, with an output file that is missing one input. Open it in a PDF reader, remove the protection, and add it again.