What this preserves, what it drops, and why you can trust it
Your file never leaves your machine. Everything — parsing, page surgery, saving — happens in this page. The upload-me PDF sites exist to harvest exactly the documents people most need edited: contracts, statements, tax forms.
What we do not touch comes back byte-for-byte. Fonts, images, page contents and anything else we do not model are copied through as their exact original bytes, not re-encoded. The test suite proves it the way that matters: every page of every saved file renders pixel-identical to its source, across a locked corpus of government PDFs from 1996 to today.
Form fields survive splitting, reordering and rotating a single document (pruned to the pages that remain). Merging two different forms into one is where every tool invents lies about field names — so on merges we keep the pages and their appearance, and drop interactivity rather than corrupt it.
Honestly not here yet: page thumbnails (we will not bolt on a rendering engine we did not write), bookmarks/outlines are not carried over, and password-protected files are refused outright rather than half-handled.
You can read the whole program. The download button on the front door hands you this entire tool as one file that is its own source.