Rotate / Organize PDF Pages

The page order box is the whole tool

Pages are numbered from 1 and the box takes the list grammar people already use: 1-5 is a range, 3, 1, 2 lists individual pages, 8- means page 8 to the end, odd and even take every other page, and the single word reverse flips the whole document. A range written backwards really does run backwards, so 5-2 gives 5, 4, 3, 2 rather than being quietly corrected.

Unlike a page selection, an order keeps duplicates: typing 4, 4, 1 gives a three-page file whose first two pages are page 4 of the original. That is how a page gets printed twice from one document - a signature page, a form, a backing sheet - and the result says how many extra copies it made rather than silently deduplicating them.

Anything not listed is left out, and the plan spells out exactly which pages: it names them, and it says when a repeated page pushes the output past the length of the source. Everything past the last page of the document is trimmed instead of failing the job, so an order copied from a longer document still works.

Turning a page writes an instruction, not a new image

Every PDF page can carry a rotation value - 0, 90, 180 or 270 degrees - and that is what a reader applies when it displays or prints the page. Turning a page here changes that value rather than redrawing anything, so the text stays text, the images keep their original resolution and the file size barely moves.

Because rotation is an instruction, it is also a promise the reader has to keep. Every mainstream viewer applies it, but some printing pipelines, older corporate viewers and a few archive systems ignore the value and show the page exactly as it was stored. If a downstream system has to receive a genuinely upright page, the page has to be re-rendered, which no page-copying tool can do.

Rotations add up rather than replace. A page that already carried 90 degrees and is turned 90 degrees more ends up at 180, and the Rotation column in the result reports the angle each output page finishes with, so the state of the file is visible instead of guessed.

What the new file keeps, and what it cannot carry

Pages are copied as objects, which keeps the text searchable, the vector art sharp and the page size and orientation of each page exactly as it was. Nothing is re-rendered and nothing is recompressed, so the output of a reorder looks like the input - just in a different order.

Document-level features do not follow the pages, because they live once per document rather than on a page: bookmarks and outlines, fillable AcroForm fields, signature fields, and the structure tags that screen readers use. A form extracted or reorganized here arrives as printed content rather than as a working field, and a reordered file is untagged.

Repeating pages multiplies size, which is why one run is capped at 2000 pages of output: every copy is a full page held in the tab memory, and 4000 pages of a scan is a different job from organizing chapters. Password-protected PDFs are refused outright, because the file never reaches a server and so there is nowhere to type a password.

The preview, drawn by the same code the job runs

The plan under the form is computed from the page count in the preview box (1 to 2000) by the same planner the real job uses, so an order can be typed, checked and corrected before a file exists. Once a document lands, the true page count replaces the guess and the plan is redrawn against it - order and rotation are never estimated.

Each output page appears as a row: its position in the result, where it came from, and the Rotation column showing the angle it finishes with, so a page that ends up at 180 degrees reads as 180 before the download. Settings that cannot build an order - an empty list, a page missing from the source - raise a note instead of failing silently, and the output is named with -organized added unless you set a name.