Rotate PDF pages

Correct sideways or upside-down pages without uploading the document. Pick the angle, then apply it to the whole file or to ranges such as 1-3, 7, 10-12. An empty range field means every page. Ranges are checked against the document first, so a typed page that does not exist comes back as a message giving the real page count rather than a half-finished download.

Rotation is recorded as an instruction on the page, not baked into it. Nothing is re-rendered, so text stays selectable, links keep working, vector drawings stay sharp, and a scan keeps the resolution it arrived with. The page also keeps its recorded width and height: a rotated A4 page is still an A4 page that a reader draws turned, which is why a preflight or print tool that reads dimensions may still report it as portrait.

Angles add to whatever rotation the page already carries. That is the behaviour to plan around in a document scanned in mixed orientations, where some pages are already turned. Rotating everything by ninety degrees fixes the sideways pages and knocks the upright ones over, so the reliable approach is to note the pages that are wrong, rotate that range alone, then check the result. Two applications of one hundred and eighty degrees return a page to where it started.

The case that sends most people here is a document feeder that took a landscape page the wrong way round, leaving pages four to six on their side in an otherwise readable file. Reading it means tilting a laptop; printing it wastes paper. A range and one angle deal with it.

Files are read and written by a worker inside the tab, so nothing is uploaded, and there is no account or queue. Encrypted documents must be unlocked before their structure can be read. On very large files the ceiling is the memory the browser has available, not a plan limit.