Recover a PDF that won't open or shows errors. The file structure is rebuilt first. Only pages whose content can't be read are redrawn as images.
Drag the file onto the page or click Upload PDF file. The arrow next to the button opens Google Drive. The repair starts as soon as the file is added.
When the repair finishes, a summary shows how many pages were recovered and lists any that were rebuilt as images or left out. If nothing could be recovered, click Try another file.
Click Download repaired file. The results screen lets you preview the PDF, download it, save it to Google Drive, open it in another tool, or delete the files.
A clean repair keeps text, links and form fields. Pages with damaged content are redrawn from what can still be displayed, so their text may no longer be selectable. Run OCR PDF on the result to make it searchable again.
After the file structure has been read, every page gets the same three checks. Its drawing instructions have to exist as a proper stream. That stream has to decompress all the way to the end without an error. And the list of fonts and images the page uses has to be a valid entry rather than something garbled. A page that fails any of these is redrawn as an image. A page with no content at all is treated as blank and kept.
What isn't checked is everything that list points to. If a page's instructions are fine but a photo or font it uses is damaged, the page passes. In the repaired file it may show a grey box where the photo was, or missing letters, even though the summary says every page was recovered. So "recovered with text, links and forms intact" means the structure is sound, not that every page looks right. Scroll through the pages with photos and unusual fonts before you rely on the file.
The tool opens the file twice: once with a strict reader that can rewrite the structure, and once with a very tolerant one that is good at displaying damaged files. The two have to agree on the number of pages. If the strict reader finds 18 pages and the tolerant one finds 20, the structure isn't trusted, and every page is redrawn from what the tolerant reader can display. The same happens if the rewritten file doesn't open again with the right page count.
That full rebuild starts from an empty PDF, so it keeps only what you can see. Bookmarks, the document title and author, attached files, links and form fields are all gone, and so is all selectable text. If that happens to a file that's mostly fine, try another copy of it before you accept the result.
Even when nothing was redrawn, the repaired file isn't the original with a patch. It's written out from scratch, and that has a few consequences:
On a page that was redrawn, links, comments and form fields on that page are removed along with the text. The image keeps the page's rotation and visible area, so it still faces the right way. It's drawn at up to 144 dpi but limited to about 16 million pixels, so a very large page such as a poster or an engineering drawing gets a lower resolution.
Other EddiesPDF tools give up on files they can't read. Protect PDF, for example, suggests trying Repair PDF first when encryption fails. A cleanly repaired file usually goes through them without trouble.
There is no fixed limit. The repair runs on your device, so very large files depend on the memory available to your browser.
Yes. It runs in any modern browser, including on phones and tablets.
Your file stays on your device. It is repaired in the browser and never uploaded.
