Extract the text from a PDF
Get the raw text of a document as a plain file you can search, paste, diff or feed to another program.
Extract the text from a PDF online
Extracting text is what you want when the formatting is in the way: you need the words themselves, to search them, to paste them into a script, to compare two revisions, or to feed them to something that only reads plain text. PDF Vision reads the document server-side with Poppler — the same pdftotext engine used across Linux — and returns the text of each page separately as well as one combined block, downloadable as a .txt file. The important condition is that the PDF must actually contain text. A file produced by Word, by an accounting package or by a web export does. A scan does not: it is a picture of words, and no extractor will find characters that were never written into the file. When the result comes back empty or nearly so, that is why — and the tool says so and points you to OCR, which recognises the letters in the image first. Files up to 10 MB and 100 pages, processed on our servers, so an account is required.
How to extract the text from a PDF
Upload the PDF
Up to 10 MB and 100 pages. The document should be one with selectable text — if you can highlight a word in a reader, it will extract.
The text is read out
Poppler reads the document server-side and returns the text of every page, kept separate, plus one combined block.
Download the .txt
Save the plain text file, or copy what you need straight from the page.
Why use it
Plain text, not a re-layout
Converting to Word tries to rebuild the design and can rearrange things. Extraction gives you the words as they are, which is what a script, a search or a comparison actually needs.
Page by page as well as whole
Each page is returned separately, so you can quote a specific page or process a document one page at a time instead of untangling one long block.
Tells you when the PDF is a scan
An empty result is diagnosed rather than left as a mystery: a document with no text layer is flagged and you are sent to OCR instead of trying the same thing again.