Extract text from a PDF — without uploading

Pulling the text out of a PDF is how you quote a contract clause, feed a document into another tool, check what a file actually contains, or get content out of a format that fights copy-paste. The documents involved are frequently confidential, and the usual online extractors want the whole file uploaded to return what is often a page of plain text.

Extraction here is done by Mozilla's pdf.js — the engine Firefox uses to render PDFs — running in your browser. The document is read directly on your device; there is no backend on this site to send it to, and a Content-Security-Policy blocks third-party requests. You get plain text with page separators, ready to copy or download as a .txt file.

If extraction returns nothing, the PDF almost certainly has no text layer. This is the single most common surprise with PDFs, and the reason is simple: a scanned document is a photograph of a page. To software it is one large image, and there are no characters in it to extract — no matter how clearly a human can read it on screen.

The fix is optical character recognition, which looks at the picture and works out which characters it depicts. OCR a scanned PDF does that on your device with Tesseract, and there is a walkthrough in extracting text from a scanned PDF.

A quick way to tell the two apart before you start: open the PDF in any viewer and try to select a line of text with your cursor. If the selection highlights individual words, there is a text layer and this tool will find it. If you can only drag a rectangle over the page, it is a scan and you need OCR.

You get the words, with page markers so you can tell where each page began. What you do not get is layout — PDF stores text as positioned fragments rather than as a document structure, so extraction cannot reliably reconstruct what was a heading, a caption, or a table cell.

In practice that means multi-column layouts can interleave, tables tend to come out as sequences of values rather than rows, and headers and footers appear once per page in the middle of the flow. This is inherent to the format, not a shortcoming of any particular extractor. For prose documents the output is usually clean; for a densely formatted report, expect to tidy it.

Useful next steps, all client-side: word and character counter for the extracted text, diff checker to compare two versions of a document, and text sorter to sort or deduplicate the lines.