PDF to Text
Text is pulled out of the PDF in your browser — the file is never uploaded. Scanned pages have no text layer, so they come out empty; that needs OCR, which this tool does not do.
Drop a PDF here, or click to choose
Single file, up to 100 MB
Layout is reconstructed from the coordinates of each text run: runs on the same baseline join into one line, close runs get a space. Multi-column pages and tables will interleave — that is a limit of the PDF text layer, not a bug.
About PDF to Text
Pulls the text out of a PDF as plain text: page by page, rebuilt into lines from coordinates, then joined together. Copy it or download a .txt.
Why rebuild from coordinates instead of just concatenating. A PDF has no concept of a paragraph — only positioned text runs. Concatenating them in storage order falls apart the moment there are columns, tables or running headers. Here, runs sharing a baseline become one line, runs are sorted by x within it, and a visible gap becomes a space. That recovers most documents with normal layout.
Two things it cannot do, up front. ① Scanned pages have no text layer, so they come out empty. That takes OCR, which in a browser means downloading tens of megabytes of language data — this tool doesn't do it and won't pretend to. ② Multi-column layouts and complex tables still interleave, because a visual column is not a structure in the PDF, only guessable from coordinates — and when the guess is wrong, the left column's sentence runs into the right one.
Common artefacts: ligatures (fi, fl) turning into odd characters, stray spaces around CJK punctuation. Those come from how fonts are encoded in the PDF, not from the extraction logic. For clean output, start from a PDF exported by Word or LaTeX.
Frequently asked questions
Why is the output empty?
That PDF has no text layer — most likely a scan or a PDF made of images. That needs OCR, which this tool does not do.
Do sentences from different columns run together?
Yes, on multi-column documents. A PDF has no notion of a column, so it is guessed from coordinates — and a wrong guess joins the left column's sentence to the right one. Single-column documents are unaffected.
Will tables come out as tables?
No. Table rules are often just decorative lines, and text runs carry no structural relationship. You get lines of text.
Is my PDF uploaded?
No. Parsing happens in your browser with pdf.js; the server never receives the file.
Why did some characters turn into odd symbols?
Font encoding. When a PDF uses a custom glyph mapping, ligatures (fi, fl) and some symbols decode incorrectly. That is a defect in the source file, not in the extraction logic.
Can I extract just one page?
Yes. Put something like 1-3 or 1,5,8 in the page range field; leave it empty for all pages.
