PDF to Text Converter

How it works

1. Drop in a PDF.
2. Optionally name the pages you want, leave it blank for the whole document.
3. The text is pulled out on your own machine, page by page.
4. Download it as a plain .txt file.

About this tool

Extract the text stored in a PDF, with paragraphs and line breaks reconstructed from where the words sit on the page. Useful when you need to edit, search or quote the content rather than look at it. It cannot read a scanned document, a scan is a photograph of a page, where the words are pixels rather than characters, and pulling text out of those needs OCR. If you cannot select the words in a PDF reader, there is nothing here to extract either.

What this does, and what it does not

This extracts the words that are stored in the PDF. It reconstructs paragraphs and line breaks from where the text sits on the page, which works well for ordinary documents. It is not a layout conversion: columns, tables and text boxes come out in roughly the order the file stores them rather than the order you read them, and images, fonts and positioning are not preserved. If you need the page to look the same, render it as an image instead.

Why is my PDF empty?

Because it has no text in it. A PDF made by scanning paper is a photograph of a document, the words are pixels, not characters, and there is nothing to extract. Recognising text in an image needs OCR, which is a genuinely different and much heavier job than this. A quick test: if you cannot select the words with your cursor in a PDF reader, this tool will not find them either.

Is the order always right?

Usually, not always. A PDF stores text as positioned fragments with no notion of reading order, so anything reconstructing it is making an educated guess. Single-column documents come out reliably. Two-column academic papers, newsletters and anything with sidebars can interleave. Check the result before relying on it.

What happens to tables

They stop being tables. A table in a PDF is not a grid with cells in it; it is a set of lines drawn on the page and a set of text fragments positioned between them, and the file records no relationship whatsoever between the two. What comes out is the cell contents in the order the file happens to store them, usually row by row but sometimes column by column, with the rules and borders gone entirely. For a two-column price list that is often still usable. For a financial statement with merged cells and a total row it is not, and the honest advice is to open the original and retype the numbers rather than trust a reconstruction nobody can verify at a glance.

Hyphens, ligatures and the other small damage

Three things routinely survive extraction in a state that surprises people. Words broken across a line end keep their hyphen, because the PDF stores exactly what was printed and nothing records that manage- and ment were one word. Ligatures - the single glyphs a typesetter uses for fi and fl - sometimes extract as one character that a later search for "file" will not match. And quotation marks and dashes come out as the typographic characters the document actually used, not the ASCII ones a keyboard produces. None of these is a fault in the extraction; all three are the document being reported accurately. A find-and-replace pass afterwards clears all of them in seconds, and it is worth doing before the text goes anywhere that will be searched. A .docx has none of these problems, because it stores words rather than glyph positions — Word to text is the same job on a document that kept its structure.

Why this runs on your machine

Because of what people extract text from. The PDFs that arrive at a converter are contracts, medical letters, bank statements and legal filings - documents whose contents are the entire reason somebody wants them as text, and the last thing they would post to a stranger's server if they thought about it for a second. The extraction here is done by your browser, page by page, and nothing about the file is transmitted anywhere. That is also why there is no page limit and no queue: there is no server to be busy.