Why copying from a PDF gives you a mess
A PDF does not store paragraphs. It stores pieces of text and the exact spot on the page where each one is drawn, one line at a time. Select everything in a PDF reader and paste it, and you get what the file holds: a hard line break at the end of every line, words cut in half where they were hyphenated, and the running header, footer and page number repeated in the middle of your text every time a page ends.
We measured it on a three-page report printed from a web page. Copying line by line gives 170 lines of text; the same content is 16 paragraphs. Along the way 36 words were split across two lines with a hyphen, and the header and “Page 1 of 3” footer appeared three times each. The Paragraphs layout here returned all 1,452 words with every paragraph whole, the hyphenated words joined, and no header or page number in sight. The test file and the script that checks it word for word against the original text are part of our test suite.
Three layouts for three jobs
- Paragraphs — for pasting into Word, Google Docs, an email or a translation tool. Lines are joined into paragraphs using the spacing between them, indents and short last lines; list items stay separate; and a sentence cut off at the bottom of a page is joined to its end on the next page.
- Lines as in the PDF — every line on its own line, with a blank line between paragraphs. Useful for poems, addresses, code listings and anything where the line breaks mean something.
- Keep columns and table spacing — text is placed by its position
on the page, with spaces lining up the columns, much like
pdftotext -layout. Choose this for tables, invoices and forms, where reading order matters less than what sits next to what.
Two-column pages read in the right order
Many tools sort the text on a page from top to bottom, which mixes the first line of the left column with the first line of the right column, and so on down the page. The Paragraphs layout follows the order the text was written into the file, which for articles, papers and newsletters is column by column. In our two-column test all four column paragraphs came out whole and word-for-word identical to the source.
Scanned pages are named, not skipped
If a page is a scan — a photograph or scanner image of paper — there is no text in the file to extract, only a picture of text. Most online extractors either return nothing without saying why or quietly upload the file to run OCR on a server. This tool checks every page, and the report lists the ones that are images by page number, such as “Page 4 is a scanned image”, so you know exactly which pages still need OCR. It does not guess at text it cannot read.
The same check catches a rarer problem: a PDF whose font has no map from its shapes back to letters. Such pages copy as rows of symbols in every reader. The report names those pages too, instead of handing you nonsense as if it were text.
Headers, footers and page numbers
A line near the top or bottom of the page that repeats on at least half of the selected pages is treated as a header or footer and left out; numbers inside it are ignored when comparing, so “Page 3 of 12” and “Page 4 of 12” count as the same line. A line that is only a page number is dropped when it is the first or last line on the page. Numbers inside a table near the bottom of a page are not mistaken for page numbers — an early version of this tool did exactly that, and the test that caught it still runs. The report tells you how many lines were removed and shows one of them.
Just some of the pages
Type pages and ranges such as 1-3, 7 in the Pages box to extract only those,
and tick “Mark where each page starts” to put a — Page 7 — line in front
of each page’s text, which makes it easy to quote with a page reference. Copy the result
with one click or download it as a UTF-8 .txt file.
Private by design
Text is read from the PDF inside your browser tab. The file is never uploaded, which is the point for contracts, payslips, medical letters and anything else you would not email to a stranger. There is no account and no page limit. Password-protected PDFs open once you type the password, and PDFs that block copying can still be read, because the text is taken from the file rather than copied through a reader.