When a PDF's words come out cleanly, and when they cannot
A PDF that was made from a document carries its words as text, and those words can be read straight back out. A PDF that was made by scanning paper carries a picture of the words instead, and there is nothing in it to read. Which one you have decides whether this tool can help, and the layout of the page decides which of its two orders to choose.
Updated
Which tool you need
A report, an article or a letter
PDF to Text in Reading order. Columns are read one after the other and paragraphs kept apart, which is what you want to paste into an email or a document.
A table, an invoice or a form
PDF to Text with Keep the layout. Each line of the page stays a line and each piece of text keeps its column, so the rows still line up in a plain-text editor.
Text you want to paste somewhere that wraps it for you
Reading order with Join lines into paragraphs turned on. Each paragraph comes out as one line, and words split across lines by a hyphen are put back together.
A scanned document or a photo of a page
Not this tool, yet. A scan has no words in it to read, only a picture of them, and text recognition (OCR) is not available here. The result names every page that came out empty.
You need the pages themselves, not their words
Extract PDF Pages keeps the pages as a PDF, with their fonts, pictures and layout, for someone who needs the document rather than its text.
How to tell a scan from a real PDF
Open the PDF in any viewer and try to select a word. If the word highlights on its own, the page carries text and this tool will read it. If the whole page highlights as one block, or nothing highlights at all, the page is a picture.
Many documents are both: a typed report with a scanned signature page, say. Every page that has text is read, and the result lists the ones that had none.
Why the order can come out wrong
A PDF does not record the order its words are read in, only where each one is drawn, and the program that made it may have drawn them in any order at all. The order here is worked out from position: columns where there is a clear gap running down the page, lines from top to bottom, words from left to right.
That is right for ordinary pages. A page with text boxes scattered over a picture, a magazine spread, or a sidebar that runs into the main text can read differently from how you would read it. Keep the layout sidesteps the question by keeping the text where it sits.
What the .txt file holds
Plain UTF-8 text and nothing else: no type sizes, no bold, no pictures, no links. The pages follow one another with a blank line between them, and with page marking on, every page gets its numbered line even when it had no text, so the numbering never skips.
Text set sideways or vertically, such as a label running up the margin or a stamp at an angle, is not mixed into the lines beside it. It comes after the rest of that page's text, as a block of its own. If none of the pages you chose has any text at all, no file is offered, rather than an empty one.
Plain text, Markdown or a summary
Plain text suits a destination that does its own formatting: an email, a word processor, a search index or a script. PDF to Markdown reads the same words in the same order and writes headings and lists into them, which suits a wiki or a notes app. Summarize PDF reads them too and keeps only the sentences that carry the most, each with its page.
Keep the layout is the one mode neither of those offers. It is as close as a text file gets to the look of the page, and the price is that a page of prose keeps the line breaks and indents it was set with.
What to know before you start
- A scanned page has no text to read and comes out empty; the result says which pages those were.
- Text drawn as shapes, which some design programs do to headings and logos, is not text and cannot be read.
- A locked PDF has to be unlocked first, and the text is read in your browser rather than anywhere else.
- A page carrying an unusually large amount of text can be cut short, and the result says so when that happens.