Lines come out in reading order — top to bottom, and left to right within a line — with a blank line between pages, so what you paste still reads like the document. What you get is the text layer already inside the PDF. A scan does not have one, and when almost nothing comes out you are told so.
The PDF reader loads only when you drop a file in, so the page itself stays light. Your file is read on your own machine and never sent anywhere. Open your browser’s network tab and watch while you use it.
No. A scan is a picture of a page with no text underneath it. When fewer than five words a page turn up, the line under the button says the file looks scanned. You still get whatever little was found, rather than a refusal. Reading a scan needs OCR, and there is none on this site.
Line breaks are. Columns are not: anything sitting side by side at the same height — two columns, or the cells of a table row — comes out as one line, read left to right. It is for getting at the words, not for rebuilding the page.
Yes, all of them, in the position they sit on the page. Nothing here decides what is furniture and what is content, so expect a running header and a page number around every page break.
Spacing in a PDF is drawn rather than stored — the text is placed piece by piece at coordinates. Pieces on one line are joined with a single space, so a word split across two pieces gains one, and a wide gap collapses to one space. Pieces are grouped by their exact height as well, so a fragment sitting a hair off the line can end up on a line of its own.
These all work the same way, in your browser. Most are free; the ones marked as needing a plan are not.
A new DeskKit tool lands every week or two. Leave an address to hear about it. Nothing else is ever sent, and one click removes you.