EZPZ FileDrop a PDF and the text inside it is rebuilt as paragraphs and tables in a Word file. Nothing is pasted in as a picture of a page, so what you download is a document you can edit straight away.
No upload
PDF to turn back
The PDF needs real text in it. A scanned PDF is a picture of a page, so there is nothing to pull out. Files open inside this browser and are never uploaded.
Reading order, paragraphs and tables come across. The exact placement on the page does not. A PDF never records what was a paragraph or a table, so both are reconstructed from where the characters sit.
Images, stamps and signatures printed on the page are pulled out and placed in the document too. They land in reading order rather than at their original coordinates.
A scanned PDF holds photographs rather than text, so there is nothing to convert. Drop one in and the tool will say so.
Whole pages are not pasted in as pictures. What you download opens as an editable document in Word, Excel or Hangul.
The file needs real text in it. Several at once is fine.
Each page is read for where its characters sit, and those are grouped into paragraphs and tables. Progress is shown by page number.
It opens in Word ready to edit. Your original PDF is untouched.
A PDF is not a format for holding text. It is a format for drawing a page. What sits in the file is a long list of instructions saying print this character here, and nothing anywhere records where a paragraph ended or which block of numbers was a table. That information is thrown away the moment a document is exported to PDF.
So this is reconstruction rather than conversion. Where the characters sit, how far apart the lines are and how large the type is are all read back to work out the original shape. A gap wider than usual, or a line that stopped well short of the right margin, means a paragraph ended there. Cells whose left edges stand in the same place across several lines mean a table.
Being a guess, it can be wrong. What it will not do is invent a value it did not find. Pages with no text are reported as empty, and the finished document is read back and checked before you get it.
Reading order, paragraphs, headings and tables come across, and so do page breaks. Images, stamps and signatures are pulled out of the page and placed in too, though they land in reading order rather than at their original coordinates. What does not come across is the exact placement on the page, the fonts and the text colour.
The layout is deliberately not imitated. Reproducing positions would mean scattering text boxes at fixed coordinates, and a document built that way opens fine but falls apart the moment you edit one line. The point of this tool is a document you can work in.
| Item | Does it come across? |
|---|---|
| Reading order and wording | Yes |
| Paragraph breaks | Reconstructed from position |
| Headings | Type larger than the body is treated as a heading |
| Tables | Rebuilt from values that line up |
| Page breaks | Yes |
| Fonts, colour, exact layout | No |
| Images, stamps, signatures | Yes, placed in reading order |
A PDF made by photographing paper or running it through an office scanner is a single picture per page. Your eyes see letters, but the file contains none. There is nothing to pull out, so there is nothing to rebuild.
The tool says so plainly instead of handing you an empty file. If only some pages are scans, it names which page numbers came back empty. Reading text out of pictures is not something this tool does.
The whole conversion happens inside this browser. Reading the PDF and writing the new document are both done by code running on your device. There is no upload step at all.
You can check the claim. Open the network tab in developer tools (F12) and run a conversion: an upload request the size of your file either appears or it does not.
No. Reading the PDF and writing the Word file both happen inside this browser. There is no account and nothing to install.
No. The wording, reading order, paragraphs and tables come across, but the placement on the page is not reproduced. Imitating it would mean pinning text boxes to coordinates, and such a document cannot be edited.
Tables in PDFs often have no ruling lines, and where they do, those lines are drawn on a different layer that text extraction never sees. So tables are found by checking whether cells line up down the page. Narrow column gaps and merged cells can throw that off.
No. A scan is a picture of a page and holds no text at all. Drop one in and the tool will tell you immediately.
That works if you know the password. When the tool meets an encrypted file a password box appears. Type it in and run the conversion again.
No limit is imposed. Everything runs on your device, though, so a very long document takes proportionally longer. A computer handles it faster than a phone.
You convert a PDF to Word and the table comes out crooked. You assume the converter is bad, try another one, and get much the same. Some problems do not go away by changing tools.
Two situations: you have a .docx and no Word, or Word refuses the file with "Word found unreadable content". Below are the free programs that open it, what each asks of you, and how to get the text out of a file none of them will open.
You need to change one date in a contract. Converting it to Word shifts the tables and the margins, and exporting back to PDF changes the fonts too. One line, and the whole document is different.