PDF to DOCX

Reading structure back out of coordinates

A PDF has no paragraphs. It has glyphs, each with a position, a size and a font, and everything a reader perceives as structure is an effect of where those glyphs were put. Converting one to Word means inferring the structure back — and that inference, not the file format, is the whole problem.

Characters sharing a baseline become a line. Horizontal gaps wider than a fifth of the font size become word spaces, because a PDF is free to position every word by hand and encode no spaces at all. Lines become a paragraph until something says otherwise: a bigger gap than the leading, a change of size, a new indent, or a previous line that stopped short of the column edge — which is what the last line of a paragraph looks like from the outside.

The switches in the toolbar

Rebuild paragraphs

Merge visual lines back into flowing paragraphs, rejoining words a typesetter hyphenated across a line break. Switch to one line each for poetry, code, forms and anything else where the line break is the content.

Pages

Read only a range. On a long report this is the difference between a second and a minute, since almost all of the time goes on parsing pages.

Headings

Promote short lines that are meaningfully larger than the body text to Heading 1-4, so the result has a real outline and a navigation pane in Word.

Lists

Recognise bullet glyphs and 1. / a) / iv. markers, strip them from the text and rebuild the item as a real Word list, indented to the level the layout implied.

Trim headers

Drop lines that repeat near the top or bottom of most pages, and lines that are nothing but a page number. Without this, a running header appears once per page in the middle of your document.

Page breaks

Keep an explicit break where each PDF page ended. Off by default, because a Word document reflows and a break in the middle of a sentence is worse than no break at all.

Where the inference gives up

• Scanned pages. A scan is a picture of words. Turning it back into words is OCR, which this tool does not do — those pages come across as images so the document stays complete, and each one is named in the report.
• Tables. Ruled tables usually arrive as their text in reading order. Telling a real table apart from text that merely lines up is guesswork, and guessing wrong scrambles the content rather than merely losing the grid.
• Bold and italic. Weight and slope are only recoverable when the PDF referenced a font instead of embedding one. Where it is not, headings still come out bold because they are headings, and body text comes out plain.
• The page as a picture. Word reflows and PDF does not, so the result reproduces the structure rather than the pixel positions. If the visual layout is what matters, the PDF is already doing that job.

Frequently asked questions

How can a PDF become an editable Word file at all?

A PDF does not store paragraphs — it stores glyphs at coordinates. So the converter reads the geometry back: characters at the same height become a line, the spacing between lines decides where one paragraph ends and the next begins, and larger-than-average lines become headings. The result is a real .docx with real paragraph and list styles rather than one giant text block.

Does it work on a scanned PDF?

Only partly. A scan holds pictures of words, not words, and reading them back needs OCR, which this tool does not do. Pages with no text layer are carried across as images so the document is still complete, and each one is reported so you know which pages you would have to retype.

Why does the layout not match the PDF exactly?

Because a Word document reflows and a PDF does not. The converter reproduces the structure — headings, paragraphs, lists, alignment, indents — rather than the pixel positions, which is what makes the result editable. If you need the visual layout preserved instead, the PDF already does that job.

What happens to tables and columns?

A ruled table usually comes back as its text, in reading order, rather than as a Word table: telling a real table apart from text that happens to be aligned is guesswork without the borders, and guessing wrong scrambles the content. Multi-column pages are read column by column when the spacing makes the columns obvious, and left to right otherwise.

Can I convert only part of the document?

Yes. Set a first and last page in the toolbar and only that range is read. It is worth doing on a long report: most of the conversion time is spent parsing pages you may not need.

What are the headers and page numbers switches for?

A running header or footer repeats on every page of the PDF and would otherwise appear dozens of times in the middle of your Word document. With trimming on, lines that repeat near the top or bottom of most pages, and lines that are nothing but a page number, are left out.