DOCX to TXT — the body travels, the presentation does not
A customer attaches a Word complaint and you need to search it, diff it or paste it into a ticket without opening Word. The conversion reads the document's paragraph body and writes it as plain text: the reading order of the main story survives, while everything that describes how the page looked — fonts, margins, page breaks, embedded images — has no representation in a text file.
Page layout, fonts, margins, headers, footers and page breaks are not reproduced: the text stream is reflowed, not re-laid-out.
Processed on your device. Files aren’t sent anywhere.
What the plain-text body is
The reader walks the document's main story in order and emits each paragraph as a line group. That is the same order a reader would read in, so a converted file can be searched with ordinary tools.
- Paragraphs keep their document order and their text.
- Heading text is kept, but the heading level is not marked — a heading line looks like any other line.
- Bold, italic and colour contribute nothing; the characters stay, the formatting does not.
- Table cells are emitted as separate lines, so a row of three columns becomes three lines.
- Lists become plain lines; the numbering or bullet character is not reconstructed.
- Output is UTF-8 text, so accented characters and non-Latin scripts arrive as themselves.
Parts that never reach the text
A Word file holds more than a story. The parts below are outside the body stream, which is why a converted file can look shorter than the document you opened.
| Document part | In the TXT output | Why |
|---|---|---|
| Body paragraphs | kept in order | this is the body stream |
| Table cells | kept as separate lines | cells are text, but the grid is not |
| Footnote text referenced by a marker | not part of the body stream | notes sit outside the main story; HTML and Markdown output can carry linked notes |
| Comments and text boxes | not a reliable part of the reading order | they are annotations, not story flow |
| Images, charts, objects | dropped | a text file cannot hold them |
| Headers, footers, page numbers | dropped | they are page furniture |
| Fonts, margins, page breaks | dropped | layout has no text counterpart |
Superscript markers without their notes
A body paragraph that references a footnote keeps its marker position but the note text is not stitched into the body. If the notes matter — a legal clause, a citation — convert to HTML or Markdown instead, where notes can be carried as linked content. Do not assume a single rule for all outputs: footnote handling depends on which format you ask for.
The practical symptom is a converted file that is shorter than the document. Compare the character count with the word count you expected before concluding that content is missing.
Diagnosing a suspicious conversion
When a search finds nothing, the usual cause is shape, not loss.
- Convert the .docx and download the .txt.
- Open it in a text editor rather than a word processor, so nothing is reflowed again.
- Search for one phrase you know is in the body — if it is present, the reader worked.
- Search for a value that spans two table cells; expect it to be split across lines.
- Check the character count against the 400000-character output cap.
- If notes or structure are needed, re-run the same file toward HTML or Markdown.
Limits that stop a conversion
Everything runs inside the browser tab against vendored readers. No Word installation, no server round-trip, and no remote request is made with your file.
One file at a time, up to 8 MiB, and the extracted text is capped at 400000 characters. A document that exceeds the cap is rejected rather than silently truncated, so a very long book needs to be split before conversion. Malformed archives — a .docx that is not a valid ZIP container or has no readable document part — fail with an error instead of producing empty text.