Compare two Word documents by their text

A Word document diff reads the text of two .docx files, list numbers, bullets and table rows included, and compares it line by line; it does not compare fonts, images, comments or layout, and a .docx file does not fix layout until a program lays it out.

Compare two documents

Runs entirely in your browser; nothing is uploaded.

What is read

Each paragraph is one line. A list item keeps its marker, numbered the way Word numbers it: 1., a) or ii., restarting where the list restarts, with each level of nesting indented, so two steps swapped in a procedure read as a change in their numbers and their text, and a numbered list turned into bullets reads as a change. A table row is one line with its cells separated by a tab, and a merged cell is padded so every row keeps the table's width. A field, such as a date or a page number, reads as the value shown, not as its code. Headings are their text.

Tracked changes are read as if every one were accepted: insertions in, deletions out, which is what Word shows with markup turned off. Two files can read the same this way and differ in what is still pending, so a file with tracked changes carries the flag tracked-changes-accepted and is reported as near-equivalent at best.

What is not read

Headers, footers, footnotes, endnotes, comments and text boxes are not compared, and the result says so when a file has text in any of them. Fonts, colors, images, page breaks and margins are not in the comparison at all. A .docx file does not hold pages: where a page breaks depends on the program, fonts and printer that lay the document out, and Equivl has no Word layout engine. So there is no page-by-page view for Word files; to compare how two documents look, save both as PDF from Word and compare the PDFs.

What a verdict means

• Identical: the two files are the same bytes, checked before either is read.
• Near-equivalent: the text matches, or nearly. Every Word comparison is reported this way at best, because only the text was compared and the flag document-text-extracted says so.
• Divergent: at least one line differs. The score is Text mode's, over the whole document.

A Word file can be compared with a PDF. The two are read by different readers, so that comparison carries cross-format and is also near-equivalent at best.

Limits

• Files are read up to 100 MB each, in a background worker, with caps on how many entries and how many bytes a file may unpack to.
• Only .docx is read: an old-style .doc file, a password-protected Word document and OpenDocument text are refused with a sentence saying so, since the first two are a different format and the third is not supported. Save as .docx without a password in Word and drop the file again.