Text diff
Compare two blocks of text and see additions and deletions highlighted at the word level.
What this does
This computes the difference between two blocks of text and renders it inline: removed spans struck through, added spans highlighted, everything else untouched. It uses jsdiff, which implements the standard Myers diff algorithm, at three granularities — whole lines, whole words (keeping whitespace), or individual characters. Options let you fold away case-only and surrounding-whitespace-only differences.
It runs in your browser. Two drafts of a contract clause, two versions of a log, or a config diff with credentials in it can all be compared without anything being sent anywhere.
When you'd use it
- Comparing two API responses to see exactly which field changed.
- Checking what an automated formatter or a find-and-replace actually altered.
- Reviewing an edit to a piece of copy, an email, or documentation.
- Spotting the one character that differs between a working and a broken config value.
- Diffing text you can't paste into a hosted tool for confidentiality reasons.
Worked example
Left:
The quick brown fox jumps over the lazy dog.
Right:
The quick red fox leaps over the lazy dog!
In Word mode the diff reads: "The quick
brownred fox jumpsleaps over
the lazy dog.!" — three small changes, each shown
exactly where it happens. Switch to Line mode and, because
the single line changed at all, the whole line is marked removed and the
whole new line added — more faithful to how a line-oriented tool sees it,
less useful for reading. Character mode would additionally
show that jumps → leaps keeps the s
at the end.
Limits and gotchas
- Not a merge or patch tool. No three-way merge, no conflict markers, no unified-diff export.
- Large inputs. Character-mode diffing of very large texts is O(n·d) and can be slow; drop to Word or Line mode.
- Whitespace and line endings are real differences unless you enable the ignore option. Mixed
- Moved blocks show as a deletion in one place and an insertion in another; the tool doesn't detect that a paragraph was relocated.
- Unicode. Combining characters and emoji sequences may split in Character mode in ways that look odd; Word mode is safer for those.
Frequently asked questions
- Word, line, or character — which should I pick?
- Line for source code, config files, and anything where a line is the unit of meaning. Word for paragraphs of prose, where line-level diffing would mark a whole rewrapped paragraph as changed. Character for spotting a single transposed letter or a changed digit in an otherwise identical string.
- Two visually identical texts show as different.
- Almost always an invisible character: trailing spaces, a tab vs spaces, Windows
\r\nvs Unix\nline endings, a non-breaking space pasted from a web page, or a Unicode look-alike. Try Line mode with "ignore leading/trailing whitespace", and if it still differs, use Character mode to find the exact spot. - Is this a real merge tool?
- No. It shows what changed between two texts; it does not do three-way merges, conflict resolution, or produce a patch file. It is for eyeballing differences — comparing two API responses, two versions of a message, a config before and after an edit.
- Does the diff match what git would show?
- The set of changed regions is usually the same, but the presentation differs. Git shows whole changed lines with
+/-prefixes; this tool shows changes inline, and in Word or Character mode it highlights sub-line edits that a defaultgit diffwould show as a full line replacement.