Diff Checker

The algorithm underneath is a longest common subsequence search — it finds the largest set of lines appearing in both versions in the same order, and everything else is an addition or a removal.

That framing explains some behaviour that otherwise looks wrong:

A moved block shows as a deletion plus an addition. Line diffing has no concept of "moved". A function relocated from the top of a file to the bottom appears twice, once in each colour.

Changing one character marks the whole line changed. Line-level granularity is the unit. A line is either the same or not.

Reformatting looks like a total rewrite. Re-indenting a file changes every line. Switch on "Ignore whitespace" to compare content rather than layout — that is exactly what it is for.

Removals and additions that sit adjacent are paired onto the same row here, so a modified line shows old and new side by side rather than as two unrelated entries.

Anything plain text works, but a little preparation makes the result far more readable.

JSON. Two documents can be semantically identical and textually different — key order and indentation vary by serialiser. Format both with the JSON formatter and sort keys first; the diff then shows only real changes.

CSV. Sort both files by the same key column before comparing, or a single inserted row will offset everything after it and mark the whole file changed.

Minified code. A minified file is one long line, so a line diff tells you nothing. Format it first.

Log files. Timestamps differ on every line, which drowns the signal. Strip them first with a regex, then compare.

Everything runs in your browser, so contracts, credentials and unreleased code are safe to paste.