Text Diff

How the Diff Works

Comparing two texts is a classic computer science problem: find the smallest set of edits that turns one into the other. This tool solves it with the longest common subsequence (LCS) algorithm — it finds the longest sequence of lines (or words) shared by both texts in the same order, and everything else is reported as an addition or a removal. No server, no network, just your two texts and a bit of dynamic programming in your browser.

LCS algorithm

Given two sequences, LCS asks: what is the longest subsequence of elements present in both, in the same order? It is solved with dynamic programming in O(n·m): a table where each cell stores the best answer for the prefixes up to that point. Walking the table backwards yields the alignment — matching lines are equal, the rest are additions and removals. The famous Unix diff utility uses closely related techniques.

Line level vs word level

Line level runs one LCS pass over whole lines: fast and readable, but a one-word edit on a long line shows as "the whole line changed". Word level adds a second pass: changed line pairs are re-diffed token by token, so the highlight pinpoints exactly which words moved. Think of it as zooming in on the lines that the first pass flagged.

Typical uses

Reviewing code changes between two versions, comparing drafts of a document, spotting typos between two copies of text, or verifying that a config file change did exactly what you intended. The counters (lines added / lines removed) give a quick size estimate of the change before you read the details.

Frequently Asked Questions

Reviewing a config file instead? Compare structure and values key by key with JSON Diff, or profile each version's writing with the Word Frequency Counter.