Text Diff — Compare Two Texts Side by Side Online

AT USE Text Diff takes any two blocks of text — original on the left, revised on the right — and marks every change instantly: additions in green, deletions in red, unchanged lines in neutral gray. The comparison updates as you type on either side. No Submit button, no page reload. Developers use it to inspect what changed between two versions of a function received by email, writers use it to check a document revision against a prior draft, and DevOps teams use it to identify field-level differences between production and staging configuration files before a deployment.

Two diff modes control the granularity of change reporting. Word mode tokenizes each line into words and highlights only the changed tokens. A twenty-word sentence with two words altered shows those two words in color and leaves the remaining eighteen in gray — the right mode for prose, documentation, contract text, and any natural-language comparison where word substitutions are what you are checking for. Character mode zooms further: it reports every individual character that changed within each changed token. If a JSON key changed from user_id to userId, character mode highlights the exact underscore that was removed and the capital I that was inserted — useful for code identifiers, regex patterns, URL slugs, and anywhere a one-character difference carries functional consequences. Switching modes is a single click; the diff re-renders in place without re-pasting anything.

JSON-aware mode adds a normalization step before the comparison runs. When enabled, both inputs are pretty-printed with consistent 2-space indentation. Two config.json files that contain the same logical content but different whitespace — one minified, one formatted — produce zero differences in this mode. Without normalization, comparing a minified JSON blob against a formatted version generates dozens of whitespace-only change lines that obscure the two genuine field differences you actually care about. JSON-aware mode surfaces only logical content differences. If either input fails to parse as valid JSON while this mode is active, the tool falls back to a plain text diff and displays a warning banner — no silent failure.

There is no cap on text length or line count. Entire source files, large API response exports, and multi-thousand-line log excerpts all process without hitting a limit. Everything runs in your browser using JavaScript — nothing is sent to a server. This is significant when the content includes API keys, database credentials, internal configuration values, or any private text that should not leave the device. The tool works offline after the page loads for the first time.

How the comparison works

The tool uses the Myers diff algorithm, which computes the shortest edit script that transforms the original sequence into the revised sequence. For word mode, the input is split into word tokens before the algorithm runs; for character mode, individual characters are the comparison units. The result reflects the minimum set of changes between the two versions — not a line-by-line replacement that would inflate the change count when most of a line is unchanged.

What content it handles

The tool compares plain text, source code in any language, JSON, YAML, XML, Markdown, CSV, SQL queries, and any other content that can be pasted as text. It does not handle binary data, PDF content, or Word documents without extracting the text first. For JSON specifically, enabling JSON-aware mode before comparing is almost always the right choice — it removes indentation noise before the Myers algorithm runs. For other structured formats like YAML or TOML, paste the raw text and use word mode. For large inputs such as full API exports or database snapshots, the tool handles them in-browser with no file-size limitation beyond available tab memory.