Text Diff Tool, Find Changes Between Two Documents

The word "diff" comes from the Unix world, but the need to compare two versions of a text is universal across professions and industries. Writers compare manuscript drafts to track editorial changes. Editors check that proofing revisions were applied correctly and completely. Project managers verify that the specification sent to a vendor exactly matches the internally approved version. Compliance teams confirm that policy documents were not altered after sign-off. Legal departments check that contract clauses were not changed between review rounds. In all of these cases, the challenge is identical: two texts that look similar at first glance can contain subtle differences that only a systematic, character-by-character comparison can reliably surface. A browser-based text diff tool handles all of these scenarios without requiring any software installation or technical background.

FixTools Text Diff Tool uses the Myers diff algorithm to compare input texts at the line level, then further analyses changed lines at the word and character level for fine-grained highlighting within each modified line. The algorithm identifies the longest common subsequence between the two texts, which represents all content that did not change between versions, and marks departures from that sequence as insertions or deletions. The visual output uses green for added content and red for removed content. All computation runs client-side in JavaScript in your browser, so the text you paste is never transmitted to any server or retained after you close the tab.

For the cleanest results, trim trailing whitespace and ensure consistent line endings before pasting. Writers comparing prose documents should consider whether paragraph-level or sentence-level comparison better serves their purpose. Breaking long paragraphs into individual sentences before pasting, with one sentence per line, can make word-level changes much easier to identify because each sentence becomes its own diff line rather than part of a large paragraph-level change block.

Most usability problems with text diffs come down to the same root cause: invisible characters that one side has and the other does not. CRLF endings on a Windows copy versus LF on a macOS copy will flag every single line as changed, which is almost never the user's intent. Non-breaking spaces inserted by word processors when copying from print layouts can make two sentences that read identically appear to differ. Smart-quote substitution converts straight quotes to typographic curly quotes silently in many editors. Normalise both inputs to plain ASCII quotes, LF endings, and no trailing whitespace before pasting and the diff cleans up dramatically. Performance on prose documents of any realistic length, including book chapters and full contracts in the tens of thousands of words, is effectively instant on modern hardware because line counts are usually modest compared to source code. The privacy guarantee is identical to other modes: nothing is uploaded, logged, or retained anywhere after the tab is closed.

Diff granularity is a deliberate choice with real performance and readability tradeoffs. Line-level diffing is the fastest and most scalable because the algorithm aligns whole lines as atomic units, which keeps the comparison space small even for very long documents. Word-level diffing compares sequences of whitespace-separated tokens, producing output that highlights individual changed words within otherwise unchanged sentences, which is ideal for prose review where a single replaced word can change meaning. Character-level diffing compares individual code points and is the most precise mode but also the most expensive computationally, because the comparison space grows with the total character count rather than the line count. FixTools uses a hybrid approach: line-level for the primary diff, with a secondary character-level pass applied only to lines that were already identified as changed at the line level. This combination delivers character-level precision where it matters most without paying the full character-level cost across the entire document. The result is interactive responsiveness on inputs of hundreds of thousands of words while still surfacing single-character changes inside long lines. For very large prose documents where even the line-level pass takes noticeable time, splitting the input by chapter or section is the standard remedy and produces more readable diffs as a side benefit.