#Text Diff

Text Diff is an offline two-pane text and file comparison tool. Beyond highlighting changes, it states in one sentence which line and which field the first difference starts at — so you don't have to hunt for the starting point in a screen full of red and green.

#First-difference locator

This is what sets the tool apart from a plain diff: the banner describes the first difference in natural language, phrased per format:

Pair it with the Previous / Next buttons to step through blocks; the header also shows "block i of n".

#Format detection

Detection combines file extension and content, and can always be overridden:

• Plain text: line-by-line comparison
• CSV / TSV: guesses the delimiter first (whichever of , \t ; | yields the most consistent column count), then compares by column, handling quoted fields and "" escapes
• INI: compares key/value pairs and names the key whose value changed
• JSON: currently compared as text lines (a notice is shown)

#Comparison options

• Ignore surrounding whitespace: skip differences caused by edge spacing
• Ignore case: skip letter-case differences
• Character-level diff: within a modified row, mark exactly which characters changed (off by default)
• Only differences: fold away identical content, keeping just the diff blocks
• Context lines: how many lines to keep around each difference when folding — "Only differences" or ±N lines

#Views and navigation

• Split / Unified: switch between the side-by-side and inline views at any time
• Maximize: full-screen mode for long files
• Swap A/B: exchange the two sides in one click
• Expand folds: identical runs collapse to "⋯ N identical lines hidden (click to expand)"
• Stats: similarity percentage, line counts for A and B, and "N differences · X modified / Y added / Z deleted"

#Performance guard

When the two sides diverge heavily, a line-level diff can take a long time. A 3-second timeout stops the run and shows "comparison timed out" rather than hanging the page. Click "Compare anyway" to force a full run with no time limit for that comparison. Any content change re-arms the guard.

#Encoding detection

Uploaded files are auto-detected as UTF-8, GBK, Big5, UTF-16LE, and other common encodings, with the detected value displayed; a manual switch re-decodes the file. Especially useful for CSVs exported from different systems, GBK config files in legacy projects, and Windows logs.

#Export

• Copy A / Copy B: copy either side out
• Export diff: generate a standard unified diff (.patch) for git apply or a code review tool. The filename defaults to B's filename

#Common patterns

• Verify a config change: drop in two .ini / .env files and read off which key's value moved
• Compare tabular data: two exported CSVs, located down to the exact field
• Self-review before a PR: compare local changes against the deployed version and export a .patch
• Proofread a translation: enable "Ignore surrounding whitespace" to see only substantive edits between revisions

Everything runs locally in your browser — nothing is uploaded — safe for sensitive configs and private code.