File Difference Tool

No. Both files are read directly in your browser - all comparison, sorting of differences, and copy/download output happens on your device using JavaScript. No upload, no proxy, no analytics, no tracking. The only data that ever leaves your browser is what you explicitly choose to copy or download yourself.

It runs a true Longest Common Subsequence (LCS) line diff. That means it correctly handles inserted and missing lines instead of just comparing line 1 to line 1. If File B has one extra line at the top compared to File A, only that one line is marked as different - everything below it is still recognized as matching.

Plain text only - source code, configs, logs, CSV/TSV, JSON, XML, HTML, Markdown, SQL, SVG, .env files, and similar. The file picker is scoped to ~70 common text extensions, and any file containing NUL bytes (a strong signal of binary content like images, PDFs, or ZIPs) is rejected with a clear error before comparison.

Yes - 10 MB per file. Files larger than that are rejected before being read, so the browser tab will not freeze trying to load a giant log dump. For most source files, configs, and JSON exports this is more than enough.

CRLF (Windows) and CR (classic Mac) line endings are normalized to LF (Unix) before the diff runs, so a Windows-saved file and a Unix-saved file with otherwise identical content will be reported as identical instead of all lines being marked as different.

A leading UTF-8 BOM (the invisible 0xFEFF marker that Notepad and some editors add at the start of a file) is stripped automatically so it does not show as a phantom difference on line 1. The tool reads files as UTF-8 - other encodings (like Windows-1252 or Shift-JIS) may show garbled characters and should be re-saved as UTF-8 first.

Yes. The on-screen split view, the Copy Diff output, and the Download .txt file are all generated from the same LCS diff result, so what you see is exactly what you export. Matching lines are pair-aligned on the same row, lines that exist only in File A appear on the left only, and lines that exist only in File B appear on the right only.

No. File A and File B contents are saved to your browser's localStorage and restored automatically on the next visit, and the comparison is re-run on load. Click Clear All to wipe both files, the diff, the navigation state, and the persisted data immediately.

The tool uses a Longest Common Subsequence dynamic programming algorithm to compute the diff. That means inserted and missing lines are detected correctly - if File B has one extra line at the top compared to File A, only that one line is marked as different and every line below it is still recognized as matching, instead of the whole file being marked as different.

Both files render in a two-column table with a vertical divider, a sticky header showing "File A" and "File B", and a 60vh scroll container. Rows are aligned from the LCS result so what you see on screen matches exactly what you copy or download - no mismatch between the visual view and the export.

Matching lines are tinted light green on both sides. Differing lines are tinted light red on both sides - including any empty cell on the side where the line is missing entirely. Matching blank lines are left plain white so empty space in the file does not visually clutter the view. The color scheme is intentionally directional-neutral: there is no "old" or "new" file, just File A and File B.

Each row shows the actual line number from File A on the left gutter and from File B on the right gutter. These numbers come from the diff itself, so a row containing a line that exists only in File A shows only an A number, a row containing a line that exists only in File B shows only a B number, and matching rows show both - reflecting the real positions in each file rather than the row position in the table.

When the diff contains any differing rows, an ↑ Prev / ↓ Next pair appears in the stats bar with a counter like "3/12" showing your current position out of the total number of differing rows. Clicking either button advances or rewinds the cursor and smoothly scrolls the matching row into view at the center of the split panel.

Above the split view, a stats bar shows how many lines matched, how many differed, and the total line count of each side (File A / File B). Every number comes from the LCS result so the stats always agree with the visual view and the export.

Click Copy Diff to copy the full diff to your clipboard in unified format - matching lines are prefixed with a space, lines unique to File A with a minus, and lines unique to File B with a plus. This is the same format git diff and patch files use, so it pastes cleanly into chats, tickets, code reviews, and most text editors with diff syntax highlighting.

Click Download to save the same unified-format diff as diff-output.txt. The blob is created locally and the object URL is revoked immediately after the download triggers, so there is no memory leak.

Click ⇄ Swap to instantly switch File A and File B contents. If a comparison is already active, it auto-reruns on the swapped content so the split view, stats, and navigation update in place - useful when you realize you loaded the files in the wrong order.

Each panel is a drop zone with a highlighted overlay while you drag a file over it. You can also click the Upload button next to the File A / File B label to open a file picker. Both paths share the same loader, so size, encoding, and binary checks apply consistently.

Files larger than 10 MB are rejected before being read with a clear error like "file.log is 23.4 MB - the maximum is 10 MB per file". This stops a giant log dump or accidental binary export from freezing the browser tab.

After reading a file, the tool scans the first 4 KB for NUL bytes - a strong signal of binary content. If any are found, the file is rejected with an error explaining that only plain text is supported, so dropping a PNG, PDF, or ZIP shows a polite message instead of a screen full of garbled mojibake.

CRLF (Windows) and CR (classic Mac) line endings are converted to LF (Unix) before the diff runs. A file saved on Windows and the same file saved on Linux will be reported as identical instead of every line being marked different because of an invisible \r.

If a file starts with a UTF-8 BOM (the invisible 0xFEFF marker that Windows Notepad and some editors add), it is stripped from the loaded text so it does not appear as a phantom difference on line 1.

File A and File B contents are saved to your browser's localStorage under the keys "diff-tool-a" and "diff-tool-b" and restored automatically on the next visit. If both files are restored, the comparison is re-run on load so you land back exactly where you left off.

Focus the tool and press Ctrl+Alt+C to compare, Ctrl+Alt+S to swap, or Ctrl+Alt+R to clear all - faster than reaching for the mouse when iterating on small text edits between runs.