How it works
The workflow, examples and limits behind every tool.
Text comparison ↗
A bounded longest-common-subsequence comparison produces an ordered edit sequence. Whitespace and case options affect comparison only.
Example: Compare “A / B / C” with “A / D / C” to see B removed and D added.
Up to 1,200 lines per side, or 1,500 Unicode code points in character mode. Very large comparisons are rejected to keep the page responsive.
Line cleaner ↗
The first occurrence is retained when deduplicating. Case-insensitive matching preserves the original surviving line.
Example: “Apple / apple / Pear” becomes “Apple / Pear” with case-insensitive deduplication enabled.
Up to 10,000 lines / 100,000 characters. Sorting is deterministic Unicode order, not natural-language collation.
List comparison ↗
Lists are treated as sets, using trimmed lines and optional case folding; blank lines are omitted.
Example: A/B and B/C give intersection B, left-only A and right-only C.
Matching is exact after selected normalization; there is no fuzzy matching or automatic language correction.
Invisible characters ↗
Space = ·, tab = ⇥ and each line ending = ↵. Cleanup targets U+200B/U+200C/U+200D/U+2060/U+FEFF only.
Example: A tab between two words appears as ⇥, while CRLF is counted as one line break.
Some zero-width characters affect scripts and emoji sequences. Inspect the preview and download a copy before replacing an original.
Local work, explicit limits
Each workspace reports its input and file limits. Imports validate the supported format before asking to replace saved work. Plain text is displayed as text, never executed as markup. Both languages use the same tools; the root address always opens in English, with Chinese under /zh/.