Chinese-English Typography Fixer
One click to fix common Chinese copy issues: spacing between Chinese and English/numbers, full-width punctuation, repeated marks and ellipses — each rule can be toggled.
Fix details
Why Chinese-English typography matters
Chinese technical writing widely follows the "pangu spacing" convention: keep one half-width space between Chinese and English, and between Chinese and numbers — e.g. 使用 Docker 部署 NAS or 35 度. The visual boundary between scripts becomes clear and long sentences read much better.
Punctuation matters just as much: Chinese text should use full-width marks (，。；：？！), but text pasted from English IMEs or code editors often carries half-width ones. Repeated exclamation or question marks should collapse, and three or more dots should become the proper ellipsis (…). Rules are applied in order with a per-rule fix count so you can verify each change.
Everything runs locally in your browser and nothing is uploaded. Each rule can be toggled independently — unsure? Start with only the spacing rule and iterate.
How to use
- Paste your copy into the box above; multiple paragraphs at once are fine.
- Toggle the rules (all on by default); the result and fix counts update live.
- Review, then click "Copy" to replace the original text.
FAQ
Will it break already-correct text?
No. Rules only touch actual issues: boundaries that already have spaces are untouched, full-width punctuation is never converted back, and you can always compare and discard the result.
Why are some half-width commas left alone?
Punctuation conversion only happens in unambiguous Chinese contexts (Chinese on both sides). Half-width marks between English words or in numbers like 1,000 are preserved to protect code and data.
Will 第1章 become 第 1 章?
Yes — that is standard pangu behavior, spacing digits from surrounding Chinese. If you prefer the compact style, turn off the spacing rule.
Is it safe for code or URLs?
For code and URLs, disable the spacing rule first or use the Find & Replace tool instead. This tool targets natural-language copy and does not parse code semantics.