🕵️ 文本检测Text Detector
在浏览器内离线检测文本的多类质量问题：写作/排版硬错误、错别字检查、隐私/敏感信息泄露风险、可读性启发式指标。文本不会上传，所有分析在本地完成。Detect multiple kinds of text issues offline in your browser: writing/format hard errors, typo checks, privacy/PII leakage risk, and readability heuristics. Nothing is uploaded — all analysis runs locally.
⚠ 检测说明⚠ Detection notice
本工具基于确定性规则与离线开放词库（OpenCC、pinyin-data、an-array-of-english-words、chinese-xinhua，均为宽松许可）进行辅助检测，无法确保 100% 准确，可能存在漏报或误报，也不能替代人工校对。对于合同、论文、公文等关键内容，建议使用专业校对工具或人工复核后再发布。 This tool uses deterministic rules and offline open word lists (OpenCC, pinyin-data, an-array-of-english-words, chinese-xinhua — all permissively licensed) for assistive checking. It cannot guarantee 100% accuracy and may miss or mis-flag issues; it does not replace human proofreading. For critical content (contracts, papers, official documents), use a professional proofreading tool or manual review before publishing.
① 输入文本① Input
② 检测选项② Options
③ 写作 / 排版问题③ Writing & format issues
④ 隐私 / 合规扫描④ Privacy / PII scan
⑤ 可读性指标⑤ Readability metrics
⑥ 错别字检查（本地规则）⑥ Local typo rules
⑦ 简繁 / 异体检查⑦ Traditional/variant
⑧ 同音别字（提示）⑧ Homophone (hint)
⑨ 英文拼写（提示）⑨ English spelling (hint)
⑩ 规范成语校验（专业词库）⑩ Idiom validation (pro dict)
⑪ 使用说明⑪ How to use
- 粘贴要检查的中文文本Paste Chinese text to check
- 勾选检查项：排版、错别字、隐私、可读性等Tick checks: formatting, typos, privacy, readability
- 查看问题清单与修改建议Review issues and suggestions
示例：一段文章 → 列出标点/错别字/敏感信息Example: A passage → punctuation/typo/PII issues
全程在你的浏览器本地完成，文本不会上传服务器。Runs entirely in your browser; your text is never uploaded.
勾选需要启用的检测模块（默认启用前五项；同音别字、英文拼写与规范成语校验为可选提示项，默认关闭），输入或粘贴文本后点击「开始检测」，结果会随输入实时更新。各项检测均为确定性规则，纯本地计算、不上传任何内容：Toggle the modules you need (the first five are on by default; Homophone, English spelling and Idiom validation are optional hints, off by default). Type or paste text and click “Detect” — results update live. All checks are deterministic rules computed locally with no upload:
- 写作 / 排版问题：定位首尾空白、连续空格、连续标点、中英文标点混用、不可见字符、括号/引号不匹配、重复词、超长句、中英文之间缺空格、中文标点前后多余空格、连续重复句、英文重复词、空括号、中文字符之间的多余空格（如 我们 一起去，建议删除空格）、连续英文句点（如 ... 建议改用中文省略号 ……），以及中文文本中误用的全角字母/数字（如 ＡＢＣ、１２３，建议改半角）等。Writing & format: flags leading/trailing whitespace, repeated spaces, repeated punctuation, mixed CN/EN punctuation, invisible characters, unmatched brackets/quotes, repeated words, over-long sentences, missing spaces between CJK and Latin, stray spaces around CJK punctuation, duplicate consecutive sentences, repeated English words, empty brackets, extra spaces between CJK characters (e.g. 我们 一起去, remove the stray space), consecutive ASCII periods (e.g. ... → use the Chinese ellipsis ……), and full-width Latin letters/digits in CJK text (suggest half-width).
- 隐私 / 合规扫描：扫描手机号、邮箱、身份证号、银行卡号、URL、IP 地址等可能泄露的敏感信息（结果脱敏展示），便于发布前脱敏。仅做 PII 合规，不做内容审查。Privacy / PII: scans phone, email, ID card, bank card, URL and IP patterns (masked in output) so you can redact before publishing. PII compliance only — no content censorship.
- 可读性指标：给出字符数、句数、平均句长、长句占比、段落数、标点密度、中英比例与估算阅读时长等量化参考（非 AI 通顺评判）。Readability: quantitative metrics — char count, sentence count, average sentence length, long-sentence ratio, paragraph count, punctuation density, CN/EN ratio and estimated reading time (heuristic, not an AI fluency judgement).
- 错别字检查（本地规则）：无需下载模型，直接检查常见语法混淆（的/地/得、在/再、以/已/己、做/作、那/哪）、上百组常见多字别字（如按装→安装、布署→部署、知到→知道）与数百条易错成语（如迫不急待→迫不及待）。规则保守、误报低，但覆盖范围有限。Local typo rules: no model download required. Checks common grammar confusions (的/地/得, 在/再, 以/已/己, 做/作, 那/哪), over a hundred frequent multi-character typos (e.g. 按装→安装, 布署→部署, 知到→知道) and hundreds of error-prone idioms (e.g. 迫不急待→迫不及待). Conservative with low false positives but limited coverage.
- 简繁 / 异体检查：基于 OpenCC 简繁对照表（Apache-2.0），标记文本中的繁体字与异体字并给出简体替换建议，适合发现简繁混用或复制粘贴带入的繁体内容。Traditional/variant: based on the OpenCC simplified↔traditional table (Apache-2.0), flags traditional/variant characters and suggests the simplified form — handy for catching mixed scripts or pasted traditional text.
- 同音别字（提示）：基于 pinyin-data（MIT）校验出的真·同音易混词群（如反应/反映、权利/权力、必须/必需），仅作「请确认用法」的提示，不判定为错别字，可能误报，按需开启。Homophone (hint): based on homophone groups validated from pinyin-data (MIT) (e.g. 反应/反映, 权利/权力, 必须/必需). Shows a “please confirm usage” hint rather than flagging an error; may mis-flag, enable when needed.
- 英文拼写（提示）：勾选后按需加载内置英文词表（an-array-of-english-words，MIT，约 27 万词），检查混合文本中的英文单词拼写；策略保守（跳过缩写、品牌名、含数字词等），可能误报，仅供参考。English spelling (hint): when enabled, lazily loads a built-in English word list (an-array-of-english-words, MIT, ~275k words) to check English words in mixed text. Conservative (skips acronyms, brand names, words with digits); may mis-flag, for reference only.
- 规范成语校验（专业词库）：勾选后按需加载内置成语底库（chinese-xinhua，MIT，约 2.9 万条规范成语）与用字读音表，扫描文本中「非规范成语但与某规范成语仅差一字且该字同音」的疑似误写（如 穿流不息 → 川流不息、不径而走 → 不胫而走）；高精度、仅提示，可能对专有名词 / 术语产生建议，按需开启。Idiom validation (pro dict): when enabled, lazily loads a built-in idiom corpus (chinese-xinhua, MIT, ~29k standard idioms) and a character-pinyin table, then flags 4-character strings that are not standard idioms yet one homophone-char away from a standard idiom (e.g. 穿流不息 → 川流不息). High precision, hint only; may suggest on proper nouns/terms, enable when needed.
重要提示：以上均为确定性规则与离线开放词库的辅助检测，无法确保 100% 准确，可能存在漏报或误报，也不能替代人工校对。对于合同、论文、公文等关键内容，建议使用专业校对工具或人工复核后再发布（详见上方「检测说明」）。Important: all of the above are assistive checks using deterministic rules and offline open word lists — they cannot guarantee 100% accuracy and may miss or mis-flag issues; they do not replace human proofreading. For critical content (contracts, papers, official documents), use a professional proofreading tool or manual review before publishing (see the “Detection notice” above).