离线 · 无上传 · 浏览器本地完成Offline · no uploads · runs in the browser
JSON 转 Word 文档——离线、可批量、不上传JSON to Word document offline, in batches, no upload
「把 JSON 变成 Word」在这条路上不是一句口号，而是一个明确的取舍：先由 JSON 生成可折叠、带语法着色的查看器 HTML，再把这份 HTML **原样封装**进 .docx 的一个 MHTML altChunk 部件里（实测产物是一个含 `[Content_Types].xml` 与 `/word/document.xml` 的 OPC 容器）。也就是说，扩展不自己去拼 Word 的 XML 正文，排版发生在 **Word 打开文件的那一刻**——这与 TXT 转 Word、HTML 转 Word 是同一条机制。好处是结构与层级完整带过去，且不需要在本机实现一份 OOXML 写入器；代价是最终观感由 Word 的 HTML 引擎决定，我们无法在本机替它保证颜色与折叠的表现。“JSON into Word” on this route is a concrete trade-off rather than a slogan: the file first becomes a collapsible, syntax-coloured viewer, and that HTML is packaged **as it stands** into one MHTML altChunk part inside the .docx (measured: the artefact is an OPC container holding `[Content_Types].xml` and `/word/document.xml`). In other words, this extension does not assemble WordprocessingML itself — layout happens **when Word opens the file**, the same mechanism as TXT to Word and HTML to Word. What that buys is structure and hierarchy carried across whole, without a hand-written OOXML writer on the client; what it costs is that Word’s HTML engine decides the result, and nothing on this machine can promise how colour or folding will look there.
这条链路怎么走How the route runs
JSONJSON HTMLHTML Word (.docx)Word (.docx)
共 2 步，由扩展的路径搜索自动选出，中间产物不落盘。2 steps, chosen by the extension's own path search; intermediate results never touch the disk.
会保留什么What carries over
- 结构与层级完整带走：键、值、对象与数组的嵌套关系来自同一份查看器 HTML，不会被压成一整段等宽文字Structure and hierarchy travel intact: keys, values and the nesting of objects and arrays come from the same viewer HTML rather than being flattened into one block of monospaced text
- 计数摘要以文字形式留在文档里（实测形如 `Array(3)`、`Object(4)`），所以「这里本来有几项」在纸面上也读得到The count summaries stay in the document as words (measured shapes like `Array(3)` and `Object(4)`), so “how many items were here” is still readable on paper
- 字符串带引号、数字与布尔不带，`null` 原样写出；中文与非 ASCII 按 UTF-8 保留，不会变问号Strings keep their quotes while numbers and booleans do not, `null` is written as itself, and Chinese or other non-ASCII text survives as UTF-8 instead of turning into question marks
- 产物是真正的 .docx 容器而不是改了扩展名的 HTML：Word 在「最近使用的文档」里认得它，也能直接附件发送The artefact is a real .docx container rather than HTML wearing a new extension: Word recognises it in recent documents and it attaches to mail as expected
如实说明的限制What it does not do
- **排版由 Word 决定**：altChunk 是「把 HTML 交给 Word 渲染」，所以字体、行距、分页位置都取决于打开它的那一份 Word；同一份文件在 LibreOffice 或 Google Docs 里可能不同**Word owns the layout**: an altChunk means “hand the HTML to Word to render”, so face, line spacing and where pages break depend on the Word that opens the file — the same document can look different in LibreOffice or Google Docs
- 语法配色与折叠属于尽力而为：颜色是 CSS，能不能被 Word 的 HTML 引擎采纳我们不在本机替你保证；折叠依赖 `<details>`，Word 可能按普通段落展开呈现Syntax colour and folding are best-effort: colours live in CSS, which we do not certify against Word’s HTML engine from this machine, and folding depends on `<details>`, which Word may present as ordinary expanded paragraphs
- 不做「按记录成表」：文档里是树形视图，不是一张可以筛选排序的表。要表格请用 JSON 转 CSV（对象数组按键取并集）或转 ExcelNo “records into a table”: the document holds a tree view, not something to filter and sort. For a table, take JSON to CSV (an array of objects becomes columns from the union of keys) or to Excel
- 没有文字层之外的检索保证：能在 Word 里搜到的只是文档里那些文字，结构与计数摘要也在其中；深结构文档篇幅可观，翻页是常态No search guarantee beyond words: what is findable in Word is the text in the document, count summaries included, and a deep structure runs to pages
- 解析不了的 JSON 直接报错，不产出一份空文档；单个文件上限 100 MB、一批最多 200 个，多份一起转会打包成 ZIPJSON that cannot be parsed fails instead of producing an empty document; 100 MB per file and 200 files per batch, several files handed over as one ZIP
实操提示Working with it
- 要在 Word 里继续编辑成一份正式文档：先 JSON 转 Markdown 再转 Word，或直接 JSON 转 HTML 后由 Word 打开，都能拿到更可预测的段落结构To keep editing it as a formal document in Word: JSON to Markdown to Word, or JSON to HTML opened by Word, both give more predictable paragraph structure
- 要的是一张能算的表：走 JSON 转 CSV 或转 Excel；要一份能归档的完整树：JSON 转 PDF 或转 HTMLFor a table you can compute on: take JSON to CSV or to Excel; for an archivable full tree: JSON to PDF or to HTML
- 这份 .docx 只在 Word 里才见到最终版式：交付前请自己打开确认一次，尤其是包含配色或需要折叠层级的场合The .docx only shows its final layout inside Word: open it once before delivery, especially where colour or collapsed levels matter
常见问题Questions people actually ask
为什么不在本地直接生成 Word 的 XML？Why not generate Word’s XML locally instead?
因为那要在浏览器里自己实现并长期维护一份 OOXML 写入器，而 altChunk 让 Word 用它自己的引擎渲染同一份 HTML。这一步是刻意的取舍：结构完整带走、代码不多一份，代价是版式归 Word 而不是归我们。Because that means implementing and maintaining an OOXML writer inside the browser, whereas an altChunk lets Word render the same HTML with its own engine. The trade is deliberate: structure travels whole and no extra writer ships, and in exchange layout belongs to Word rather than to us.
颜色会丢吗？Will the syntax colours be lost?
不好保证。颜色写在 CSS 里，altChunk 交给 Word 的 HTML 引擎解释，本机没有 Word 可以替你验证它到底采纳哪些——所以页面只承诺「结构与文字完整带走」，不承诺配色一致。介意这点就走 JSON 转 PDF（画面一致但不能改字）或转 HTML。It cannot be promised. Colours live in CSS and the altChunk hands them to Word’s HTML engine, and there is no Word on this machine to certify which rules it honours — so the page promises structure and text, not a matching palette. If that matters, take JSON to PDF (same picture, no editing) or to HTML.
接口返回的原始 JSON 会外发吗？Does the raw API response ever leave the machine?
不会。解析、渲染、封装 .docx 都在扩展自己的标签页里完成，只申请 storage 权限、没有 host 权限，产物里也没有把远端资源当代码执行的路径。No. Parsing, rendering and packaging the .docx all happen inside the extension’s own tab: the storage permission only, no host permissions, and no path in the bundle that executes remotely hosted code.
不需要账号，也不需要联网：装好扩展后点图标，在新标签页里打开工作台，文件从磁盘直接选。No account and no network: install the extension, click its icon, and the workbench opens in a new tab reading straight off your disk.
从 Chrome 应用商店安装Install from the Chrome Web Store 安装步骤Install steps GitHub 源码Source on GitHub