离线 · 无上传 · 浏览器本地完成Offline · no uploads · runs in the browser
PDF 转 Word——本地两步走完，中间产物不落盘PDF to Word (.docx) locally — two steps, nothing on disk
先说这条路的真相，因为它决定了产物好不好用：Word 的 .docx 本应是一棵由段落、运行与样式组成的 XML 树，而这个转换产出的是 Word 的 altChunk——文档包里嵌一份 HTML，由 Word 打开时自己排版。好处是真的能编辑、能存回 .docx；代价是它带不来 PDF 里没有的任何东西（版式、分页、字体），也别指望段落样式与手工排版的 Word 文档一致。换句话说：这是"把 PDF 里的文字变成可编辑内容"，不是"把 PDF 变成 Word"。Start with what this route actually is, because it decides how useful the file is. A .docx is meant to be an XML tree of paragraphs, runs and styles; this converter produces a Word altChunk instead — an HTML part embedded in the package, which Word lays out itself when it opens. The upside is real: the text is editable and saves back as .docx. The cost is that it cannot bring what the PDF never carried (layout, pagination, fonts), and paragraph styles will not match a hand-built Word document. In one line: this turns the words in a PDF into editable content, it does not turn a PDF into a Word document.
这条链路怎么走How the route runs
PDFPDF 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
- 文字层按阅读顺序进正文，书签按层级成为标题、没有书签时短而全大写的行成为标题，其余成段——与 PDF → HTML 同一套判断The text layer enters the body in reading order, the PDF’s own bookmarks become headings at their outline depth and, where there is no outline, a short all-caps line becomes a heading — the rest become paragraphs, on the same judgement PDF to HTML makes
- 链接标注保留为可点击超链接，并按协议白名单过滤（javascript: 与 data: 一律剔除）Link annotations survive as clickable hyperlinks, filtered by scheme allowlist (javascript: and data: removed outright)
- 每页之间的分隔线保留下来，在 Word 里仍然看得见页面边界The separator between pages carries through, so page boundaries stay visible inside Word
- 中间那份 HTML 只是内存里的过路产物，不会出现在你的下载里，也就没有"多下一个文件"的意外The intermediate HTML exists only in memory and never appears in your downloads, so there is no extra file to wonder about
- 这条链的两端各自可独立使用：只要 HTML 就停在第一步，只要可编辑文字就直接跑这条Both halves stand alone: stop at the first step if you only wanted HTML, and take this route when editable text is the goal
如实说明的限制What it does not do
- 不是原生 OOXML 段落树：altChunk 由 Word 现场排版，字体、行距、样式与真正的 Word 模板不一致，也不带页眉页脚与页面设置Not a native OOXML paragraph tree: an altChunk is laid out live by Word, so fonts, leading and styles will not match a real Word template, and there are no headers, footers or page setup
- 版式与表格不保留：PDF 里的表格只是排过版的字符，到 Word 里成不了一张表（要结构化表格请回到源文件；界面上 PDF → Excel / CSV / JSON 是置灰的）Layout and tables do not carry: a table in a PDF is typeset characters, and it does not become a table in Word (for structured data go back to the source; PDF → Excel / CSV / JSON is greyed out in the picker)
- PDF 内嵌图片不进 Word，产物里没有图像；需要页面图像走 PDF → 图片Images embedded in the PDF are not carried into the Word file, which contains no pictures; for page imagery take PDF to image
- 扫描件为空——没有 OCR，也不会替你猜字A scan comes back empty — there is no OCR and no invented text
- 打开时的排版由 Word 决定：同一份产物在 Word、LibreOffice 与在线 Office 里可能不完全一样（本项目只负责把它写对）Word decides the final typography: the same file can render slightly differently in Word, LibreOffice and Office online (this project only guarantees what it writes)
实操提示Working with it
- 单个上限 100 MB，一批最多 200 个文件；链路较长，PDF 页数多时可以先只转几页试形状100 MB per file and 200 files per batch; it is a long chain, so on a big document convert a few pages first to see the shape
- 反向也有路：这个扩展自己产出的 HTML → DOCX 同样是 altChunk，而 DOCX → HTML 会识别并救回它，所以自己的产物能原样读回来The reverse path exists too: this extension’s own HTML → DOCX output is an altChunk as well, and DOCX → HTML recognises it and recovers the content, so its own files read back
- 需要"可编辑 + 结构干净"时，通常先在 PDF → HTML 上手工清一遍，再转 Word 更省心When you want editable *and* clean structure, tidying PDF → HTML by hand and only then converting to Word usually costs less
常见问题Questions people actually ask
为什么在 Word 里段落样式跟别人的文档差很多？Why do the paragraph styles look nothing like other people’s documents?
因为产物里没有样式表——它嵌的是一份 HTML，Word 用自己的默认样式去排。想要模板外观，在 Word 里套一次样式或另存为模板，比在这里期待转换"猜对"要可靠。Because there is no stylesheet in the file — it embeds an HTML part that Word sets with its own defaults. Applying a style once inside Word, or saving it as a template, is far more reliable than hoping a converter guesses the design.
转出来的 .docx 能被别的程序读吗？Will other programs read the .docx?
主流字处理软件能打开，但 altChunk 是 Word 家族的机制，部分第三方工具与索引服务对它的解析程度不一。要跨工具长期存放，HTML 或纯文本是更稳的落点——它们不依赖任何一家软件的排版。Main word processors will open it, but altChunk is a Word-family mechanism and third-party tools and indexing services handle it to varying degrees. For long-term storage outside one tool, HTML or plain text is the sturdier landing place — neither depends on one vendor’s layout engine.
为什么不干脆做成真正的段落结构？Why not build a real paragraph tree instead?
因为 PDF 里没有那棵树可还原：它只有坐标与字形。硬造 XML 只是把"由 Word 现场排版"挪到"由我们的猜测现场排版"，多出来的自信并没有多出来的信息。所以这一条宁可把话说窄。Because no such tree exists in a PDF to recover — only coordinates and glyphs. Manufacturing XML would move "laid out by guesswork" from Word to us, adding confidence without adding information. This page would rather state the narrower claim.
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