删除元数据
清除 PDF 携带的、关于是谁制作了它的隐藏信息。
这款工具完全在你的浏览器中运行。文件从不上传，你可以在浏览器的网络标签中亲自验证这一点。 亲自验证：打开浏览器的网络标签观察。你会看到一个查询你是否还有剩余任务的小请求，只包含工具名称和一段哈希，绝不会包含文件。
这款工具的作用
This clears the information a PDF carries about itself rather than anything on its pages. That is more than the author field people expect: the document properties, the XMP packet, the private data an editor left under /PieceInfo, per-page metadata, and the identifier in the trailer. The file is read first, so the result can tell you what was actually found and what was removed.
Use it before a document leaves your organisation: a tender response that should not name the person who drafted it, a CV that would otherwise carry your employer's file server path, or a template you are publishing that still credits the client you built it for.
工作原理
- Drop the PDF onto this page.
- Leave Keep the title off unless the title is deliberate and public - it is often a file path or a working name nobody meant to publish.
- Leave Keep the dates off if you are anonymising. The timestamps are a fingerprint, and clearing them clears the document identifier with them.
- Press Remove metadata.
- Read the report. It lists what was found in the file and what was taken out, so nothing to remove is a real and useful answer.
The gap between what people expect this to remove and what a PDF actually carries is the reason the tool exists. A file routinely records the application that produced it, the operating system account signed in at the time, the full path the source document lived at, every editing session's timestamps, and - in anything that passed through Illustrator or InDesign - a private data blob under /PieceInfo that no PDF reader will ever display to you.
The trailer identifier matters more than its obscurity suggests. It is a pair of hashes derived from the file and the moment it was written, and it survives every cleanup tool that only clears the document properties dialog. Two files carrying the same first identifier are demonstrably versions of one original, which is precisely what someone anonymising a document does not want to leave behind.
Removal here means removal, not blanking. Each entry is deleted from the file and anything only that entry pointed at goes with it, so the values are not left sitting in the bytes for anyone who opens the file in a text editor. When the properties dictionary ends up empty it is dropped entirely, because an empty one is itself a small signal about how the file was made.
This does not touch what is written on the pages. A name in a letterhead, a signature block, a footer with a file path - all of that is content, and content is what Redact PDF is for. Metadata removal makes the file anonymous; it does not make the document anonymous.
The reading and the rewriting both happen in your browser with pdf-lib. There is a pleasing consistency in that: a tool whose job is to stop a document telling strangers about you would be a strange thing to run by sending the document to a stranger.
这款工具做不到的事
- Only the file's own metadata is removed. Names, addresses and paths written into the page content stay exactly where they are - use Redact PDF for those.
- Metadata stored inside embedded images, such as a photograph's EXIF block with its camera and GPS fields, is not read or stripped.
常见问题
- 具体会删除什么？
- 文档信息字典——标题、作者、主题、关键词、创建程序、生成程序和时间戳——再加上 XMP 元数据包、编辑器在文档级和页面级留下的 /PieceInfo 数据、每页的元数据流，以及尾部里的标识符。结果会列出其中哪些确实存在于你的文件里。
- 怎么核实真的删掉了？
- 在任意阅读器里打开结果，查看文档属性：这些字段都是空的。要更严格一点，可以对文件运行 exiftool，或者用文本编辑器打开、搜索作者的名字——这些项是被删除而不是清空，所以再也找不到任何残留。
- 这会把我的名字从文档正文里删掉吗？
- 不会，而且这个区别很重要。它清除的是文件关于自身记录的信息。印在信头里、打在签名栏中、放在页脚里的名字属于页面内容，删除它意味着对页面做涂黑——「涂黑删除」会把可见的像素和底下的文字一起删掉。
- 为什么日期和文档标识符会一起被清除？
- 因为它们是同一类证据。创建时间和修改时间说明文件是何时写出的；尾部标识符由文件和同一时刻推导而来，会把同一文档的各份副本关联在一起。保留日期就是保留文件的身份，所以两者一起处理，免得让你误以为文件已经干净。
- 文件会去哪里？
- 哪里都不去。它由 pdf-lib 在你设备的 Web Worker 里读取并重写，页面不会发出任何请求。对于一个专门用来不让文档泄露你信息的工具来说，这是唯一说得通的做法——你也可以在网络面板里确认。