Publisher File Inspector

What this can and cannot do

Microsoft Publisher has no public format specification. Microsoft never documented how a .pub stores a page, a text frame, a guide or a layout. Every tool that claims to open one either has Publisher installed behind it or is guessing.

So this tool is honest about its scope. It does two things, both of them on documented ground:

1. It reads the OLE compound file container — the same FAT-like structure inside .doc, .xls and .ppt, specified in MS-CFB — and its property sets, specified in MS-OLEPS and MS-OSHARED.
2. It recovers embedded pictures by scanning the streams for image signatures: a PNG header through to its IEND chunk, a JPEG FFD8 through to FFD9, a GIF header to its trailer.

It does not render a page, and it will not until someone documents the format. That is why it is called an inspector.

What you actually get

The document properties are the useful part, and they are complete: title, subject, author, last saved by, keywords, comments, revision number, editing time, created, modified and last-printed dates, page and word counts, the application that wrote the file, and from the second property set the category, manager, company, content type and status.

That is enough to answer most of the questions people have about a .pub they cannot open: who made it, when, with which version, how long it is, and what it was for.

You also get the complete storage and stream tree with each stream's size, CLSID and timestamps — Publisher's Quill, Escher and Contents streams among them — and every recoverable picture, previewed and downloadable individually or as a ZIP.

Picture recovery is best-effort, and says so

Publisher stores its pictures without per-image framing that any public document describes, so they are located by signature scanning rather than read from a manifest. That finds ordinary PNG, JPEG and GIF payloads reliably and will miss anything stored in a wrapper it does not recognise — WMF and EMF metafiles in particular. The count is what was found, not a claim about what is there.

A .pub that is not a Publisher file

The format is detected from the file's first bytes, not its name. A .pub that is really a ZIP, an MHTML archive or some other compound document is a common thing to be holding, and it is reported as such with a link to the tool that does open it.

Privacy

The file is read in your browser. Nothing is uploaded.