Open a Publisher .pub file without Publisher
Drop a .pub from Publisher 98 to 2019 or Microsoft 365 to get all its text, story by story, to copy or save as .txt; every picture stored in it, each with a Save button; the fonts it uses; and its title, author and dates. It works on a Mac, a Chromebook or a phone, where Publisher never ran.
The file is read by this tab only. It is not uploaded and not stored.
What you get
- All the text, story by story. Each text frame (or chain of linked frames) is one story, shown in the order the file stores them, with paragraphs and line breaks kept. Copy all text puts it on the clipboard; Save as .txt writes it with a
--- Story n ---line before each. - Every picture stored in the file, once each: JPEG photos, PNG and GIF pictures and Windows bitmaps (saved as PNG) shown as thumbnails with their pixel size; EMF and WMF clip art listed with their size and offered as .emf/.wmf files.
- The fonts named in the text's font table, so you know what to install before printing it elsewhere.
- The details: title, author, last saved by, created and saved dates (when Publisher recorded them), whether Publisher 98/2000 or 2002 and later wrote the file, macros and embedded objects, and the small first-page thumbnail Publisher keeps in every file, which you can save as PNG.
A worked example: a 158 KB sample brochure from Apache POI's test files gives 19 stories (the gallery's phone number, opening hours, three articles), 8 pictures (one JPEG photo, two GIFs and five WMF clip-art drawings), 22 font names and a 145 × 103 px thumbnail of its cover.
Why .pub files are getting stranded
Microsoft has announced that Publisher reaches end of support on 13 October 2026 and is then removed from Microsoft 365 subscriptions. Publisher never existed for macOS, iPad or Android, and no other Microsoft program opens a .pub: Word cannot import one. Years of newsletters, menus and flyers are kept in a format only one Windows program read fully. Getting the words and pictures out is usually what matters, and that is what this page does without installing anything.
Other ways out, honestly compared: Publisher itself, while you still have it, saves a PDF or a .docx (File › Save As) and is the only way to keep the exact layout. LibreOffice Draw (free, Windows, Mac and Linux) imports .pub files and redraws the pages, often closely, sometimes with shifted text and missing effects. Online converters do the same work on someone else's server, which means uploading the file.
How it reads the file
A .pub from Publisher 98 onwards is a Microsoft compound file (the container of old .doc and .xls files, starting D0 CF 11 E0) whose root carries the Publisher class ID {00021201-0000-0000-00C0-000000000046}. Inside it:
- Quill/QuillSub/CONTENTS: the text
- Starts with the eight bytes
CHNKINKand an index of chunks, each with a four-letter name, an offset and a length. TheTEXTchunk holds every story back to back in UTF-16 (0x0D ends a paragraph, 0x0B breaks a line); theSTRSchunk lists each story's length, which splits them;FONTlists the font names. - Escher/EscherStm and Escher/EscherDelayStm: the pictures
- Office's drawing format (the same records as in Word and PowerPoint files): a picture store whose entries point at picture records in the delay stream, each a short header and then the JPEG, PNG, GIF, bitmap or compressed metafile bytes, which are unpacked here.
- Contents: the layout
- Pages, frames, positions and formatting. This page reads only the length of its header, which tells a Publisher 98 or 2000 file from a 2002-or-later one.
- \x05SummaryInformation: the details
- The standard Office property set: title, author, dates, and the first-page thumbnail as a small bitmap inside a Windows metafile.
The reader is viewhack's own code, written from published descriptions of these structures and checked against real Publisher files; the compound-file, property and picture code is shared with this site's .doc and .msi viewers.
What this cannot do
- It does not lay out the pages as Publisher drew them. Frame positions, columns, master pages, colours, shapes and text formatting are not read: you get the words and the pictures, not the design.
- It cannot edit the file or convert it to an exact PDF. For that you need Publisher (Save As PDF) or, approximately, LibreOffice Draw.
- Very old Publisher files (Publisher 97 and earlier) are named but not read. They keep their text in a different layout; this page says what the file is and stops.
- Windows metafile pictures (EMF, WMF) are not drawn. Browsers cannot display them; they are saved as files that Windows, LibreOffice and Inkscape open.
- Story order is the file's order, usually but not always the order on the page. Table cells come one per line; WordArt and embedded objects (charts, spreadsheets) are not included, and pictures that were only linked, not embedded, are not in the file at all.