rgaudin
commented
1 year ago Indeed, with this ZIM, it looks like all epubs are stored as application/zip. This is a regression from the switch to pylibzim as the mime is now guessed by libmagic…
| mime | nb |
|---|---|
| application/javascript | 1487 |
| application/pdf | 6 |
| application/vnd.ms-fontobject | 1 |
| application/vnd.ms-opentype | 1 |
| application/zip | 794 |
| audio/midi | 7 |
| audio/mpeg | 21 |
| font/sfnt | 10 |
| font/woff | 1 |
| image/gif | 50 |
| image/jpeg | 7318 |
| image/png | 3722 |
| image/svg+xml | 1 |
| image/vnd.microsoft.icon | 1 |
| text/css | 13 |
| text/html | 1958 |
| text/plain | 1 |
Fixing it now
Jaifroid
I was confused by the fact that, recently, downloading EPUBs from a Gutenberg ZIM in Kiwix JS-family apps is producing a ZIP archive instead of an EPUB. This never used to happen. I've now found an older Gutenberg ZIM (from 2018), and have compared the stated MIME types. EPUB dirEntries in the old (version 0) ZIM are
application/epub+zip(see BOTTOM screenshot) -- this is correct. However, in the same language (version 1) ZIM scraped last month, the MIME type isapplication/zip(see TOP screenshot + accompanying EPUB dirEntry).Because we use the MIME type to determine the download type, the incorrect MIME type causes an issue for us. Possibly this issue hasn't been picked up in other readers because they may rely on the file extension, or give more weight to it?
The main difference between these two ZIMs is that the old one is a version 0 ZIM, and the new one is a version 1 ZIM. That may or may not be relevant.
INCORRECT MIME TYPE in 2023 ZIM:
CORRECT MIME TYPE in 2018 ZIM: