Wallabag: epubs being converted to htm in 2024.03+

[dlgoodr]

• KOReader version: 2024.03
• Device: android (leaf2 and inkpalm) (can test kindle if needed)

Issue

EPUB files downloaded through the wallabag plugin are being converted to HTML and lots of the cruft wallabag strips is restored, along with guesses about chapter marks.

Steps to reproduce

Install latest, download article epub, weep. Restore to 2024.01 for the expected behavior.

This isn't a crash, just an unwelcome change in experience that I can't figure out how to revert without going back to 2024.01

[poire-z]

Probably #11492 (not using Wallabag, so no idea of the context).
Try reverting manually the small changes in 3e7ab19.

[dlgoodr]

Based on my poor understanding of LUA, we shouldn't be affected, we're correctly returning application/epub+zip for the file:

direct wget of epub

dlg@manila ~ % wget https://wb.example.org/api/entries/1668524955/export.epub
--2024-03-09 10:09:13--  https://wb.example.org/api/entries/1668524955/export.epub
Resolving wb.example.org (wb.example.org)... 100.12.12.12
Connecting to wb.example.org (wb.example.org)| 100.12.12.12 |:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 890444 (870K) [application/epub+zip]
Saving to: ‘export.epub’

export.epub         100%[===================>] 869.57K   711KB/s    in 1.2s    

2024-03-09 10:09:15 (711 KB/s) - ‘export.epub’ saved [890444/890444]

dlg@manila ~ % file export.epub 
export.epub: EPUB document
dlg@manila ~ % 

but reverting 3e7ab19 definitely fixes the issue for me.

[poire-z]

Quick reading, and it feels this change negate the original aim:

-    if article.mimetype ~= "text/html" then -- different from text/html
+    if type(article.mimetype) == "string" and article.mimetype:find("^text/html") then -- starts/equal to text/html

@Frenzie ?

[Frenzie]

No time today but based purely on the line you posted there a not is missing

…

On Sat, Mar 9, 2024, 15:50 poire-z ***@***.***> wrote: Probably #11492 <#11492> (not using Wallabag, so no idea of the context). Try reverting manually the small changes in 3e7ab19 <3e7ab19> . — Reply to this email directly, view it on GitHub <#11528 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABRQBJ2FXHIZ4POWYHOTJTYXMOT3AVCNFSM6AAAAABEOFCCQ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBWHA3TMNBUHE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[dlgoodr]

Just adding a not does not fix the issue.

[Frenzie]

@dlgoodr Could you share a URL where you see this issue and your Wallabag version for good measure?

[dlgoodr]

@Frenzie
https://wbip.dsrw.org/api/entries/1668524955/export.epub

[Frenzie]

@dlgoodr That's an export. Please provide the original URL.

[dlgoodr]

@Frenzie sorry, the original article URL?

https://www.theatlantic.com/magazine/archive/2024/04/secret-gospel-mark-controversial-forgery-jesus/677472/

[Frenzie]

Curious, my Wallabag gives text/html; utf-8 for that. In any case, so will have simply been the obvious case of an exact text/html match, thanks.

[Frenzie]

Tests welcome.