[GUI Plugin] epubveri library

[DNSB]

Could you try turning off the Usage Notes in the configuration so you don't see those messages or uncheck the checkbox for Usage Notes in the results?

Otherwise, I'm tempted to suggest that anyone using epubveri and unable to tell the difference between a usage obsolete item and a warning/error should perhaps not be using the utility.

After all, epubcheck with usage enabled will show you the same message (as you can see in my message #42 above) and part of the design philosophy for epubveri was to match epubcheck.

[Kayadelenium]

OBS-001 is the only "outdated" message, and it is a usage note. In this plugin usage notes are off by default (Preferences → Usage notes, and the same checkbox in the results window), so unless that box was ticked, those lines are not shown. They never affect the verdict either: a book whose only findings are OBS-001 is in the "no error" list. The message says "outdated", not "remove". An NCX in an EPUB 3 is still valid, and some reading systems still use it. If you saw them without ticking the box, tell me where (library or editor plugin) and I'll look.

[JSWolf]

Originally Posted by DNSB

Could you try turning off the Usage Notes in the configuration so you don't see those messages or uncheck the checkbox for Usage Notes in the results?

Otherwise, I'm tempted to suggest that anyone using epubveri and unable to tell the difference between a usage obsolete item and a warning/error should perhaps not be using the utility.

After all, epubcheck with usage enabled will show you the same message (as you can see in my message #42 above) and part of the design philosophy for epubveri was to match epubcheck.

I think the obsolete messages in both programs should be removed. The only messages that should show is what's lacking to make the ePub backwards compatible enough and/or what doesn't work in ePub2.

[Kayadelenium]

epubveri library 0.5.2: a safer install of the validator it downloads. Release notes

The plugin downloads epubveri from its GitHub release and checks the file against the checksum list published with it. Both come from the same place, so that check proves the download arrived intact. It cannot prove the release itself is sound. If a release were ever tampered with, the old unpacking step would have let a crafted archive write files outside the plugin's folder before anything was run. It now refuses any archive entry that would land outside that folder. Nothing suggests any release was ever tampered with, and you will not see any difference: this closes a door rather than fixing something you have met.

calibre offers the update itself. You can also take it from Preferences / Plugins / Get new plugins. The zip on the first post is the same file as the release, and so is its checksum.

Separately, the validator the plugin runs moves to epubveri 0.17.1 by itself the next time you use the plugin, unless you have switched off its automatic update. That release fixes a crash and a false error and closes a way to exhaust memory. What changed there is described in the epubveri thread, which is where the validator itself is discussed.

[Kayadelenium]

Originally Posted by DNSB

I left the cores at automatic and the total time for 17703 books dropped to 10 min 48 sec using 14 cores.

@DNSB, if you have the time, I would love to see that run again. epubveri 0.17.5, out today, spends much less time repeating work it has already done, with the same reports, and the plugin picks it up by itself. How much faster that makes a scan depends on the books in the library, so my own numbers say little about yours. Your 17,703 books on 14 cores are a far better test than my 474, and your 10 min 48 sec is the perfect number to compare against.

The easiest way to share the result is Copy / "What this scan was" in the report window: it puts the time, the epubveri version, the core count and your system on one line, so we can both see the new version did the work. If you have scanned the whole library since mid-September, "Compare with the previous scan" will also show which rules moved.

The findings won't match your September run exactly: three weeks of new and corrected rules sit between the two. If any book gets something that looks wrong to you, please tell me, because that is exactly what I most want to hear about.

[DNSB]

The new scan on my main library took 185 seconds so much faster. The Copy results are below. Since I had run several scans on my Intake library, the results for my Main library were not available. Since I now have usage turned on, the number of errors was a lot higher and I was also noticing more errors overall. Looking at a couple of the books, the errors were mostly in older books from the 2010-2015.

scanned: 2026-09-23 16:52
books: 17703 of 17703
took: 185 s (0.010 s/book)
at once: 14
epubveri: 0.17.5+ec09da4
plugin: 0.5.2
system: Windows 11 (10.0.26200)
processor: AMD64
cores: 16 physical, 24 logical
calibre: 9.15.100

[DNSB]

I took a look at a repeated error message on some of the older books. One oddity was that what epubveri and epubcheck named correctly under Sigil showed a remote resources error using epubveri library under calibre.

epubveri_library/calibre:

OPF-014	opf.content_document.property_used_undeclared			error	1	stylesheet uses a remote resource but doesn't declare the "remote-resources" property - add it to this stylesheet's manifest item in the OPF

epubcheck/Sigil:

OEBPS/cover.xhtml	N/A	N/A	Col: -1: ERROR(OPF-014): The property "svg" should be declared in the OPF file.

epubveri/Sigil:

OEBPS/cover.xhtml	13	460	ERROR OPF-014: content document uses svg but doesn't declare the "svg" property - add it to this document's manifest item in the OPF

[Kayadelenium]

Thank you. I am not adding ADV-013. The discussion in the Sigil plugin thread settled it: DNSB tested a 1.3 MB single-file book on current readers without trouble, and KevinH pointed out that the size limit belongs to very old readers, not to EPUB. calibre's Check Book and the Quality Check plugin already report large HTML files, so nothing changes in this plugin.

@JSWolf, switching advisory notes on and off by group is a separate idea, and I have noted it.

[Kayadelenium]

epubveri library 0.5.3: each book in the per-book export now gets its own message. Release notes

@DNSB, thank you for #52. You found a real bug, and it was in this plugin, not in epubveri. epubveri reported your cover.xhtml correctly, as the missing "svg" property, which is why Sigil showed the right message. The report groups findings by rule, and one rule (OPF-014's property_used_undeclared) has several different sentences: "svg" for a content document, "remote resource" for a stylesheet, and so on. The per-book export then took the first sentence the scan met for that rule and wrote it against every book in the row. So your book was listed with another book's sentence.

Now each book is listed with its own message, and with "e.g." in front when that book has more than one message for the rule. The comparison export and the tooltips in the report window show "e.g." too wherever a row holds different sentences, as the table already did. A scan you saved with 0.5.2 has no per-book messages, so its export falls back to the row's "e.g." label rather than to another book's sentence. Scan again to get each book's own message.

calibre will offer the update itself, or you can take it from Preferences / Plugins / Get new plugins. The zip on the first post is the same file as the release, and so is its checksum.

[slimcat]

Small feature request: would you consider supporting custom icon overrides through calibre's resources folder?

Thank you, by the way, for this plugin. It's a game changer!