EPUBCheck 5.0 preview for EPUB 3.3

[jhowell]

A preview release of EPUBCheck 5.0 has been posted at https://github.com/w3c/epubcheck/releases
(Look under Assets for the download link.)

This is an early preview release of the next major version of EPUBCheck, v5.0.0. It provides initial (partial) support for checking conformance to EPUB 3.3.

This EPUBCheck version is not available in the Maven Central Repository. It is only available (temporarily) as a binary on the GitHub release page.

This release was made by the DAISY Consortium for the W3C. Many thanks to everyone who contributed and reported issues!

Runtime

• require Java 8 (9435936)

Features

• refactored API and internals (39888e2)
• add new Core Media Types (ECMAScript, OPUS, WebP) (166256a), closes #1249 #1189
• allow 'auto' value for the 'dir' attribute of Package Documents (a2e3a77), closes #1220
• allow 'hreflang' attribute on Package Document link elements (e39a801), closes #1219
• allow informative schema checking (reported as new RSC-024 and RSC-025) (a2516f0)
• check that item URLs have no fragment (ec28b59), closes #1250
• do not check conformance of SVG content (ba48aaa)
• enable informative (USAGE) checking of SVG content conformance (b9ddf8f)
• new check (OPF-092) for language tags well-formedness (52ebd80), closes #1221 #702
• update checking of the Package Document 'refines' attribute (72366b4), closes #1226
• update to latest schemas from the Nu HTML Checker (af7a5f7)

BREAKING CHANGES
the required Java runtime environment is now Java 8 or greater

[KevinH]

FWIW, It appears to be an alpha 1 release so it is probably not for production use.

[Doitsu]

Originally Posted by KevinH

FWIW, It appears to be an alpha 1 release so it is probably not for production use.

FYI: Since the latest version doesn't have a proper version number, the updater of the Sigil EPUBCheck plugin choked on it.
I released a new version that'll prevent the download

The Calibre EPUBCheck plugin doesn't have this limitation and will offer the EPUBCheck 5.0 preview as an update.
I posted instructions on how to disable update checks in the Calibre forum.

[jhowell]

There is a new EPUBCheck preview. Version v5.0.0-beta-1 is available from https://github.com/w3c/epubcheck/releases

Features

• allow SVG/MathML doctype declarations (6e44b39), closes #1192 #1114
• allow any file extension (bdee846)
• check discouraged HTML elements (base, rp, embed) (afb28cb), closes #1271
• check Media Overlays total duration consistency (ee9c720), closes #1217
• check that 'page-list' and 'landmarks' nav are flat lists (15825a5), closes #1279
• check that publication resources are not in META-INF (61dbf53), closes #1227
• HTML link element resources do not require fallbacks (8d77b0f), closes #1247
• ignore Nav Doc nav elements with no epub:type (f989588), closes #1222
• loosen restriction on where remote resources can be referenced (73fe57c), closes #1288
• page-list nav does not have to match reading order (d0f12a9), closes #1237
• report reserved string in XHTML custom attribute namespaces (bc86db8), closes #1190
• restrict obfuscation to font core media types (a229edf), closes #1291

Minor Changes

• remove "under review" notice for toc nav ordering warning (fe08c59), closes #1194

[jhowell]

EPUBCheck version v5.0.0-beta-2 is available from https://github.com/w3c/epubcheck/releases

This is a preview release of the next major version of EPUBCheck, v5.0.0. It keeps on improving support for checking conformance to EPUB 3.3.

This EPUBCheck version is not available in the Maven Central Repository. It is only available (temporarily) as a binary on the GitHub release page.

This release was made by the DAISY Consortium for the W3C. Many thanks to everyone who contributed and reported issues!

Features

• proper handling of URL parsing throughout all checks, based on the URL standard (using the Galimatias library)
• reorganized tests for EPUB 3.3
• allow SVG/MathML doctype declarations (6e44b39), closes #1192 #1114
• allow any file extension (bdee846)
• check discouraged HTML elements (base, rp, embed) (afb28cb), closes #1271
• check Media Overlays total duration consistency (ee9c720), closes #1217
• check that 'page-list' and 'landmarks' nav are flat lists (15825a5), closes #1279
• check that publication resources are not in META-INF (61dbf53), closes #1227
• HTML link element resources do not require fallbacks (8d77b0f), closes #1247
• ignore Nav Doc nav elements with no epub:type (f989588), closes #1222
• loosen restriction on where remote resources can be referenced (73fe57c), closes #1288
• page-list nav does not have to match reading order (d0f12a9), closes #1237
• report reserved string in XHTML custom attribute namespaces (bc86db8), closes #1190
• restrict obfuscation to font core media types (a229edf), closes #1291
• refactored API and internals (39888e2)
• add new Core Media Types (ECMAScript, OPUS, WebP) (166256a), closes #1249 #1189
• allow 'auto' value for the 'dir' attribute of Package Documents (a2e3a77), closes #1220
• allow 'hreflang' attribute on Package Document link elements (e39a801), closes #1219
• allow informative schema checking (reported as new RSC-024 and RSC-025) (a2516f0)
• check that item URLs have no fragment (ec28b59), closes #1250
• do not check conformance of SVG content (ba48aaa)
• enable informative (USAGE) checking of SVG content conformance (b9ddf8f)
• new check (OPF-092) for language tags well-formedness (52ebd80), closes #1221 #702
• update checking of the Package Document 'refines' attribute (72366b4), closes #1226
• update to latest schemas from the Nu HTML Checker (af7a5f7)
• remove "under review" notice for toc nav ordering warning (fe08c59), closes #1194

[jhowell]

EPUBCheck version v5.0.0-beta-3 is available from https://github.com/w3c/epubcheck/releases

This is the latest preview release of the next major version of EPUBCheck, v5.0.0. This version fully implements the new rules in EPUB 3.3.

We recommend users to effictively test this EPUBCheck version in their production workflow, to get ready for EPUB 3.3! 🚀

Note: This EPUBCheck version is not available in the Maven Central Repository. It is only available (temporarily) as a binary on the GitHub release page.

This release was made by the DAISY Consortium for the W3C. Many thanks to everyone who contributed and reported issues!

Features

• allow 1s tolerance for media overlays duration sum check (e64c7c5), closes #1328
• allow NCX documents to declare a doctype (f3e5f67), closes #1329
• better check media overlays styling properties (6a58b8e)
• better parse URL fragment micro syntaxes (bec390e)
• check epub:type restrictions in XHTML and SVG (0fab5c8), closes #1348
• check file name uniqueness wiht Unicode canonical case fold normalization (111e772), closes #1246
• check fragment requirements on overlays links to text content (2091d14), closes #1248 #1301
• check that container-relative URLs have no query (a4fed67)
• check that overlays audio file URLs have no fragments (fdabe66), closes #1251
• check the formal viewport meta tag syntax (01a7758), closes #1214 #1347
• extend proper URL checking to more places (like CSS) (a3c736d)
• improve checking of data URLs (cbc0b2a), closes #1238 #1239
• improve checking of OCF file name characters (b6ac8ea), closes #1240 #1292 #1302
• improve detection of non-UTF-8 file names (5248914), closes #1236
• improve fallback detection check (545b7f7), closes #1304 #1298
• new requirements of package link 'media-type' attribute (3199b81), closes #1307
• no longer check collection role (8eb8492), closes #1350
• remove checks related to deprecated bindings feature (7ca1edf)
• report 'file' URLs as errors (0f6b509), closes #1270
• report empty URL as a schema error instead of HTM-008 (e3d3afe)
• restrict SVG title element to HTML elements (8631b7d), closes #1342
• update checking of the package rendering vocabulary (3086258), closes #1327
• update the reporting of file encoding issues (0d6f927), closes #1245
• warn about non-HTTPS remote resource references (7eb5ff2), closes #1337

Bug Fixes

• consider HTML form elements as scripted content (1bae7f7), closes #1282
• declaring prefixes for default vocabs is an error (035effa), closes #1306
• fix CVE-2021-23792 vulnerability from imageio-jpeg (50b847f), closes #1336
• improve checking of missing SVG link labels, now USAGE (5903626), closes #1353
• OPF-018 was incorrectly reported with inline CSS (c3d767c), closes #1335
• regression that lost line and column numbers in messages (20b5142)
• remove recursion in reading order checks (6cbbefa), closes #1358 #1356
• report fatal XML parsing errors only once as RSC-016 (62d67e7)
• support CSS logical combination pseudo-classes (5635807), closes #1289 #1354

[Quoth]

What difference for epub 2 users?

[Karellen]

Originally Posted by jhowell

EPUBCheck version v5.0.0-beta-3 is available from https://github.com/w3c/epubcheck/releases

Seems it won't install in Calibre. How/where do you use the zip file?

[Doitsu]

Originally Posted by Karellen

Seems it won't install in Calibre. How/where do you use the zip file?

If you're using my Calibre Editor plugin, you'll need to manually upgrade the EPUBCheck Java files.
Preferences > Miscellaneous > Open Calibre Configuration folder > plugins > EPUBCheck.
Replace epubcheck.jar and the lib folder.
(The plugin won't download beta versions by design.)

[Karellen]

Originally Posted by Doitsu

If you're using my Calibre Editor plugin, you'll need to manually upgrade the EPUBCheck Java files.
Preferences > Miscellaneous > Open Calibre Configuration folder > plugins > EPUBCheck.
Replace epubcheck.jar and the lib folder.
(The plugin won't download beta versions by design.)

Great, thank you.
Installed and working fine.

[jhowell]

Originally Posted by Quoth

What difference for epub 2 users?

Most of the effort is going to checking conformance with the upcoming EPUB 3.3 specification. It is unclear which changes, if any, impact EPUB 2 conformance testing.

[DNSB]

Also works under Sigil using the replace epubcheck.jar and the lib directory. Still no version number but Running EpubCheck HEAD... does give a gentle hint that you are using the beta.

[Quoth]

Originally Posted by jhowell

Most of the effort is going to checking conformance with the upcoming EPUB 3.3 specification. It is unclear which changes, if any, impact EPUB 2 conformance testing.

Thanks.

[DNSB]

I did a comparison between the two versions (4 and 5) using a trio of epub2 files that had errors when purchased and I saw no real differences other than some error messages only appearing once with the beta. One of files was a puzzle since it was downloaded from Google and says it's an epub2 but quite a few epub3 specific bits and bobs (epub:type, role, etc.) are in it which triggered errors. I keep that original of that one around just for testing.

[Doitsu]

EPUBCheck version v5.0.0 is available from https://github.com/w3c/epubcheck/releases/tag/v5.0.0.

My Sigil and Calibre EPUBCheck plugins should automatically download this version.