v0.5.7
Two content-model reporting improvements, both grounded in forum feedback.
Corpus recall is unchanged (600/607, 1.1% false positives) — these change
what and how much is reported, not the valid/invalid verdict.
Added
- Bare text directly in
<body>is now flagged in EPUB 2. XHTML 1.1'sbody
content model is block-level only, so loose text there is a content-model error
(RSC-005) — one per text run, with a realline:column. epubcheck reports
this; epubveri was silently missing it. (EPUB 3, whose HTML5 body allows flow
content including text, is unaffected.) The unambiguous EPUB 2 half of the
bare-text discussion on the MobileRead forum.
Changed
- Content-model (
RSC-005) failures now report every offending node, not just
the first. A list like<ol><p>…</p><p>…</p></ol>used to draw one finding;
it now reports each misplaced child, each with its ownline:columnand element
path — matching epubcheck's per-node output. (Reported by Doitsu on the
MobileRead forum. A valid-but-empty list is still fine, so only the misplaced
children are flagged, not the list element itself.)
Full notes: CHANGELOG.md. Also on crates.io.