Various XML, text and FB2 fixes #438
Conversation
In case footnotes' <body> don't come with a proper <title> that would have ensured this page break.
Increase the buffer size for recognition of named character references to handle all those expected to be recognised. The longest name in the def_entity_table[] appears to be "CounterClockwiseContourIntegral" so including a trailing null a buffer size of 32 is needed.
Rework 89b0650: ignore the space even if followed by another text node; just don't ignore it for the first line where it could have some purpose (eg. to add some indentation).
| lChar32 entname[16]; | ||
| for ( k = 0; k < 16; k++ ) { | ||
| lChar32 entname[32]; | ||
| for ( k = 0; k < 32; k++ ) { |
There was a problem hiding this comment.
I'd possibly replace the subsequent 32 occurrences with sizeof(entname)
EDIT: Oh, wait, not a byte array. Never mind me.
There was a problem hiding this comment.
Unless CRe already has an ARRAY_SIZE-like macro somewhere?
Otherwise: koreader/koreader-base@a01e142
There was a problem hiding this comment.
(I would have been embarassed :) I try to not change commits marked "(Upstream)", by principle.
Sometimes, I get a chance to tweak them in a followup commit touching some related code.
But as we're diverging again :/, this may change.)
|
Bumping this after 2021.05. When is the release planned? |
|
Hopefully tomorrow. ^_^ |
|
Mentionning a XML parsing issue, that I did not have the courage to fix in #555: <div>test XML parser and unquoted last attribute:</diV>
1:<img hspace="20" border=5 src="lights.gif"/>
2:<img hspace="20" border=10 src=lights.gif />
3:<img hspace="20" border=10 src=lights.gif/> <!-- this fails -->
4:<img hspace="20" src=lights.gif border=15/>
edit: beware when/if we fix this: if some elements did disappear, and fixing it would make it reappear, it could shift xpath/xpointers, screwing past highlights. So, it would need a new DOM_VERSION_CURRENT, that would need to propagate to the XML parser (which we up to now never had to do). |
|
I will say it's slightly surprising that 2 and 4 work while 3 doesn't, but eh, they're all supposed to fail in XML really. :-) |
fb2.css: ensure page break after
In case footnotes'
<body>don't come with a proper<title>that would have ensured this page break.See koreader/koreader#7625 (comment)
(Upstream) XML parsing: fix long named character references
Increase the buffer size for recognition of named character references to handle all those expected to be recognised. The longest name in the
def_entity_table[]appears to be CounterClockwiseContourIntegral" so including a trailing null a buffer size of 32 is needed.From buggins/coolreader#287
XML parsing: don't trim double spaces in attributes
Should allow closing koreader/koreader#7661 and hopefully won't cause much harm.
Fix ignore occasional space at start of line
Rework 89b0650: ignore the space even if followed by another text node; just don't ignore it for the first line where it could have some purpose (eg. to add some indentation).
See #364 (comment)
This change is