Accents #2404
Accents #2404
Conversation
… standalone char and for use in math
…g) chars for accents, leveraging data from Util::Unicode module
| if (my $entry = unicode_accent($char)) { | ||
| applyAccent($stomach, $letter, $$entry{combiner}, $$entry{standalone}, | ||
| Invocation(T_CS('\accent'), $num, $letter)); } | ||
| else { # Unknown accent ? Really should OVERLAY it on top of $letter??? |
There was a problem hiding this comment.
Do we have a small test for using an unknown accent? Might be useful to keep checking as things move forward.
There was a problem hiding this comment.
OK, put in a simplistic handling for this case, along w/CSS and an extra line in test.
lib/LaTeXML/Util/Unicode.pm
Outdated
| @@ -24,6 +23,73 @@ sub UTF { | |||
| my ($code) = @_; | |||
| return pack('U', $code); } | |||
|
|
|||
| my $NBSP = UTF(0xA0); | |||
| sub NBSP { return $NBSP; } | |||
There was a problem hiding this comment.
Is this attempting to guard the lexical $NBSP from some sort of external redefinition?
Surprised it wasn't introduced as a package-level our $NSBP; declaration. That can be added to @EXPORT too, if desired.
There was a problem hiding this comment.
Thanks, switched over to cleaner \N{NBSP} approach (along with use charnames ':full'; for older perls)
| foreach my $entry (@accent_data) { | ||
| $accent_data{ $$entry{standalone} } = $entry; | ||
| $accent_data{ $$entry{combiner} } = $entry; | ||
| } |
There was a problem hiding this comment.
Oh, this is rather dangerous. Could you change the name of either @accent_data or %accent_data?
I was thrown for a bit not knowing what happened, seeing the hash indexing syntax in $accent_data{$char}.
Such a setup is bound to generate a subtle bug one day, when someone uses the wrong delimiters in haste:
my @a = ('x','y','z');
my %a = ('1'=>'a', '2'=>'b', '3'=>'c');
print $a[2]; # z
print $a{2}; # bThere was a problem hiding this comment.
I think the hash variant can be kept, and you can even fully expand the loop that adds standalone and combiner in code, to avoid looping the setters every time latexml is ran...
Simplest is to run it once, print with Dumper, then copy the result back into the file. And on second read - aren't standalone and combiner already set in the explicit hashes? Maybe the loop is a leftover? Except that the loop sets them as equal, while the hash often has the values as different. Hm, I wish I understood a little more...
There was a problem hiding this comment.
Renamed a bit to make it less scary.
There was a problem hiding this comment.
Thank you for the updates, the PR looks ready to merge. ✅
I tried another two alternatives for the "unknown accent" markup, visible in this codepen. Not a focus at the moment, but something to mull over in the background. <ruby> and display: inline-grid; are curious tools.
This PR is another round in improving the fidelity of lower-level TeX and plain.tex. It deals with accents. Accents in TeX fonts basically show an accented space which gets overlaid on top of a character to be accented (potentially with some positioning tweaks). To deal with Unicode and MathML, we need to distinguish 3 kinds of thing:
So, we've set up a table within
LaTeXML::Util::Unicodeto store this info, and use it wherever accenting operations are needed.