263411 - ::first-letter pseudo-element doesn't include punctuation marks :

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Jungshik Shin]

This is a regression due to the patch for bug 180266. A testcase is available in
bug 54467. Somehow, I made a mistake of not including all characters with
Unicode character class Ps, Pe and Po. I'm gonna generate a new list out of the
Unicode 4.0.1 data file.

Comment 1

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

FWIW, CSS 2.1 says:

  Punctuation (i.e, characters defined in Unicode [UNICODE] in the "open" (Ps),
  "close" (Pe), "initial" (Pi). "final" (Pf) and "other" (Po) punctuation
  classes), that precedes or follows the first letter should be included

  -- http://www.w3.org/TR/2004/CR-CSS21-20040225/selector.html#first-letter

Two classes were added between last call and CR as a result of updating the
Unicode reference to Unicode 4.0:
http://www.w3.org/Style/css21-updates/CR-CSS21-20040225-20030915-diff/selector.html#first-letter

Comment 2

[Jungshik Shin]

Attachment: patch (Unicode 4.0.1, CSS 2.1 update)

I generated a new list out of the Unicode 4.0.1 data file. It includes
characters classified as Ps,Po,Pe, Pi, and Pf. 
(cut -d ';' -f 1-3 UnicodeData-4.0.1.txt | egrep 'Ps|Pe|Po|Pf|Pi' | cut -d ';'
-f 1-2	| sed -e 's/;/ : /'  -e 's/^/  0X/')
I excluded three non-BMP characters from the list because nsTextFrame.cpp
currently doesn't handle non-BMP characters correctly when it comes to identify
'first-letter'. 

Thanks for saving me some digging up. I was wondering why U00AB was not
included in my list generated from the Unicode 4.0.1 data file and it turned
out to be classified as Pi.

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Could you update the comment at the top of punct_marks.ccmap to describe how it
was generated?

Also, there are a lot of new characters added that seem a bit surprising, but
after thinking about them I think a lot of them make sense, e.g., for:

*A*
[B]
¿Y?
¡O!

Comment 4

[Jungshik Shin]

fix checked in with a comment added about the way the file was generated. 

Comment 5

[Jungshik Shin]

Comment on attachment 161430 [details] [diff] [review]
patch (Unicode 4.0.1, CSS 2.1 update)

asking for a to branches.
This patch fixes the regression introduced between 1.3 and 1.4 by my mistake in
generating 'punct_marks.ccmap'. It doesn't change any C++ code except for
several lines of comments.

Comment 6

[Mike Kaply [:mkaply]]

Comment on attachment 161430 [details] [diff] [review]
patch (Unicode 4.0.1, CSS 2.1 update)

a=mkaply for branches

Comment 7

[Jungshik Shin]

thanks for r/sr/a