400826 - text-transform ::first-letter and ::first-line affects whole element instead of only the first letter/line

Status: RESOLVED FIXED

(Core :: Layout: Text and Fonts, defect, P3)

Description

[Raphael Schweikert]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a9pre) Gecko/2007102205 Minefield/3.0a9pre
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a9pre) Gecko/2007102205 Minefield/3.0a9pre

text-transform uppercase/lowercase in a ::first-letter/::first-line rule is applied to the whole element instead of only the first letter/line. Other formatting rules are correctly only applied to the first elements. The problem disappears if any font-* rules are used (see second paragraph in the testcase).

Reproducible: Always

Steps to Reproduce:
1. Open the testcase with a current nightly build
2. Open the testcase in Firefox 2
3. Compare the rendering (Firefox 2 renders it correctly)
Actual Results:  
The first line of the first paragraph is red and capitalized, the rest of the paragraph is black but also in uppercase.

Expected Results:  
The first line of the first paragraph is red and capitalized, the rest of the paragraph is in lowercase and black.

I have reproduced this bug in current nightly builds on both mac and windows.

Comment 1

[Raphael Schweikert]

Attachment: Simple testcase

Comment 2

[Elmar Ludwig]

Duplicate of bug 287088?

Comment 3

[Elmar Ludwig]

Confiming for now and reassigning.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Reassigning back.

Comment 5

[Raphael Schweikert]

I've spent about half an hour searching for this bug before filing it to make absolutely sure it was not a duplicate… Oh well, sorry.

Comment 6

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Is this a regression?

Comment 7

[philippe (part-time)]

Well, Fx 2.0.0.8 renders this correctly. That makes it a regression.

This bug is a bit similar to bug 392435 (now fixed) which was a regression from bug 385270. When testing the patch in bug 392435, iirc I tested also with text-transform. Maybe it regressed again. I'll check.

Comment 8

[philippe (part-time)]

Ok, this regressed again:
(Mac OS X builds)
2007091716 Minefield/3.0a8pre works (tinderbox build)
2007091804 Minefield/3.0a8pre fails

http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2007-09-17+15%3A30%3A00&maxdate=2007-09-18+03%3A30%3A00&cvsroot=%2Fcvsroot

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Looks like the testcase in bug 392435 that I checked in doesn't test text-transform. Even if it did, the fact that changing the font properties makes the text-transform work as well probably would have masked this bug.

Comment 10

[philippe (part-time)]

Attachment: testcase with font-variant

Interesting test case using font-variant: small-caps;
lifted from the CSS 2.1 test suite.

Resize the window to see the changes in the first line (contains <strong> elements)

Comment 11

[Simon Montagu :smontagu]

Backing out bug 393796 fixes this.

Comment 12

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

We shouldn't build the styles array for transformed textruns assuming every frame in a flow has the same style; instead we should build it frame by frame.

Comment 13

[Reed Loden [:reed]]

Checking in layout/generic/nsTextFrameThebes.cpp;
/cvsroot/mozilla/layout/generic/nsTextFrameThebes.cpp,v  <--  nsTextFrameThebes.cpp
new revision: 3.118; previous revision: 3.117
done

Comment 14

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: reftest

I'll commit this when I get a chance

Comment 15

[Robert O'Callahan (:roc) (email my personal email if necessary)]

checked in reftest