Bug 114721 – REGRESSION (r147588): Line breaks occur in the middle of Hebrew words at haaertz.co.il and other websites

• Status: RESOLVED
• Resolution: FIXED
• Component: Text

Created attachment 198464 [details]
Test case - Hebrew - with empty inline

<rdar://problem/13670430>

To reproduce: open the attached test case. The second line breaks in the middle of the word צרור. It should break before that word. This was caused by <http://trac.webkit.org/r147588>, the fix for bug 105692.

[Glenn Adams]

Will work on this later today. Please hold off on a revert.

[Glenn Adams]

Created attachment 198625 [details]
Patch

[Philippe Wittenbergh]

I suspect this might be the same as what I see on some French pages.
Example:
http://www.lemonde.fr/societe/article/2013/04/16/la-cedh-refuse-l-extradition-vers-les-etats-unis-d-un-schizophrene-accuse-de-terrorisme_3160563_3224.html

In the last paragraph, in the sentence 'Il fait partie d'un groupe de six islamistes radicaux dont le célèbre imam londonien Abou Hamza Masri …', the word 'célèbre' is split after 'cé'. The line break should happen before or after the whole word (before based on Safari 6.04).

[Glenn Adams]

Created attachment 198626 [details]
Patch

[Glenn Adams]

(In reply to comment #3)
> I suspect this might be the same as what I see on some French pages.
> Example:
> http://www.lemonde.fr/societe/article/2013/04/16/la-cedh-refuse-l-extradition-vers-les-etats-unis-d-un-schizophrene-accuse-de-terrorisme_3160563_3224.html
> 
> In the last paragraph, in the sentence 'Il fait partie d'un groupe de six islamistes radicaux dont le célèbre imam londonien Abou Hamza Masri …', the word 'célèbre' is split after 'cé'. The line break should happen before or after the whole word (before based on Safari 6.04).

I'll investigate, but at first glance, it appears to be a different case, since there is no empty inline involved (as was the case with the hebrew text). The patch I just posted fixes the hebrew test case, but not the example you cite.

(In reply to comment #5)
> (In reply to comment #3)
> > I suspect this might be the same as what I see on some French pages.
> > Example:
> > http://www.lemonde.fr/societe/article/2013/04/16/la-cedh-refuse-l-extradition-vers-les-etats-unis-d-un-schizophrene-accuse-de-terrorisme_3160563_3224.html
> > 
> > In the last paragraph, in the sentence 'Il fait partie d'un groupe de six islamistes radicaux dont le célèbre imam londonien Abou Hamza Masri …', the word 'célèbre' is split after 'cé'. The line break should happen before or after the whole word (before based on Safari 6.04).
> 
> I'll investigate, but at first glance, it appears to be a different case, since there is no empty inline involved (as was the case with the hebrew text). The patch I just posted fixes the hebrew test case, but not the example you cite.

The attached test case is based on an example where the inline wasn’t empty. If your patch doesn’t fix that, then I suggest that you rework it to completely fix the regression.

Added a URL that shows the bug and doesn’t involve empty inlines.

[Philippe Wittenbergh]

(In reply to comment #5)
 
> I'll investigate, but at first glance, it appears to be a different case, since there is no empty inline involved (as was the case with the hebrew text). The patch I just posted fixes the hebrew test case, but not the example you cite.

It has the same regression range: between r147562 and r147603 (based on publicly available builds)

[Glenn Adams]

Created attachment 198629 [details]
Test case - French

This test case fragment is based on http://www.lemonde.fr/societe/article/2013/04/16/la-cedh-refuse-l-extradition-vers-les-etats-unis-d-un-schizophrene-accuse-de-terrorisme_3160563_3224.html.

[Glenn Adams]

(In reply to comment #8)
> (In reply to comment #5)
> 
> > I'll investigate, but at first glance, it appears to be a different case, since there is no empty inline involved (as was the case with the hebrew text). The patch I just posted fixes the hebrew test case, but not the example you cite.
> 
> It has the same regression range: between r147562 and r147603 (based on publicly available builds)

I agree it is a regression from r147588. I was just saying that it didn't seem the same mechanism as the empty inline. In any case, I will do as mitz suggests and create a patch that fixes both cases.

[Glenn Adams]

Created attachment 198669 [details]
Test case - Latin1 Reduction

Reduced case of French example of regression. Here showing alternatives that do not entail regression.

[Glenn Adams]

Created attachment 198740 [details]
Test case - Latin1 Reduction - Improved

A few more variations.

[Glenn Adams]

Created attachment 198791 [details]
Patch

[Glenn Adams]

(In reply to comment #13)
> Created an attachment (id=198791) [details]
> Patch

This patch corrects both the original problem reported by mitt (comment #1) and the additional problem reported by Philippe (comment #3). Before requesting review, I want to run some performance tests.

[Glenn Adams]

(In reply to comment #14)
> (In reply to comment #13)
> > Created an attachment (id=198791) [details] [details]
> > Patch
> 
> This patch corrects both the original problem reported by mitt (comment #1) and the additional problem reported by Philippe (comment #3). Before requesting review, I want to run some performance tests.

s/mitt/mitz/

[Build Bot]

Comment on attachment 198791 [details]
Patch

Attachment 198791 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/11201

New failing tests:
fast/writing-mode/Kusa-Makura-background-canvas.html

[Build Bot]

Created attachment 198796 [details]
Archive of layout-test-results from webkit-ews-05 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-05  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Build Bot]

Comment on attachment 198791 [details]
Patch

Attachment 198791 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/170168

New failing tests:
fast/writing-mode/Kusa-Makura-background-canvas.html

[Build Bot]

Created attachment 198800 [details]
Archive of layout-test-results from webkit-ews-01 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-01  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Glenn Adams]

Created attachment 198801 [details]
Patch

[Glenn Adams]

Created attachment 198922 [details]
Performance Results

No apparent degradation.

[Glenn Adams]

Comment on attachment 198801 [details]
Patch

No performance degradation indicated; ready for review.

[Dean Jackson]

Comment on attachment 198801 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=198801&action=review

> LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:18
> +    testRunner.dumpAsText();
> +function getLineWidths(paragraphNumber) {

Nit: Needs blank line

> LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:33
> +}
> +var widths1 = getLineWidths(0);

Here too.

> LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:38
> +if (widths1.length != widths2.length) {
> +    results += 'FAIL: different number of lines, got ' + widths2.length + ', expected ' + widths1.length;
> +} else {

Single line block -> no braces.

> LayoutTests/fast/text/line-break-after-inline-latin1.html:10
> +<p>With span inline on first word, Latitn1 in first position of second word.</p>

Typo Latin1

> LayoutTests/fast/text/line-break-after-inline-latin1.html:16
> +<p>Without span, Latitn1 in first position of second word.</p>

Typo Latin1

> LayoutTests/fast/text/line-break-after-inline-latin1.html:25
> +    testRunner.dumpAsText();
> +function mergeRect(rects, rect) {

Nit: needs blank line

> LayoutTests/fast/text/line-break-after-inline-latin1.html:29
> +    if (!rects.length) {
> +        newRects.push(rect);
> +    } else {

Single line block -> no braces

> LayoutTests/fast/text/line-break-after-inline-latin1.html:33
> +            if (!rect) {
> +                newRects.push(r);

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:41
> +                if (rect.left + rect.width > r.left + r.width) {
> +                    r.width = rect.left + rect.width - r.left;
> +                }

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:46
> +            } else {
> +                newRects.push(r);
> +            }

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:50
> +        if (rect) {
> +            newRects.push(rect);
> +        }

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:57
> +    for (var i = 0; i < newRects.length; ++i) {
> +        rects = mergeRect(rects, newRects[i]);
> +    }

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:85
> +    if (rects1.length != rects2.length) {
> +        results = appendResults(results, 'FAIL: different number of lines, got ' + rects1.length + ', expected ' + rects2.length);
> +    } else {

Ditto

> LayoutTests/fast/text/line-break-after-inline-latin1.html:99
> +for (var i = 0; i < 3; i++ ) {
> +    results = compareParagraphLineRects(i*2 + 1, i*2 + 1 + 6, results);
> +}

Ditto

[Glenn Adams]

Thanks. I'll fix these tomorrow before landing. I didn't realize folks were using the WK C++ style convention for JS. Will keep that in mind in future.

(In reply to comment #23)
> (From update of attachment 198801 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=198801&action=review
> 
> > LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:18
> > +    testRunner.dumpAsText();
> > +function getLineWidths(paragraphNumber) {
> 
> Nit: Needs blank line
> 
> > LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:33
> > +}
> > +var widths1 = getLineWidths(0);
> 
> Here too.
> 
> > LayoutTests/fast/text/line-break-after-empty-inline-hebrew.html:38
> > +if (widths1.length != widths2.length) {
> > +    results += 'FAIL: different number of lines, got ' + widths2.length + ', expected ' + widths1.length;
> > +} else {
> 
> Single line block -> no braces.
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:10
> > +<p>With span inline on first word, Latitn1 in first position of second word.</p>
> 
> Typo Latin1
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:16
> > +<p>Without span, Latitn1 in first position of second word.</p>
> 
> Typo Latin1
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:25
> > +    testRunner.dumpAsText();
> > +function mergeRect(rects, rect) {
> 
> Nit: needs blank line
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:29
> > +    if (!rects.length) {
> > +        newRects.push(rect);
> > +    } else {
> 
> Single line block -> no braces
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:33
> > +            if (!rect) {
> > +                newRects.push(r);
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:41
> > +                if (rect.left + rect.width > r.left + r.width) {
> > +                    r.width = rect.left + rect.width - r.left;
> > +                }
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:46
> > +            } else {
> > +                newRects.push(r);
> > +            }
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:50
> > +        if (rect) {
> > +            newRects.push(rect);
> > +        }
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:57
> > +    for (var i = 0; i < newRects.length; ++i) {
> > +        rects = mergeRect(rects, newRects[i]);
> > +    }
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:85
> > +    if (rects1.length != rects2.length) {
> > +        results = appendResults(results, 'FAIL: different number of lines, got ' + rects1.length + ', expected ' + rects2.length);
> > +    } else {
> 
> Ditto
> 
> > LayoutTests/fast/text/line-break-after-inline-latin1.html:99
> > +for (var i = 0; i < 3; i++ ) {
> > +    results = compareParagraphLineRects(i*2 + 1, i*2 + 1 + 6, results);
> > +}
> 
> Ditto

[Glenn Adams]

Created attachment 198945 [details]
Patch for landing

[Glenn Adams]

(In reply to comment #25)
> Created an attachment (id=198945) [details]
> Patch for landing

With JS style fixes.

[WebKit Commit Bot]

Comment on attachment 198945 [details]
Patch for landing

Clearing flags on attachment: 198945

Committed r148791: <http://trac.webkit.org/changeset/148791>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Glenn Adams]

Note that fast/writing-mode/Kusa-Makura-background-canvas.html will require rebaselining (text and image).

[Glenn Adams]

(In reply to comment #29)
> Note that fast/writing-mode/Kusa-Makura-background-canvas.html will require rebaselining (text and image).

P.S. I'm collecting results now to do this rebaseline very shortly.

[Glenn Adams]

(In reply to comment #30)
> (In reply to comment #29)
> > Note that fast/writing-mode/Kusa-Makura-background-canvas.html will require rebaselining (text and image).
> 
> P.S. I'm collecting results now to do this rebaseline very shortly.

Done.

http://trac.webkit.org/changeset/148792
http://trac.webkit.org/changeset/148797