RESOLVED CONFIGURATION CHANGED 105171
Custom Arabic Fonts Rendered Incorrectly
https://bugs.webkit.org/show_bug.cgi?id=105171
Summary Custom Arabic Fonts Rendered Incorrectly
Wu Chiang
Reported 2012-12-17 06:20:52 PST
Created attachment 179731 [details] Arabic font rendering on chrome Some Arabic fonts render incorrectly, the letters are disconnected, higher than where they should be, or lower than where they should be. Tashkeel (symbols above and below letters) are also extremely out of place. I've tested this on the latest nightly build, chrome, and the latest Safari build on Mac OS X 10.8, the issue is present in all of them. On Firefox, all the text is rendered correctly.
Attachments
Arabic font rendering on chrome (447.46 KB, image/png)
2012-12-17 06:20 PST, Wu Chiang
no flags
Arabic font rendering on Safari (494.73 KB, image/png)
2012-12-17 06:21 PST, Wu Chiang
no flags
Arabic font rendering on Firefox (521.05 KB, image/png)
2012-12-17 06:22 PST, Wu Chiang
no flags
Correct rendering of compact exampe on Firefox (50.15 KB, image/png)
2012-12-19 00:39 PST, Wu Chiang
no flags
Incorrect rendering of compact example on latest WebKit nightly (51.29 KB, image/png)
2012-12-19 00:43 PST, Wu Chiang
no flags
Wu Chiang
Comment 1 2012-12-17 06:21:47 PST
Created attachment 179732 [details] Arabic font rendering on Safari
Wu Chiang
Comment 2 2012-12-17 06:22:35 PST
Created attachment 179733 [details] Arabic font rendering on Firefox
Wu Chiang
Comment 3 2012-12-17 06:28:38 PST
I've attached screenshots that show incorrect rendering in Safari and Chrome in Mac OS X 10.8 and one that shows correct rendering in Firefox. I've tested this on the latest version of Chrome on Windows 8, as well as the latest nightly version for Windows, and the text renders correctly as it does in Firefox. This is the URL I've been using for the rendering tests: http://www.quranexplorer.com/quran/
Wu Chiang
Comment 4 2012-12-17 23:06:00 PST
I've made a much more compact example at: https://dl.dropbox.com/u/9658048/bugs/webkit/index.html
Alexey Proskuryakov
Comment 5 2012-12-18 10:02:44 PST
Radar WebKit Bug Importer
Comment 6 2012-12-18 14:01:41 PST
Dean Jackson
Comment 7 2012-12-18 14:06:15 PST
Can you upload a screenshot of the correct rendering for your compact example? Like Alexey, quranexplorer is definitely reproducing. The compact example seems to have the Tashkeel in the correct location, but maybe some of the glyphs are disconnected.
Wu Chiang
Comment 8 2012-12-19 00:39:07 PST
Created attachment 180109 [details] Correct rendering of compact exampe on Firefox
Wu Chiang
Comment 9 2012-12-19 00:43:54 PST
Created attachment 180113 [details] Incorrect rendering of compact example on latest WebKit nightly
Wu Chiang
Comment 10 2012-12-19 00:49:57 PST
You are correct, the tashkeel IS in the correct place in the compact example, however some letters are shifted up as you can see in the images. It also seems to get worse if I add more text. The only problem is the vertical positioning of the glyphs in my compact example. On quranexplorer, it's the vertical positioning of parts of lines, the tashkeel, and the glyphs.
Dean Jackson
Comment 11 2012-12-19 10:28:43 PST
(In reply to comment #10) > You are correct, the tashkeel IS in the correct place in the compact example, however some letters are shifted up as you can see in the images. It also seems to get worse if I add more text. The only problem is the vertical positioning of the glyphs in my compact example. On quranexplorer, it's the vertical positioning of parts of lines, the tashkeel, and the glyphs. Thanks!
Wu Chiang
Comment 12 2012-12-28 22:35:28 PST
Further testing shows that the problem goes away if Tashkeel ( diacritics ) is removed from the text.
Wu Chiang
Comment 13 2012-12-30 04:11:39 PST
Is the text rendering in WebKit for Mac OS X based on the Quartz engine? I've looked through the wiki and documentation and couldn't find an answer for this. All I've managed to find is that RenderText is in charge of rendering raw text. Does WebKit use its own rendering logic or does it rely on Quartz? If it doesn't rely on Quartz, then where are the glyphs for complex shapes determined? and where are their positions calculated?
Glenn Adams
Comment 14 2012-12-30 05:06:28 PST
(In reply to comment #13) > Is the text rendering in WebKit for Mac OS X based on the Quartz engine? I've looked through the wiki and documentation and couldn't find an answer for this. All I've managed to find is that RenderText is in charge of rendering raw text. Does WebKit use its own rendering logic or does it rely on Quartz? If it doesn't rely on Quartz, then where are the glyphs for complex shapes determined? and where are their positions calculated? for apple-mac port, see WebCore/platform/graphics/mac/ComplexTextController*
Wu Chiang
Comment 15 2012-12-30 23:10:43 PST
Looks like WebKit is relying on CoreText for complex text, CoreText is what is at fault and not WebKit. I tried making a quick application using CoreText and I immediately saw the same issues. Chromium is using HarfBuzz and having the same issues as well, so I'm not entirely sure.
Alexey Proskuryakov
Comment 16 2012-12-31 17:38:00 PST
(In reply to comment #15) > I tried making a quick application using CoreText and I immediately saw the same issues. Can you please file a bug with this test application attached via <http://bugreport.apple.com>?
Dean Jackson
Comment 17 2013-01-15 13:50:59 PST
(In reply to comment #16) > (In reply to comment #15) > > I tried making a quick application using CoreText and I immediately saw the same issues. > > Can you please file a bug with this test application attached via <http://bugreport.apple.com>? I believe there already is a bug for this (other than ours above): <rdar://problem/13004368> But, yes, if you could file another with your test application, that would be great.
Cameron McCormack (:heycam)
Comment 18 2024-11-30 13:27:57 PST
The reported content is rendering correctly these days.
Note You need to log in before you can comment on or make changes to this bug.