Bug 228097 – Characters with no fonts that support them are drawn as two .notdef glyphs in the fast text codepath

• Component: New Bugs
• Status: RESOLVED FIXED

[Myles C. Maxfield]

Characters with no fonts that support them are drawn as two .notdef glyphs in the fast text codepath

[Myles C. Maxfield]

Created attachment 433844 [details]
Patch

[Myles C. Maxfield]

<rdar://problem/80798113>

[Myles C. Maxfield]

Comment on attachment 433844 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433844&action=review

> Source/WebCore/platform/graphics/WidthIterator.cpp:274
> +        if (!U_IS_BMP(character) && glyph)

I'm not sure this is correct. Shaping expects the number of glyphs to equal the length of the string.

[Myles C. Maxfield]

Created attachment 433847 [details]
Patch

[Myles C. Maxfield]

Created attachment 433848 [details]
Patch

[Myles C. Maxfield]

Comment on attachment 433848 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433848&action=review

> Source/WebCore/ChangeLog:3
> +        Characters with no fonts that support them are drawn as two .notdef glyphs in the fast text codepath

REGRESSION(r272379):

> LayoutTests/ChangeLog:3
> +        Characters with no fonts that support them are drawn as two .notdef glyphs in the fast text codepath

REGRESSION(r272379):

[Myles C. Maxfield]

Created attachment 433849 [details]
Patch

[alan]

Comment on attachment 433849 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433849&action=review

> Source/WebCore/platform/graphics/WidthIterator.cpp:115
> +            if (!glyphBuffer.glyphAt(i) && !glyphBuffer.glyphAt(i + 1)) {

I assume these surrogates come in pairs and the i + 1 derefs are safe.

[Simon Fraser (smfr)]

Comment on attachment 433849 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433849&action=review

> Source/WebCore/platform/graphics/WidthIterator.cpp:121
> +                            continue;

This continue does nothing.

[Myles C. Maxfield]

Committed r280103 (239820@main): <https://commits.webkit.org/239820@main>

[Myles C. Maxfield]

Comment on attachment 433849 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433849&action=review

>> Source/WebCore/platform/graphics/WidthIterator.cpp:115
>> +            if (!glyphBuffer.glyphAt(i) && !glyphBuffer.glyphAt(i + 1)) {
> 
> I assume these surrogates come in pairs and the i + 1 derefs are safe.

The for loop has "i < glyphBuffer.size() - 1"

They're not guaranteed to come in pairs, but when they do, that's the case where we get the double .notdef glyphs. If they don't come in pairs, we don't need to strip them out.

[Myles C. Maxfield]

*** Bug 223308 has been marked as a duplicate of this bug. ***