Avoiding multiple text/string parsing while creating TextRun

Source/core/rendering/InlineTextBox.cpp

Index: Source/core/rendering/InlineTextBox.cpp
diff --git a/Source/core/rendering/InlineTextBox.cpp b/Source/core/rendering/InlineTextBox.cpp
index 60a35deccf1253ac5df7bbcae293e5ccc1ae2019..d63e63e8a77577094a9b4a275f5af06a41ec2b06 100644
--- a/Source/core/rendering/InlineTextBox.cpp
+++ b/Source/core/rendering/InlineTextBox.cpp
@@ -1523,6 +1523,7 @@ TextRun InlineTextBox::constructTextRun(RenderStyle* style, const Font& font, St
 
     TextRun run(string, textPos(), expansion(), expansionBehavior(), direction(), dirOverride() || style->rtlOrdering() == VisualOrder, !textRenderer->canUseSimpleFontCodePath());
     run.setTabSize(!style->collapseWhiteSpace(), style->tabSize());
+    run.setCharacterScanForCodePath(!textRenderer->canUseSimpleFontCodePath());

[eae, 2013/12/30 19:29:14]

Why do we only need to do this for the complex code path?

[h.joshi, 2013/12/31 06:07:43]

We can have three cases for text, One; String with only Simple Code point,
Second String with Only Complex Code Point and third String with Mix of Simple
and Complex Code points. While creating Text Object we parse the string and
store if its Simple of not. As for other two cases Blink share "to" and "from"
location and we need to judge string Code Points based on "to" and "from". For
case 1 it will always be Simple but for other two cases it can land in Simple or
Complex.

     if (textRunNeedsRenderingContext(font))
         run.setRenderingContext(SVGTextRunRenderingContext::create(textRenderer));