Implement CSS4 pseudo-class :dir()

third_party/WebKit/Source/core/css/SelectorChecker.cpp

Index: third_party/WebKit/Source/core/css/SelectorChecker.cpp
diff --git a/third_party/WebKit/Source/core/css/SelectorChecker.cpp b/third_party/WebKit/Source/core/css/SelectorChecker.cpp
index 497e1722f3105f2352964ee1340c0149557e1bb2..f82eb6ce5792caf511e1b6360acf4b795687a839 100644
--- a/third_party/WebKit/Source/core/css/SelectorChecker.cpp
+++ b/third_party/WebKit/Source/core/css/SelectorChecker.cpp
@@ -949,6 +949,30 @@ bool SelectorChecker::checkPseudoClass(const SelectorCheckingContext& context, M
                 break;
             return true;
         }
+    case CSSSelector::PseudoDir:
+        {
+            // TODO(ramya.v): Need to check the exact direction element is rendered.
+            AtomicString value;
+            TextDirection dir = LTR;
+            if (m_mode == ResolvingStyle)
+                dir = context.elementStyle->direction();

[esprehn, 2016/02/12 05:13:41]

how does this work? The elementStyle is the one being created right now, we
haven't figured out what values apply yet.

+            else if (element.ensureComputedStyle())
+                dir = element.ensureComputedStyle()->direction();

[esprehn, 2016/02/12 05:13:41]

ensureComputedStyle() is non-trivial, always do 

if (ComputedStyle* style = element.ensureComputedStyle()) {
  // use style here
}

I don't think this is right though, you can't compute the style in a detached
subtree, and this would cause a style resolve from inside querySelector for
display none elements, or use stale style for displayed elements giving you back
the wrong value.

+            switch (dir) {
+            case LTR:
+                value = "ltr";

[esprehn, 2016/02/12 05:13:41]

we should just atomize these with a DEFINE_STATIC_LOCAL(AtomicString, ltrAtom,
"ltr") and then the check below will usually be very fast since authors will
write dir(ltr)

+                break;
+            case RTL:
+                value = "rtl";
+                break;
+            default:
+                break;
+            }
+            const AtomicString& argument = selector.argument();
+            if (value.isEmpty() || !equalIgnoringCase(value, argument))

[esprehn, 2016/02/12 05:13:41]

should this use ascii case? this file is inconsistent, I'm not sure when we're
supposed to be using ascii these days.

+                break;
+            return true;
+        }
     case CSSSelector::PseudoFullScreen:
         // While a Document is in the fullscreen state, and the document's current fullscreen
         // element is an element in the document, the 'full-screen' pseudoclass applies to