Changeset 153007 in webkit

Timestamp:

Jul 22, 2013 3:29:03 PM (11 years ago)

Author:

benjamin@webkit.org

Message:

String::lower() - Skip to slow path on the first failure
​https://bugs.webkit.org/show_bug.cgi?id=118885

Reviewed by Andreas Kling.

In the 8 bits case, we don't need to know the state of the full string before changing characters
to their lowercase variant.
Just fail immediately and start transforming characters from the point of failure.

This avoid reading the string twice when the uppercase character is not at the end of the string.

• wtf/text/StringImpl.cpp:

(WTF::StringImpl::lower):

Location:

trunk/Source/WTF

Files:

• ChangeLog (modified) (1 diff)
• wtf/text/StringImpl.cpp (modified) (1 diff)

trunk/Source/WTF/ChangeLog

@@ +1 @@
+2013-07-22  Benjamin Poulain  <benjamin@webkit.org>
+
+        String::lower() - Skip to slow path on the first failure
+        https://bugs.webkit.org/show_bug.cgi?id=118885
+
+        Reviewed by Andreas Kling.
+
+        In the 8 bits case, we don't need to know the state of the full string before changing characters
+        to their lowercase variant.
+        Just fail immediately and start transforming characters from the point of failure.
+
+        This avoid reading the string twice when the uppercase character is not at the end of the string.
+
+        * wtf/text/StringImpl.cpp:
+        (WTF::StringImpl::lower):
+
 2013-07-22  Brent Fulgham  <bfulgham@apple.com>
 

trunk/Source/WTF/wtf/text/StringImpl.cpp

@@ -393 +393 @@
     // First scan the string for uppercase and non-ASCII characters:
     bool noUpper = true;
-    unsigned ored = 0;
     if (is8Bit()) {
+        unsigned failingIndex;
         for (unsigned i = 0; i < m_length; ++i) {
             LChar character = m_data8[i];
-            if (UNLIKELY(isASCIIUpper(character)))
-                noUpper = false;
-            ored |= character;
-        }
-        // Nothing to do if the string is all ASCII with no uppercase.
-        if (noUpper && !(ored & ~0x7F))
-            return this;
-
+            if (UNLIKELY((character & ~0x7F) || isASCIIUpper(character))) {
+                failingIndex = i;
+                goto SlowPath8bitLower;
+            }
+        }
+        return this;
+
+SlowPath8bitLower:
         LChar* data8;
         RefPtr<StringImpl> newImpl = createUninitialized(m_length, data8);
 
-        if (!(ored & ~0x7F)) {
-            for (unsigned i = 0; i < m_length; ++i)
-                data8[i] = toASCIILower(m_data8[i]);
-
-            return newImpl.release();
-        }
-
-        // Do a slower implementation for cases that include non-ASCII Latin-1 characters.
-        for (unsigned i = 0; i < m_length; ++i)
-            data8[i] = static_cast<LChar>(Unicode::toLower(m_data8[i]));
+        for (unsigned i = 0; i < failingIndex; ++i)
+            data8[i] = m_data8[i];
+
+        for (unsigned i = failingIndex; i < m_length; ++i) {
+            LChar character = m_data8[i];
+            if (!(character & ~0x7F))
+                data8[i] = toASCIILower(character);
+            else
+                data8[i] = static_cast<LChar>(Unicode::toLower(character));
+        }
 
         return newImpl.release();
     }
+    unsigned ored = 0;
 
     for (unsigned i = 0; i < m_length; ++i) {