Remove SegmentedString::m_pushedChar{1,2} optimization

Source/platform/text/SegmentedString.h

 /*
     Copyright (C) 2004, 2005, 2006, 2007, 2008 Apple Inc. All rights reserved.
 
     This library is free software; you can redistribute it and/or
     modify it under the terms of the GNU Library General Public
     License as published by the Free Software Foundation; either
     version 2 of the License, or (at your option) any later version.
 
     This library is distributed in the hope that it will be useful,
     but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ skipping 33 matching lines @@
         : m_length(str.length())
         , m_doNotExcludeLineNumbers(true)
         , m_string(str)
     {
         if (m_length) {
             if (m_string.is8Bit()) {
                 m_is8Bit = true;
                 m_data.string8Ptr = m_string.characters8();
             } else {
                 m_is8Bit = false;
                 m_data.string16Ptr = m_string.characters16();

[Yoav Weiss, 2015/08/27 13:08:12]

Unrelated to current CL, but when str.length() is zero, m_data is not
initialized!

[kouhei (in TOK), 2015/08/28 05:38:47]

Done.

             }
         } else {
             m_is8Bit = false;
         }
     }
 
     void clear() { m_length = 0; m_data.string16Ptr = 0; m_is8Bit = false;}
 
     bool is8Bit() { return m_is8Bit; }
 
     bool excludeLineNumbers() const { return !m_doNotExcludeLineNumbers; }
     bool doNotExcludeLineNumbers() const { return m_doNotExcludeLineNumbers; }
 
     void setExcludeLineNumbers() { m_doNotExcludeLineNumbers = false; }
 
     int numberOfCharactersConsumed() const { return m_string.length() - m_length; }
 
     void appendTo(StringBuilder& builder) const
     {
         int offset = m_string.length() - m_length;
 
         if (!offset) {
             if (m_length)
                 builder.append(m_string);
         } else {
             builder.append(m_string.substring(offset, m_length));
         }
     }
 
+    void push(UChar c)
+    {
+        if (m_is8Bit) {
+            if (m_data.string8Ptr == m_string.characters8()) {
+                m_string.insert(&c, 1, 0);

[Yoav Weiss, 2015/08/27 13:19:03]

AFAICT this goes back to String::append which looks pretty awful in terms of
performance :/
Can we somehow use a StringBuilder for that?

[kouhei (in TOK), 2015/08/28 05:38:47]

I replaced this codepath with prepend() call in SegmentedString::push(). Not
optimal, but better than insert() call.

SegmentedString::push() is only used for undo-ing advance(), so chances are that
--m_data.string{8,16}Ptr path is taken almost all of the time. I have never seen
this codepath being taken except for some unittests/LayoutTests.

+                m_data.string8Ptr = m_string.characters8();
+                m_length = m_string.length();
+            } else {
+                --m_data.string8Ptr;
+                ASSERT(*m_data.string8Ptr == c);
+            }
+        } else {
+            if (m_data.string16Ptr == m_string.characters16()) {
+                m_string.insert(&c, 1, 0);
+                m_data.string16Ptr = m_string.characters16();
+                m_length = m_string.length();
+            } else {
+                --m_data.string16Ptr;
+                ASSERT(*m_data.string16Ptr == static_cast<LChar>(c));
+            }
+        }
+    }
+
     UChar getCurrentChar8()
     {
         return *m_data.string8Ptr;
     }
 
     UChar getCurrentChar16()
     {
         return m_data.string16Ptr ? *m_data.string16Ptr : 0;
     }
 
@@ skipping 47 matching lines @@
     } m_data;
     int m_length;
     bool m_doNotExcludeLineNumbers;
     bool m_is8Bit;
     String m_string;
 };
 
 class PLATFORM_EXPORT SegmentedString {
 public:
     SegmentedString()
-        : m_pushedChar1(0)
-        , m_pushedChar2(0)
-        , m_currentChar(0)
+        : m_currentChar(0)
         , m_numberOfCharactersConsumedPriorToCurrentString(0)
         , m_numberOfCharactersConsumedPriorToCurrentLine(0)
         , m_currentLine(0)
         , m_closed(false)
         , m_empty(true)
         , m_fastPathFlags(NoFastPath)
         , m_advanceFunc(&SegmentedString::advanceEmpty)
         , m_advanceAndUpdateLineNumberFunc(&SegmentedString::advanceEmpty)
     {
     }
 
     SegmentedString(const String& str)
-        : m_pushedChar1(0)
-        , m_pushedChar2(0)
-        , m_currentString(str)
+        : m_currentString(str)
         , m_currentChar(0)
         , m_numberOfCharactersConsumedPriorToCurrentString(0)
         , m_numberOfCharactersConsumedPriorToCurrentLine(0)
         , m_currentLine(0)
         , m_closed(false)
         , m_empty(!str.length())
         , m_fastPathFlags(NoFastPath)
     {
         if (m_currentString.length())
             m_currentChar = m_currentString.getCurrentChar();
         updateAdvanceFunctionPointers();
     }
 
     void clear();
     void close();
 
     void append(const SegmentedString&);
     void prepend(const SegmentedString&);
 
     bool excludeLineNumbers() const { return m_currentString.excludeLineNumbers(); }
     void setExcludeLineNumbers();
 
-    void push(UChar c)
-    {
-        ASSERT(c);
-
-        if (!m_pushedChar1) {
-            m_currentChar = m_pushedChar1 = c;
-            updateSlowCaseFunctionPointers();
-        } else {
-            ASSERT(!m_pushedChar2);
-            m_pushedChar2 = m_pushedChar1;
-            m_currentChar = m_pushedChar1 = c;
-        }
-    }
+    void push(UChar);
 
     bool isEmpty() const { return m_empty; }
     unsigned length() const;
 
     bool isClosed() const { return m_closed; }
 
     enum LookAheadResult {
         DidNotMatch,
         DidMatch,
         NotEnoughCharacters,
     };
 
     LookAheadResult lookAhead(const String& string) { return lookAheadInline(string, TextCaseSensitive); }
     LookAheadResult lookAheadIgnoringCase(const String& string) { return lookAheadInline(string, TextCaseInsensitive); }
 
     void advance()
     {
         if (m_fastPathFlags & Use8BitAdvance) {
-            ASSERT(!m_pushedChar1);
             m_currentChar = m_currentString.incrementAndGetCurrentChar8();
             decrementAndCheckLength();
             return;
         }
 
         (this->*m_advanceFunc)();
     }
 
     inline void advanceAndUpdateLineNumber()
     {
         if (m_fastPathFlags & Use8BitAdvance) {
-            ASSERT(!m_pushedChar1);
-
             bool haveNewLine = (m_currentChar == '\n') & !!(m_fastPathFlags & Use8BitAdvanceAndUpdateLineNumbers);
             m_currentChar = m_currentString.incrementAndGetCurrentChar8();
             decrementAndCheckLength();
 
             if (haveNewLine) {
                 ++m_currentLine;
                 m_numberOfCharactersConsumedPriorToCurrentLine =  m_numberOfCharactersConsumedPriorToCurrentString + m_currentString.numberOfCharactersConsumed();
             }
 
             return;
@@ skipping 16 matching lines @@
 
     void advancePastNonNewline()
     {
         ASSERT(currentChar() != '\n');
         advance();
     }
 
     void advancePastNewlineAndUpdateLineNumber()
     {
         ASSERT(currentChar() == '\n');
-        if (!m_pushedChar1 && m_currentString.length() > 1) {
+        if (m_currentString.length() > 1) {
             int newLineFlag = m_currentString.doNotExcludeLineNumbers();
             m_currentLine += newLineFlag;
             if (newLineFlag)
                 m_numberOfCharactersConsumedPriorToCurrentLine = numberOfCharactersConsumed() + 1;
             decrementAndCheckLength();
             m_currentChar = m_currentString.incrementAndGetCurrentChar();
             return;
         }
         advanceAndUpdateLineNumberSlowCase();
     }
 
     // Writes the consumed characters into consumedCharacters, which must
     // have space for at least |count| characters.
     void advance(unsigned count, UChar* consumedCharacters);
 
-    bool escaped() const { return m_pushedChar1; }
+    bool escaped() const { return false; }
 
     int numberOfCharactersConsumed() const
     {
         int numberOfPushedCharacters = 0;
-        if (m_pushedChar1) {
-            ++numberOfPushedCharacters;
-            if (m_pushedChar2)
-                ++numberOfPushedCharacters;
-        }
         return m_numberOfCharactersConsumedPriorToCurrentString + m_currentString.numberOfCharactersConsumed() - numberOfPushedCharacters;
     }
 
     String toString() const;
 
     UChar currentChar() const { return m_currentChar; }
 
     // The method is moderately slow, comparing to currentLine method.
     OrdinalNumber currentColumn() const;
     OrdinalNumber currentLine() const;
@@ skipping 25 matching lines @@
     void decrementAndCheckLength()
     {
         ASSERT(m_currentString.length() > 1);
         m_currentString.decrementLength();
         if (m_currentString.haveOneCharacterLeft())
             updateSlowCaseFunctionPointers();
     }
 
     void updateAdvanceFunctionPointers()
     {
-        if ((m_currentString.length() > 1) && !m_pushedChar1) {
+        if (m_currentString.length() > 1) {
             if (m_currentString.is8Bit()) {
                 m_advanceFunc = &SegmentedString::advance8;
                 m_fastPathFlags = Use8BitAdvance;
                 if (m_currentString.doNotExcludeLineNumbers()) {
                     m_advanceAndUpdateLineNumberFunc = &SegmentedString::advanceAndUpdateLineNumber8;
                     m_fastPathFlags |= Use8BitAdvanceAndUpdateLineNumbers;
                 } else {
                     m_advanceAndUpdateLineNumberFunc = &SegmentedString::advance8;
                 }
                 return;
@@ skipping 12 matching lines @@
             m_advanceFunc = &SegmentedString::advanceEmpty;
             m_fastPathFlags = NoFastPath;
             m_advanceAndUpdateLineNumberFunc = &SegmentedString::advanceEmpty;
         }
 
         updateSlowCaseFunctionPointers();
     }
 
     inline LookAheadResult lookAheadInline(const String& string, TextCaseSensitivity caseSensitivity)
     {
-        if (!m_pushedChar1 && string.length() <= static_cast<unsigned>(m_currentString.length())) {
+        if (string.length() <= static_cast<unsigned>(m_currentString.length())) {
             String currentSubstring = m_currentString.currentSubString(string.length());
             if (currentSubstring.startsWith(string, caseSensitivity))
                 return DidMatch;
             return DidNotMatch;
         }
         return lookAheadSlowCase(string, caseSensitivity);
     }
 
     LookAheadResult lookAheadSlowCase(const String& string, TextCaseSensitivity caseSensitivity)
     {
         unsigned count = string.length();
         if (count > length())
             return NotEnoughCharacters;
         UChar* consumedCharacters;
         String consumedString = String::createUninitialized(count, consumedCharacters);
         advance(count, consumedCharacters);
         LookAheadResult result = DidNotMatch;
         if (consumedString.startsWith(string, caseSensitivity))
             result = DidMatch;
         prepend(SegmentedString(consumedString));
         return result;
     }
 
     bool isComposite() const { return !m_substrings.isEmpty(); }
 
-    UChar m_pushedChar1;
-    UChar m_pushedChar2;
     SegmentedSubstring m_currentString;
     UChar m_currentChar;
     int m_numberOfCharactersConsumedPriorToCurrentString;
     int m_numberOfCharactersConsumedPriorToCurrentLine;
     int m_currentLine;
     Deque<SegmentedSubstring> m_substrings;
     bool m_closed;
     bool m_empty;
     unsigned char m_fastPathFlags;
     void (SegmentedString::*m_advanceFunc)();
     void (SegmentedString::*m_advanceAndUpdateLineNumberFunc)();
 };
 
 }
 
 #endif