Changeset 64827 in webkit for trunk/JavaScriptCore/parser/Lexer.cpp

Timestamp:

Aug 6, 2010, 3:33:10 AM (15 years ago)

Author:

[email protected]

Message:

Refactor identifier parsing in lexer
​https://bugs.webkit.org/show_bug.cgi?id=41845

Reviewed by Darin Adler.

The code is refactored to avoid gotos. The new code
has the same performance as the old one.

SunSpider --parse-only: no change (from 34.0ms to 33.6ms)
SunSpider: no change (from 523.2ms to 523.5ms)

• parser/Lexer.cpp:

(JSC::Lexer::parseIdent):
(JSC::Lexer::lex):

• parser/Lexer.h:

File:

• trunk/JavaScriptCore/parser/Lexer.cpp (modified) (6 diffs)

trunk/JavaScriptCore/parser/Lexer.cpp

@@ -400 +400 @@
 }
 
+ALWAYS_INLINE JSTokenType Lexer::parseIdentifier(JSTokenData* lvalp, LexType lexType)
+{
+    bool bufferRequired = false;
+    const UChar* identifierStart = currentCharacter();
+    int identifierLength;
+
+    while (true) {
+        if (LIKELY(isIdentPart(m_current))) {
+            shift();
+            continue;
+        }
+        if (LIKELY(m_current != '\\'))
+            break;
+
+        // \uXXXX unicode characters.
+        bufferRequired = true;
+        if (identifierStart != currentCharacter())
+            m_buffer16.append(identifierStart, currentCharacter() - identifierStart);
+        shift();
+        if (UNLIKELY(m_current != 'u'))
+            return ERRORTOK;
+        shift();
+        int character = getUnicodeCharacter();
+        if (UNLIKELY(character == -1))
+            return ERRORTOK;
+        if (UNLIKELY(m_buffer16.size() ? !isIdentPart(character) : !isIdentStart(character)))
+            return ERRORTOK;
+        record16(character);
+        identifierStart = currentCharacter();
+    }
+
+    if (!bufferRequired)
+        identifierLength = currentCharacter() - identifierStart;
+    else {
+        if (identifierStart != currentCharacter())
+            m_buffer16.append(identifierStart, currentCharacter() - identifierStart);
+        identifierStart = m_buffer16.data();
+        identifierLength = m_buffer16.size();
+    }
+
+    const Identifier* ident = makeIdentifier(identifierStart, identifierLength);
+    lvalp->ident = ident;
+    m_delimited = false;
+
+    if (LIKELY(!bufferRequired && lexType == IdentifyReservedWords)) {
+        // Keywords must not be recognized if there was an \uXXXX in the identifier.
+        const HashEntry* entry = m_keywordTable.entry(m_globalData, *ident);
+        return entry ? static_cast<JSTokenType>(entry->lexerValue()) : IDENT;
+    }
+
+    m_buffer16.resize(0);
+    return IDENT;
+}
+
 ALWAYS_INLINE bool Lexer::parseString(JSTokenData* lvalp)
 {
@@ -489 +543 @@
 
     JSTokenType token = ERRORTOK;
-    int identChar = 0;
     m_terminator = false;
 
@@ -754 +807 @@
         token = CLOSEBRACE;
         break;
-    case CharacterBackSlash:
-        goto startIdentifierWithBackslash;
     case CharacterZero:
         goto startNumberWithZeroDigit;
@@ -769 +820 @@
     case CharacterIdentifierStart:
         ASSERT(isIdentStart(m_current));
-        goto startIdentifierOrKeyword;
+        // Fall through into CharacterBackSlash.
+    case CharacterBackSlash:
+        token = parseIdentifier(lvalp, lexType);
+        break;
     case CharacterLineTerminator:
         ASSERT(isLineTerminator(m_current));
@@ -790 +844 @@
     goto returnToken;
 
-startIdentifierWithBackslash: {
-    shift();
-    if (UNLIKELY(m_current != 'u'))
-        goto returnError;
-    shift();
-
-    identChar = getUnicodeCharacter();
-    if (UNLIKELY(identChar == -1))
-        goto returnError;
-    if (UNLIKELY(!isIdentStart(identChar)))
-        goto returnError;
-    goto inIdentifierAfterCharacterCheck;
-}
-
-startIdentifierOrKeyword: {
-    const UChar* identifierStart = currentCharacter();
-    shift();
-    while (isIdentPart(m_current))
-        shift();
-    if (LIKELY(m_current != '\\')) {
-        // Fast case for idents which does not contain \uCCCC characters
-        lvalp->ident = makeIdentifier(identifierStart, currentCharacter() - identifierStart);
-        goto doneIdentifierOrKeyword;
-    }
-    m_buffer16.append(identifierStart, currentCharacter() - identifierStart);
-}
-
-    do {
-        shift();
-        if (UNLIKELY(m_current != 'u'))
-            goto returnError;
-        shift();
-        identChar = getUnicodeCharacter();
-        if (UNLIKELY(identChar == -1))
-            goto returnError;
-        if (UNLIKELY(!isIdentPart(identChar)))
-            goto returnError;
-inIdentifierAfterCharacterCheck:
-        record16(identChar);
-
-        while (isIdentPart(m_current)) {
-            record16(m_current);
-            shift();
-        }
-    } while (UNLIKELY(m_current == '\\'));
-    goto doneIdentifier;
-
 inSingleLineComment:
     while (!isLineTerminator(m_current)) {
@@ -1009 +1016 @@
     token = SEMICOLON;
     m_delimited = true;
-    goto returnToken;
-
-doneIdentifier:
-    m_atLineStart = false;
-    m_delimited = false;
-    lvalp->ident = makeIdentifier(m_buffer16.data(), m_buffer16.size());
-    m_buffer16.resize(0);
-    token = IDENT;
-    goto returnToken;
-
-doneIdentifierOrKeyword: {
-    m_atLineStart = false;
-    m_delimited = false;
-    m_buffer16.resize(0);
-    if (lexType == IdentifyReservedWords) {
-        const HashEntry* entry = m_keywordTable.entry(m_globalData, *lvalp->ident);
-        token = entry ? static_cast<JSTokenType>(entry->lexerValue()) : IDENT;
-    } else
-        token = IDENT;
     // Fall through into returnToken.
-}
 
 returnToken: {