X-Git-Url: https://git.saurik.com/apple/javascriptcore.git/blobdiff_plain/2d39b0e377c0896910ee49ae70082ba665faf986..refs/heads/master:/parser/Lexer.cpp diff --git a/parser/Lexer.cpp b/parser/Lexer.cpp index 9ca7610..53aa9b0 100644 --- a/parser/Lexer.cpp +++ b/parser/Lexer.cpp @@ -30,7 +30,6 @@ #include "BuiltinNames.h" #include "JSGlobalObjectFunctions.h" #include "Identifier.h" -#include "NodeInfo.h" #include "Nodes.h" #include "JSCInlines.h" #include @@ -72,6 +71,7 @@ enum CharacterType { CharacterQuestion, CharacterTilde, CharacterQuote, + CharacterBackQuote, CharacterDot, CharacterSlash, CharacterBackSlash, @@ -193,7 +193,11 @@ static const unsigned short typesOfLatin1Characters[256] = { /* 93 - ] */ CharacterCloseBracket, /* 94 - ^ */ CharacterXor, /* 95 - _ */ CharacterIdentifierStart, +#if ENABLE(ES6_TEMPLATE_LITERAL_SYNTAX) +/* 96 - ` */ CharacterBackQuote, +#else /* 96 - ` */ CharacterInvalid, +#endif /* 97 - a */ CharacterIdentifierStart, /* 98 - b */ CharacterIdentifierStart, /* 99 - c */ CharacterIdentifierStart, @@ -489,13 +493,20 @@ static const LChar singleCharacterEscapeValuesForASCII[128] = { }; template -Lexer::Lexer(VM* vm, JSParserStrictness strictness) +Lexer::Lexer(VM* vm, JSParserBuiltinMode builtinMode) : m_isReparsing(false) , m_vm(vm) - , m_parsingBuiltinFunction(strictness == JSParseBuiltin) + , m_parsingBuiltinFunction(builtinMode == JSParserBuiltinMode::Builtin) { } +static inline JSTokenType tokenTypeForIntegerLikeToken(double doubleValue) +{ + if ((doubleValue || !std::signbit(doubleValue)) && static_cast(doubleValue) == doubleValue) + return INTEGER; + return DOUBLE; +} + template Lexer::~Lexer() { @@ -506,21 +517,21 @@ String Lexer::invalidCharacterMessage() const { switch (m_current) { case 0: - return "Invalid character: '\\0'"; + return ASCIILiteral("Invalid character: '\\0'"); case 10: - return "Invalid character: '\\n'"; + return ASCIILiteral("Invalid character: '\\n'"); case 11: - return "Invalid character: '\\v'"; + return ASCIILiteral("Invalid character: '\\v'"); case 13: - return "Invalid character: '\\r'"; + return ASCIILiteral("Invalid character: '\\r'"); case 35: - return "Invalid character: '#'"; + return ASCIILiteral("Invalid character: '#'"); case 64: - return "Invalid character: '@'"; + return ASCIILiteral("Invalid character: '@'"); case 96: - return "Invalid character: '`'"; + return ASCIILiteral("Invalid character: '`'"); default: - return String::format("Invalid character '\\u%04u'", static_cast(m_current)).impl(); + return String::format("Invalid character '\\u%04u'", static_cast(m_current)); } } @@ -558,6 +569,7 @@ void Lexer::setCode(const SourceCode& source, ParserArena* arena) m_buffer8.reserveInitialCapacity(initialReadBufferCapacity); m_buffer16.reserveInitialCapacity((m_codeEnd - m_code) / 2); + m_bufferForRawTemplateString16.reserveInitialCapacity(initialReadBufferCapacity); if (LIKELY(m_code < m_codeEnd)) m_current = *m_code; @@ -599,22 +611,60 @@ ALWAYS_INLINE T Lexer::peek(int offset) const return (code < m_codeEnd) ? *code : 0; } -template -typename Lexer::UnicodeHexValue Lexer::parseFourDigitUnicodeHex() -{ - T char1 = peek(1); - T char2 = peek(2); - T char3 = peek(3); +struct ParsedUnicodeEscapeValue { + ParsedUnicodeEscapeValue(UChar32 value) + : m_value(value) + { + ASSERT(isValid()); + } + + enum SpecialValueType { Incomplete = -2, Invalid = -1 }; + ParsedUnicodeEscapeValue(SpecialValueType type) + : m_value(type) + { + } + + bool isValid() const { return m_value >= 0; } + bool isIncomplete() const { return m_value == Incomplete; } - if (UNLIKELY(!isASCIIHexDigit(m_current) || !isASCIIHexDigit(char1) || !isASCIIHexDigit(char2) || !isASCIIHexDigit(char3))) - return UnicodeHexValue((m_code + 4) >= m_codeEnd ? UnicodeHexValue::IncompleteHex : UnicodeHexValue::InvalidHex); + UChar32 value() const + { + ASSERT(isValid()); + return m_value; + } + +private: + UChar32 m_value; +}; + +template ParsedUnicodeEscapeValue Lexer::parseUnicodeEscape() +{ + if (m_current == '{') { + shift(); + UChar32 codePoint = 0; + do { + if (!isASCIIHexDigit(m_current)) + return m_current ? ParsedUnicodeEscapeValue::Invalid : ParsedUnicodeEscapeValue::Incomplete; + codePoint = (codePoint << 4) | toASCIIHexValue(m_current); + if (codePoint > UCHAR_MAX_VALUE) + return ParsedUnicodeEscapeValue::Invalid; + shift(); + } while (m_current != '}'); + shift(); + return codePoint; + } - int result = convertUnicode(m_current, char1, char2, char3); + auto character2 = peek(1); + auto character3 = peek(2); + auto character4 = peek(3); + if (UNLIKELY(!isASCIIHexDigit(m_current) || !isASCIIHexDigit(character2) || !isASCIIHexDigit(character3) || !isASCIIHexDigit(character4))) + return (m_code + 4) >= m_codeEnd ? ParsedUnicodeEscapeValue::Incomplete : ParsedUnicodeEscapeValue::Invalid; + auto result = convertUnicode(m_current, character2, character3, character4); shift(); shift(); shift(); shift(); - return UnicodeHexValue(result); + return result; } template @@ -654,18 +704,24 @@ static ALWAYS_INLINE bool isLatin1(UChar c) return c < 256; } +static ALWAYS_INLINE bool isLatin1(UChar32 c) +{ + return !(c & ~0xFF); +} + static inline bool isIdentStart(LChar c) { return typesOfLatin1Characters[c] == CharacterIdentifierStart; } -static inline bool isIdentStart(UChar c) +static inline bool isIdentStart(UChar32 c) { return isLatin1(c) ? isIdentStart(static_cast(c)) : isNonLatin1IdentStart(c); } -static NEVER_INLINE bool isNonLatin1IdentPart(int c) +static NEVER_INLINE bool isNonLatin1IdentPart(UChar32 c) { + // FIXME: ES6 says this should be based on the Unicode property ID_Continue now instead. return (U_GET_GC_MASK(c) & (U_GC_L_MASK | U_GC_MN_MASK | U_GC_MC_MASK | U_GC_ND_MASK | U_GC_PC_MASK)) || c == 0x200C || c == 0x200D; } @@ -677,11 +733,61 @@ static ALWAYS_INLINE bool isIdentPart(LChar c) return typesOfLatin1Characters[c] <= CharacterNumber; } -static ALWAYS_INLINE bool isIdentPart(UChar c) +static ALWAYS_INLINE bool isIdentPart(UChar32 c) { return isLatin1(c) ? isIdentPart(static_cast(c)) : isNonLatin1IdentPart(c); } +static ALWAYS_INLINE bool isIdentPart(UChar c) +{ + return isIdentPart(static_cast(c)); +} + +template ALWAYS_INLINE bool isIdentPartIncludingEscapeTemplate(const CharacterType* code, const CharacterType* codeEnd) +{ + if (isIdentPart(code[0])) + return true; + + // Shortest sequence handled below is \u{0}, which is 5 characters. + if (!(code[0] == '\\' && codeEnd - code >= 5 && code[1] == 'u')) + return false; + + if (code[2] == '{') { + UChar32 codePoint = 0; + const CharacterType* pointer; + for (pointer = &code[3]; pointer < codeEnd; ++pointer) { + auto digit = *pointer; + if (!isASCIIHexDigit(digit)) + break; + codePoint = (codePoint << 4) | toASCIIHexValue(digit); + if (codePoint > UCHAR_MAX_VALUE) + return false; + } + return isIdentPart(codePoint) && pointer < codeEnd && *pointer == '}'; + } + + // Shortest sequence handled below is \uXXXX, which is 6 characters. + if (codeEnd - code < 6) + return false; + + auto character1 = code[2]; + auto character2 = code[3]; + auto character3 = code[4]; + auto character4 = code[5]; + return isASCIIHexDigit(character1) && isASCIIHexDigit(character2) && isASCIIHexDigit(character3) && isASCIIHexDigit(character4) + && isIdentPart(Lexer::convertUnicode(character1, character2, character3, character4)); +} + +static ALWAYS_INLINE bool isIdentPartIncludingEscape(const LChar* code, const LChar* codeEnd) +{ + return isIdentPartIncludingEscapeTemplate(code, codeEnd); +} + +static ALWAYS_INLINE bool isIdentPartIncludingEscape(const UChar* code, const UChar* codeEnd) +{ + return isIdentPartIncludingEscapeTemplate(code, codeEnd); +} + static inline LChar singleEscape(int c) { if (c < 128) { @@ -758,6 +864,18 @@ inline void Lexer::record16(int c) m_buffer16.append(static_cast(c)); } +template inline void Lexer::recordUnicodeCodePoint(UChar32 codePoint) +{ + ASSERT(codePoint >= 0); + ASSERT(codePoint <= UCHAR_MAX_VALUE); + if (U_IS_BMP(codePoint)) + record16(codePoint); + else { + UChar codeUnits[2] = { U16_LEAD(codePoint), U16_TRAIL(codePoint) }; + append16(codeUnits, 2); + } +} + #if !ASSERT_DISABLED bool isSafeBuiltinIdentifier(VM& vm, const Identifier* ident) { @@ -766,6 +884,7 @@ bool isSafeBuiltinIdentifier(VM& vm, const Identifier* ident) /* Just block any use of suspicious identifiers. This is intended to * be used as a safety net while implementing builtins. */ + // FIXME: How can a debug-only assertion be a safety net? if (*ident == vm.propertyNames->builtinNames().callPublicName()) return false; if (*ident == vm.propertyNames->builtinNames().applyPublicName()) @@ -919,11 +1038,9 @@ template ALWAYS_INLINE JSTokenType Lexer::p return IDENT; } -template -template JSTokenType Lexer::parseIdentifierSlowCase(JSTokenData* tokenData, unsigned lexerFlags, bool strictMode) +template template JSTokenType Lexer::parseIdentifierSlowCase(JSTokenData* tokenData, unsigned lexerFlags, bool strictMode) { - const ptrdiff_t remaining = m_codeEnd - m_code; - const T* identifierStart = currentSourcePtr(); + auto identifierStart = currentSourcePtr(); bool bufferRequired = false; while (true) { @@ -942,19 +1059,18 @@ template JSTokenType Lexer::parseIdentifierSlow if (UNLIKELY(m_current != 'u')) return atEnd() ? UNTERMINATED_IDENTIFIER_ESCAPE_ERRORTOK : INVALID_IDENTIFIER_ESCAPE_ERRORTOK; shift(); - UnicodeHexValue character = parseFourDigitUnicodeHex(); + auto character = parseUnicodeEscape(); if (UNLIKELY(!character.isValid())) - return character.valueType() == UnicodeHexValue::IncompleteHex ? UNTERMINATED_IDENTIFIER_UNICODE_ESCAPE_ERRORTOK : INVALID_IDENTIFIER_UNICODE_ESCAPE_ERRORTOK; - UChar ucharacter = static_cast(character.value()); - if (UNLIKELY(m_buffer16.size() ? !isIdentPart(ucharacter) : !isIdentStart(ucharacter))) + return character.isIncomplete() ? UNTERMINATED_IDENTIFIER_UNICODE_ESCAPE_ERRORTOK : INVALID_IDENTIFIER_UNICODE_ESCAPE_ERRORTOK; + if (UNLIKELY(m_buffer16.size() ? !isIdentPart(character.value()) : !isIdentStart(character.value()))) return INVALID_IDENTIFIER_UNICODE_ESCAPE_ERRORTOK; if (shouldCreateIdentifier) - record16(ucharacter); + recordUnicodeCodePoint(character.value()); identifierStart = currentSourcePtr(); } int identifierLength; - const Identifier* ident = 0; + const Identifier* ident = nullptr; if (shouldCreateIdentifier) { if (!bufferRequired) { identifierLength = currentSourcePtr() - identifierStart; @@ -967,23 +1083,19 @@ template JSTokenType Lexer::parseIdentifierSlow tokenData->ident = ident; } else - tokenData->ident = 0; + tokenData->ident = nullptr; - if (LIKELY(!bufferRequired && !(lexerFlags & LexerFlagsIgnoreReservedWords))) { + m_buffer16.shrink(0); + + if (LIKELY(!(lexerFlags & LexerFlagsIgnoreReservedWords))) { ASSERT(shouldCreateIdentifier); - // Keywords must not be recognized if there was an \uXXXX in the identifier. - if (remaining < maxTokenLength) { - const HashTableValue* entry = m_vm->keywords->getKeyword(*ident); - ASSERT((remaining < maxTokenLength) || !entry); - if (!entry) - return IDENT; - JSTokenType token = static_cast(entry->lexerValue()); - return (token != RESERVED_IF_STRICT) || strictMode ? token : IDENT; - } - return IDENT; + const HashTableValue* entry = m_vm->keywords->getKeyword(*ident); + if (!entry) + return IDENT; + JSTokenType token = static_cast(entry->lexerValue()); + return (token != RESERVED_IF_STRICT) || strictMode ? token : IDENT; } - m_buffer16.resize(0); return IDENT; } @@ -1016,7 +1128,7 @@ template ALWAYS_INLINE typename Lexer::StringParseR LChar escape = singleEscape(m_current); - // Most common escape sequences first + // Most common escape sequences first. if (escape) { if (shouldBuildStrings) record8(escape); @@ -1026,7 +1138,7 @@ template ALWAYS_INLINE typename Lexer::StringParseR else if (m_current == 'x') { shift(); if (!isASCIIHexDigit(m_current) || !isASCIIHexDigit(peek(1))) { - m_lexErrorMessage = "\\x can only be followed by a hex character sequence"; + m_lexErrorMessage = ASCIILiteral("\\x can only be followed by a hex character sequence"); return (atEnd() || (isASCIIHexDigit(m_current) && (m_code + 1 == m_codeEnd))) ? StringUnterminated : StringCannotBeParsed; } T prev = m_current; @@ -1037,7 +1149,7 @@ template ALWAYS_INLINE typename Lexer::StringParseR } else { setOffset(startingOffset, startingLineStartOffset); setLineNumber(startingLineNumber); - m_buffer8.resize(0); + m_buffer8.shrink(0); return parseStringSlowCase(tokenData, strictMode); } stringStart = currentSourcePtr(); @@ -1047,7 +1159,7 @@ template ALWAYS_INLINE typename Lexer::StringParseR if (UNLIKELY(characterRequiresParseStringSlowCase(m_current))) { setOffset(startingOffset, startingLineStartOffset); setLineNumber(startingLineNumber); - m_buffer8.resize(0); + m_buffer8.shrink(0); return parseStringSlowCase(tokenData, strictMode); } @@ -1058,7 +1170,7 @@ template ALWAYS_INLINE typename Lexer::StringParseR append8(stringStart, currentSourcePtr() - stringStart); if (shouldBuildStrings) { tokenData->ident = makeIdentifier(m_buffer8.data(), m_buffer8.size()); - m_buffer8.resize(0); + m_buffer8.shrink(0); } else tokenData->ident = 0; @@ -1066,7 +1178,93 @@ template ALWAYS_INLINE typename Lexer::StringParseR } template -template typename Lexer::StringParseResult Lexer::parseStringSlowCase(JSTokenData* tokenData, bool strictMode) +template ALWAYS_INLINE auto Lexer::parseComplexEscape(EscapeParseMode escapeParseMode, bool strictMode, T stringQuoteCharacter) -> StringParseResult +{ + if (m_current == 'x') { + shift(); + if (!isASCIIHexDigit(m_current) || !isASCIIHexDigit(peek(1))) { + m_lexErrorMessage = ASCIILiteral("\\x can only be followed by a hex character sequence"); + return StringCannotBeParsed; + } + T prev = m_current; + shift(); + if (shouldBuildStrings) + record16(convertHex(prev, m_current)); + shift(); + return StringParsedSuccessfully; + } + + if (m_current == 'u') { + shift(); + + if (escapeParseMode == EscapeParseMode::String && m_current == stringQuoteCharacter) { + if (shouldBuildStrings) + record16('u'); + return StringParsedSuccessfully; + } + + auto character = parseUnicodeEscape(); + if (character.isValid()) { + if (shouldBuildStrings) + recordUnicodeCodePoint(character.value()); + return StringParsedSuccessfully; + } + + m_lexErrorMessage = ASCIILiteral("\\u can only be followed by a Unicode character sequence"); + return character.isIncomplete() ? StringUnterminated : StringCannotBeParsed; + } + + if (strictMode) { + if (isASCIIDigit(m_current)) { + // The only valid numeric escape in strict mode is '\0', and this must not be followed by a decimal digit. + int character1 = m_current; + shift(); + if (character1 != '0' || isASCIIDigit(m_current)) { + m_lexErrorMessage = ASCIILiteral("The only valid numeric escape in strict mode is '\\0'"); + return StringCannotBeParsed; + } + if (shouldBuildStrings) + record16(0); + return StringParsedSuccessfully; + } + } else { + if (isASCIIOctalDigit(m_current)) { + // Octal character sequences + T character1 = m_current; + shift(); + if (isASCIIOctalDigit(m_current)) { + // Two octal characters + T character2 = m_current; + shift(); + if (character1 >= '0' && character1 <= '3' && isASCIIOctalDigit(m_current)) { + if (shouldBuildStrings) + record16((character1 - '0') * 64 + (character2 - '0') * 8 + m_current - '0'); + shift(); + } else { + if (shouldBuildStrings) + record16((character1 - '0') * 8 + character2 - '0'); + } + } else { + if (shouldBuildStrings) + record16(character1 - '0'); + } + return StringParsedSuccessfully; + } + } + + if (!atEnd()) { + if (shouldBuildStrings) + record16(m_current); + shift(); + return StringParsedSuccessfully; + } + + m_lexErrorMessage = ASCIILiteral("Unterminated string constant"); + return StringUnterminated; +} + +template +template auto Lexer::parseStringSlowCase(JSTokenData* tokenData, bool strictMode) -> StringParseResult { T stringQuoteCharacter = m_current; shift(); @@ -1088,67 +1286,10 @@ template typename Lexer::StringParseResult Lexer shift(); } else if (UNLIKELY(isLineTerminator(m_current))) shiftLineTerminator(); - else if (m_current == 'x') { - shift(); - if (!isASCIIHexDigit(m_current) || !isASCIIHexDigit(peek(1))) { - m_lexErrorMessage = "\\x can only be followed by a hex character sequence"; - return StringCannotBeParsed; - } - T prev = m_current; - shift(); - if (shouldBuildStrings) - record16(convertHex(prev, m_current)); - shift(); - } else if (m_current == 'u') { - shift(); - UnicodeHexValue character = parseFourDigitUnicodeHex(); - if (character.isValid()) { - if (shouldBuildStrings) - record16(character.value()); - } else if (m_current == stringQuoteCharacter) { - if (shouldBuildStrings) - record16('u'); - } else { - m_lexErrorMessage = "\\u can only be followed by a Unicode character sequence"; - return character.valueType() == UnicodeHexValue::IncompleteHex ? StringUnterminated : StringCannotBeParsed; - } - } else if (strictMode && isASCIIDigit(m_current)) { - // The only valid numeric escape in strict mode is '\0', and this must not be followed by a decimal digit. - int character1 = m_current; - shift(); - if (character1 != '0' || isASCIIDigit(m_current)) { - m_lexErrorMessage = "The only valid numeric escape in strict mode is '\\0'"; - return StringCannotBeParsed; - } - if (shouldBuildStrings) - record16(0); - } else if (!strictMode && isASCIIOctalDigit(m_current)) { - // Octal character sequences - T character1 = m_current; - shift(); - if (isASCIIOctalDigit(m_current)) { - // Two octal characters - T character2 = m_current; - shift(); - if (character1 >= '0' && character1 <= '3' && isASCIIOctalDigit(m_current)) { - if (shouldBuildStrings) - record16((character1 - '0') * 64 + (character2 - '0') * 8 + m_current - '0'); - shift(); - } else { - if (shouldBuildStrings) - record16((character1 - '0') * 8 + character2 - '0'); - } - } else { - if (shouldBuildStrings) - record16(character1 - '0'); - } - } else if (!atEnd()) { - if (shouldBuildStrings) - record16(m_current); - shift(); - } else { - m_lexErrorMessage = "Unterminated string constant"; - return StringUnterminated; + else { + StringParseResult result = parseComplexEscape(EscapeParseMode::String, strictMode, stringQuoteCharacter); + if (result != StringParsedSuccessfully) + return result; } stringStart = currentSourcePtr(); @@ -1160,7 +1301,7 @@ template typename Lexer::StringParseResult Lexer if (UNLIKELY(((static_cast(m_current) - 0xE) & 0x2000))) { // New-line or end of input is not allowed if (atEnd() || isLineTerminator(m_current)) { - m_lexErrorMessage = "Unexpected EOF"; + m_lexErrorMessage = ASCIILiteral("Unexpected EOF"); return atEnd() ? StringUnterminated : StringCannotBeParsed; } // Anything else is just a normal character @@ -1175,10 +1316,190 @@ template typename Lexer::StringParseResult Lexer else tokenData->ident = 0; - m_buffer16.resize(0); + m_buffer16.shrink(0); return StringParsedSuccessfully; } +#if ENABLE(ES6_TEMPLATE_LITERAL_SYNTAX) +// While the lexer accepts (not ) sequence +// as one line terminator and increments one line number, +// TemplateLiteral considers it as two line terminators and . +// +// TemplateLiteral normalizes line terminators as follows. +// +// => +// => +// => +// <\u2028> => <\u2028> +// <\u2029> => <\u2029> +// +// So, should be normalized to . +// However, the lexer should increment the line number only once for . +// +// To achieve this, LineNumberAdder holds the current status of line terminator sequence. +// When TemplateLiteral lexer encounters a line terminator, it notifies to LineNumberAdder. +// LineNumberAdder maintains the status and increments the line number when it's necessary. +// For example, LineNumberAdder increments the line number only once for and . +template +class LineNumberAdder { +public: + LineNumberAdder(int& lineNumber) + : m_lineNumber(lineNumber) + { + } + + void clear() + { + m_previous = 0; + } + + void add(CharacterType character) + { + ASSERT(Lexer::isLineTerminator(character)); + if ((character + m_previous) == ('\n' + '\r')) + m_previous = 0; + else { + ++m_lineNumber; + m_previous = character; + } + } + +private: + int& m_lineNumber; + CharacterType m_previous { 0 }; +}; + +template +template typename Lexer::StringParseResult Lexer::parseTemplateLiteral(JSTokenData* tokenData, RawStringsBuildMode rawStringsBuildMode) +{ + const T* stringStart = currentSourcePtr(); + const T* rawStringStart = currentSourcePtr(); + + LineNumberAdder lineNumberAdder(m_lineNumber); + + while (m_current != '`') { + if (UNLIKELY(m_current == '\\')) { + lineNumberAdder.clear(); + if (stringStart != currentSourcePtr() && shouldBuildStrings) + append16(stringStart, currentSourcePtr() - stringStart); + shift(); + + LChar escape = singleEscape(m_current); + + // Most common escape sequences first. + if (escape) { + if (shouldBuildStrings) + record16(escape); + shift(); + } else if (UNLIKELY(isLineTerminator(m_current))) { + if (m_current == '\r') { + lineNumberAdder.add(m_current); + shift(); + if (m_current == '\n') { + lineNumberAdder.add(m_current); + shift(); + } + } else { + lineNumberAdder.add(m_current); + shift(); + } + } else { + bool strictMode = true; + StringParseResult result = parseComplexEscape(EscapeParseMode::Template, strictMode, '`'); + if (result != StringParsedSuccessfully) + return result; + } + + stringStart = currentSourcePtr(); + continue; + } + + if (m_current == '$' && peek(1) == '{') + break; + + // Fast check for characters that require special handling. + // Catches 0, \n, \r, 0x2028, and 0x2029 as efficiently + // as possible, and lets through all common ASCII characters. + if (UNLIKELY(((static_cast(m_current) - 0xE) & 0x2000))) { + // End of input is not allowed. + // Unlike String, line terminator is allowed. + if (atEnd()) { + m_lexErrorMessage = ASCIILiteral("Unexpected EOF"); + return atEnd() ? StringUnterminated : StringCannotBeParsed; + } + + if (isLineTerminator(m_current)) { + if (m_current == '\r') { + // Normalize , to . + if (shouldBuildStrings) { + if (stringStart != currentSourcePtr()) + append16(stringStart, currentSourcePtr() - stringStart); + if (rawStringStart != currentSourcePtr() && rawStringsBuildMode == RawStringsBuildMode::BuildRawStrings) + m_bufferForRawTemplateString16.append(rawStringStart, currentSourcePtr() - rawStringStart); + + record16('\n'); + if (rawStringsBuildMode == RawStringsBuildMode::BuildRawStrings) + m_bufferForRawTemplateString16.append('\n'); + } + lineNumberAdder.add(m_current); + shift(); + if (m_current == '\n') { + lineNumberAdder.add(m_current); + shift(); + } + stringStart = currentSourcePtr(); + rawStringStart = currentSourcePtr(); + } else { + lineNumberAdder.add(m_current); + shift(); + } + continue; + } + // Anything else is just a normal character + } + + lineNumberAdder.clear(); + shift(); + } + + bool isTail = m_current == '`'; + + if (shouldBuildStrings) { + if (currentSourcePtr() != stringStart) + append16(stringStart, currentSourcePtr() - stringStart); + if (rawStringStart != currentSourcePtr() && rawStringsBuildMode == RawStringsBuildMode::BuildRawStrings) + m_bufferForRawTemplateString16.append(rawStringStart, currentSourcePtr() - rawStringStart); + } + + if (shouldBuildStrings) { + tokenData->cooked = makeIdentifier(m_buffer16.data(), m_buffer16.size()); + // Line terminator normalization (e.g. => ) should be applied to both the raw and cooked representations. + if (rawStringsBuildMode == RawStringsBuildMode::BuildRawStrings) + tokenData->raw = makeIdentifier(m_bufferForRawTemplateString16.data(), m_bufferForRawTemplateString16.size()); + else + tokenData->raw = makeEmptyIdentifier(); + } else { + tokenData->cooked = makeEmptyIdentifier(); + tokenData->raw = makeEmptyIdentifier(); + } + tokenData->isTail = isTail; + + m_buffer16.shrink(0); + m_bufferForRawTemplateString16.shrink(0); + + if (isTail) { + // Skip ` + shift(); + } else { + // Skip $ and { + shift(); + shift(); + } + + return StringParsedSuccessfully; +} +#endif + template ALWAYS_INLINE void Lexer::parseHex(double& returnValue) { @@ -1186,9 +1507,6 @@ ALWAYS_INLINE void Lexer::parseHex(double& returnValue) uint32_t hexValue = 0; int maximumDigits = 7; - // Shift out the 'x' prefix. - shift(); - do { hexValue = (hexValue << 4) + toASCIIHexValue(m_current); shift(); @@ -1219,29 +1537,68 @@ ALWAYS_INLINE void Lexer::parseHex(double& returnValue) returnValue = parseIntOverflow(m_buffer8.data(), m_buffer8.size(), 16); } +template +ALWAYS_INLINE bool Lexer::parseBinary(double& returnValue) +{ + // Optimization: most binary values fit into 4 bytes. + uint32_t binaryValue = 0; + const unsigned maximumDigits = 32; + int digit = maximumDigits - 1; + // Temporary buffer for the digits. Makes easier + // to reconstruct the input characters when needed. + LChar digits[maximumDigits]; + + do { + binaryValue = (binaryValue << 1) + (m_current - '0'); + digits[digit] = m_current; + shift(); + --digit; + } while (isASCIIBinaryDigit(m_current) && digit >= 0); + + if (!isASCIIDigit(m_current) && digit >= 0) { + returnValue = binaryValue; + return true; + } + + for (int i = maximumDigits - 1; i > digit; --i) + record8(digits[i]); + + while (isASCIIBinaryDigit(m_current)) { + record8(m_current); + shift(); + } + + if (isASCIIDigit(m_current)) + return false; + + returnValue = parseIntOverflow(m_buffer8.data(), m_buffer8.size(), 2); + return true; +} + template ALWAYS_INLINE bool Lexer::parseOctal(double& returnValue) { // Optimization: most octal values fit into 4 bytes. uint32_t octalValue = 0; - int maximumDigits = 9; + const unsigned maximumDigits = 10; + int digit = maximumDigits - 1; // Temporary buffer for the digits. Makes easier // to reconstruct the input characters when needed. - LChar digits[10]; + LChar digits[maximumDigits]; do { octalValue = octalValue * 8 + (m_current - '0'); - digits[maximumDigits] = m_current; + digits[digit] = m_current; shift(); - --maximumDigits; - } while (isASCIIOctalDigit(m_current) && maximumDigits >= 0); + --digit; + } while (isASCIIOctalDigit(m_current) && digit >= 0); - if (!isASCIIDigit(m_current) && maximumDigits >= 0) { + if (!isASCIIDigit(m_current) && digit >= 0) { returnValue = octalValue; return true; } - for (int i = 9; i > maximumDigits; --i) + for (int i = maximumDigits - 1; i > digit; --i) record8(digits[i]); while (isASCIIOctalDigit(m_current)) { @@ -1265,24 +1622,25 @@ ALWAYS_INLINE bool Lexer::parseDecimal(double& returnValue) // Since parseOctal may be executed before parseDecimal, // the m_buffer8 may hold ascii digits. if (!m_buffer8.size()) { - int maximumDigits = 9; + const unsigned maximumDigits = 10; + int digit = maximumDigits - 1; // Temporary buffer for the digits. Makes easier // to reconstruct the input characters when needed. - LChar digits[10]; + LChar digits[maximumDigits]; do { decimalValue = decimalValue * 10 + (m_current - '0'); - digits[maximumDigits] = m_current; + digits[digit] = m_current; shift(); - --maximumDigits; - } while (isASCIIDigit(m_current) && maximumDigits >= 0); + --digit; + } while (isASCIIDigit(m_current) && digit >= 0); - if (maximumDigits >= 0 && m_current != '.' && (m_current | 0x20) != 'e') { + if (digit >= 0 && m_current != '.' && (m_current | 0x20) != 'e') { returnValue = decimalValue; return true; } - for (int i = 9; i > maximumDigits; --i) + for (int i = maximumDigits - 1; i > digit; --i) record8(digits[i]); } @@ -1357,11 +1715,25 @@ bool Lexer::nextTokenIsColon() return code < m_codeEnd && *code == ':'; } +#if ENABLE(ES6_ARROWFUNCTION_SYNTAX) +template +void Lexer::setTokenPosition(JSToken* tokenRecord) +{ + JSTokenData* tokenData = &tokenRecord->m_data; + tokenData->line = lineNumber(); + tokenData->offset = currentOffset(); + tokenData->lineStartOffset = currentLineStartOffset(); + ASSERT(tokenData->offset >= tokenData->lineStartOffset); +} +#endif + template JSTokenType Lexer::lex(JSToken* tokenRecord, unsigned lexerFlags, bool strictMode) { JSTokenData* tokenData = &tokenRecord->m_data; JSTokenLocation* tokenLocation = &tokenRecord->m_location; + m_lastTockenLocation = JSTokenLocation(tokenRecord->m_location); + ASSERT(!m_error); ASSERT(m_buffer8.isEmpty()); ASSERT(m_buffer16.isEmpty()); @@ -1420,7 +1792,19 @@ start: } token = GT; break; - case CharacterEqual: + case CharacterEqual: { +#if ENABLE(ES6_ARROWFUNCTION_SYNTAX) + if (peek(1) == '>') { + token = ARROWFUNCTION; + tokenData->line = lineNumber(); + tokenData->offset = currentOffset(); + tokenData->lineStartOffset = currentLineStartOffset(); + ASSERT(tokenData->offset >= tokenData->lineStartOffset); + shift(); + shift(); + break; + } +#endif shift(); if (m_current == '=') { shift(); @@ -1434,6 +1818,7 @@ start: } token = EQUAL; break; + } case CharacterLess: shift(); if (m_current == '!' && peek(1) == '-' && peek(2) == '-') { @@ -1522,7 +1907,7 @@ start: shift(); if (parseMultilineComment()) goto start; - m_lexErrorMessage = "Multiline comment was not closed properly"; + m_lexErrorMessage = ASCIILiteral("Multiline comment was not closed properly"); token = UNTERMINATED_MULTILINE_COMMENT_ERRORTOK; goto returnError; } @@ -1648,79 +2033,144 @@ start: shift(); if ((m_current | 0x20) == 'x') { if (!isASCIIHexDigit(peek(1))) { - m_lexErrorMessage = "No hexadecimal digits after '0x'"; + m_lexErrorMessage = ASCIILiteral("No hexadecimal digits after '0x'"); token = INVALID_HEX_NUMBER_ERRORTOK; goto returnError; } + + // Shift out the 'x' prefix. + shift(); + parseHex(tokenData->doubleValue); if (isIdentStart(m_current)) { - m_lexErrorMessage = "No space between hexadecimal literal and identifier"; + m_lexErrorMessage = ASCIILiteral("No space between hexadecimal literal and identifier"); token = INVALID_HEX_NUMBER_ERRORTOK; goto returnError; } - token = NUMBER; - m_buffer8.resize(0); + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); + m_buffer8.shrink(0); + break; + } + if ((m_current | 0x20) == 'b') { + if (!isASCIIBinaryDigit(peek(1))) { + m_lexErrorMessage = ASCIILiteral("No binary digits after '0b'"); + token = INVALID_BINARY_NUMBER_ERRORTOK; + goto returnError; + } + + // Shift out the 'b' prefix. + shift(); + + parseBinary(tokenData->doubleValue); + if (isIdentStart(m_current)) { + m_lexErrorMessage = ASCIILiteral("No space between binary literal and identifier"); + token = INVALID_BINARY_NUMBER_ERRORTOK; + goto returnError; + } + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); + m_buffer8.shrink(0); + break; + } + + if ((m_current | 0x20) == 'o') { + if (!isASCIIOctalDigit(peek(1))) { + m_lexErrorMessage = ASCIILiteral("No octal digits after '0o'"); + token = INVALID_OCTAL_NUMBER_ERRORTOK; + goto returnError; + } + + // Shift out the 'o' prefix. + shift(); + + parseOctal(tokenData->doubleValue); + if (isIdentStart(m_current)) { + m_lexErrorMessage = ASCIILiteral("No space between octal literal and identifier"); + token = INVALID_OCTAL_NUMBER_ERRORTOK; + goto returnError; + } + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); + m_buffer8.shrink(0); break; } record8('0'); + if (strictMode && isASCIIDigit(m_current)) { + m_lexErrorMessage = ASCIILiteral("Decimal integer literals with a leading zero are forbidden in strict mode"); + token = INVALID_OCTAL_NUMBER_ERRORTOK; + goto returnError; + } if (isASCIIOctalDigit(m_current)) { if (parseOctal(tokenData->doubleValue)) { - if (strictMode) { - m_lexErrorMessage = "Octal escapes are forbidden in strict mode"; - token = INVALID_OCTAL_NUMBER_ERRORTOK; - goto returnError; - } - token = NUMBER; + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); } } FALLTHROUGH; case CharacterNumber: - if (LIKELY(token != NUMBER)) { + if (LIKELY(token != INTEGER && token != DOUBLE)) { if (!parseDecimal(tokenData->doubleValue)) { + token = INTEGER; if (m_current == '.') { shift(); inNumberAfterDecimalPoint: parseNumberAfterDecimalPoint(); + token = DOUBLE; } if ((m_current | 0x20) == 'e') { if (!parseNumberAfterExponentIndicator()) { - m_lexErrorMessage = "Non-number found after exponent indicator"; + m_lexErrorMessage = ASCIILiteral("Non-number found after exponent indicator"); token = atEnd() ? UNTERMINATED_NUMERIC_LITERAL_ERRORTOK : INVALID_NUMERIC_LITERAL_ERRORTOK; goto returnError; } } size_t parsedLength; tokenData->doubleValue = parseDouble(m_buffer8.data(), m_buffer8.size(), parsedLength); - } - token = NUMBER; + if (token == INTEGER) + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); + } else + token = tokenTypeForIntegerLikeToken(tokenData->doubleValue); } // No identifiers allowed directly after numeric literal, e.g. "3in" is bad. if (UNLIKELY(isIdentStart(m_current))) { - m_lexErrorMessage = "At least one digit must occur after a decimal point"; + m_lexErrorMessage = ASCIILiteral("At least one digit must occur after a decimal point"); token = atEnd() ? UNTERMINATED_NUMERIC_LITERAL_ERRORTOK : INVALID_NUMERIC_LITERAL_ERRORTOK; goto returnError; } - m_buffer8.resize(0); + m_buffer8.shrink(0); break; - case CharacterQuote: - if (lexerFlags & LexerFlagsDontBuildStrings) { - StringParseResult result = parseString(tokenData, strictMode); - if (UNLIKELY(result != StringParsedSuccessfully)) { - token = result == StringUnterminated ? UNTERMINATED_STRING_LITERAL_ERRORTOK : INVALID_STRING_LITERAL_ERRORTOK; - goto returnError; - } - } else { - StringParseResult result = parseString(tokenData, strictMode); - if (UNLIKELY(result != StringParsedSuccessfully)) { - token = result == StringUnterminated ? UNTERMINATED_STRING_LITERAL_ERRORTOK : INVALID_STRING_LITERAL_ERRORTOK; - goto returnError; - } + case CharacterQuote: { + StringParseResult result = StringCannotBeParsed; + if (lexerFlags & LexerFlagsDontBuildStrings) + result = parseString(tokenData, strictMode); + else + result = parseString(tokenData, strictMode); + + if (UNLIKELY(result != StringParsedSuccessfully)) { + token = result == StringUnterminated ? UNTERMINATED_STRING_LITERAL_ERRORTOK : INVALID_STRING_LITERAL_ERRORTOK; + goto returnError; } shift(); token = STRING; break; + } +#if ENABLE(ES6_TEMPLATE_LITERAL_SYNTAX) + case CharacterBackQuote: { + // Skip backquote. + shift(); + StringParseResult result = StringCannotBeParsed; + if (lexerFlags & LexerFlagsDontBuildStrings) + result = parseTemplateLiteral(tokenData, RawStringsBuildMode::BuildRawStrings); + else + result = parseTemplateLiteral(tokenData, RawStringsBuildMode::BuildRawStrings); + + if (UNLIKELY(result != StringParsedSuccessfully)) { + token = result == StringUnterminated ? UNTERMINATED_TEMPLATE_LITERAL_ERRORTOK : INVALID_TEMPLATE_LITERAL_ERRORTOK; + goto returnError; + } + token = TEMPLATE; + break; + } +#endif case CharacterIdentifierStart: ASSERT(isIdentStart(m_current)); FALLTHROUGH; @@ -1749,7 +2199,7 @@ inNumberAfterDecimalPoint: goto returnError; default: RELEASE_ASSERT_NOT_REACHED(); - m_lexErrorMessage = "Internal Error"; + m_lexErrorMessage = ASCIILiteral("Internal Error"); token = ERRORTOK; goto returnError; } @@ -1823,7 +2273,7 @@ bool Lexer::scanRegExp(const Identifier*& pattern, const Identifier*& flags, while (true) { if (isLineTerminator(m_current) || atEnd()) { - m_buffer16.resize(0); + m_buffer16.shrink(0); return false; } @@ -1857,7 +2307,7 @@ bool Lexer::scanRegExp(const Identifier*& pattern, const Identifier*& flags, pattern = makeRightSizedIdentifier(m_buffer16.data(), m_buffer16.size(), charactersOredTogether); - m_buffer16.resize(0); + m_buffer16.shrink(0); charactersOredTogether = 0; while (isIdentPart(m_current)) { @@ -1867,7 +2317,7 @@ bool Lexer::scanRegExp(const Identifier*& pattern, const Identifier*& flags, } flags = makeRightSizedIdentifier(m_buffer16.data(), m_buffer16.size(), charactersOredTogether); - m_buffer16.resize(0); + m_buffer16.shrink(0); return true; } @@ -1913,6 +2363,40 @@ bool Lexer::skipRegExp() return true; } +#if ENABLE(ES6_TEMPLATE_LITERAL_SYNTAX) +template +JSTokenType Lexer::scanTrailingTemplateString(JSToken* tokenRecord, RawStringsBuildMode rawStringsBuildMode) +{ + JSTokenData* tokenData = &tokenRecord->m_data; + JSTokenLocation* tokenLocation = &tokenRecord->m_location; + ASSERT(!m_error); + ASSERT(m_buffer16.isEmpty()); + + // Leading closing brace } is already shifted in the previous token scan. + // So in this re-scan phase, shift() is not needed here. + StringParseResult result = parseTemplateLiteral(tokenData, rawStringsBuildMode); + JSTokenType token = ERRORTOK; + if (UNLIKELY(result != StringParsedSuccessfully)) { + token = result == StringUnterminated ? UNTERMINATED_TEMPLATE_LITERAL_ERRORTOK : INVALID_TEMPLATE_LITERAL_ERRORTOK; + m_error = true; + } else { + token = TEMPLATE; + m_lastToken = token; + } + + // Since TemplateString always ends with ` or }, m_atLineStart always becomes false. + m_atLineStart = false; + + // Adjust current tokenLocation data for TemplateString. + tokenLocation->line = m_lineNumber; + tokenLocation->endOffset = currentOffset(); + tokenLocation->lineStartOffset = currentLineStartOffset(); + ASSERT(tokenLocation->endOffset >= tokenLocation->lineStartOffset); + tokenRecord->m_endPosition = currentPosition(); + return token; +} +#endif + template void Lexer::clear() { @@ -1924,6 +2408,9 @@ void Lexer::clear() Vector newBuffer16; m_buffer16.swap(newBuffer16); + Vector newBufferForRawTemplateString16; + m_bufferForRawTemplateString16.swap(newBufferForRawTemplateString16); + m_isReparsing = false; }