namespace v8 {
namespace internal {
-int HexValue(uc32 c) {
- if ('0' <= c && c <= '9')
- return c - '0';
- if ('a' <= c && c <= 'f')
- return c - 'a' + 10;
- if ('A' <= c && c <= 'F')
- return c - 'A' + 10;
- return -1;
-}
-
namespace {
// C++-style iterator adaptor for StringInputBuffer
}
-// Returns the value (0 .. 15) of a hexadecimal character c.
-// If c is not a legal hexadecimal character, returns a value < 0.
-int HexValue(uc32 c);
-
-
// Enumeration for allowing octals and ignoring junk when converting
// strings to numbers.
enum ConversionFlags {
// Initialize parser state.
source->TryFlatten();
- scanner_.Initialize(source, JAVASCRIPT);
+ scanner_.Initialize(source);
ASSERT(target_stack_ == NULL);
if (pre_data_ != NULL) pre_data_->Initialize();
// Initialize parser state.
source->TryFlatten();
- scanner_.Initialize(source, info->start_position(), info->end_position(),
- JAVASCRIPT);
+ scanner_.Initialize(source, info->start_position(), info->end_position());
ASSERT(target_stack_ == NULL);
mode_ = PARSE_EAGERLY;
Handle<Object> JsonParser::ParseJson(Handle<String> source) {
source->TryFlatten();
- scanner_.Initialize(source, JSON);
+ scanner_.Initialize(source);
Handle<Object> result = ParseJsonValue();
if (result.is_null() || scanner_.Next() != Token::EOS) {
if (scanner_.stack_overflow()) {
static ScriptDataImpl* DoPreParse(UTF16Buffer* stream,
bool allow_lazy,
PartialParserRecorder* recorder) {
- typedef preparser::Scanner<UTF16Buffer, UTF8Buffer> PreScanner;
- PreScanner scanner;
+ preparser::Scanner scanner;
scanner.Initialize(stream);
- preparser::PreParser<PreScanner, PartialParserRecorder> preparser;
+ preparser::PreParser<preparser::Scanner, PartialParserRecorder> preparser;
if (!preparser.PreParseProgram(&scanner, recorder, allow_lazy)) {
Top::StackOverflow();
return NULL;
Expression* ParseV8Intrinsic(bool* ok);
INLINE(Token::Value peek()) { return scanner_.peek(); }
- INLINE(Token::Value Next()) { return scanner_.Next(); }
+ INLINE(Token::Value Next()) { return scanner_.NextCheckStack(); }
INLINE(void Consume(Token::Value token));
void Expect(Token::Value token, bool* ok);
bool Check(Token::Value token);
ZoneList<Handle<String> > symbol_cache_;
Handle<Script> script_;
- Scanner scanner_;
+ V8JavaScriptScanner scanner_;
Scope* top_scope_;
int with_nesting_level_;
// Converts the currently parsed literal to a JavaScript String.
Handle<String> GetString();
- Scanner scanner_;
+ JsonScanner scanner_;
};
} } // namespace v8::internal
typedef int uc32;
-int HexValue(uc32 c) {
- int res = c | 0x20; // Uppercase letters.
- int is_digit = (c & 0x10) >> 4; // 0 if non-digit, 1 if digit.
- // What to add to digits to make them consecutive with 'a'-'f' letters.
- int kDelta = 'a' - '9' - 1;
- // What to subtract to digits and letters to get them back to the range 0..15.
- int kStart = '0' + kDelta;
- res -= kStart;
- res += kDelta * is_digit;
- return res;
-}
-
-
class PreScannerStackGuard {
public:
explicit PreScannerStackGuard(int max_size)
// Scanner for preparsing.
// InputStream is a source of UC16 characters with limited push-back.
// LiteralsBuffer is a collector of (UTF-8) characters used to capture literals.
-template <typename InputStream, typename LiteralsBuffer>
class Scanner {
public:
enum LiteralType {
Scanner();
- void Initialize(InputStream* stream);
+ void Initialize(i::UTF16Buffer* stream);
// Returns the next token.
i::Token::Value Next();
return next_.literal_chars;
}
-
// Returns the length of the next token (that would be returned if
// Next() were called).
int next_literal_length() const {
bool has_line_terminator_before_next_;
// Source.
- InputStream* source_;
+ i::UTF16Buffer* source_;
// Buffer to hold literal values (identifiers, strings, numerals, regexps and
// regexp flags) using '\x00'-terminated UTF-8 encoding.
// Handles allocation internally.
// Notice that the '\x00' termination is meaningless for strings and regexps
// which may contain the zero-character, but can be used as terminator for
- // identifiers, numerals and regexp flags.
- LiteralsBuffer literal_buffer_;
+ // identifiers, numerals and regexp flags.Collector
+ i::LiteralCollector literal_buffer_;
bool stack_overflow_;
// ----------------------------------------------------------------------------
// Scanner::LiteralScope
-template <typename InputStream, typename LiteralsBuffer>
-Scanner<InputStream, LiteralsBuffer>::LiteralScope::LiteralScope(
+Scanner::LiteralScope::LiteralScope(
Scanner* self, LiteralType type)
: scanner_(self), complete_(false) {
self->StartLiteral(type);
}
-template <typename InputStream, typename LiteralsBuffer>
-Scanner<InputStream, LiteralsBuffer>::LiteralScope::~LiteralScope() {
+Scanner::LiteralScope::~LiteralScope() {
if (!complete_) scanner_->DropLiteral();
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::LiteralScope::Complete() {
+void Scanner::LiteralScope::Complete() {
scanner_->TerminateLiteral();
complete_ = true;
}
// ----------------------------------------------------------------------------
// Scanner.
-template <typename InputStream, typename LiteralsBuffer>
-Scanner<InputStream, LiteralsBuffer>::Scanner()
+Scanner::Scanner()
: stack_guard_(kMaxStackSize),
has_line_terminator_before_next_(false),
source_(NULL),
stack_overflow_(false) {}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::Initialize(InputStream* stream) {
+void Scanner::Initialize(i::UTF16Buffer* stream) {
source_ = stream;
// Initialize current_ to not refer to a literal.
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::Next() {
+i::Token::Value Scanner::Next() {
// BUG 1215673: Find a thread safe way to set a stack limit in
// pre-parse mode. Otherwise, we cannot safely pre-parse from other
// threads.
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::StartLiteral(LiteralType type) {
+void Scanner::StartLiteral(LiteralType type) {
// Only record string and literal identifiers when preparsing.
// Those are the ones that are recorded as symbols. Numbers and
// regexps are not recorded.
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::AddLiteralChar(uc32 c) {
+void Scanner::AddLiteralChar(uc32 c) {
literal_buffer_.AddChar(c);
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::TerminateLiteral() {
+void Scanner::TerminateLiteral() {
i::Vector<const char> chars = literal_buffer_.EndLiteral();
next_.literal_chars = chars.start();
next_.literal_length = chars.length();
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::DropLiteral() {
+void Scanner::DropLiteral() {
literal_buffer_.DropLiteral();
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::AddLiteralCharAdvance() {
+void Scanner::AddLiteralCharAdvance() {
AddLiteralChar(c0_);
Advance();
}
}
-template <typename InputStream, typename LiteralsBuffer>
-bool Scanner<InputStream, LiteralsBuffer>::SkipWhiteSpace() {
+bool Scanner::SkipWhiteSpace() {
int start_position = source_pos();
while (true) {
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::SkipSingleLineComment() {
+i::Token::Value Scanner::SkipSingleLineComment() {
Advance();
// The line terminator at the end of the line is not considered
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::SkipMultiLineComment() {
+i::Token::Value Scanner::SkipMultiLineComment() {
ASSERT(c0_ == '*');
Advance();
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::ScanHtmlComment() {
+i::Token::Value Scanner::ScanHtmlComment() {
// Check for <!-- comments.
ASSERT(c0_ == '!');
Advance();
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::Scan() {
+void Scanner::Scan() {
next_.literal_length = 0;
i::Token::Value token;
do {
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::SeekForward(int pos) {
+void Scanner::SeekForward(int pos) {
source_->SeekForward(pos - 1);
Advance();
// This function is only called to seek to the location
}
-template <typename InputStream, typename LiteralsBuffer>
-uc32 Scanner<InputStream, LiteralsBuffer>::ScanHexEscape(uc32 c, int length) {
+uc32 Scanner::ScanHexEscape(uc32 c, int length) {
ASSERT(length <= 4); // prevent overflow
uc32 digits[4];
uc32 x = 0;
for (int i = 0; i < length; i++) {
digits[i] = c0_;
- int d = HexValue(c0_);
+ int d = i::HexValue(c0_);
if (d < 0) {
// According to ECMA-262, 3rd, 7.8.4, page 18, these hex escapes
// should be illegal, but other JS VMs just return the
// Octal escapes of the forms '\0xx' and '\xxx' are not a part of
// ECMA-262. Other JS VMs support them.
-template <typename InputStream, typename LiteralsBuffer>
-uc32 Scanner<InputStream, LiteralsBuffer>::ScanOctalEscape(
+uc32 Scanner::ScanOctalEscape(
uc32 c, int length) {
uc32 x = c - '0';
for (int i = 0; i < length; i++) {
}
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::ScanEscape() {
+void Scanner::ScanEscape() {
uc32 c = c0_;
Advance();
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::ScanString() {
+i::Token::Value Scanner::ScanString() {
uc32 quote = c0_;
Advance(); // consume quote
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::Select(
+i::Token::Value Scanner::Select(
i::Token::Value tok) {
Advance();
return tok;
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::Select(
+i::Token::Value Scanner::Select(
uc32 next,
i::Token::Value then,
i::Token::Value else_) {
// Returns true if any decimal digits were scanned, returns false otherwise.
-template <typename InputStream, typename LiteralsBuffer>
-void Scanner<InputStream, LiteralsBuffer>::ScanDecimalDigits() {
+void Scanner::ScanDecimalDigits() {
while (i::IsDecimalDigit(c0_))
AddLiteralCharAdvance();
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::ScanNumber(
+i::Token::Value Scanner::ScanNumber(
bool seen_period) {
// c0_ is the first digit of the number or the fraction.
ASSERT(i::IsDecimalDigit(c0_));
}
-template <typename InputStream, typename LiteralsBuffer>
-uc32 Scanner<InputStream, LiteralsBuffer>::ScanIdentifierUnicodeEscape() {
+uc32 Scanner::ScanIdentifierUnicodeEscape() {
Advance();
if (c0_ != 'u') return unibrow::Utf8::kBadChar;
Advance();
}
-template <typename InputStream, typename LiteralsBuffer>
-i::Token::Value Scanner<InputStream, LiteralsBuffer>::ScanIdentifier() {
+i::Token::Value Scanner::ScanIdentifier() {
ASSERT(i::ScannerConstants::kIsIdentifierStart.get(c0_));
LiteralScope literal(this, kLiteralIdentifier);
}
-template <typename InputStream, typename LiteralsBuffer>
-bool Scanner<InputStream, LiteralsBuffer>::ScanRegExpPattern(bool seen_equal) {
+bool Scanner::ScanRegExpPattern(bool seen_equal) {
// Scan: ('/' | '/=') RegularExpressionBody '/' RegularExpressionFlags
bool in_character_class = false;
return true;
}
-template <typename InputStream, typename LiteralsBuffer>
-bool Scanner<InputStream, LiteralsBuffer>::ScanRegExpFlags() {
+bool Scanner::ScanRegExpFlags() {
// Scan regular expression flags.
LiteralScope literal(this, kLiteralRegExpFlags);
while (i::ScannerConstants::kIsIdentifierPart.get(c0_)) {
#include "../include/v8stdint.h"
#include "scanner-base.h"
+#include "char-predicates-inl.h"
namespace v8 {
namespace internal {
// ----------------------------------------------------------------------------
+// UTF16Buffer
+
+UTF16Buffer::UTF16Buffer()
+ : pos_(0), end_(kNoEndPosition) { }
+
+// ----------------------------------------------------------------------------
+// LiteralCollector
+
+LiteralCollector::LiteralCollector()
+ : buffer_(kInitialCapacity), recording_(false) { }
+
+
+LiteralCollector::~LiteralCollector() {}
+
+
+void LiteralCollector::AddCharSlow(uc32 c) {
+ ASSERT(static_cast<unsigned>(c) > unibrow::Utf8::kMaxOneByteChar);
+ int length = unibrow::Utf8::Length(c);
+ Vector<char> block = buffer_.AddBlock(length, '\0');
+#ifdef DEBUG
+ int written_length = unibrow::Utf8::Encode(block.start(), c);
+ CHECK_EQ(length, written_length);
+#else
+ unibrow::Utf8::Encode(block.start(), c);
+#endif
+}
+
+// ----------------------------------------------------------------------------
// Character predicates
unibrow::Predicate<IdentifierStart, 128> ScannerConstants::kIsIdentifierStart;
}
// ----------------------------------------------------------------------------
+// Scanner
+
+Scanner::Scanner() : source_(NULL), stack_overflow_(false) {}
+
+
+uc32 Scanner::ScanHexEscape(uc32 c, int length) {
+ ASSERT(length <= 4); // prevent overflow
+
+ uc32 digits[4];
+ uc32 x = 0;
+ for (int i = 0; i < length; i++) {
+ digits[i] = c0_;
+ int d = HexValue(c0_);
+ if (d < 0) {
+ // According to ECMA-262, 3rd, 7.8.4, page 18, these hex escapes
+ // should be illegal, but other JS VMs just return the
+ // non-escaped version of the original character.
+
+ // Push back digits read, except the last one (in c0_).
+ for (int j = i-1; j >= 0; j--) {
+ PushBack(digits[j]);
+ }
+ // Notice: No handling of error - treat it as "\u"->"u".
+ return c;
+ }
+ x = x * 16 + d;
+ Advance();
+ }
+
+ return x;
+}
+
+
+// Octal escapes of the forms '\0xx' and '\xxx' are not a part of
+// ECMA-262. Other JS VMs support them.
+uc32 Scanner::ScanOctalEscape(uc32 c, int length) {
+ uc32 x = c - '0';
+ for (int i = 0; i < length; i++) {
+ int d = c0_ - '0';
+ if (d < 0 || d > 7) break;
+ int nx = x * 8 + d;
+ if (nx >= 256) break;
+ x = nx;
+ Advance();
+ }
+ return x;
+}
+
+
+// ----------------------------------------------------------------------------
+// JavaScriptScanner
+
+JavaScriptScanner::JavaScriptScanner()
+ : has_line_terminator_before_next_(false) {}
+
+
+Token::Value JavaScriptScanner::Next() {
+ current_ = next_;
+ has_line_terminator_before_next_ = false;
+ Scan();
+ return current_.token;
+}
+
+
+static inline bool IsByteOrderMark(uc32 c) {
+ // The Unicode value U+FFFE is guaranteed never to be assigned as a
+ // Unicode character; this implies that in a Unicode context the
+ // 0xFF, 0xFE byte pattern can only be interpreted as the U+FEFF
+ // character expressed in little-endian byte order (since it could
+ // not be a U+FFFE character expressed in big-endian byte
+ // order). Nevertheless, we check for it to be compatible with
+ // Spidermonkey.
+ return c == 0xFEFF || c == 0xFFFE;
+}
+
+
+bool JavaScriptScanner::SkipWhiteSpace() {
+ int start_position = source_pos();
+
+ while (true) {
+ // We treat byte-order marks (BOMs) as whitespace for better
+ // compatibility with Spidermonkey and other JavaScript engines.
+ while (ScannerConstants::kIsWhiteSpace.get(c0_) || IsByteOrderMark(c0_)) {
+ // IsWhiteSpace() includes line terminators!
+ if (ScannerConstants::kIsLineTerminator.get(c0_)) {
+ // Ignore line terminators, but remember them. This is necessary
+ // for automatic semicolon insertion.
+ has_line_terminator_before_next_ = true;
+ }
+ Advance();
+ }
+
+ // If there is an HTML comment end '-->' at the beginning of a
+ // line (with only whitespace in front of it), we treat the rest
+ // of the line as a comment. This is in line with the way
+ // SpiderMonkey handles it.
+ if (c0_ == '-' && has_line_terminator_before_next_) {
+ Advance();
+ if (c0_ == '-') {
+ Advance();
+ if (c0_ == '>') {
+ // Treat the rest of the line as a comment.
+ SkipSingleLineComment();
+ // Continue skipping white space after the comment.
+ continue;
+ }
+ PushBack('-'); // undo Advance()
+ }
+ PushBack('-'); // undo Advance()
+ }
+ // Return whether or not we skipped any characters.
+ return source_pos() != start_position;
+ }
+}
+
+
+Token::Value JavaScriptScanner::SkipSingleLineComment() {
+ Advance();
+
+ // The line terminator at the end of the line is not considered
+ // to be part of the single-line comment; it is recognized
+ // separately by the lexical grammar and becomes part of the
+ // stream of input elements for the syntactic grammar (see
+ // ECMA-262, section 7.4, page 12).
+ while (c0_ >= 0 && !ScannerConstants::kIsLineTerminator.get(c0_)) {
+ Advance();
+ }
+
+ return Token::WHITESPACE;
+}
+
+
+Token::Value JavaScriptScanner::SkipMultiLineComment() {
+ ASSERT(c0_ == '*');
+ Advance();
+
+ while (c0_ >= 0) {
+ char ch = c0_;
+ Advance();
+ // If we have reached the end of the multi-line comment, we
+ // consume the '/' and insert a whitespace. This way all
+ // multi-line comments are treated as whitespace - even the ones
+ // containing line terminators. This contradicts ECMA-262, section
+ // 7.4, page 12, that says that multi-line comments containing
+ // line terminators should be treated as a line terminator, but it
+ // matches the behaviour of SpiderMonkey and KJS.
+ if (ch == '*' && c0_ == '/') {
+ c0_ = ' ';
+ return Token::WHITESPACE;
+ }
+ }
+
+ // Unterminated multi-line comment.
+ return Token::ILLEGAL;
+}
+
+
+Token::Value JavaScriptScanner::ScanHtmlComment() {
+ // Check for <!-- comments.
+ ASSERT(c0_ == '!');
+ Advance();
+ if (c0_ == '-') {
+ Advance();
+ if (c0_ == '-') return SkipSingleLineComment();
+ PushBack('-'); // undo Advance()
+ }
+ PushBack('!'); // undo Advance()
+ ASSERT(c0_ == '!');
+ return Token::LT;
+}
+
+
+void JavaScriptScanner::Scan() {
+ next_.literal_chars = Vector<const char>();
+ Token::Value token;
+ do {
+ // Remember the position of the next token
+ next_.location.beg_pos = source_pos();
+
+ switch (c0_) {
+ case ' ':
+ case '\t':
+ Advance();
+ token = Token::WHITESPACE;
+ break;
+
+ case '\n':
+ Advance();
+ has_line_terminator_before_next_ = true;
+ token = Token::WHITESPACE;
+ break;
+
+ case '"': case '\'':
+ token = ScanString();
+ break;
+
+ case '<':
+ // < <= << <<= <!--
+ Advance();
+ if (c0_ == '=') {
+ token = Select(Token::LTE);
+ } else if (c0_ == '<') {
+ token = Select('=', Token::ASSIGN_SHL, Token::SHL);
+ } else if (c0_ == '!') {
+ token = ScanHtmlComment();
+ } else {
+ token = Token::LT;
+ }
+ break;
+
+ case '>':
+ // > >= >> >>= >>> >>>=
+ Advance();
+ if (c0_ == '=') {
+ token = Select(Token::GTE);
+ } else if (c0_ == '>') {
+ // >> >>= >>> >>>=
+ Advance();
+ if (c0_ == '=') {
+ token = Select(Token::ASSIGN_SAR);
+ } else if (c0_ == '>') {
+ token = Select('=', Token::ASSIGN_SHR, Token::SHR);
+ } else {
+ token = Token::SAR;
+ }
+ } else {
+ token = Token::GT;
+ }
+ break;
+
+ case '=':
+ // = == ===
+ Advance();
+ if (c0_ == '=') {
+ token = Select('=', Token::EQ_STRICT, Token::EQ);
+ } else {
+ token = Token::ASSIGN;
+ }
+ break;
+
+ case '!':
+ // ! != !==
+ Advance();
+ if (c0_ == '=') {
+ token = Select('=', Token::NE_STRICT, Token::NE);
+ } else {
+ token = Token::NOT;
+ }
+ break;
+
+ case '+':
+ // + ++ +=
+ Advance();
+ if (c0_ == '+') {
+ token = Select(Token::INC);
+ } else if (c0_ == '=') {
+ token = Select(Token::ASSIGN_ADD);
+ } else {
+ token = Token::ADD;
+ }
+ break;
+
+ case '-':
+ // - -- --> -=
+ Advance();
+ if (c0_ == '-') {
+ Advance();
+ if (c0_ == '>' && has_line_terminator_before_next_) {
+ // For compatibility with SpiderMonkey, we skip lines that
+ // start with an HTML comment end '-->'.
+ token = SkipSingleLineComment();
+ } else {
+ token = Token::DEC;
+ }
+ } else if (c0_ == '=') {
+ token = Select(Token::ASSIGN_SUB);
+ } else {
+ token = Token::SUB;
+ }
+ break;
+
+ case '*':
+ // * *=
+ token = Select('=', Token::ASSIGN_MUL, Token::MUL);
+ break;
+
+ case '%':
+ // % %=
+ token = Select('=', Token::ASSIGN_MOD, Token::MOD);
+ break;
+
+ case '/':
+ // / // /* /=
+ Advance();
+ if (c0_ == '/') {
+ token = SkipSingleLineComment();
+ } else if (c0_ == '*') {
+ token = SkipMultiLineComment();
+ } else if (c0_ == '=') {
+ token = Select(Token::ASSIGN_DIV);
+ } else {
+ token = Token::DIV;
+ }
+ break;
+
+ case '&':
+ // & && &=
+ Advance();
+ if (c0_ == '&') {
+ token = Select(Token::AND);
+ } else if (c0_ == '=') {
+ token = Select(Token::ASSIGN_BIT_AND);
+ } else {
+ token = Token::BIT_AND;
+ }
+ break;
+
+ case '|':
+ // | || |=
+ Advance();
+ if (c0_ == '|') {
+ token = Select(Token::OR);
+ } else if (c0_ == '=') {
+ token = Select(Token::ASSIGN_BIT_OR);
+ } else {
+ token = Token::BIT_OR;
+ }
+ break;
+
+ case '^':
+ // ^ ^=
+ token = Select('=', Token::ASSIGN_BIT_XOR, Token::BIT_XOR);
+ break;
+
+ case '.':
+ // . Number
+ Advance();
+ if (IsDecimalDigit(c0_)) {
+ token = ScanNumber(true);
+ } else {
+ token = Token::PERIOD;
+ }
+ break;
+
+ case ':':
+ token = Select(Token::COLON);
+ break;
+
+ case ';':
+ token = Select(Token::SEMICOLON);
+ break;
+
+ case ',':
+ token = Select(Token::COMMA);
+ break;
+
+ case '(':
+ token = Select(Token::LPAREN);
+ break;
+
+ case ')':
+ token = Select(Token::RPAREN);
+ break;
+
+ case '[':
+ token = Select(Token::LBRACK);
+ break;
+
+ case ']':
+ token = Select(Token::RBRACK);
+ break;
+
+ case '{':
+ token = Select(Token::LBRACE);
+ break;
+
+ case '}':
+ token = Select(Token::RBRACE);
+ break;
+
+ case '?':
+ token = Select(Token::CONDITIONAL);
+ break;
+
+ case '~':
+ token = Select(Token::BIT_NOT);
+ break;
+
+ default:
+ if (ScannerConstants::kIsIdentifierStart.get(c0_)) {
+ token = ScanIdentifier();
+ } else if (IsDecimalDigit(c0_)) {
+ token = ScanNumber(false);
+ } else if (SkipWhiteSpace()) {
+ token = Token::WHITESPACE;
+ } else if (c0_ < 0) {
+ token = Token::EOS;
+ } else {
+ token = Select(Token::ILLEGAL);
+ }
+ break;
+ }
+
+ // Continue scanning for tokens as long as we're just skipping
+ // whitespace.
+ } while (token == Token::WHITESPACE);
+
+ next_.location.end_pos = source_pos();
+ next_.token = token;
+}
+
+
+void JavaScriptScanner::SeekForward(int pos) {
+ source_->SeekForward(pos - 1);
+ Advance();
+ // This function is only called to seek to the location
+ // of the end of a function (at the "}" token). It doesn't matter
+ // whether there was a line terminator in the part we skip.
+ has_line_terminator_before_next_ = false;
+ Scan();
+}
+
+
+void JavaScriptScanner::ScanEscape() {
+ uc32 c = c0_;
+ Advance();
+
+ // Skip escaped newlines.
+ if (ScannerConstants::kIsLineTerminator.get(c)) {
+ // Allow CR+LF newlines in multiline string literals.
+ if (IsCarriageReturn(c) && IsLineFeed(c0_)) Advance();
+ // Allow LF+CR newlines in multiline string literals.
+ if (IsLineFeed(c) && IsCarriageReturn(c0_)) Advance();
+ return;
+ }
+
+ switch (c) {
+ case '\'': // fall through
+ case '"' : // fall through
+ case '\\': break;
+ case 'b' : c = '\b'; break;
+ case 'f' : c = '\f'; break;
+ case 'n' : c = '\n'; break;
+ case 'r' : c = '\r'; break;
+ case 't' : c = '\t'; break;
+ case 'u' : c = ScanHexEscape(c, 4); break;
+ case 'v' : c = '\v'; break;
+ case 'x' : c = ScanHexEscape(c, 2); break;
+ case '0' : // fall through
+ case '1' : // fall through
+ case '2' : // fall through
+ case '3' : // fall through
+ case '4' : // fall through
+ case '5' : // fall through
+ case '6' : // fall through
+ case '7' : c = ScanOctalEscape(c, 2); break;
+ }
+
+ // According to ECMA-262, 3rd, 7.8.4 (p 18ff) these
+ // should be illegal, but they are commonly handled
+ // as non-escaped characters by JS VMs.
+ AddLiteralChar(c);
+}
+
+
+Token::Value JavaScriptScanner::ScanString() {
+ uc32 quote = c0_;
+ Advance(); // consume quote
+
+ LiteralScope literal(this);
+ while (c0_ != quote && c0_ >= 0
+ && !ScannerConstants::kIsLineTerminator.get(c0_)) {
+ uc32 c = c0_;
+ Advance();
+ if (c == '\\') {
+ if (c0_ < 0) return Token::ILLEGAL;
+ ScanEscape();
+ } else {
+ AddLiteralChar(c);
+ }
+ }
+ if (c0_ != quote) return Token::ILLEGAL;
+ literal.Complete();
+
+ Advance(); // consume quote
+ return Token::STRING;
+}
+
+
+void JavaScriptScanner::ScanDecimalDigits() {
+ while (IsDecimalDigit(c0_))
+ AddLiteralCharAdvance();
+}
+
+
+Token::Value JavaScriptScanner::ScanNumber(bool seen_period) {
+ ASSERT(IsDecimalDigit(c0_)); // the first digit of the number or the fraction
+
+ enum { DECIMAL, HEX, OCTAL } kind = DECIMAL;
+
+ LiteralScope literal(this);
+ if (seen_period) {
+ // we have already seen a decimal point of the float
+ AddLiteralChar('.');
+ ScanDecimalDigits(); // we know we have at least one digit
+
+ } else {
+ // if the first character is '0' we must check for octals and hex
+ if (c0_ == '0') {
+ AddLiteralCharAdvance();
+
+ // either 0, 0exxx, 0Exxx, 0.xxx, an octal number, or a hex number
+ if (c0_ == 'x' || c0_ == 'X') {
+ // hex number
+ kind = HEX;
+ AddLiteralCharAdvance();
+ if (!IsHexDigit(c0_)) {
+ // we must have at least one hex digit after 'x'/'X'
+ return Token::ILLEGAL;
+ }
+ while (IsHexDigit(c0_)) {
+ AddLiteralCharAdvance();
+ }
+ } else if ('0' <= c0_ && c0_ <= '7') {
+ // (possible) octal number
+ kind = OCTAL;
+ while (true) {
+ if (c0_ == '8' || c0_ == '9') {
+ kind = DECIMAL;
+ break;
+ }
+ if (c0_ < '0' || '7' < c0_) break;
+ AddLiteralCharAdvance();
+ }
+ }
+ }
+
+ // Parse decimal digits and allow trailing fractional part.
+ if (kind == DECIMAL) {
+ ScanDecimalDigits(); // optional
+ if (c0_ == '.') {
+ AddLiteralCharAdvance();
+ ScanDecimalDigits(); // optional
+ }
+ }
+ }
+
+ // scan exponent, if any
+ if (c0_ == 'e' || c0_ == 'E') {
+ ASSERT(kind != HEX); // 'e'/'E' must be scanned as part of the hex number
+ if (kind == OCTAL) return Token::ILLEGAL; // no exponent for octals allowed
+ // scan exponent
+ AddLiteralCharAdvance();
+ if (c0_ == '+' || c0_ == '-')
+ AddLiteralCharAdvance();
+ if (!IsDecimalDigit(c0_)) {
+ // we must have at least one decimal digit after 'e'/'E'
+ return Token::ILLEGAL;
+ }
+ ScanDecimalDigits();
+ }
+
+ // The source character immediately following a numeric literal must
+ // not be an identifier start or a decimal digit; see ECMA-262
+ // section 7.8.3, page 17 (note that we read only one decimal digit
+ // if the value is 0).
+ if (IsDecimalDigit(c0_) || ScannerConstants::kIsIdentifierStart.get(c0_))
+ return Token::ILLEGAL;
+
+ literal.Complete();
+
+ return Token::NUMBER;
+}
+
+
+uc32 JavaScriptScanner::ScanIdentifierUnicodeEscape() {
+ Advance();
+ if (c0_ != 'u') return unibrow::Utf8::kBadChar;
+ Advance();
+ uc32 c = ScanHexEscape('u', 4);
+ // We do not allow a unicode escape sequence to start another
+ // unicode escape sequence.
+ if (c == '\\') return unibrow::Utf8::kBadChar;
+ return c;
+}
+
+
+Token::Value JavaScriptScanner::ScanIdentifier() {
+ ASSERT(ScannerConstants::kIsIdentifierStart.get(c0_));
+
+ LiteralScope literal(this);
+ KeywordMatcher keyword_match;
+
+ // Scan identifier start character.
+ if (c0_ == '\\') {
+ uc32 c = ScanIdentifierUnicodeEscape();
+ // Only allow legal identifier start characters.
+ if (!ScannerConstants::kIsIdentifierStart.get(c)) return Token::ILLEGAL;
+ AddLiteralChar(c);
+ keyword_match.Fail();
+ } else {
+ AddLiteralChar(c0_);
+ keyword_match.AddChar(c0_);
+ Advance();
+ }
+
+ // Scan the rest of the identifier characters.
+ while (ScannerConstants::kIsIdentifierPart.get(c0_)) {
+ if (c0_ == '\\') {
+ uc32 c = ScanIdentifierUnicodeEscape();
+ // Only allow legal identifier part characters.
+ if (!ScannerConstants::kIsIdentifierPart.get(c)) return Token::ILLEGAL;
+ AddLiteralChar(c);
+ keyword_match.Fail();
+ } else {
+ AddLiteralChar(c0_);
+ keyword_match.AddChar(c0_);
+ Advance();
+ }
+ }
+ literal.Complete();
+
+ return keyword_match.token();
+}
+
+
+
+bool JavaScriptScanner::ScanRegExpPattern(bool seen_equal) {
+ // Scan: ('/' | '/=') RegularExpressionBody '/' RegularExpressionFlags
+ bool in_character_class = false;
+
+ // Previous token is either '/' or '/=', in the second case, the
+ // pattern starts at =.
+ next_.location.beg_pos = source_pos() - (seen_equal ? 2 : 1);
+ next_.location.end_pos = source_pos() - (seen_equal ? 1 : 0);
+
+ // Scan regular expression body: According to ECMA-262, 3rd, 7.8.5,
+ // the scanner should pass uninterpreted bodies to the RegExp
+ // constructor.
+ LiteralScope literal(this);
+ if (seen_equal)
+ AddLiteralChar('=');
+
+ while (c0_ != '/' || in_character_class) {
+ if (ScannerConstants::kIsLineTerminator.get(c0_) || c0_ < 0) return false;
+ if (c0_ == '\\') { // escaped character
+ AddLiteralCharAdvance();
+ if (ScannerConstants::kIsLineTerminator.get(c0_) || c0_ < 0) return false;
+ AddLiteralCharAdvance();
+ } else { // unescaped character
+ if (c0_ == '[') in_character_class = true;
+ if (c0_ == ']') in_character_class = false;
+ AddLiteralCharAdvance();
+ }
+ }
+ Advance(); // consume '/'
+
+ literal.Complete();
+
+ return true;
+}
+
+bool JavaScriptScanner::ScanRegExpFlags() {
+ // Scan regular expression flags.
+ LiteralScope literal(this);
+ while (ScannerConstants::kIsIdentifierPart.get(c0_)) {
+ if (c0_ == '\\') {
+ uc32 c = ScanIdentifierUnicodeEscape();
+ if (c != static_cast<uc32>(unibrow::Utf8::kBadChar)) {
+ // We allow any escaped character, unlike the restriction on
+ // IdentifierPart when it is used to build an IdentifierName.
+ AddLiteralChar(c);
+ continue;
+ }
+ }
+ AddLiteralCharAdvance();
+ }
+ literal.Complete();
+
+ next_.location.end_pos = source_pos() - 1;
+ return true;
+}
+
+// ----------------------------------------------------------------------------
// Keyword Matcher
KeywordMatcher::FirstState KeywordMatcher::first_states_[] = {
#include "unicode-inl.h"
#include "char-predicates.h"
#include "utils.h"
+#include "list-inl.h"
namespace v8 {
namespace internal {
-// Interface through which the scanner reads characters from the input source.
+// Returns the value (0 .. 15) of a hexadecimal character c.
+// If c is not a legal hexadecimal character, returns a value < 0.
+inline int HexValue(uc32 c) {
+ c -= '0';
+ if (static_cast<unsigned>(c) <= 9) return c;
+ c = (c | 0x20) - ('a' - '0'); // detect 0x11..0x16 and 0x31..0x36.
+ if (static_cast<unsigned>(c) <= 6) return c + 10;
+ return -1;
+}
+
+// ----------------------------------------------------------------------------
+// UTF16Buffer - scanner input source with pushback.
+
class UTF16Buffer {
public:
UTF16Buffer();
int pos() const { return pos_; }
+ static const int kNoEndPosition = 1;
+
protected:
+ // Initial value of end_ before the input stream is initialized.
+
int pos_; // Current position in the buffer.
int end_; // Position where scanning should stop (EOF).
};
static StaticResource<Utf8Decoder> utf8_decoder_;
};
+// ----------------------------------------------------------------------------
+// LiteralCollector - Collector of chars of literals.
+
+class LiteralCollector {
+ public:
+ LiteralCollector();
+ ~LiteralCollector();
+
+ inline void AddChar(uc32 c) {
+ if (recording_) {
+ if (static_cast<unsigned>(c) <= unibrow::Utf8::kMaxOneByteChar) {
+ buffer_.Add(static_cast<char>(c));
+ } else {
+ AddCharSlow(c);
+ }
+ }
+ }
+
+ void StartLiteral() {
+ buffer_.StartSequence();
+ recording_ = true;
+ }
+
+ Vector<const char> EndLiteral() {
+ if (recording_) {
+ recording_ = false;
+ buffer_.Add(kEndMarker);
+ Vector<char> sequence = buffer_.EndSequence();
+ return Vector<const char>(sequence.start(), sequence.length());
+ }
+ return Vector<const char>();
+ }
+
+ void DropLiteral() {
+ if (recording_) {
+ recording_ = false;
+ buffer_.DropSequence();
+ }
+ }
+
+ void Reset() {
+ buffer_.Reset();
+ }
+
+ // The end marker added after a parsed literal.
+ // Using zero allows the usage of strlen and similar functions on
+ // identifiers and numbers (but not strings, since they may contain zero
+ // bytes).
+ static const char kEndMarker = '\x00';
+ private:
+ static const int kInitialCapacity = 256;
+ SequenceCollector<char, 4> buffer_;
+ bool recording_;
+ void AddCharSlow(uc32 c);
+};
+
+// ----------------------------------------------------------------------------
+// Scanner base-class.
+
+// Generic functionality used by both JSON and JavaScript scanners.
+class Scanner {
+ public:
+ typedef unibrow::Utf8InputBuffer<1024> Utf8Decoder;
+
+ class LiteralScope {
+ public:
+ explicit LiteralScope(Scanner* self);
+ ~LiteralScope();
+ void Complete();
+
+ private:
+ Scanner* scanner_;
+ bool complete_;
+ };
+
+ Scanner();
+
+ // Returns the current token again.
+ Token::Value current_token() { return current_.token; }
+
+ // One token look-ahead (past the token returned by Next()).
+ Token::Value peek() const { return next_.token; }
+
+ struct Location {
+ Location(int b, int e) : beg_pos(b), end_pos(e) { }
+ Location() : beg_pos(0), end_pos(0) { }
+ int beg_pos;
+ int end_pos;
+ };
+
+ // Returns the location information for the current token
+ // (the token returned by Next()).
+ Location location() const { return current_.location; }
+ Location peek_location() const { return next_.location; }
+
+ // Returns the literal string, if any, for the current token (the
+ // token returned by Next()). The string is 0-terminated and in
+ // UTF-8 format; they may contain 0-characters. Literal strings are
+ // collected for identifiers, strings, and numbers.
+ // These functions only give the correct result if the literal
+ // was scanned between calls to StartLiteral() and TerminateLiteral().
+ const char* literal_string() const {
+ return current_.literal_chars.start();
+ }
+
+ int literal_length() const {
+ // Excluding terminal '\x00' added by TerminateLiteral().
+ return current_.literal_chars.length() - 1;
+ }
+
+ Vector<const char> literal() const {
+ return Vector<const char>(literal_string(), literal_length());
+ }
+
+ // Returns the literal string for the next token (the token that
+ // would be returned if Next() were called).
+ const char* next_literal_string() const {
+ return next_.literal_chars.start();
+ }
+
+
+ // Returns the length of the next token (that would be returned if
+ // Next() were called).
+ int next_literal_length() const {
+ // Excluding terminal '\x00' added by TerminateLiteral().
+ return next_.literal_chars.length() - 1;
+ }
+
+ Vector<const char> next_literal() const {
+ return Vector<const char>(next_literal_string(), next_literal_length());
+ }
+
+ bool stack_overflow() { return stack_overflow_; }
+
+ static const int kCharacterLookaheadBufferSize = 1;
+
+ protected:
+ // The current and look-ahead token.
+ struct TokenDesc {
+ Token::Value token;
+ Location location;
+ Vector<const char> literal_chars;
+ };
+
+ // Call this after setting source_ to the input.
+ void Init() {
+ // Set c0_ (one character ahead)
+ ASSERT(kCharacterLookaheadBufferSize == 1);
+ Advance();
+ // Initialize current_ to not refer to a literal.
+ current_.literal_chars = Vector<const char>();
+ // Reset literal buffer.
+ literal_buffer_.Reset();
+ }
+
+ // Literal buffer support
+ inline void StartLiteral() {
+ literal_buffer_.StartLiteral();
+ }
+
+ inline void AddLiteralChar(uc32 c) {
+ literal_buffer_.AddChar(c);
+ }
+
+ // Complete scanning of a literal.
+ inline void TerminateLiteral() {
+ next_.literal_chars = literal_buffer_.EndLiteral();
+ }
+
+ // Stops scanning of a literal and drop the collected characters,
+ // e.g., due to an encountered error.
+ inline void DropLiteral() {
+ literal_buffer_.DropLiteral();
+ }
+
+ inline void AddLiteralCharAdvance() {
+ AddLiteralChar(c0_);
+ Advance();
+ }
+
+ // Low-level scanning support.
+ void Advance() { c0_ = source_->Advance(); }
+ void PushBack(uc32 ch) {
+ source_->PushBack(ch);
+ c0_ = ch;
+ }
+
+ inline Token::Value Select(Token::Value tok) {
+ Advance();
+ return tok;
+ }
+
+ inline Token::Value Select(uc32 next, Token::Value then, Token::Value else_) {
+ Advance();
+ if (c0_ == next) {
+ Advance();
+ return then;
+ } else {
+ return else_;
+ }
+ }
+
+ uc32 ScanHexEscape(uc32 c, int length);
+ uc32 ScanOctalEscape(uc32 c, int length);
+
+ // Return the current source position.
+ int source_pos() {
+ return source_->pos() - kCharacterLookaheadBufferSize;
+ }
+
+ TokenDesc current_; // desc for current token (as returned by Next())
+ TokenDesc next_; // desc for next token (one token look-ahead)
+
+ // Input stream. Must be initialized to an UTF16Buffer.
+ UTF16Buffer* source_;
+
+ // Buffer to hold literal values (identifiers, strings, numbers)
+ // using '\x00'-terminated UTF-8 encoding. Handles allocation internally.
+ LiteralCollector literal_buffer_;
+
+ bool stack_overflow_;
+
+ // One Unicode character look-ahead; c0_ < 0 at the end of the input.
+ uc32 c0_;
+};
+
+// ----------------------------------------------------------------------------
+// JavaScriptScanner - base logic for JavaScript scanning.
+
+class JavaScriptScanner : public Scanner {
+ public:
+ JavaScriptScanner();
+
+ // Returns the next token.
+ Token::Value Next();
+
+ // Returns true if there was a line terminator before the peek'ed token.
+ bool has_line_terminator_before_next() const {
+ return has_line_terminator_before_next_;
+ }
+
+ // Scans the input as a regular expression pattern, previous
+ // character(s) must be /(=). Returns true if a pattern is scanned.
+ bool ScanRegExpPattern(bool seen_equal);
+ // Returns true if regexp flags are scanned (always since flags can
+ // be empty).
+ bool ScanRegExpFlags();
+
+ // Tells whether the buffer contains an identifier (no escapes).
+ // Used for checking if a property name is an identifier.
+ static bool IsIdentifier(unibrow::CharacterStream* buffer);
+
+ // Seek forward to the given position. This operation does not
+ // work in general, for instance when there are pushed back
+ // characters, but works for seeking forward until simple delimiter
+ // tokens, which is what it is used for.
+ void SeekForward(int pos);
+
+ protected:
+ bool SkipWhiteSpace();
+ Token::Value SkipSingleLineComment();
+ Token::Value SkipMultiLineComment();
+
+ // Scans a single JavaScript token.
+ void Scan();
+
+ void ScanDecimalDigits();
+ Token::Value ScanNumber(bool seen_period);
+ Token::Value ScanIdentifier();
+
+ void ScanEscape();
+ Token::Value ScanString();
+
+ // Scans a possible HTML comment -- begins with '<!'.
+ Token::Value ScanHtmlComment();
+
+ // Decodes a unicode escape-sequence which is part of an identifier.
+ // If the escape sequence cannot be decoded the result is kBadChar.
+ uc32 ScanIdentifierUnicodeEscape();
+
+ bool has_line_terminator_before_next_;
+};
+
+
+// ----------------------------------------------------------------------------
+// Keyword matching state machine.
class KeywordMatcher {
// Incrementally recognize keywords.
namespace internal {
// ----------------------------------------------------------------------------
-// UTF8Buffer
-
-UTF8Buffer::UTF8Buffer() : buffer_(kInitialCapacity), recording_(false) { }
-
-
-UTF8Buffer::~UTF8Buffer() {}
-
-
-void UTF8Buffer::AddCharSlow(uc32 c) {
- ASSERT(static_cast<unsigned>(c) > unibrow::Utf8::kMaxOneByteChar);
- int length = unibrow::Utf8::Length(c);
- Vector<char> block = buffer_.AddBlock(length, '\0');
-#ifdef DEBUG
- int written_length = unibrow::Utf8::Encode(block.start(), c);
- CHECK_EQ(length, written_length);
-#else
- unibrow::Utf8::Encode(block.start(), c);
-#endif
-}
-
-
-// ----------------------------------------------------------------------------
// UTF16Buffer
-
-UTF16Buffer::UTF16Buffer()
- : pos_(0), end_(Scanner::kNoEndPosition) { }
-
-
// CharacterStreamUTF16Buffer
CharacterStreamUTF16Buffer::CharacterStreamUTF16Buffer()
: pushback_buffer_(0), last_(0), stream_(NULL) { }
if (start_position > 0) {
SeekForward(start_position);
}
- end_ = end_position != Scanner::kNoEndPosition ? end_position : kMaxInt;
+ end_ = end_position != kNoEndPosition ? end_position : kMaxInt;
}
uc32 CharacterStreamUTF16Buffer::Advance() {
- ASSERT(end_ != Scanner::kNoEndPosition);
+ ASSERT(end_ != kNoEndPosition);
ASSERT(end_ >= 0);
// NOTE: It is of importance to Persian / Farsi resources that we do
// *not* strip format control characters in the scanner; see
}
// ----------------------------------------------------------------------------
-// Scanner
+// V8JavaScriptScanner
-Scanner::Scanner()
- : has_line_terminator_before_next_(false),
- is_parsing_json_(false),
- source_(NULL),
- stack_overflow_(false) {}
+void V8JavaScriptScanner::Initialize(Handle<String> source) {
+ source_ = stream_initializer_.Init(source, NULL, 0, source->length());
+ Init();
+ // Skip initial whitespace allowing HTML comment ends just like
+ // after a newline and scan first token.
+ has_line_terminator_before_next_ = true;
+ SkipWhiteSpace();
+ Scan();
+}
-void Scanner::Initialize(Handle<String> source,
- ParserLanguage language) {
- Init(source, NULL, 0, source->length(), language);
+void V8JavaScriptScanner::Initialize(Handle<String> source,
+ unibrow::CharacterStream* stream) {
+ source_ = stream_initializer_.Init(source, stream,
+ 0, UTF16Buffer::kNoEndPosition);
+ Init();
+ // Skip initial whitespace allowing HTML comment ends just like
+ // after a newline and scan first token.
+ has_line_terminator_before_next_ = true;
+ SkipWhiteSpace();
+ Scan();
}
-void Scanner::Initialize(Handle<String> source,
- unibrow::CharacterStream* stream,
- ParserLanguage language) {
- Init(source, stream, 0, kNoEndPosition, language);
+void V8JavaScriptScanner::Initialize(Handle<String> source,
+ int start_position,
+ int end_position) {
+ source_ = stream_initializer_.Init(source, NULL,
+ start_position, end_position);
+ Init();
+ // Skip initial whitespace allowing HTML comment ends just like
+ // after a newline and scan first token.
+ has_line_terminator_before_next_ = true;
+ SkipWhiteSpace();
+ Scan();
}
-void Scanner::Initialize(Handle<String> source,
- int start_position,
- int end_position,
- ParserLanguage language) {
- Init(source, NULL, start_position, end_position, language);
+Token::Value V8JavaScriptScanner::NextCheckStack() {
+ // BUG 1215673: Find a thread safe way to set a stack limit in
+ // pre-parse mode. Otherwise, we cannot safely pre-parse from other
+ // threads.
+ StackLimitCheck check;
+ if (check.HasOverflowed()) {
+ stack_overflow_ = true;
+ current_ = next_;
+ next_.token = Token::ILLEGAL;
+ return current_.token;
+ } else {
+ return Next();
+ }
}
-void Scanner::Init(Handle<String> source,
- unibrow::CharacterStream* stream,
- int start_position,
- int end_position,
- ParserLanguage language) {
+UTF16Buffer* StreamInitializer::Init(Handle<String> source,
+ unibrow::CharacterStream* stream,
+ int start_position,
+ int end_position) {
// Either initialize the scanner from a character stream or from a
// string.
ASSERT(source.is_null() || stream == NULL);
Handle<ExternalTwoByteString>::cast(source),
start_position,
end_position);
- source_ = &two_byte_string_buffer_;
+ return &two_byte_string_buffer_;
} else if (!source.is_null() && StringShape(*source).IsExternalAscii()) {
ascii_string_buffer_.Initialize(
Handle<ExternalAsciiString>::cast(source),
start_position,
end_position);
- source_ = &ascii_string_buffer_;
+ return &ascii_string_buffer_;
} else {
if (!source.is_null()) {
safe_string_input_buffer_.Reset(source.location());
stream,
start_position,
end_position);
- source_ = &char_stream_buffer_;
+ return &char_stream_buffer_;
}
+}
- is_parsing_json_ = (language == JSON);
+// ----------------------------------------------------------------------------
+// JsonScanner
- // Set c0_ (one character ahead)
- ASSERT(kCharacterLookaheadBufferSize == 1);
- Advance();
- // Initialize current_ to not refer to a literal.
- current_.literal_chars = Vector<const char>();
- // Reset literal buffer.
- literal_buffer_.Reset();
+JsonScanner::JsonScanner() {}
- // Skip initial whitespace allowing HTML comment ends just like
- // after a newline and scan first token.
- has_line_terminator_before_next_ = true;
- SkipWhiteSpace();
- Scan();
+
+void JsonScanner::Initialize(Handle<String> source) {
+ source_ = stream_initializer_.Init(source, NULL, 0, source->length());
+ Init();
+ // Skip initial whitespace.
+ SkipJsonWhiteSpace();
+ // Preload first token as look-ahead.
+ ScanJson();
}
-Token::Value Scanner::Next() {
+Token::Value JsonScanner::Next() {
// BUG 1215673: Find a thread safe way to set a stack limit in
// pre-parse mode. Otherwise, we cannot safely pre-parse from other
// threads.
stack_overflow_ = true;
next_.token = Token::ILLEGAL;
} else {
- has_line_terminator_before_next_ = false;
- Scan();
+ ScanJson();
}
return current_.token;
}
-void Scanner::StartLiteral() {
- literal_buffer_.StartLiteral();
-}
-
-
-void Scanner::AddLiteralChar(uc32 c) {
- literal_buffer_.AddChar(c);
-}
-
-
-void Scanner::TerminateLiteral() {
- next_.literal_chars = literal_buffer_.EndLiteral();
-}
-
-
-void Scanner::DropLiteral() {
- literal_buffer_.DropLiteral();
-}
-
-
-void Scanner::AddLiteralCharAdvance() {
- AddLiteralChar(c0_);
- Advance();
-}
-
-
-static inline bool IsByteOrderMark(uc32 c) {
- // The Unicode value U+FFFE is guaranteed never to be assigned as a
- // Unicode character; this implies that in a Unicode context the
- // 0xFF, 0xFE byte pattern can only be interpreted as the U+FEFF
- // character expressed in little-endian byte order (since it could
- // not be a U+FFFE character expressed in big-endian byte
- // order). Nevertheless, we check for it to be compatible with
- // Spidermonkey.
- return c == 0xFEFF || c == 0xFFFE;
-}
-
-
-bool Scanner::SkipJsonWhiteSpace() {
+bool JsonScanner::SkipJsonWhiteSpace() {
int start_position = source_pos();
// JSON WhiteSpace is tab, carrige-return, newline and space.
while (c0_ == ' ' || c0_ == '\n' || c0_ == '\r' || c0_ == '\t') {
}
-bool Scanner::SkipJavaScriptWhiteSpace() {
- int start_position = source_pos();
-
- while (true) {
- // We treat byte-order marks (BOMs) as whitespace for better
- // compatibility with Spidermonkey and other JavaScript engines.
- while (ScannerConstants::kIsWhiteSpace.get(c0_) || IsByteOrderMark(c0_)) {
- // IsWhiteSpace() includes line terminators!
- if (ScannerConstants::kIsLineTerminator.get(c0_)) {
- // Ignore line terminators, but remember them. This is necessary
- // for automatic semicolon insertion.
- has_line_terminator_before_next_ = true;
- }
- Advance();
- }
-
- // If there is an HTML comment end '-->' at the beginning of a
- // line (with only whitespace in front of it), we treat the rest
- // of the line as a comment. This is in line with the way
- // SpiderMonkey handles it.
- if (c0_ == '-' && has_line_terminator_before_next_) {
- Advance();
- if (c0_ == '-') {
- Advance();
- if (c0_ == '>') {
- // Treat the rest of the line as a comment.
- SkipSingleLineComment();
- // Continue skipping white space after the comment.
- continue;
- }
- PushBack('-'); // undo Advance()
- }
- PushBack('-'); // undo Advance()
- }
- // Return whether or not we skipped any characters.
- return source_pos() != start_position;
- }
-}
-
-
-Token::Value Scanner::SkipSingleLineComment() {
- Advance();
-
- // The line terminator at the end of the line is not considered
- // to be part of the single-line comment; it is recognized
- // separately by the lexical grammar and becomes part of the
- // stream of input elements for the syntactic grammar (see
- // ECMA-262, section 7.4, page 12).
- while (c0_ >= 0 && !ScannerConstants::kIsLineTerminator.get(c0_)) {
- Advance();
- }
-
- return Token::WHITESPACE;
-}
-
-
-Token::Value Scanner::SkipMultiLineComment() {
- ASSERT(c0_ == '*');
- Advance();
-
- while (c0_ >= 0) {
- char ch = c0_;
- Advance();
- // If we have reached the end of the multi-line comment, we
- // consume the '/' and insert a whitespace. This way all
- // multi-line comments are treated as whitespace - even the ones
- // containing line terminators. This contradicts ECMA-262, section
- // 7.4, page 12, that says that multi-line comments containing
- // line terminators should be treated as a line terminator, but it
- // matches the behaviour of SpiderMonkey and KJS.
- if (ch == '*' && c0_ == '/') {
- c0_ = ' ';
- return Token::WHITESPACE;
- }
- }
-
- // Unterminated multi-line comment.
- return Token::ILLEGAL;
-}
-
-
-Token::Value Scanner::ScanHtmlComment() {
- // Check for <!-- comments.
- ASSERT(c0_ == '!');
- Advance();
- if (c0_ == '-') {
- Advance();
- if (c0_ == '-') return SkipSingleLineComment();
- PushBack('-'); // undo Advance()
- }
- PushBack('!'); // undo Advance()
- ASSERT(c0_ == '!');
- return Token::LT;
-}
-
-
-
-void Scanner::ScanJson() {
+void JsonScanner::ScanJson() {
next_.literal_chars = Vector<const char>();
Token::Value token;
- has_line_terminator_before_next_ = false;
do {
// Remember the position of the next token
next_.location.beg_pos = source_pos();
}
-Token::Value Scanner::ScanJsonString() {
+Token::Value JsonScanner::ScanJsonString() {
ASSERT_EQ('"', c0_);
Advance();
LiteralScope literal(this);
}
-Token::Value Scanner::ScanJsonNumber() {
+Token::Value JsonScanner::ScanJsonNumber() {
LiteralScope literal(this);
if (c0_ == '-') AddLiteralCharAdvance();
if (c0_ == '0') {
}
-Token::Value Scanner::ScanJsonIdentifier(const char* text,
- Token::Value token) {
+Token::Value JsonScanner::ScanJsonIdentifier(const char* text,
+ Token::Value token) {
LiteralScope literal(this);
while (*text != '\0') {
if (c0_ != *text) return Token::ILLEGAL;
}
-void Scanner::ScanJavaScript() {
- next_.literal_chars = Vector<const char>();
- Token::Value token;
- do {
- // Remember the position of the next token
- next_.location.beg_pos = source_pos();
-
- switch (c0_) {
- case ' ':
- case '\t':
- Advance();
- token = Token::WHITESPACE;
- break;
-
- case '\n':
- Advance();
- has_line_terminator_before_next_ = true;
- token = Token::WHITESPACE;
- break;
-
- case '"': case '\'':
- token = ScanString();
- break;
-
- case '<':
- // < <= << <<= <!--
- Advance();
- if (c0_ == '=') {
- token = Select(Token::LTE);
- } else if (c0_ == '<') {
- token = Select('=', Token::ASSIGN_SHL, Token::SHL);
- } else if (c0_ == '!') {
- token = ScanHtmlComment();
- } else {
- token = Token::LT;
- }
- break;
-
- case '>':
- // > >= >> >>= >>> >>>=
- Advance();
- if (c0_ == '=') {
- token = Select(Token::GTE);
- } else if (c0_ == '>') {
- // >> >>= >>> >>>=
- Advance();
- if (c0_ == '=') {
- token = Select(Token::ASSIGN_SAR);
- } else if (c0_ == '>') {
- token = Select('=', Token::ASSIGN_SHR, Token::SHR);
- } else {
- token = Token::SAR;
- }
- } else {
- token = Token::GT;
- }
- break;
-
- case '=':
- // = == ===
- Advance();
- if (c0_ == '=') {
- token = Select('=', Token::EQ_STRICT, Token::EQ);
- } else {
- token = Token::ASSIGN;
- }
- break;
-
- case '!':
- // ! != !==
- Advance();
- if (c0_ == '=') {
- token = Select('=', Token::NE_STRICT, Token::NE);
- } else {
- token = Token::NOT;
- }
- break;
-
- case '+':
- // + ++ +=
- Advance();
- if (c0_ == '+') {
- token = Select(Token::INC);
- } else if (c0_ == '=') {
- token = Select(Token::ASSIGN_ADD);
- } else {
- token = Token::ADD;
- }
- break;
-
- case '-':
- // - -- --> -=
- Advance();
- if (c0_ == '-') {
- Advance();
- if (c0_ == '>' && has_line_terminator_before_next_) {
- // For compatibility with SpiderMonkey, we skip lines that
- // start with an HTML comment end '-->'.
- token = SkipSingleLineComment();
- } else {
- token = Token::DEC;
- }
- } else if (c0_ == '=') {
- token = Select(Token::ASSIGN_SUB);
- } else {
- token = Token::SUB;
- }
- break;
-
- case '*':
- // * *=
- token = Select('=', Token::ASSIGN_MUL, Token::MUL);
- break;
-
- case '%':
- // % %=
- token = Select('=', Token::ASSIGN_MOD, Token::MOD);
- break;
-
- case '/':
- // / // /* /=
- Advance();
- if (c0_ == '/') {
- token = SkipSingleLineComment();
- } else if (c0_ == '*') {
- token = SkipMultiLineComment();
- } else if (c0_ == '=') {
- token = Select(Token::ASSIGN_DIV);
- } else {
- token = Token::DIV;
- }
- break;
-
- case '&':
- // & && &=
- Advance();
- if (c0_ == '&') {
- token = Select(Token::AND);
- } else if (c0_ == '=') {
- token = Select(Token::ASSIGN_BIT_AND);
- } else {
- token = Token::BIT_AND;
- }
- break;
-
- case '|':
- // | || |=
- Advance();
- if (c0_ == '|') {
- token = Select(Token::OR);
- } else if (c0_ == '=') {
- token = Select(Token::ASSIGN_BIT_OR);
- } else {
- token = Token::BIT_OR;
- }
- break;
-
- case '^':
- // ^ ^=
- token = Select('=', Token::ASSIGN_BIT_XOR, Token::BIT_XOR);
- break;
-
- case '.':
- // . Number
- Advance();
- if (IsDecimalDigit(c0_)) {
- token = ScanNumber(true);
- } else {
- token = Token::PERIOD;
- }
- break;
-
- case ':':
- token = Select(Token::COLON);
- break;
-
- case ';':
- token = Select(Token::SEMICOLON);
- break;
-
- case ',':
- token = Select(Token::COMMA);
- break;
-
- case '(':
- token = Select(Token::LPAREN);
- break;
-
- case ')':
- token = Select(Token::RPAREN);
- break;
-
- case '[':
- token = Select(Token::LBRACK);
- break;
-
- case ']':
- token = Select(Token::RBRACK);
- break;
-
- case '{':
- token = Select(Token::LBRACE);
- break;
-
- case '}':
- token = Select(Token::RBRACE);
- break;
-
- case '?':
- token = Select(Token::CONDITIONAL);
- break;
-
- case '~':
- token = Select(Token::BIT_NOT);
- break;
-
- default:
- if (ScannerConstants::kIsIdentifierStart.get(c0_)) {
- token = ScanIdentifier();
- } else if (IsDecimalDigit(c0_)) {
- token = ScanNumber(false);
- } else if (SkipWhiteSpace()) {
- token = Token::WHITESPACE;
- } else if (c0_ < 0) {
- token = Token::EOS;
- } else {
- token = Select(Token::ILLEGAL);
- }
- break;
- }
-
- // Continue scanning for tokens as long as we're just skipping
- // whitespace.
- } while (token == Token::WHITESPACE);
-
- next_.location.end_pos = source_pos();
- next_.token = token;
-}
-
-
-void Scanner::SeekForward(int pos) {
- source_->SeekForward(pos - 1);
- Advance();
- // This function is only called to seek to the location
- // of the end of a function (at the "}" token). It doesn't matter
- // whether there was a line terminator in the part we skip.
- has_line_terminator_before_next_ = false;
- Scan();
-}
-
-
-uc32 Scanner::ScanHexEscape(uc32 c, int length) {
- ASSERT(length <= 4); // prevent overflow
-
- uc32 digits[4];
- uc32 x = 0;
- for (int i = 0; i < length; i++) {
- digits[i] = c0_;
- int d = HexValue(c0_);
- if (d < 0) {
- // According to ECMA-262, 3rd, 7.8.4, page 18, these hex escapes
- // should be illegal, but other JS VMs just return the
- // non-escaped version of the original character.
-
- // Push back digits read, except the last one (in c0_).
- for (int j = i-1; j >= 0; j--) {
- PushBack(digits[j]);
- }
- // Notice: No handling of error - treat it as "\u"->"u".
- return c;
- }
- x = x * 16 + d;
- Advance();
- }
-
- return x;
-}
-
-
-// Octal escapes of the forms '\0xx' and '\xxx' are not a part of
-// ECMA-262. Other JS VMs support them.
-uc32 Scanner::ScanOctalEscape(uc32 c, int length) {
- uc32 x = c - '0';
- for (int i = 0; i < length; i++) {
- int d = c0_ - '0';
- if (d < 0 || d > 7) break;
- int nx = x * 8 + d;
- if (nx >= 256) break;
- x = nx;
- Advance();
- }
- return x;
-}
-
-
-void Scanner::ScanEscape() {
- uc32 c = c0_;
- Advance();
-
- // Skip escaped newlines.
- if (ScannerConstants::kIsLineTerminator.get(c)) {
- // Allow CR+LF newlines in multiline string literals.
- if (IsCarriageReturn(c) && IsLineFeed(c0_)) Advance();
- // Allow LF+CR newlines in multiline string literals.
- if (IsLineFeed(c) && IsCarriageReturn(c0_)) Advance();
- return;
- }
-
- switch (c) {
- case '\'': // fall through
- case '"' : // fall through
- case '\\': break;
- case 'b' : c = '\b'; break;
- case 'f' : c = '\f'; break;
- case 'n' : c = '\n'; break;
- case 'r' : c = '\r'; break;
- case 't' : c = '\t'; break;
- case 'u' : c = ScanHexEscape(c, 4); break;
- case 'v' : c = '\v'; break;
- case 'x' : c = ScanHexEscape(c, 2); break;
- case '0' : // fall through
- case '1' : // fall through
- case '2' : // fall through
- case '3' : // fall through
- case '4' : // fall through
- case '5' : // fall through
- case '6' : // fall through
- case '7' : c = ScanOctalEscape(c, 2); break;
- }
-
- // According to ECMA-262, 3rd, 7.8.4 (p 18ff) these
- // should be illegal, but they are commonly handled
- // as non-escaped characters by JS VMs.
- AddLiteralChar(c);
-}
-
-
-Token::Value Scanner::ScanString() {
- uc32 quote = c0_;
- Advance(); // consume quote
-
- LiteralScope literal(this);
- while (c0_ != quote && c0_ >= 0
- && !ScannerConstants::kIsLineTerminator.get(c0_)) {
- uc32 c = c0_;
- Advance();
- if (c == '\\') {
- if (c0_ < 0) return Token::ILLEGAL;
- ScanEscape();
- } else {
- AddLiteralChar(c);
- }
- }
- if (c0_ != quote) return Token::ILLEGAL;
- literal.Complete();
-
- Advance(); // consume quote
- return Token::STRING;
-}
-
-
-Token::Value Scanner::Select(Token::Value tok) {
- Advance();
- return tok;
-}
-
-
-Token::Value Scanner::Select(uc32 next, Token::Value then, Token::Value else_) {
- Advance();
- if (c0_ == next) {
- Advance();
- return then;
- } else {
- return else_;
- }
-}
-
-
-// Returns true if any decimal digits were scanned, returns false otherwise.
-void Scanner::ScanDecimalDigits() {
- while (IsDecimalDigit(c0_))
- AddLiteralCharAdvance();
-}
-
-
-Token::Value Scanner::ScanNumber(bool seen_period) {
- ASSERT(IsDecimalDigit(c0_)); // the first digit of the number or the fraction
-
- enum { DECIMAL, HEX, OCTAL } kind = DECIMAL;
-
- LiteralScope literal(this);
- if (seen_period) {
- // we have already seen a decimal point of the float
- AddLiteralChar('.');
- ScanDecimalDigits(); // we know we have at least one digit
-
- } else {
- // if the first character is '0' we must check for octals and hex
- if (c0_ == '0') {
- AddLiteralCharAdvance();
-
- // either 0, 0exxx, 0Exxx, 0.xxx, an octal number, or a hex number
- if (c0_ == 'x' || c0_ == 'X') {
- // hex number
- kind = HEX;
- AddLiteralCharAdvance();
- if (!IsHexDigit(c0_)) {
- // we must have at least one hex digit after 'x'/'X'
- return Token::ILLEGAL;
- }
- while (IsHexDigit(c0_)) {
- AddLiteralCharAdvance();
- }
- } else if ('0' <= c0_ && c0_ <= '7') {
- // (possible) octal number
- kind = OCTAL;
- while (true) {
- if (c0_ == '8' || c0_ == '9') {
- kind = DECIMAL;
- break;
- }
- if (c0_ < '0' || '7' < c0_) break;
- AddLiteralCharAdvance();
- }
- }
- }
-
- // Parse decimal digits and allow trailing fractional part.
- if (kind == DECIMAL) {
- ScanDecimalDigits(); // optional
- if (c0_ == '.') {
- AddLiteralCharAdvance();
- ScanDecimalDigits(); // optional
- }
- }
- }
-
- // scan exponent, if any
- if (c0_ == 'e' || c0_ == 'E') {
- ASSERT(kind != HEX); // 'e'/'E' must be scanned as part of the hex number
- if (kind == OCTAL) return Token::ILLEGAL; // no exponent for octals allowed
- // scan exponent
- AddLiteralCharAdvance();
- if (c0_ == '+' || c0_ == '-')
- AddLiteralCharAdvance();
- if (!IsDecimalDigit(c0_)) {
- // we must have at least one decimal digit after 'e'/'E'
- return Token::ILLEGAL;
- }
- ScanDecimalDigits();
- }
-
- // The source character immediately following a numeric literal must
- // not be an identifier start or a decimal digit; see ECMA-262
- // section 7.8.3, page 17 (note that we read only one decimal digit
- // if the value is 0).
- if (IsDecimalDigit(c0_) || ScannerConstants::kIsIdentifierStart.get(c0_))
- return Token::ILLEGAL;
-
- literal.Complete();
-
- return Token::NUMBER;
-}
-
-
-uc32 Scanner::ScanIdentifierUnicodeEscape() {
- Advance();
- if (c0_ != 'u') return unibrow::Utf8::kBadChar;
- Advance();
- uc32 c = ScanHexEscape('u', 4);
- // We do not allow a unicode escape sequence to start another
- // unicode escape sequence.
- if (c == '\\') return unibrow::Utf8::kBadChar;
- return c;
-}
-
-
-Token::Value Scanner::ScanIdentifier() {
- ASSERT(ScannerConstants::kIsIdentifierStart.get(c0_));
-
- LiteralScope literal(this);
- KeywordMatcher keyword_match;
-
- // Scan identifier start character.
- if (c0_ == '\\') {
- uc32 c = ScanIdentifierUnicodeEscape();
- // Only allow legal identifier start characters.
- if (!ScannerConstants::kIsIdentifierStart.get(c)) return Token::ILLEGAL;
- AddLiteralChar(c);
- keyword_match.Fail();
- } else {
- AddLiteralChar(c0_);
- keyword_match.AddChar(c0_);
- Advance();
- }
-
- // Scan the rest of the identifier characters.
- while (ScannerConstants::kIsIdentifierPart.get(c0_)) {
- if (c0_ == '\\') {
- uc32 c = ScanIdentifierUnicodeEscape();
- // Only allow legal identifier part characters.
- if (!ScannerConstants::kIsIdentifierPart.get(c)) return Token::ILLEGAL;
- AddLiteralChar(c);
- keyword_match.Fail();
- } else {
- AddLiteralChar(c0_);
- keyword_match.AddChar(c0_);
- Advance();
- }
- }
- literal.Complete();
-
- return keyword_match.token();
-}
-
-
-
-bool Scanner::ScanRegExpPattern(bool seen_equal) {
- // Scan: ('/' | '/=') RegularExpressionBody '/' RegularExpressionFlags
- bool in_character_class = false;
-
- // Previous token is either '/' or '/=', in the second case, the
- // pattern starts at =.
- next_.location.beg_pos = source_pos() - (seen_equal ? 2 : 1);
- next_.location.end_pos = source_pos() - (seen_equal ? 1 : 0);
-
- // Scan regular expression body: According to ECMA-262, 3rd, 7.8.5,
- // the scanner should pass uninterpreted bodies to the RegExp
- // constructor.
- LiteralScope literal(this);
- if (seen_equal)
- AddLiteralChar('=');
-
- while (c0_ != '/' || in_character_class) {
- if (ScannerConstants::kIsLineTerminator.get(c0_) || c0_ < 0) return false;
- if (c0_ == '\\') { // escaped character
- AddLiteralCharAdvance();
- if (ScannerConstants::kIsLineTerminator.get(c0_) || c0_ < 0) return false;
- AddLiteralCharAdvance();
- } else { // unescaped character
- if (c0_ == '[') in_character_class = true;
- if (c0_ == ']') in_character_class = false;
- AddLiteralCharAdvance();
- }
- }
- Advance(); // consume '/'
-
- literal.Complete();
-
- return true;
-}
-
-bool Scanner::ScanRegExpFlags() {
- // Scan regular expression flags.
- LiteralScope literal(this);
- while (ScannerConstants::kIsIdentifierPart.get(c0_)) {
- if (c0_ == '\\') {
- uc32 c = ScanIdentifierUnicodeEscape();
- if (c != static_cast<uc32>(unibrow::Utf8::kBadChar)) {
- // We allow any escaped character, unlike the restriction on
- // IdentifierPart when it is used to build an IdentifierName.
- AddLiteralChar(c);
- continue;
- }
- }
- AddLiteralCharAdvance();
- }
- literal.Complete();
-
- next_.location.end_pos = source_pos() - 1;
- return true;
-}
} } // namespace v8::internal
namespace v8 {
namespace internal {
-
-class UTF8Buffer {
- public:
- UTF8Buffer();
- ~UTF8Buffer();
-
- inline void AddChar(uc32 c) {
- if (recording_) {
- if (static_cast<unsigned>(c) <= unibrow::Utf8::kMaxOneByteChar) {
- buffer_.Add(static_cast<char>(c));
- } else {
- AddCharSlow(c);
- }
- }
- }
-
- void StartLiteral() {
- buffer_.StartSequence();
- recording_ = true;
- }
-
- Vector<const char> EndLiteral() {
- if (recording_) {
- recording_ = false;
- buffer_.Add(kEndMarker);
- Vector<char> sequence = buffer_.EndSequence();
- return Vector<const char>(sequence.start(), sequence.length());
- }
- return Vector<const char>();
- }
-
- void DropLiteral() {
- if (recording_) {
- recording_ = false;
- buffer_.DropSequence();
- }
- }
-
- void Reset() {
- buffer_.Reset();
- }
-
- // The end marker added after a parsed literal.
- // Using zero allows the usage of strlen and similar functions on
- // identifiers and numbers (but not strings, since they may contain zero
- // bytes).
- // TODO(lrn): Use '\xff' as end marker, since it cannot occur inside
- // an utf-8 string. This requires changes in all places that uses
- // str-functions on the literals, but allows a single pointer to represent
- // the literal, even if it contains embedded zeros.
- static const char kEndMarker = '\x00';
- private:
- static const int kInitialCapacity = 256;
- SequenceCollector<char, 4> buffer_;
- bool recording_;
- void AddCharSlow(uc32 c);
-};
-
-
// UTF16 buffer to read characters from a character stream.
class CharacterStreamUTF16Buffer: public UTF16Buffer {
public:
};
-enum ParserLanguage { JAVASCRIPT, JSON };
-
-
-class Scanner {
+// Initializes a UTF16Buffer as input stream, using one of a number
+// of strategies depending on the available character sources.
+class StreamInitializer {
public:
- typedef unibrow::Utf8InputBuffer<1024> Utf8Decoder;
+ UTF16Buffer* Init(Handle<String> source,
+ unibrow::CharacterStream* stream,
+ int start_position,
+ int end_position);
+ private:
+ // Different UTF16 buffers used to pull characters from. Based on input one of
+ // these will be initialized as the actual data source.
+ CharacterStreamUTF16Buffer char_stream_buffer_;
+ ExternalStringUTF16Buffer<ExternalTwoByteString, uint16_t>
+ two_byte_string_buffer_;
+ ExternalStringUTF16Buffer<ExternalAsciiString, char> ascii_string_buffer_;
- class LiteralScope {
- public:
- explicit LiteralScope(Scanner* self);
- ~LiteralScope();
- void Complete();
+ // Used to convert the source string into a character stream when a stream
+ // is not passed to the scanner.
+ SafeStringInputBuffer safe_string_input_buffer_;
+};
+
+// ----------------------------------------------------------------------------
+// V8JavaScriptScanner
+// JavaScript scanner getting its input from either a V8 String or a unicode
+// CharacterStream.
- private:
- Scanner* scanner_;
- bool complete_;
- };
+class V8JavaScriptScanner : public JavaScriptScanner {
+ public:
+ V8JavaScriptScanner() {}
- Scanner();
+ Token::Value NextCheckStack();
// Initialize the Scanner to scan source.
+ void Initialize(Handle<String> source);
void Initialize(Handle<String> source,
- ParserLanguage language);
- void Initialize(Handle<String> source,
- unibrow::CharacterStream* stream,
- ParserLanguage language);
+ unibrow::CharacterStream* stream);
void Initialize(Handle<String> source,
- int start_position, int end_position,
- ParserLanguage language);
+ int start_position, int end_position);
- // Returns the next token.
- Token::Value Next();
-
- // Returns the current token again.
- Token::Value current_token() { return current_.token; }
-
- // One token look-ahead (past the token returned by Next()).
- Token::Value peek() const { return next_.token; }
-
- // Returns true if there was a line terminator before the peek'ed token.
- bool has_line_terminator_before_next() const {
- return has_line_terminator_before_next_;
- }
-
- struct Location {
- Location(int b, int e) : beg_pos(b), end_pos(e) { }
- Location() : beg_pos(0), end_pos(0) { }
- int beg_pos;
- int end_pos;
- };
-
- // Returns the location information for the current token
- // (the token returned by Next()).
- Location location() const { return current_.location; }
- Location peek_location() const { return next_.location; }
-
- // Returns the literal string, if any, for the current token (the
- // token returned by Next()). The string is 0-terminated and in
- // UTF-8 format; they may contain 0-characters. Literal strings are
- // collected for identifiers, strings, and numbers.
- // These functions only give the correct result if the literal
- // was scanned between calls to StartLiteral() and TerminateLiteral().
- const char* literal_string() const {
- return current_.literal_chars.start();
- }
-
- int literal_length() const {
- // Excluding terminal '\x00' added by TerminateLiteral().
- return current_.literal_chars.length() - 1;
- }
-
- Vector<const char> literal() const {
- return Vector<const char>(literal_string(), literal_length());
- }
-
- // Returns the literal string for the next token (the token that
- // would be returned if Next() were called).
- const char* next_literal_string() const {
- return next_.literal_chars.start();
- }
-
-
- // Returns the length of the next token (that would be returned if
- // Next() were called).
- int next_literal_length() const {
- // Excluding terminal '\x00' added by TerminateLiteral().
- return next_.literal_chars.length() - 1;
- }
-
- Vector<const char> next_literal() const {
- return Vector<const char>(next_literal_string(), next_literal_length());
- }
-
- // Scans the input as a regular expression pattern, previous
- // character(s) must be /(=). Returns true if a pattern is scanned.
- bool ScanRegExpPattern(bool seen_equal);
- // Returns true if regexp flags are scanned (always since flags can
- // be empty).
- bool ScanRegExpFlags();
-
- // Seek forward to the given position. This operation does not
- // work in general, for instance when there are pushed back
- // characters, but works for seeking forward until simple delimiter
- // tokens, which is what it is used for.
- void SeekForward(int pos);
-
- bool stack_overflow() { return stack_overflow_; }
+ protected:
+ StreamInitializer stream_initializer_;
+};
- // Tells whether the buffer contains an identifier (no escapes).
- // Used for checking if a property name is an identifier.
- static bool IsIdentifier(unibrow::CharacterStream* buffer);
- static const int kCharacterLookaheadBufferSize = 1;
- static const int kNoEndPosition = 1;
+class JsonScanner : public Scanner {
+ public:
+ JsonScanner();
- private:
- // The current and look-ahead token.
- struct TokenDesc {
- Token::Value token;
- Location location;
- Vector<const char> literal_chars;
- };
-
- void Init(Handle<String> source,
- unibrow::CharacterStream* stream,
- int start_position, int end_position,
- ParserLanguage language);
-
- // Literal buffer support
- inline void StartLiteral();
- inline void AddLiteralChar(uc32 ch);
- inline void AddLiteralCharAdvance();
- inline void TerminateLiteral();
- // Stops scanning of a literal, e.g., due to an encountered error.
- inline void DropLiteral();
-
- // Low-level scanning support.
- void Advance() { c0_ = source_->Advance(); }
- void PushBack(uc32 ch) {
- source_->PushBack(ch);
- c0_ = ch;
- }
+ // Initialize the Scanner to scan source.
+ void Initialize(Handle<String> source);
- bool SkipWhiteSpace() {
- if (is_parsing_json_) {
- return SkipJsonWhiteSpace();
- } else {
- return SkipJavaScriptWhiteSpace();
- }
- }
+ // Returns the next token.
+ Token::Value Next();
- bool SkipJavaScriptWhiteSpace();
+ protected:
+ // Skip past JSON whitespace (only space, tab, newline and carrige-return).
bool SkipJsonWhiteSpace();
- Token::Value SkipSingleLineComment();
- Token::Value SkipMultiLineComment();
-
- inline Token::Value Select(Token::Value tok);
- inline Token::Value Select(uc32 next, Token::Value then, Token::Value else_);
-
- inline void Scan() {
- if (is_parsing_json_) {
- ScanJson();
- } else {
- ScanJavaScript();
- }
- }
-
- // Scans a single JavaScript token.
- void ScanJavaScript();
// Scan a single JSON token. The JSON lexical grammar is specified in the
// ECMAScript 5 standard, section 15.12.1.1.
// JSONNullLiteral).
Token::Value ScanJsonIdentifier(const char* text, Token::Value token);
- void ScanDecimalDigits();
- Token::Value ScanNumber(bool seen_period);
- Token::Value ScanIdentifier();
- uc32 ScanHexEscape(uc32 c, int length);
- uc32 ScanOctalEscape(uc32 c, int length);
- void ScanEscape();
- Token::Value ScanString();
-
- // Scans a possible HTML comment -- begins with '<!'.
- Token::Value ScanHtmlComment();
-
- // Return the current source position.
- int source_pos() {
- return source_->pos() - kCharacterLookaheadBufferSize;
- }
-
- // Decodes a unicode escape-sequence which is part of an identifier.
- // If the escape sequence cannot be decoded the result is kBadRune.
- uc32 ScanIdentifierUnicodeEscape();
-
- TokenDesc current_; // desc for current token (as returned by Next())
- TokenDesc next_; // desc for next token (one token look-ahead)
- bool has_line_terminator_before_next_;
- bool is_parsing_json_;
-
- // Different UTF16 buffers used to pull characters from. Based on input one of
- // these will be initialized as the actual data source.
- CharacterStreamUTF16Buffer char_stream_buffer_;
- ExternalStringUTF16Buffer<ExternalTwoByteString, uint16_t>
- two_byte_string_buffer_;
- ExternalStringUTF16Buffer<ExternalAsciiString, char> ascii_string_buffer_;
-
- // Source. Will point to one of the buffers declared above.
- UTF16Buffer* source_;
-
- // Used to convert the source string into a character stream when a stream
- // is not passed to the scanner.
- SafeStringInputBuffer safe_string_input_buffer_;
-
- // Buffer to hold literal values (identifiers, strings, numbers)
- // using '\x00'-terminated UTF-8 encoding. Handles allocation internally.
- UTF8Buffer literal_buffer_;
-
- bool stack_overflow_;
-
- // One Unicode character look-ahead; c0_ < 0 at the end of the input.
- uc32 c0_;
+ StreamInitializer stream_initializer_;
};
SeekForward(start_position);
}
end_ =
- end_position != Scanner::kNoEndPosition ? end_position : data->length();
+ end_position != kNoEndPosition ? end_position : data->length();
}
#include "parser.h"
#include "utils.h"
#include "execution.h"
-#include "scanner.h"
#include "preparser.h"
#include "cctest.h"
const char* program = programs[i];
unibrow::Utf8InputBuffer<256> stream(program, strlen(program));
i::CompleteParserRecorder log;
- i::Scanner scanner;
- scanner.Initialize(i::Handle<i::String>::null(), &stream, i::JAVASCRIPT);
- v8::preparser::PreParser<i::Scanner, i::CompleteParserRecorder> preparser;
+ i::V8JavaScriptScanner scanner;
+ scanner.Initialize(i::Handle<i::String>::null(), &stream);
+ v8::preparser::PreParser<i::V8JavaScriptScanner,
+ i::CompleteParserRecorder> preparser;
bool result = preparser.PreParseProgram(&scanner, &log, true);
CHECK(result);
i::ScriptDataImpl data(log.ExtractData());