1 // Copyright 2011 the V8 project authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4 
5 // Features shared by parsing and pre-parsing scanners.
6 
7 #include "src/parsing/scanner.h"
8 
9 #include <stdint.h>
10 
11 #include <cmath>
12 
13 #include "src/ast/ast-value-factory.h"
14 #include "src/char-predicates-inl.h"
15 #include "src/conversions-inl.h"
16 #include "src/list-inl.h"
17 #include "src/parsing/parser.h"
18 
19 namespace v8 {
20 namespace internal {
21 
22 
Internalize(Isolate * isolate) const23 Handle<String> LiteralBuffer::Internalize(Isolate* isolate) const {
24   if (is_one_byte()) {
25     return isolate->factory()->InternalizeOneByteString(one_byte_literal());
26   }
27   return isolate->factory()->InternalizeTwoByteString(two_byte_literal());
28 }
29 
30 
31 // Default implementation for streams that do not support bookmarks.
SetBookmark()32 bool Utf16CharacterStream::SetBookmark() { return false; }
ResetToBookmark()33 void Utf16CharacterStream::ResetToBookmark() { UNREACHABLE(); }
34 
35 
36 // ----------------------------------------------------------------------------
37 // Scanner
38 
Scanner(UnicodeCache * unicode_cache)39 Scanner::Scanner(UnicodeCache* unicode_cache)
40     : unicode_cache_(unicode_cache),
41       bookmark_c0_(kNoBookmark),
42       octal_pos_(Location::invalid()) {
43   bookmark_current_.literal_chars = &bookmark_current_literal_;
44   bookmark_current_.raw_literal_chars = &bookmark_current_raw_literal_;
45   bookmark_next_.literal_chars = &bookmark_next_literal_;
46   bookmark_next_.raw_literal_chars = &bookmark_next_raw_literal_;
47 }
48 
49 
Initialize(Utf16CharacterStream * source)50 void Scanner::Initialize(Utf16CharacterStream* source) {
51   source_ = source;
52   // Need to capture identifiers in order to recognize "get" and "set"
53   // in object literals.
54   Init();
55   // Skip initial whitespace allowing HTML comment ends just like
56   // after a newline and scan first token.
57   has_line_terminator_before_next_ = true;
58   SkipWhiteSpace();
59   Scan();
60 }
61 
62 
63 template <bool capture_raw>
ScanHexNumber(int expected_length)64 uc32 Scanner::ScanHexNumber(int expected_length) {
65   DCHECK(expected_length <= 4);  // prevent overflow
66 
67   uc32 x = 0;
68   for (int i = 0; i < expected_length; i++) {
69     int d = HexValue(c0_);
70     if (d < 0) {
71       return -1;
72     }
73     x = x * 16 + d;
74     Advance<capture_raw>();
75   }
76 
77   return x;
78 }
79 
80 
81 template <bool capture_raw>
ScanUnlimitedLengthHexNumber(int max_value)82 uc32 Scanner::ScanUnlimitedLengthHexNumber(int max_value) {
83   uc32 x = 0;
84   int d = HexValue(c0_);
85   if (d < 0) {
86     return -1;
87   }
88   while (d >= 0) {
89     x = x * 16 + d;
90     if (x > max_value) return -1;
91     Advance<capture_raw>();
92     d = HexValue(c0_);
93   }
94   return x;
95 }
96 
97 
98 // Ensure that tokens can be stored in a byte.
99 STATIC_ASSERT(Token::NUM_TOKENS <= 0x100);
100 
101 // Table of one-character tokens, by character (0x00..0x7f only).
102 static const byte one_char_tokens[] = {
103   Token::ILLEGAL,
104   Token::ILLEGAL,
105   Token::ILLEGAL,
106   Token::ILLEGAL,
107   Token::ILLEGAL,
108   Token::ILLEGAL,
109   Token::ILLEGAL,
110   Token::ILLEGAL,
111   Token::ILLEGAL,
112   Token::ILLEGAL,
113   Token::ILLEGAL,
114   Token::ILLEGAL,
115   Token::ILLEGAL,
116   Token::ILLEGAL,
117   Token::ILLEGAL,
118   Token::ILLEGAL,
119   Token::ILLEGAL,
120   Token::ILLEGAL,
121   Token::ILLEGAL,
122   Token::ILLEGAL,
123   Token::ILLEGAL,
124   Token::ILLEGAL,
125   Token::ILLEGAL,
126   Token::ILLEGAL,
127   Token::ILLEGAL,
128   Token::ILLEGAL,
129   Token::ILLEGAL,
130   Token::ILLEGAL,
131   Token::ILLEGAL,
132   Token::ILLEGAL,
133   Token::ILLEGAL,
134   Token::ILLEGAL,
135   Token::ILLEGAL,
136   Token::ILLEGAL,
137   Token::ILLEGAL,
138   Token::ILLEGAL,
139   Token::ILLEGAL,
140   Token::ILLEGAL,
141   Token::ILLEGAL,
142   Token::ILLEGAL,
143   Token::LPAREN,       // 0x28
144   Token::RPAREN,       // 0x29
145   Token::ILLEGAL,
146   Token::ILLEGAL,
147   Token::COMMA,        // 0x2c
148   Token::ILLEGAL,
149   Token::ILLEGAL,
150   Token::ILLEGAL,
151   Token::ILLEGAL,
152   Token::ILLEGAL,
153   Token::ILLEGAL,
154   Token::ILLEGAL,
155   Token::ILLEGAL,
156   Token::ILLEGAL,
157   Token::ILLEGAL,
158   Token::ILLEGAL,
159   Token::ILLEGAL,
160   Token::ILLEGAL,
161   Token::COLON,        // 0x3a
162   Token::SEMICOLON,    // 0x3b
163   Token::ILLEGAL,
164   Token::ILLEGAL,
165   Token::ILLEGAL,
166   Token::CONDITIONAL,  // 0x3f
167   Token::ILLEGAL,
168   Token::ILLEGAL,
169   Token::ILLEGAL,
170   Token::ILLEGAL,
171   Token::ILLEGAL,
172   Token::ILLEGAL,
173   Token::ILLEGAL,
174   Token::ILLEGAL,
175   Token::ILLEGAL,
176   Token::ILLEGAL,
177   Token::ILLEGAL,
178   Token::ILLEGAL,
179   Token::ILLEGAL,
180   Token::ILLEGAL,
181   Token::ILLEGAL,
182   Token::ILLEGAL,
183   Token::ILLEGAL,
184   Token::ILLEGAL,
185   Token::ILLEGAL,
186   Token::ILLEGAL,
187   Token::ILLEGAL,
188   Token::ILLEGAL,
189   Token::ILLEGAL,
190   Token::ILLEGAL,
191   Token::ILLEGAL,
192   Token::ILLEGAL,
193   Token::ILLEGAL,
194   Token::LBRACK,     // 0x5b
195   Token::ILLEGAL,
196   Token::RBRACK,     // 0x5d
197   Token::ILLEGAL,
198   Token::ILLEGAL,
199   Token::ILLEGAL,
200   Token::ILLEGAL,
201   Token::ILLEGAL,
202   Token::ILLEGAL,
203   Token::ILLEGAL,
204   Token::ILLEGAL,
205   Token::ILLEGAL,
206   Token::ILLEGAL,
207   Token::ILLEGAL,
208   Token::ILLEGAL,
209   Token::ILLEGAL,
210   Token::ILLEGAL,
211   Token::ILLEGAL,
212   Token::ILLEGAL,
213   Token::ILLEGAL,
214   Token::ILLEGAL,
215   Token::ILLEGAL,
216   Token::ILLEGAL,
217   Token::ILLEGAL,
218   Token::ILLEGAL,
219   Token::ILLEGAL,
220   Token::ILLEGAL,
221   Token::ILLEGAL,
222   Token::ILLEGAL,
223   Token::ILLEGAL,
224   Token::ILLEGAL,
225   Token::ILLEGAL,
226   Token::LBRACE,       // 0x7b
227   Token::ILLEGAL,
228   Token::RBRACE,       // 0x7d
229   Token::BIT_NOT,      // 0x7e
230   Token::ILLEGAL
231 };
232 
233 
Next()234 Token::Value Scanner::Next() {
235   if (next_.token == Token::EOS) {
236     next_.location.beg_pos = current_.location.beg_pos;
237     next_.location.end_pos = current_.location.end_pos;
238   }
239   current_ = next_;
240   if (V8_UNLIKELY(next_next_.token != Token::UNINITIALIZED)) {
241     next_ = next_next_;
242     next_next_.token = Token::UNINITIALIZED;
243     return current_.token;
244   }
245   has_line_terminator_before_next_ = false;
246   has_multiline_comment_before_next_ = false;
247   if (static_cast<unsigned>(c0_) <= 0x7f) {
248     Token::Value token = static_cast<Token::Value>(one_char_tokens[c0_]);
249     if (token != Token::ILLEGAL) {
250       int pos = source_pos();
251       next_.token = token;
252       next_.location.beg_pos = pos;
253       next_.location.end_pos = pos + 1;
254       Advance();
255       return current_.token;
256     }
257   }
258   Scan();
259   return current_.token;
260 }
261 
262 
PeekAhead()263 Token::Value Scanner::PeekAhead() {
264   if (next_next_.token != Token::UNINITIALIZED) {
265     return next_next_.token;
266   }
267   TokenDesc prev = current_;
268   Next();
269   Token::Value ret = next_.token;
270   next_next_ = next_;
271   next_ = current_;
272   current_ = prev;
273   return ret;
274 }
275 
276 
277 // TODO(yangguo): check whether this is actually necessary.
IsLittleEndianByteOrderMark(uc32 c)278 static inline bool IsLittleEndianByteOrderMark(uc32 c) {
279   // The Unicode value U+FFFE is guaranteed never to be assigned as a
280   // Unicode character; this implies that in a Unicode context the
281   // 0xFF, 0xFE byte pattern can only be interpreted as the U+FEFF
282   // character expressed in little-endian byte order (since it could
283   // not be a U+FFFE character expressed in big-endian byte
284   // order). Nevertheless, we check for it to be compatible with
285   // Spidermonkey.
286   return c == 0xFFFE;
287 }
288 
289 
SkipWhiteSpace()290 bool Scanner::SkipWhiteSpace() {
291   int start_position = source_pos();
292 
293   while (true) {
294     while (true) {
295       // The unicode cache accepts unsigned inputs.
296       if (c0_ < 0) break;
297       // Advance as long as character is a WhiteSpace or LineTerminator.
298       // Remember if the latter is the case.
299       if (unicode_cache_->IsLineTerminator(c0_)) {
300         has_line_terminator_before_next_ = true;
301       } else if (!unicode_cache_->IsWhiteSpace(c0_) &&
302                  !IsLittleEndianByteOrderMark(c0_)) {
303         break;
304       }
305       Advance();
306     }
307 
308     // If there is an HTML comment end '-->' at the beginning of a
309     // line (with only whitespace in front of it), we treat the rest
310     // of the line as a comment. This is in line with the way
311     // SpiderMonkey handles it.
312     if (c0_ == '-' && has_line_terminator_before_next_) {
313       Advance();
314       if (c0_ == '-') {
315         Advance();
316         if (c0_ == '>') {
317           // Treat the rest of the line as a comment.
318           SkipSingleLineComment();
319           // Continue skipping white space after the comment.
320           continue;
321         }
322         PushBack('-');  // undo Advance()
323       }
324       PushBack('-');  // undo Advance()
325     }
326     // Return whether or not we skipped any characters.
327     return source_pos() != start_position;
328   }
329 }
330 
331 
SkipSingleLineComment()332 Token::Value Scanner::SkipSingleLineComment() {
333   Advance();
334 
335   // The line terminator at the end of the line is not considered
336   // to be part of the single-line comment; it is recognized
337   // separately by the lexical grammar and becomes part of the
338   // stream of input elements for the syntactic grammar (see
339   // ECMA-262, section 7.4).
340   while (c0_ >= 0 && !unicode_cache_->IsLineTerminator(c0_)) {
341     Advance();
342   }
343 
344   return Token::WHITESPACE;
345 }
346 
347 
SkipSourceURLComment()348 Token::Value Scanner::SkipSourceURLComment() {
349   TryToParseSourceURLComment();
350   while (c0_ >= 0 && !unicode_cache_->IsLineTerminator(c0_)) {
351     Advance();
352   }
353 
354   return Token::WHITESPACE;
355 }
356 
357 
TryToParseSourceURLComment()358 void Scanner::TryToParseSourceURLComment() {
359   // Magic comments are of the form: //[#@]\s<name>=\s*<value>\s*.* and this
360   // function will just return if it cannot parse a magic comment.
361   if (c0_ < 0 || !unicode_cache_->IsWhiteSpace(c0_)) return;
362   Advance();
363   LiteralBuffer name;
364   while (c0_ >= 0 && !unicode_cache_->IsWhiteSpaceOrLineTerminator(c0_) &&
365          c0_ != '=') {
366     name.AddChar(c0_);
367     Advance();
368   }
369   if (!name.is_one_byte()) return;
370   Vector<const uint8_t> name_literal = name.one_byte_literal();
371   LiteralBuffer* value;
372   if (name_literal == STATIC_CHAR_VECTOR("sourceURL")) {
373     value = &source_url_;
374   } else if (name_literal == STATIC_CHAR_VECTOR("sourceMappingURL")) {
375     value = &source_mapping_url_;
376   } else {
377     return;
378   }
379   if (c0_ != '=')
380     return;
381   Advance();
382   value->Reset();
383   while (c0_ >= 0 && unicode_cache_->IsWhiteSpace(c0_)) {
384     Advance();
385   }
386   while (c0_ >= 0 && !unicode_cache_->IsLineTerminator(c0_)) {
387     // Disallowed characters.
388     if (c0_ == '"' || c0_ == '\'') {
389       value->Reset();
390       return;
391     }
392     if (unicode_cache_->IsWhiteSpace(c0_)) {
393       break;
394     }
395     value->AddChar(c0_);
396     Advance();
397   }
398   // Allow whitespace at the end.
399   while (c0_ >= 0 && !unicode_cache_->IsLineTerminator(c0_)) {
400     if (!unicode_cache_->IsWhiteSpace(c0_)) {
401       value->Reset();
402       break;
403     }
404     Advance();
405   }
406 }
407 
408 
SkipMultiLineComment()409 Token::Value Scanner::SkipMultiLineComment() {
410   DCHECK(c0_ == '*');
411   Advance();
412 
413   while (c0_ >= 0) {
414     uc32 ch = c0_;
415     Advance();
416     if (c0_ >= 0 && unicode_cache_->IsLineTerminator(ch)) {
417       // Following ECMA-262, section 7.4, a comment containing
418       // a newline will make the comment count as a line-terminator.
419       has_multiline_comment_before_next_ = true;
420     }
421     // If we have reached the end of the multi-line comment, we
422     // consume the '/' and insert a whitespace. This way all
423     // multi-line comments are treated as whitespace.
424     if (ch == '*' && c0_ == '/') {
425       c0_ = ' ';
426       return Token::WHITESPACE;
427     }
428   }
429 
430   // Unterminated multi-line comment.
431   return Token::ILLEGAL;
432 }
433 
434 
ScanHtmlComment()435 Token::Value Scanner::ScanHtmlComment() {
436   // Check for <!-- comments.
437   DCHECK(c0_ == '!');
438   Advance();
439   if (c0_ == '-') {
440     Advance();
441     if (c0_ == '-') return SkipSingleLineComment();
442     PushBack('-');  // undo Advance()
443   }
444   PushBack('!');  // undo Advance()
445   DCHECK(c0_ == '!');
446   return Token::LT;
447 }
448 
449 
Scan()450 void Scanner::Scan() {
451   next_.literal_chars = NULL;
452   next_.raw_literal_chars = NULL;
453   Token::Value token;
454   do {
455     // Remember the position of the next token
456     next_.location.beg_pos = source_pos();
457 
458     switch (c0_) {
459       case ' ':
460       case '\t':
461         Advance();
462         token = Token::WHITESPACE;
463         break;
464 
465       case '\n':
466         Advance();
467         has_line_terminator_before_next_ = true;
468         token = Token::WHITESPACE;
469         break;
470 
471       case '"': case '\'':
472         token = ScanString();
473         break;
474 
475       case '<':
476         // < <= << <<= <!--
477         Advance();
478         if (c0_ == '=') {
479           token = Select(Token::LTE);
480         } else if (c0_ == '<') {
481           token = Select('=', Token::ASSIGN_SHL, Token::SHL);
482         } else if (c0_ == '!') {
483           token = ScanHtmlComment();
484         } else {
485           token = Token::LT;
486         }
487         break;
488 
489       case '>':
490         // > >= >> >>= >>> >>>=
491         Advance();
492         if (c0_ == '=') {
493           token = Select(Token::GTE);
494         } else if (c0_ == '>') {
495           // >> >>= >>> >>>=
496           Advance();
497           if (c0_ == '=') {
498             token = Select(Token::ASSIGN_SAR);
499           } else if (c0_ == '>') {
500             token = Select('=', Token::ASSIGN_SHR, Token::SHR);
501           } else {
502             token = Token::SAR;
503           }
504         } else {
505           token = Token::GT;
506         }
507         break;
508 
509       case '=':
510         // = == === =>
511         Advance();
512         if (c0_ == '=') {
513           token = Select('=', Token::EQ_STRICT, Token::EQ);
514         } else if (c0_ == '>') {
515           token = Select(Token::ARROW);
516         } else {
517           token = Token::ASSIGN;
518         }
519         break;
520 
521       case '!':
522         // ! != !==
523         Advance();
524         if (c0_ == '=') {
525           token = Select('=', Token::NE_STRICT, Token::NE);
526         } else {
527           token = Token::NOT;
528         }
529         break;
530 
531       case '+':
532         // + ++ +=
533         Advance();
534         if (c0_ == '+') {
535           token = Select(Token::INC);
536         } else if (c0_ == '=') {
537           token = Select(Token::ASSIGN_ADD);
538         } else {
539           token = Token::ADD;
540         }
541         break;
542 
543       case '-':
544         // - -- --> -=
545         Advance();
546         if (c0_ == '-') {
547           Advance();
548           if (c0_ == '>' && has_line_terminator_before_next_) {
549             // For compatibility with SpiderMonkey, we skip lines that
550             // start with an HTML comment end '-->'.
551             token = SkipSingleLineComment();
552           } else {
553             token = Token::DEC;
554           }
555         } else if (c0_ == '=') {
556           token = Select(Token::ASSIGN_SUB);
557         } else {
558           token = Token::SUB;
559         }
560         break;
561 
562       case '*':
563         // * *=
564         token = Select('=', Token::ASSIGN_MUL, Token::MUL);
565         break;
566 
567       case '%':
568         // % %=
569         token = Select('=', Token::ASSIGN_MOD, Token::MOD);
570         break;
571 
572       case '/':
573         // /  // /* /=
574         Advance();
575         if (c0_ == '/') {
576           Advance();
577           if (c0_ == '#' || c0_ == '@') {
578             Advance();
579             token = SkipSourceURLComment();
580           } else {
581             PushBack(c0_);
582             token = SkipSingleLineComment();
583           }
584         } else if (c0_ == '*') {
585           token = SkipMultiLineComment();
586         } else if (c0_ == '=') {
587           token = Select(Token::ASSIGN_DIV);
588         } else {
589           token = Token::DIV;
590         }
591         break;
592 
593       case '&':
594         // & && &=
595         Advance();
596         if (c0_ == '&') {
597           token = Select(Token::AND);
598         } else if (c0_ == '=') {
599           token = Select(Token::ASSIGN_BIT_AND);
600         } else {
601           token = Token::BIT_AND;
602         }
603         break;
604 
605       case '|':
606         // | || |=
607         Advance();
608         if (c0_ == '|') {
609           token = Select(Token::OR);
610         } else if (c0_ == '=') {
611           token = Select(Token::ASSIGN_BIT_OR);
612         } else {
613           token = Token::BIT_OR;
614         }
615         break;
616 
617       case '^':
618         // ^ ^=
619         token = Select('=', Token::ASSIGN_BIT_XOR, Token::BIT_XOR);
620         break;
621 
622       case '.':
623         // . Number
624         Advance();
625         if (IsDecimalDigit(c0_)) {
626           token = ScanNumber(true);
627         } else {
628           token = Token::PERIOD;
629           if (c0_ == '.') {
630             Advance();
631             if (c0_ == '.') {
632               Advance();
633               token = Token::ELLIPSIS;
634             } else {
635               PushBack('.');
636             }
637           }
638         }
639         break;
640 
641       case ':':
642         token = Select(Token::COLON);
643         break;
644 
645       case ';':
646         token = Select(Token::SEMICOLON);
647         break;
648 
649       case ',':
650         token = Select(Token::COMMA);
651         break;
652 
653       case '(':
654         token = Select(Token::LPAREN);
655         break;
656 
657       case ')':
658         token = Select(Token::RPAREN);
659         break;
660 
661       case '[':
662         token = Select(Token::LBRACK);
663         break;
664 
665       case ']':
666         token = Select(Token::RBRACK);
667         break;
668 
669       case '{':
670         token = Select(Token::LBRACE);
671         break;
672 
673       case '}':
674         token = Select(Token::RBRACE);
675         break;
676 
677       case '?':
678         token = Select(Token::CONDITIONAL);
679         break;
680 
681       case '~':
682         token = Select(Token::BIT_NOT);
683         break;
684 
685       case '`':
686         token = ScanTemplateStart();
687         break;
688 
689       default:
690         if (c0_ < 0) {
691           token = Token::EOS;
692         } else if (unicode_cache_->IsIdentifierStart(c0_)) {
693           token = ScanIdentifierOrKeyword();
694         } else if (IsDecimalDigit(c0_)) {
695           token = ScanNumber(false);
696         } else if (SkipWhiteSpace()) {
697           token = Token::WHITESPACE;
698         } else {
699           token = Select(Token::ILLEGAL);
700         }
701         break;
702     }
703 
704     // Continue scanning for tokens as long as we're just skipping
705     // whitespace.
706   } while (token == Token::WHITESPACE);
707 
708   next_.location.end_pos = source_pos();
709   next_.token = token;
710 }
711 
712 
SeekForward(int pos)713 void Scanner::SeekForward(int pos) {
714   // After this call, we will have the token at the given position as
715   // the "next" token. The "current" token will be invalid.
716   if (pos == next_.location.beg_pos) return;
717   int current_pos = source_pos();
718   DCHECK_EQ(next_.location.end_pos, current_pos);
719   // Positions inside the lookahead token aren't supported.
720   DCHECK(pos >= current_pos);
721   if (pos != current_pos) {
722     source_->SeekForward(pos - source_->pos());
723     Advance();
724     // This function is only called to seek to the location
725     // of the end of a function (at the "}" token). It doesn't matter
726     // whether there was a line terminator in the part we skip.
727     has_line_terminator_before_next_ = false;
728     has_multiline_comment_before_next_ = false;
729   }
730   Scan();
731 }
732 
733 
734 template <bool capture_raw, bool in_template_literal>
ScanEscape()735 bool Scanner::ScanEscape() {
736   uc32 c = c0_;
737   Advance<capture_raw>();
738 
739   // Skip escaped newlines.
740   if (!in_template_literal && c0_ >= 0 && unicode_cache_->IsLineTerminator(c)) {
741     // Allow CR+LF newlines in multiline string literals.
742     if (IsCarriageReturn(c) && IsLineFeed(c0_)) Advance<capture_raw>();
743     // Allow LF+CR newlines in multiline string literals.
744     if (IsLineFeed(c) && IsCarriageReturn(c0_)) Advance<capture_raw>();
745     return true;
746   }
747 
748   switch (c) {
749     case '\'':  // fall through
750     case '"' :  // fall through
751     case '\\': break;
752     case 'b' : c = '\b'; break;
753     case 'f' : c = '\f'; break;
754     case 'n' : c = '\n'; break;
755     case 'r' : c = '\r'; break;
756     case 't' : c = '\t'; break;
757     case 'u' : {
758       c = ScanUnicodeEscape<capture_raw>();
759       if (c < 0) return false;
760       break;
761     }
762     case 'v':
763       c = '\v';
764       break;
765     case 'x': {
766       c = ScanHexNumber<capture_raw>(2);
767       if (c < 0) return false;
768       break;
769     }
770     case '0':  // Fall through.
771     case '1':  // fall through
772     case '2':  // fall through
773     case '3':  // fall through
774     case '4':  // fall through
775     case '5':  // fall through
776     case '6':  // fall through
777     case '7':
778       c = ScanOctalEscape<capture_raw>(c, 2);
779       break;
780   }
781 
782   // According to ECMA-262, section 7.8.4, characters not covered by the
783   // above cases should be illegal, but they are commonly handled as
784   // non-escaped characters by JS VMs.
785   AddLiteralChar(c);
786   return true;
787 }
788 
789 
790 // Octal escapes of the forms '\0xx' and '\xxx' are not a part of
791 // ECMA-262. Other JS VMs support them.
792 template <bool capture_raw>
ScanOctalEscape(uc32 c,int length)793 uc32 Scanner::ScanOctalEscape(uc32 c, int length) {
794   uc32 x = c - '0';
795   int i = 0;
796   for (; i < length; i++) {
797     int d = c0_ - '0';
798     if (d < 0 || d > 7) break;
799     int nx = x * 8 + d;
800     if (nx >= 256) break;
801     x = nx;
802     Advance<capture_raw>();
803   }
804   // Anything except '\0' is an octal escape sequence, illegal in strict mode.
805   // Remember the position of octal escape sequences so that an error
806   // can be reported later (in strict mode).
807   // We don't report the error immediately, because the octal escape can
808   // occur before the "use strict" directive.
809   if (c != '0' || i > 0) {
810     octal_pos_ = Location(source_pos() - i - 1, source_pos() - 1);
811   }
812   return x;
813 }
814 
815 
816 const int kMaxAscii = 127;
817 
818 
ScanString()819 Token::Value Scanner::ScanString() {
820   uc32 quote = c0_;
821   Advance<false, false>();  // consume quote
822 
823   LiteralScope literal(this);
824   while (true) {
825     if (c0_ > kMaxAscii) {
826       HandleLeadSurrogate();
827       break;
828     }
829     if (c0_ < 0 || c0_ == '\n' || c0_ == '\r') return Token::ILLEGAL;
830     if (c0_ == quote) {
831       literal.Complete();
832       Advance<false, false>();
833       return Token::STRING;
834     }
835     uc32 c = c0_;
836     if (c == '\\') break;
837     Advance<false, false>();
838     AddLiteralChar(c);
839   }
840 
841   while (c0_ != quote && c0_ >= 0
842          && !unicode_cache_->IsLineTerminator(c0_)) {
843     uc32 c = c0_;
844     Advance();
845     if (c == '\\') {
846       if (c0_ < 0 || !ScanEscape<false, false>()) return Token::ILLEGAL;
847     } else {
848       AddLiteralChar(c);
849     }
850   }
851   if (c0_ != quote) return Token::ILLEGAL;
852   literal.Complete();
853 
854   Advance();  // consume quote
855   return Token::STRING;
856 }
857 
858 
ScanTemplateSpan()859 Token::Value Scanner::ScanTemplateSpan() {
860   // When scanning a TemplateSpan, we are looking for the following construct:
861   // TEMPLATE_SPAN ::
862   //     ` LiteralChars* ${
863   //   | } LiteralChars* ${
864   //
865   // TEMPLATE_TAIL ::
866   //     ` LiteralChars* `
867   //   | } LiteralChar* `
868   //
869   // A TEMPLATE_SPAN should always be followed by an Expression, while a
870   // TEMPLATE_TAIL terminates a TemplateLiteral and does not need to be
871   // followed by an Expression.
872 
873   Token::Value result = Token::TEMPLATE_SPAN;
874   LiteralScope literal(this);
875   StartRawLiteral();
876   const bool capture_raw = true;
877   const bool in_template_literal = true;
878 
879   while (true) {
880     uc32 c = c0_;
881     Advance<capture_raw>();
882     if (c == '`') {
883       result = Token::TEMPLATE_TAIL;
884       ReduceRawLiteralLength(1);
885       break;
886     } else if (c == '$' && c0_ == '{') {
887       Advance<capture_raw>();  // Consume '{'
888       ReduceRawLiteralLength(2);
889       break;
890     } else if (c == '\\') {
891       if (c0_ > 0 && unicode_cache_->IsLineTerminator(c0_)) {
892         // The TV of LineContinuation :: \ LineTerminatorSequence is the empty
893         // code unit sequence.
894         uc32 lastChar = c0_;
895         Advance<capture_raw>();
896         if (lastChar == '\r') {
897           ReduceRawLiteralLength(1);  // Remove \r
898           if (c0_ == '\n') {
899             Advance<capture_raw>();  // Adds \n
900           } else {
901             AddRawLiteralChar('\n');
902           }
903         }
904       } else if (!ScanEscape<capture_raw, in_template_literal>()) {
905         return Token::ILLEGAL;
906       }
907     } else if (c < 0) {
908       // Unterminated template literal
909       PushBack(c);
910       break;
911     } else {
912       // The TRV of LineTerminatorSequence :: <CR> is the CV 0x000A.
913       // The TRV of LineTerminatorSequence :: <CR><LF> is the sequence
914       // consisting of the CV 0x000A.
915       if (c == '\r') {
916         ReduceRawLiteralLength(1);  // Remove \r
917         if (c0_ == '\n') {
918           Advance<capture_raw>();  // Adds \n
919         } else {
920           AddRawLiteralChar('\n');
921         }
922         c = '\n';
923       }
924       AddLiteralChar(c);
925     }
926   }
927   literal.Complete();
928   next_.location.end_pos = source_pos();
929   next_.token = result;
930   return result;
931 }
932 
933 
ScanTemplateStart()934 Token::Value Scanner::ScanTemplateStart() {
935   DCHECK(c0_ == '`');
936   next_.location.beg_pos = source_pos();
937   Advance();  // Consume `
938   return ScanTemplateSpan();
939 }
940 
941 
ScanTemplateContinuation()942 Token::Value Scanner::ScanTemplateContinuation() {
943   DCHECK_EQ(next_.token, Token::RBRACE);
944   next_.location.beg_pos = source_pos() - 1;  // We already consumed }
945   return ScanTemplateSpan();
946 }
947 
948 
ScanDecimalDigits()949 void Scanner::ScanDecimalDigits() {
950   while (IsDecimalDigit(c0_))
951     AddLiteralCharAdvance();
952 }
953 
954 
ScanNumber(bool seen_period)955 Token::Value Scanner::ScanNumber(bool seen_period) {
956   DCHECK(IsDecimalDigit(c0_));  // the first digit of the number or the fraction
957 
958   enum { DECIMAL, HEX, OCTAL, IMPLICIT_OCTAL, BINARY } kind = DECIMAL;
959 
960   LiteralScope literal(this);
961   bool at_start = !seen_period;
962   if (seen_period) {
963     // we have already seen a decimal point of the float
964     AddLiteralChar('.');
965     ScanDecimalDigits();  // we know we have at least one digit
966 
967   } else {
968     // if the first character is '0' we must check for octals and hex
969     if (c0_ == '0') {
970       int start_pos = source_pos();  // For reporting octal positions.
971       AddLiteralCharAdvance();
972 
973       // either 0, 0exxx, 0Exxx, 0.xxx, a hex number, a binary number or
974       // an octal number.
975       if (c0_ == 'x' || c0_ == 'X') {
976         // hex number
977         kind = HEX;
978         AddLiteralCharAdvance();
979         if (!IsHexDigit(c0_)) {
980           // we must have at least one hex digit after 'x'/'X'
981           return Token::ILLEGAL;
982         }
983         while (IsHexDigit(c0_)) {
984           AddLiteralCharAdvance();
985         }
986       } else if (c0_ == 'o' || c0_ == 'O') {
987         kind = OCTAL;
988         AddLiteralCharAdvance();
989         if (!IsOctalDigit(c0_)) {
990           // we must have at least one octal digit after 'o'/'O'
991           return Token::ILLEGAL;
992         }
993         while (IsOctalDigit(c0_)) {
994           AddLiteralCharAdvance();
995         }
996       } else if (c0_ == 'b' || c0_ == 'B') {
997         kind = BINARY;
998         AddLiteralCharAdvance();
999         if (!IsBinaryDigit(c0_)) {
1000           // we must have at least one binary digit after 'b'/'B'
1001           return Token::ILLEGAL;
1002         }
1003         while (IsBinaryDigit(c0_)) {
1004           AddLiteralCharAdvance();
1005         }
1006       } else if ('0' <= c0_ && c0_ <= '7') {
1007         // (possible) octal number
1008         kind = IMPLICIT_OCTAL;
1009         while (true) {
1010           if (c0_ == '8' || c0_ == '9') {
1011             at_start = false;
1012             kind = DECIMAL;
1013             break;
1014           }
1015           if (c0_  < '0' || '7'  < c0_) {
1016             // Octal literal finished.
1017             octal_pos_ = Location(start_pos, source_pos());
1018             break;
1019           }
1020           AddLiteralCharAdvance();
1021         }
1022       }
1023     }
1024 
1025     // Parse decimal digits and allow trailing fractional part.
1026     if (kind == DECIMAL) {
1027       if (at_start) {
1028         uint64_t value = 0;
1029         while (IsDecimalDigit(c0_)) {
1030           value = 10 * value + (c0_ - '0');
1031 
1032           uc32 first_char = c0_;
1033           Advance<false, false>();
1034           AddLiteralChar(first_char);
1035         }
1036 
1037         if (next_.literal_chars->one_byte_literal().length() <= 10 &&
1038             value <= Smi::kMaxValue && c0_ != '.' && c0_ != 'e' && c0_ != 'E') {
1039           next_.smi_value_ = static_cast<int>(value);
1040           literal.Complete();
1041           HandleLeadSurrogate();
1042 
1043           return Token::SMI;
1044         }
1045         HandleLeadSurrogate();
1046       }
1047 
1048       ScanDecimalDigits();  // optional
1049       if (c0_ == '.') {
1050         AddLiteralCharAdvance();
1051         ScanDecimalDigits();  // optional
1052       }
1053     }
1054   }
1055 
1056   // scan exponent, if any
1057   if (c0_ == 'e' || c0_ == 'E') {
1058     DCHECK(kind != HEX);  // 'e'/'E' must be scanned as part of the hex number
1059     if (kind != DECIMAL) return Token::ILLEGAL;
1060     // scan exponent
1061     AddLiteralCharAdvance();
1062     if (c0_ == '+' || c0_ == '-')
1063       AddLiteralCharAdvance();
1064     if (!IsDecimalDigit(c0_)) {
1065       // we must have at least one decimal digit after 'e'/'E'
1066       return Token::ILLEGAL;
1067     }
1068     ScanDecimalDigits();
1069   }
1070 
1071   // The source character immediately following a numeric literal must
1072   // not be an identifier start or a decimal digit; see ECMA-262
1073   // section 7.8.3, page 17 (note that we read only one decimal digit
1074   // if the value is 0).
1075   if (IsDecimalDigit(c0_) ||
1076       (c0_ >= 0 && unicode_cache_->IsIdentifierStart(c0_)))
1077     return Token::ILLEGAL;
1078 
1079   literal.Complete();
1080 
1081   return Token::NUMBER;
1082 }
1083 
1084 
ScanIdentifierUnicodeEscape()1085 uc32 Scanner::ScanIdentifierUnicodeEscape() {
1086   Advance();
1087   if (c0_ != 'u') return -1;
1088   Advance();
1089   return ScanUnicodeEscape<false>();
1090 }
1091 
1092 
1093 template <bool capture_raw>
ScanUnicodeEscape()1094 uc32 Scanner::ScanUnicodeEscape() {
1095   // Accept both \uxxxx and \u{xxxxxx}. In the latter case, the number of
1096   // hex digits between { } is arbitrary. \ and u have already been read.
1097   if (c0_ == '{') {
1098     Advance<capture_raw>();
1099     uc32 cp = ScanUnlimitedLengthHexNumber<capture_raw>(0x10ffff);
1100     if (cp < 0) {
1101       return -1;
1102     }
1103     if (c0_ != '}') {
1104       return -1;
1105     }
1106     Advance<capture_raw>();
1107     return cp;
1108   }
1109   return ScanHexNumber<capture_raw>(4);
1110 }
1111 
1112 
1113 // ----------------------------------------------------------------------------
1114 // Keyword Matcher
1115 
1116 #define KEYWORDS(KEYWORD_GROUP, KEYWORD)                    \
1117   KEYWORD_GROUP('b')                                        \
1118   KEYWORD("break", Token::BREAK)                            \
1119   KEYWORD_GROUP('c')                                        \
1120   KEYWORD("case", Token::CASE)                              \
1121   KEYWORD("catch", Token::CATCH)                            \
1122   KEYWORD("class", Token::CLASS)                            \
1123   KEYWORD("const", Token::CONST)                            \
1124   KEYWORD("continue", Token::CONTINUE)                      \
1125   KEYWORD_GROUP('d')                                        \
1126   KEYWORD("debugger", Token::DEBUGGER)                      \
1127   KEYWORD("default", Token::DEFAULT)                        \
1128   KEYWORD("delete", Token::DELETE)                          \
1129   KEYWORD("do", Token::DO)                                  \
1130   KEYWORD_GROUP('e')                                        \
1131   KEYWORD("else", Token::ELSE)                              \
1132   KEYWORD("enum", Token::FUTURE_RESERVED_WORD)              \
1133   KEYWORD("export", Token::EXPORT)                          \
1134   KEYWORD("extends", Token::EXTENDS)                        \
1135   KEYWORD_GROUP('f')                                        \
1136   KEYWORD("false", Token::FALSE_LITERAL)                    \
1137   KEYWORD("finally", Token::FINALLY)                        \
1138   KEYWORD("for", Token::FOR)                                \
1139   KEYWORD("function", Token::FUNCTION)                      \
1140   KEYWORD_GROUP('i')                                        \
1141   KEYWORD("if", Token::IF)                                  \
1142   KEYWORD("implements", Token::FUTURE_STRICT_RESERVED_WORD) \
1143   KEYWORD("import", Token::IMPORT)                          \
1144   KEYWORD("in", Token::IN)                                  \
1145   KEYWORD("instanceof", Token::INSTANCEOF)                  \
1146   KEYWORD("interface", Token::FUTURE_STRICT_RESERVED_WORD)  \
1147   KEYWORD_GROUP('l')                                        \
1148   KEYWORD("let", Token::LET)                                \
1149   KEYWORD_GROUP('n')                                        \
1150   KEYWORD("new", Token::NEW)                                \
1151   KEYWORD("null", Token::NULL_LITERAL)                      \
1152   KEYWORD_GROUP('p')                                        \
1153   KEYWORD("package", Token::FUTURE_STRICT_RESERVED_WORD)    \
1154   KEYWORD("private", Token::FUTURE_STRICT_RESERVED_WORD)    \
1155   KEYWORD("protected", Token::FUTURE_STRICT_RESERVED_WORD)  \
1156   KEYWORD("public", Token::FUTURE_STRICT_RESERVED_WORD)     \
1157   KEYWORD_GROUP('r')                                        \
1158   KEYWORD("return", Token::RETURN)                          \
1159   KEYWORD_GROUP('s')                                        \
1160   KEYWORD("static", Token::STATIC)                          \
1161   KEYWORD("super", Token::SUPER)                            \
1162   KEYWORD("switch", Token::SWITCH)                          \
1163   KEYWORD_GROUP('t')                                        \
1164   KEYWORD("this", Token::THIS)                              \
1165   KEYWORD("throw", Token::THROW)                            \
1166   KEYWORD("true", Token::TRUE_LITERAL)                      \
1167   KEYWORD("try", Token::TRY)                                \
1168   KEYWORD("typeof", Token::TYPEOF)                          \
1169   KEYWORD_GROUP('v')                                        \
1170   KEYWORD("var", Token::VAR)                                \
1171   KEYWORD("void", Token::VOID)                              \
1172   KEYWORD_GROUP('w')                                        \
1173   KEYWORD("while", Token::WHILE)                            \
1174   KEYWORD("with", Token::WITH)                              \
1175   KEYWORD_GROUP('y')                                        \
1176   KEYWORD("yield", Token::YIELD)
1177 
1178 
KeywordOrIdentifierToken(const uint8_t * input,int input_length,bool escaped)1179 static Token::Value KeywordOrIdentifierToken(const uint8_t* input,
1180                                              int input_length, bool escaped) {
1181   DCHECK(input_length >= 1);
1182   const int kMinLength = 2;
1183   const int kMaxLength = 10;
1184   if (input_length < kMinLength || input_length > kMaxLength) {
1185     return Token::IDENTIFIER;
1186   }
1187   switch (input[0]) {
1188     default:
1189 #define KEYWORD_GROUP_CASE(ch)                                \
1190       break;                                                  \
1191     case ch:
1192 #define KEYWORD(keyword, token)                                     \
1193   {                                                                 \
1194     /* 'keyword' is a char array, so sizeof(keyword) is */          \
1195     /* strlen(keyword) plus 1 for the NUL char. */                  \
1196     const int keyword_length = sizeof(keyword) - 1;                 \
1197     STATIC_ASSERT(keyword_length >= kMinLength);                    \
1198     STATIC_ASSERT(keyword_length <= kMaxLength);                    \
1199     if (input_length == keyword_length && input[1] == keyword[1] && \
1200         (keyword_length <= 2 || input[2] == keyword[2]) &&          \
1201         (keyword_length <= 3 || input[3] == keyword[3]) &&          \
1202         (keyword_length <= 4 || input[4] == keyword[4]) &&          \
1203         (keyword_length <= 5 || input[5] == keyword[5]) &&          \
1204         (keyword_length <= 6 || input[6] == keyword[6]) &&          \
1205         (keyword_length <= 7 || input[7] == keyword[7]) &&          \
1206         (keyword_length <= 8 || input[8] == keyword[8]) &&          \
1207         (keyword_length <= 9 || input[9] == keyword[9])) {          \
1208       if (escaped) {                                                \
1209         return token == Token::FUTURE_STRICT_RESERVED_WORD          \
1210                    ? Token::ESCAPED_STRICT_RESERVED_WORD            \
1211                    : Token::ESCAPED_KEYWORD;                        \
1212       }                                                             \
1213       return token;                                                 \
1214     }                                                               \
1215   }
1216     KEYWORDS(KEYWORD_GROUP_CASE, KEYWORD)
1217   }
1218   return Token::IDENTIFIER;
1219 }
1220 
1221 
IdentifierIsFutureStrictReserved(const AstRawString * string) const1222 bool Scanner::IdentifierIsFutureStrictReserved(
1223     const AstRawString* string) const {
1224   // Keywords are always 1-byte strings.
1225   if (!string->is_one_byte()) return false;
1226   if (string->IsOneByteEqualTo("let") || string->IsOneByteEqualTo("static") ||
1227       string->IsOneByteEqualTo("yield")) {
1228     return true;
1229   }
1230   return Token::FUTURE_STRICT_RESERVED_WORD ==
1231          KeywordOrIdentifierToken(string->raw_data(), string->length(), false);
1232 }
1233 
1234 
ScanIdentifierOrKeyword()1235 Token::Value Scanner::ScanIdentifierOrKeyword() {
1236   DCHECK(unicode_cache_->IsIdentifierStart(c0_));
1237   LiteralScope literal(this);
1238   if (IsInRange(c0_, 'a', 'z')) {
1239     do {
1240       uc32 first_char = c0_;
1241       Advance<false, false>();
1242       AddLiteralChar(first_char);
1243     } while (IsInRange(c0_, 'a', 'z'));
1244 
1245     if (IsDecimalDigit(c0_) || IsInRange(c0_, 'A', 'Z') || c0_ == '_' ||
1246         c0_ == '$') {
1247       // Identifier starting with lowercase.
1248       uc32 first_char = c0_;
1249       Advance<false, false>();
1250       AddLiteralChar(first_char);
1251       while (IsAsciiIdentifier(c0_)) {
1252         uc32 first_char = c0_;
1253         Advance<false, false>();
1254         AddLiteralChar(first_char);
1255       }
1256       if (c0_ <= kMaxAscii && c0_ != '\\') {
1257         literal.Complete();
1258         return Token::IDENTIFIER;
1259       }
1260     } else if (c0_ <= kMaxAscii && c0_ != '\\') {
1261       // Only a-z+: could be a keyword or identifier.
1262       literal.Complete();
1263       Vector<const uint8_t> chars = next_.literal_chars->one_byte_literal();
1264       return KeywordOrIdentifierToken(chars.start(), chars.length(), false);
1265     }
1266 
1267     HandleLeadSurrogate();
1268   } else if (IsInRange(c0_, 'A', 'Z') || c0_ == '_' || c0_ == '$') {
1269     do {
1270       uc32 first_char = c0_;
1271       Advance<false, false>();
1272       AddLiteralChar(first_char);
1273     } while (IsAsciiIdentifier(c0_));
1274 
1275     if (c0_ <= kMaxAscii && c0_ != '\\') {
1276       literal.Complete();
1277       return Token::IDENTIFIER;
1278     }
1279 
1280     HandleLeadSurrogate();
1281   } else if (c0_ == '\\') {
1282     // Scan identifier start character.
1283     uc32 c = ScanIdentifierUnicodeEscape();
1284     // Only allow legal identifier start characters.
1285     if (c < 0 ||
1286         c == '\\' ||  // No recursive escapes.
1287         !unicode_cache_->IsIdentifierStart(c)) {
1288       return Token::ILLEGAL;
1289     }
1290     AddLiteralChar(c);
1291     return ScanIdentifierSuffix(&literal, true);
1292   } else {
1293     uc32 first_char = c0_;
1294     Advance();
1295     AddLiteralChar(first_char);
1296   }
1297 
1298   // Scan the rest of the identifier characters.
1299   while (c0_ >= 0 && unicode_cache_->IsIdentifierPart(c0_)) {
1300     if (c0_ != '\\') {
1301       uc32 next_char = c0_;
1302       Advance();
1303       AddLiteralChar(next_char);
1304       continue;
1305     }
1306     // Fallthrough if no longer able to complete keyword.
1307     return ScanIdentifierSuffix(&literal, false);
1308   }
1309 
1310   literal.Complete();
1311 
1312   if (next_.literal_chars->is_one_byte()) {
1313     Vector<const uint8_t> chars = next_.literal_chars->one_byte_literal();
1314     return KeywordOrIdentifierToken(chars.start(), chars.length(), false);
1315   }
1316   return Token::IDENTIFIER;
1317 }
1318 
1319 
ScanIdentifierSuffix(LiteralScope * literal,bool escaped)1320 Token::Value Scanner::ScanIdentifierSuffix(LiteralScope* literal,
1321                                            bool escaped) {
1322   // Scan the rest of the identifier characters.
1323   while (c0_ >= 0 && unicode_cache_->IsIdentifierPart(c0_)) {
1324     if (c0_ == '\\') {
1325       uc32 c = ScanIdentifierUnicodeEscape();
1326       escaped = true;
1327       // Only allow legal identifier part characters.
1328       if (c < 0 ||
1329           c == '\\' ||
1330           !unicode_cache_->IsIdentifierPart(c)) {
1331         return Token::ILLEGAL;
1332       }
1333       AddLiteralChar(c);
1334     } else {
1335       AddLiteralChar(c0_);
1336       Advance();
1337     }
1338   }
1339   literal->Complete();
1340 
1341   if (escaped && next_.literal_chars->is_one_byte()) {
1342     Vector<const uint8_t> chars = next_.literal_chars->one_byte_literal();
1343     return KeywordOrIdentifierToken(chars.start(), chars.length(), true);
1344   }
1345   return Token::IDENTIFIER;
1346 }
1347 
1348 
ScanRegExpPattern(bool seen_equal)1349 bool Scanner::ScanRegExpPattern(bool seen_equal) {
1350   // Scan: ('/' | '/=') RegularExpressionBody '/' RegularExpressionFlags
1351   bool in_character_class = false;
1352 
1353   // Previous token is either '/' or '/=', in the second case, the
1354   // pattern starts at =.
1355   next_.location.beg_pos = source_pos() - (seen_equal ? 2 : 1);
1356   next_.location.end_pos = source_pos() - (seen_equal ? 1 : 0);
1357 
1358   // Scan regular expression body: According to ECMA-262, 3rd, 7.8.5,
1359   // the scanner should pass uninterpreted bodies to the RegExp
1360   // constructor.
1361   LiteralScope literal(this);
1362   if (seen_equal) {
1363     AddLiteralChar('=');
1364   }
1365 
1366   while (c0_ != '/' || in_character_class) {
1367     if (c0_ < 0 || unicode_cache_->IsLineTerminator(c0_)) return false;
1368     if (c0_ == '\\') {  // Escape sequence.
1369       AddLiteralCharAdvance();
1370       if (c0_ < 0 || unicode_cache_->IsLineTerminator(c0_)) return false;
1371       AddLiteralCharAdvance();
1372       // If the escape allows more characters, i.e., \x??, \u????, or \c?,
1373       // only "safe" characters are allowed (letters, digits, underscore),
1374       // otherwise the escape isn't valid and the invalid character has
1375       // its normal meaning. I.e., we can just continue scanning without
1376       // worrying whether the following characters are part of the escape
1377       // or not, since any '/', '\\' or '[' is guaranteed to not be part
1378       // of the escape sequence.
1379 
1380       // TODO(896): At some point, parse RegExps more throughly to capture
1381       // octal esacpes in strict mode.
1382     } else {  // Unescaped character.
1383       if (c0_ == '[') in_character_class = true;
1384       if (c0_ == ']') in_character_class = false;
1385       AddLiteralCharAdvance();
1386     }
1387   }
1388   Advance();  // consume '/'
1389 
1390   literal.Complete();
1391 
1392   return true;
1393 }
1394 
1395 
ScanRegExpFlags()1396 Maybe<RegExp::Flags> Scanner::ScanRegExpFlags() {
1397   // Scan regular expression flags.
1398   LiteralScope literal(this);
1399   int flags = 0;
1400   while (c0_ >= 0 && unicode_cache_->IsIdentifierPart(c0_)) {
1401     RegExp::Flags flag = RegExp::kNone;
1402     switch (c0_) {
1403       case 'g':
1404         flag = RegExp::kGlobal;
1405         break;
1406       case 'i':
1407         flag = RegExp::kIgnoreCase;
1408         break;
1409       case 'm':
1410         flag = RegExp::kMultiline;
1411         break;
1412       case 'u':
1413         if (!FLAG_harmony_unicode_regexps) return Nothing<RegExp::Flags>();
1414         flag = RegExp::kUnicode;
1415         break;
1416       case 'y':
1417         if (!FLAG_harmony_regexps) return Nothing<RegExp::Flags>();
1418         flag = RegExp::kSticky;
1419         break;
1420       default:
1421         return Nothing<RegExp::Flags>();
1422     }
1423     if (flags & flag) return Nothing<RegExp::Flags>();
1424     AddLiteralCharAdvance();
1425     flags |= flag;
1426   }
1427   literal.Complete();
1428 
1429   next_.location.end_pos = source_pos();
1430   return Just(RegExp::Flags(flags));
1431 }
1432 
1433 
CurrentSymbol(AstValueFactory * ast_value_factory)1434 const AstRawString* Scanner::CurrentSymbol(AstValueFactory* ast_value_factory) {
1435   if (is_literal_one_byte()) {
1436     return ast_value_factory->GetOneByteString(literal_one_byte_string());
1437   }
1438   return ast_value_factory->GetTwoByteString(literal_two_byte_string());
1439 }
1440 
1441 
NextSymbol(AstValueFactory * ast_value_factory)1442 const AstRawString* Scanner::NextSymbol(AstValueFactory* ast_value_factory) {
1443   if (is_next_literal_one_byte()) {
1444     return ast_value_factory->GetOneByteString(next_literal_one_byte_string());
1445   }
1446   return ast_value_factory->GetTwoByteString(next_literal_two_byte_string());
1447 }
1448 
1449 
CurrentRawSymbol(AstValueFactory * ast_value_factory)1450 const AstRawString* Scanner::CurrentRawSymbol(
1451     AstValueFactory* ast_value_factory) {
1452   if (is_raw_literal_one_byte()) {
1453     return ast_value_factory->GetOneByteString(raw_literal_one_byte_string());
1454   }
1455   return ast_value_factory->GetTwoByteString(raw_literal_two_byte_string());
1456 }
1457 
1458 
DoubleValue()1459 double Scanner::DoubleValue() {
1460   DCHECK(is_literal_one_byte());
1461   return StringToDouble(
1462       unicode_cache_,
1463       literal_one_byte_string(),
1464       ALLOW_HEX | ALLOW_OCTAL | ALLOW_IMPLICIT_OCTAL | ALLOW_BINARY);
1465 }
1466 
1467 
ContainsDot()1468 bool Scanner::ContainsDot() {
1469   DCHECK(is_literal_one_byte());
1470   Vector<const uint8_t> str = literal_one_byte_string();
1471   return std::find(str.begin(), str.end(), '.') != str.end();
1472 }
1473 
1474 
FindSymbol(DuplicateFinder * finder,int value)1475 int Scanner::FindSymbol(DuplicateFinder* finder, int value) {
1476   if (is_literal_one_byte()) {
1477     return finder->AddOneByteSymbol(literal_one_byte_string(), value);
1478   }
1479   return finder->AddTwoByteSymbol(literal_two_byte_string(), value);
1480 }
1481 
1482 
SetBookmark()1483 bool Scanner::SetBookmark() {
1484   if (c0_ != kNoBookmark && bookmark_c0_ == kNoBookmark &&
1485       next_next_.token == Token::UNINITIALIZED && source_->SetBookmark()) {
1486     bookmark_c0_ = c0_;
1487     CopyTokenDesc(&bookmark_current_, &current_);
1488     CopyTokenDesc(&bookmark_next_, &next_);
1489     return true;
1490   }
1491   return false;
1492 }
1493 
1494 
ResetToBookmark()1495 void Scanner::ResetToBookmark() {
1496   DCHECK(BookmarkHasBeenSet());  // Caller hasn't called SetBookmark.
1497 
1498   source_->ResetToBookmark();
1499   c0_ = bookmark_c0_;
1500   StartLiteral();
1501   StartRawLiteral();
1502   CopyTokenDesc(&next_, &bookmark_current_);
1503   current_ = next_;
1504   StartLiteral();
1505   StartRawLiteral();
1506   CopyTokenDesc(&next_, &bookmark_next_);
1507 
1508   bookmark_c0_ = kBookmarkWasApplied;
1509 }
1510 
1511 
BookmarkHasBeenSet()1512 bool Scanner::BookmarkHasBeenSet() { return bookmark_c0_ >= 0; }
1513 
1514 
BookmarkHasBeenReset()1515 bool Scanner::BookmarkHasBeenReset() {
1516   return bookmark_c0_ == kBookmarkWasApplied;
1517 }
1518 
1519 
DropBookmark()1520 void Scanner::DropBookmark() { bookmark_c0_ = kNoBookmark; }
1521 
1522 
CopyTokenDesc(TokenDesc * to,TokenDesc * from)1523 void Scanner::CopyTokenDesc(TokenDesc* to, TokenDesc* from) {
1524   DCHECK_NOT_NULL(to);
1525   DCHECK_NOT_NULL(from);
1526   to->token = from->token;
1527   to->location = from->location;
1528   to->literal_chars->CopyFrom(from->literal_chars);
1529   to->raw_literal_chars->CopyFrom(from->raw_literal_chars);
1530 }
1531 
1532 
AddOneByteSymbol(Vector<const uint8_t> key,int value)1533 int DuplicateFinder::AddOneByteSymbol(Vector<const uint8_t> key, int value) {
1534   return AddSymbol(key, true, value);
1535 }
1536 
1537 
AddTwoByteSymbol(Vector<const uint16_t> key,int value)1538 int DuplicateFinder::AddTwoByteSymbol(Vector<const uint16_t> key, int value) {
1539   return AddSymbol(Vector<const uint8_t>::cast(key), false, value);
1540 }
1541 
1542 
AddSymbol(Vector<const uint8_t> key,bool is_one_byte,int value)1543 int DuplicateFinder::AddSymbol(Vector<const uint8_t> key,
1544                                bool is_one_byte,
1545                                int value) {
1546   uint32_t hash = Hash(key, is_one_byte);
1547   byte* encoding = BackupKey(key, is_one_byte);
1548   HashMap::Entry* entry = map_.LookupOrInsert(encoding, hash);
1549   int old_value = static_cast<int>(reinterpret_cast<intptr_t>(entry->value));
1550   entry->value =
1551     reinterpret_cast<void*>(static_cast<intptr_t>(value | old_value));
1552   return old_value;
1553 }
1554 
1555 
AddNumber(Vector<const uint8_t> key,int value)1556 int DuplicateFinder::AddNumber(Vector<const uint8_t> key, int value) {
1557   DCHECK(key.length() > 0);
1558   // Quick check for already being in canonical form.
1559   if (IsNumberCanonical(key)) {
1560     return AddOneByteSymbol(key, value);
1561   }
1562 
1563   int flags = ALLOW_HEX | ALLOW_OCTAL | ALLOW_IMPLICIT_OCTAL | ALLOW_BINARY;
1564   double double_value = StringToDouble(
1565       unicode_constants_, key, flags, 0.0);
1566   int length;
1567   const char* string;
1568   if (!std::isfinite(double_value)) {
1569     string = "Infinity";
1570     length = 8;  // strlen("Infinity");
1571   } else {
1572     string = DoubleToCString(double_value,
1573                              Vector<char>(number_buffer_, kBufferSize));
1574     length = StrLength(string);
1575   }
1576   return AddSymbol(Vector<const byte>(reinterpret_cast<const byte*>(string),
1577                                       length), true, value);
1578 }
1579 
1580 
IsNumberCanonical(Vector<const uint8_t> number)1581 bool DuplicateFinder::IsNumberCanonical(Vector<const uint8_t> number) {
1582   // Test for a safe approximation of number literals that are already
1583   // in canonical form: max 15 digits, no leading zeroes, except an
1584   // integer part that is a single zero, and no trailing zeros below
1585   // the decimal point.
1586   int pos = 0;
1587   int length = number.length();
1588   if (number.length() > 15) return false;
1589   if (number[pos] == '0') {
1590     pos++;
1591   } else {
1592     while (pos < length &&
1593            static_cast<unsigned>(number[pos] - '0') <= ('9' - '0')) pos++;
1594   }
1595   if (length == pos) return true;
1596   if (number[pos] != '.') return false;
1597   pos++;
1598   bool invalid_last_digit = true;
1599   while (pos < length) {
1600     uint8_t digit = number[pos] - '0';
1601     if (digit > '9' - '0') return false;
1602     invalid_last_digit = (digit == 0);
1603     pos++;
1604   }
1605   return !invalid_last_digit;
1606 }
1607 
1608 
Hash(Vector<const uint8_t> key,bool is_one_byte)1609 uint32_t DuplicateFinder::Hash(Vector<const uint8_t> key, bool is_one_byte) {
1610   // Primitive hash function, almost identical to the one used
1611   // for strings (except that it's seeded by the length and representation).
1612   int length = key.length();
1613   uint32_t hash = (length << 1) | (is_one_byte ? 1 : 0);
1614   for (int i = 0; i < length; i++) {
1615     uint32_t c = key[i];
1616     hash = (hash + c) * 1025;
1617     hash ^= (hash >> 6);
1618   }
1619   return hash;
1620 }
1621 
1622 
Match(void * first,void * second)1623 bool DuplicateFinder::Match(void* first, void* second) {
1624   // Decode lengths.
1625   // Length + representation is encoded as base 128, most significant heptet
1626   // first, with a 8th bit being non-zero while there are more heptets.
1627   // The value encodes the number of bytes following, and whether the original
1628   // was Latin1.
1629   byte* s1 = reinterpret_cast<byte*>(first);
1630   byte* s2 = reinterpret_cast<byte*>(second);
1631   uint32_t length_one_byte_field = 0;
1632   byte c1;
1633   do {
1634     c1 = *s1;
1635     if (c1 != *s2) return false;
1636     length_one_byte_field = (length_one_byte_field << 7) | (c1 & 0x7f);
1637     s1++;
1638     s2++;
1639   } while ((c1 & 0x80) != 0);
1640   int length = static_cast<int>(length_one_byte_field >> 1);
1641   return memcmp(s1, s2, length) == 0;
1642 }
1643 
1644 
BackupKey(Vector<const uint8_t> bytes,bool is_one_byte)1645 byte* DuplicateFinder::BackupKey(Vector<const uint8_t> bytes,
1646                                  bool is_one_byte) {
1647   uint32_t one_byte_length = (bytes.length() << 1) | (is_one_byte ? 1 : 0);
1648   backing_store_.StartSequence();
1649   // Emit one_byte_length as base-128 encoded number, with the 7th bit set
1650   // on the byte of every heptet except the last, least significant, one.
1651   if (one_byte_length >= (1 << 7)) {
1652     if (one_byte_length >= (1 << 14)) {
1653       if (one_byte_length >= (1 << 21)) {
1654         if (one_byte_length >= (1 << 28)) {
1655           backing_store_.Add(
1656               static_cast<uint8_t>((one_byte_length >> 28) | 0x80));
1657         }
1658         backing_store_.Add(
1659             static_cast<uint8_t>((one_byte_length >> 21) | 0x80u));
1660       }
1661       backing_store_.Add(
1662           static_cast<uint8_t>((one_byte_length >> 14) | 0x80u));
1663     }
1664     backing_store_.Add(static_cast<uint8_t>((one_byte_length >> 7) | 0x80u));
1665   }
1666   backing_store_.Add(static_cast<uint8_t>(one_byte_length & 0x7f));
1667 
1668   backing_store_.AddBlock(bytes);
1669   return backing_store_.EndSequence().start();
1670 }
1671 
1672 }  // namespace internal
1673 }  // namespace v8
1674