summaryrefslogtreecommitdiffstats
path: root/js/src/frontend/TokenStream.h
blob: 18e9cb3ca45a8a3d9e3edbf8c496bcd09bc84511 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
 * vim: set ts=8 sts=4 et sw=4 tw=99:
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#ifndef frontend_TokenStream_h
#define frontend_TokenStream_h

// JS lexical scanner interface.

#include "mozilla/ArrayUtils.h"
#include "mozilla/Assertions.h"
#include "mozilla/Attributes.h"
#include "mozilla/DebugOnly.h"
#include "mozilla/PodOperations.h"

#include <stdarg.h>
#include <stddef.h>
#include <stdio.h>

#include "jscntxt.h"
#include "jspubtd.h"

#include "frontend/TokenKind.h"
#include "js/UniquePtr.h"
#include "js/Vector.h"
#include "vm/RegExpObject.h"

struct KeywordInfo;

namespace js {
namespace frontend {

class AutoAwaitIsKeyword;

struct TokenPos {
    uint32_t    begin;  // Offset of the token's first char.
    uint32_t    end;    // Offset of 1 past the token's last char.

    TokenPos() {}
    TokenPos(uint32_t begin, uint32_t end) : begin(begin), end(end) {}

    // Return a TokenPos that covers left, right, and anything in between.
    static TokenPos box(const TokenPos& left, const TokenPos& right) {
        MOZ_ASSERT(left.begin <= left.end);
        MOZ_ASSERT(left.end <= right.begin);
        MOZ_ASSERT(right.begin <= right.end);
        return TokenPos(left.begin, right.end);
    }

    bool operator==(const TokenPos& bpos) const {
        return begin == bpos.begin && end == bpos.end;
    }

    bool operator!=(const TokenPos& bpos) const {
        return begin != bpos.begin || end != bpos.end;
    }

    bool operator <(const TokenPos& bpos) const {
        return begin < bpos.begin;
    }

    bool operator <=(const TokenPos& bpos) const {
        return begin <= bpos.begin;
    }

    bool operator >(const TokenPos& bpos) const {
        return !(*this <= bpos);
    }

    bool operator >=(const TokenPos& bpos) const {
        return !(*this < bpos);
    }

    bool encloses(const TokenPos& pos) const {
        return begin <= pos.begin && pos.end <= end;
    }
};

enum DecimalPoint { NoDecimal = false, HasDecimal = true };

class TokenStream;

struct Token
{
  private:
    // Sometimes the parser needs to inform the tokenizer to interpret
    // subsequent text in a particular manner: for example, to tokenize a
    // keyword as an identifier, not as the actual keyword, on the right-hand
    // side of a dotted property access.  Such information is communicated to
    // the tokenizer as a Modifier when getting the next token.
    //
    // Ideally this definition would reside in TokenStream as that's the real
    // user, but the debugging-use of it here causes a cyclic dependency (and
    // C++ provides no way to forward-declare an enum inside a class).  So
    // define it here, then typedef it into TokenStream with static consts to
    // bring the initializers into scope.
    enum Modifier
    {
        // Normal operation.
        None,

        // Looking for an operand, not an operator.  In practice, this means
        // that when '/' is seen, we look for a regexp instead of just returning
        // TOK_DIV.
        Operand,

        // Treat keywords as names by returning TOK_NAME.
        KeywordIsName,

        // Treat subsequent characters as the tail of a template literal, after
        // a template substitution, beginning with a "}", continuing with zero
        // or more template literal characters, and ending with either "${" or
        // the end of the template literal.  For example:
        //
        //   var entity = "world";
        //   var s = `Hello ${entity}!`;
        //                          ^ TemplateTail context
        TemplateTail,
    };
    enum ModifierException
    {
        NoException,

        // Used in following 2 cases:
        // a) After |yield| we look for a token on the same line that starts an
        // expression (Operand): |yield <expr>|.  If no token is found, the
        // |yield| stands alone, and the next token on a subsequent line must
        // be: a comma continuing a comma expression, a semicolon terminating
        // the statement that ended with |yield|, or the start of another
        // statement (possibly an expression statement).  The comma/semicolon
        // cases are gotten as operators (None), contrasting with Operand
        // earlier.
        // b) After an arrow function with a block body in an expression
        // statement, the next token must be: a colon in a conditional
        // expression, a comma continuing a comma expression, a semicolon
        // terminating the statement, or the token on a subsequent line that is
        // the start of another statement (possibly an expression statement).
        // Colon is gotten as operator (None), and it should only be gotten in
        // conditional expression and missing it results in SyntaxError.
        // Comma/semicolon cases are also gotten as operators (None), and 4th
        // case is gotten after them.  If no comma/semicolon found but EOL,
        // the next token should be gotten as operand in 4th case (especially if
        // '/' is the first character).  So we should peek the token as
        // operand before try getting colon/comma/semicolon.
        // See also the comment in Parser::assignExpr().
        NoneIsOperand,

        // If a semicolon is inserted automatically, the next token is already
        // gotten with None, but we expect Operand.
        OperandIsNone,

        // If name of method definition is `get` or `set`, the next token is
        // already gotten with KeywordIsName, but we expect None.
        NoneIsKeywordIsName,
    };
    friend class TokenStream;

  public:
    TokenKind           type;           // char value or above enumerator
    TokenPos            pos;            // token position in file
    union {
      private:
        friend struct Token;
        PropertyName*   name;          // non-numeric atom
        JSAtom*         atom;          // potentially-numeric atom
        struct {
            double      value;          // floating point number
            DecimalPoint decimalPoint;  // literal contains '.'
        } number;
        RegExpFlag      reflags;        // regexp flags; use tokenbuf to access
                                        //   regexp chars
    } u;
#ifdef DEBUG
    Modifier modifier;                  // Modifier used to get this token
    ModifierException modifierException; // Exception for this modifier
#endif

    // Mutators

    void setName(PropertyName* name) {
        MOZ_ASSERT(type == TOK_NAME);
        u.name = name;
    }

    void setAtom(JSAtom* atom) {
        MOZ_ASSERT(type == TOK_STRING ||
                   type == TOK_TEMPLATE_HEAD ||
                   type == TOK_NO_SUBS_TEMPLATE);
        u.atom = atom;
    }

    void setRegExpFlags(js::RegExpFlag flags) {
        MOZ_ASSERT(type == TOK_REGEXP);
        MOZ_ASSERT((flags & AllFlags) == flags);
        u.reflags = flags;
    }

    void setNumber(double n, DecimalPoint decimalPoint) {
        MOZ_ASSERT(type == TOK_NUMBER);
        u.number.value = n;
        u.number.decimalPoint = decimalPoint;
    }

    // Type-safe accessors

    PropertyName* name() const {
        MOZ_ASSERT(type == TOK_NAME);
        return u.name->JSAtom::asPropertyName(); // poor-man's type verification
    }

    bool nameContainsEscape() const {
        PropertyName* n = name();
        return pos.begin + n->length() != pos.end;
    }

    JSAtom* atom() const {
        MOZ_ASSERT(type == TOK_STRING ||
                   type == TOK_TEMPLATE_HEAD ||
                   type == TOK_NO_SUBS_TEMPLATE);
        return u.atom;
    }

    js::RegExpFlag regExpFlags() const {
        MOZ_ASSERT(type == TOK_REGEXP);
        MOZ_ASSERT((u.reflags & AllFlags) == u.reflags);
        return u.reflags;
    }

    double number() const {
        MOZ_ASSERT(type == TOK_NUMBER);
        return u.number.value;
    }

    DecimalPoint decimalPoint() const {
        MOZ_ASSERT(type == TOK_NUMBER);
        return u.number.decimalPoint;
    }
};

class CompileError : public JSErrorReport {
public:
    void throwError(JSContext* cx);
};

// Ideally, tokenizing would be entirely independent of context.  But the
// strict mode flag, which is in SharedContext, affects tokenizing, and
// TokenStream needs to see it.
//
// This class is a tiny back-channel from TokenStream to the strict mode flag
// that avoids exposing the rest of SharedContext to TokenStream.
//
class StrictModeGetter {
  public:
    virtual bool strictMode() = 0;
};

// TokenStream is the lexical scanner for Javascript source text.
//
// It takes a buffer of char16_t characters and linearly scans it into |Token|s.
// Internally the class uses a four element circular buffer |tokens| of
// |Token|s. As an index for |tokens|, the member |cursor| points to the
// current token.
// Calls to getToken() increase |cursor| by one and return the new current
// token. If a TokenStream was just created, the current token is initialized
// with random data (i.e. not initialized). It is therefore important that
// one of the first four member functions listed below is called first.
// The circular buffer lets us go back up to two tokens from the last
// scanned token. Internally, the relative number of backward steps that were
// taken (via ungetToken()) after the last token was scanned is stored in
// |lookahead|.
//
// The following table lists in which situations it is safe to call each listed
// function. No checks are made by the functions in non-debug builds.
//
// Function Name     | Precondition; changes to |lookahead|
// ------------------+---------------------------------------------------------
// getToken          | none; if |lookahead > 0| then |lookahead--|
// peekToken         | none; if |lookahead == 0| then |lookahead == 1|
// peekTokenSameLine | none; if |lookahead == 0| then |lookahead == 1|
// matchToken        | none; if |lookahead > 0| and the match succeeds then
//                   |       |lookahead--|
// consumeKnownToken | none; if |lookahead > 0| then |lookahead--|
// ungetToken        | 0 <= |lookahead| <= |maxLookahead - 1|; |lookahead++|
//
// The behavior of the token scanning process (see getTokenInternal()) can be
// modified by calling one of the first four above listed member functions with
// an optional argument of type Modifier.  However, the modifier will be
// ignored unless |lookahead == 0| holds.  Due to constraints of the grammar,
// this turns out not to be a problem in practice. See the
// mozilla.dev.tech.js-engine.internals thread entitled 'Bug in the scanner?'
// for more details:
// https://groups.google.com/forum/?fromgroups=#!topic/mozilla.dev.tech.js-engine.internals/2JLH5jRcr7E).
//
// The methods seek() and tell() allow to rescan from a previous visited
// location of the buffer.
//
class MOZ_STACK_CLASS TokenStream
{
    // Unicode separators that are treated as line terminators, in addition to \n, \r.
    enum {
        LINE_SEPARATOR = 0x2028,
        PARA_SEPARATOR = 0x2029
    };

    static const size_t ntokens = 4;                // 1 current + 2 lookahead, rounded
                                                    // to power of 2 to avoid divmod by 3
    static const unsigned maxLookahead = 2;
    static const unsigned ntokensMask = ntokens - 1;

  public:
    typedef Vector<char16_t, 32> CharBuffer;

    TokenStream(ExclusiveContext* cx, const ReadOnlyCompileOptions& options,
                const char16_t* base, size_t length, StrictModeGetter* smg);

    ~TokenStream();

    MOZ_MUST_USE bool checkOptions();

    // Accessors.
    const Token& currentToken() const { return tokens[cursor]; }
    bool isCurrentTokenType(TokenKind type) const {
        return currentToken().type == type;
    }
    const CharBuffer& getTokenbuf() const { return tokenbuf; }
    const char* getFilename() const { return filename; }
    bool getMutedErrors() const { return mutedErrors; }
    JSVersion versionNumber() const { return VersionNumber(options().version); }
    JSVersion versionWithFlags() const { return options().version; }

    PropertyName* currentName() const {
        if (isCurrentTokenType(TOK_YIELD))
            return cx->names().yield;
        MOZ_ASSERT(isCurrentTokenType(TOK_NAME));
        return currentToken().name();
    }

    PropertyName* nextName() const {
        if (nextToken().type == TOK_YIELD)
            return cx->names().yield;
        MOZ_ASSERT(nextToken().type == TOK_NAME);
        return nextToken().name();
    }

    bool nextNameContainsEscape() const {
        if (nextToken().type == TOK_YIELD)
            return false;
        MOZ_ASSERT(nextToken().type == TOK_NAME);
        return nextToken().nameContainsEscape();
    }

    bool isCurrentTokenAssignment() const {
        return TokenKindIsAssignment(currentToken().type);
    }

    // Flag methods.
    bool isEOF() const { return flags.isEOF; }
    bool sawOctalEscape() const { return flags.sawOctalEscape; }
    bool hadError() const { return flags.hadError; }
    void clearSawOctalEscape() { flags.sawOctalEscape = false; }

    // TokenStream-specific error reporters.
    bool reportError(unsigned errorNumber, ...);
    bool reportErrorNoOffset(unsigned errorNumber, ...);

    // Report the given error at the current offset.
    void error(unsigned errorNumber, ...);

    // Warn at the current offset.
    MOZ_MUST_USE bool warning(unsigned errorNumber, ...);

    static const uint32_t NoOffset = UINT32_MAX;

    // General-purpose error reporters.  You should avoid calling these
    // directly, and instead use the more succinct alternatives (error(),
    // warning(), &c.) in TokenStream, Parser, and BytecodeEmitter.
    bool reportCompileErrorNumberVA(uint32_t offset, unsigned flags, unsigned errorNumber,
                                    va_list args);
    bool reportStrictModeErrorNumberVA(uint32_t offset, bool strictMode, unsigned errorNumber,
                                       va_list args);
    bool reportExtraWarningErrorNumberVA(uint32_t offset, unsigned errorNumber, va_list args);

    // asm.js reporter
    void reportAsmJSError(uint32_t offset, unsigned errorNumber, ...);

    JSAtom* getRawTemplateStringAtom() {
        MOZ_ASSERT(currentToken().type == TOK_TEMPLATE_HEAD ||
                   currentToken().type == TOK_NO_SUBS_TEMPLATE);
        const char16_t* cur = userbuf.rawCharPtrAt(currentToken().pos.begin + 1);
        const char16_t* end;
        if (currentToken().type == TOK_TEMPLATE_HEAD) {
            // Of the form    |`...${|   or   |}...${|
            end = userbuf.rawCharPtrAt(currentToken().pos.end - 2);
        } else {
            // NO_SUBS_TEMPLATE is of the form   |`...`|   or   |}...`|
            end = userbuf.rawCharPtrAt(currentToken().pos.end - 1);
        }

        CharBuffer charbuf(cx);
        while (cur < end) {
            int32_t ch = *cur;
            if (ch == '\r') {
                ch = '\n';
                if ((cur + 1 < end) && (*(cur + 1) == '\n'))
                    cur++;
            }
            if (!charbuf.append(ch))
                return nullptr;
            cur++;
        }
        return AtomizeChars(cx, charbuf.begin(), charbuf.length());
    }

  private:
    // These are private because they should only be called by the tokenizer
    // while tokenizing not by, for example, BytecodeEmitter.
    bool reportStrictModeError(unsigned errorNumber, ...);
    bool strictMode() const { return strictModeGetter && strictModeGetter->strictMode(); }

    static JSAtom* atomize(ExclusiveContext* cx, CharBuffer& cb);
    MOZ_MUST_USE bool putIdentInTokenbuf(const char16_t* identStart);

    struct Flags
    {
        bool isEOF:1;           // Hit end of file.
        bool isDirtyLine:1;     // Non-whitespace since start of line.
        bool sawOctalEscape:1;  // Saw an octal character escape.
        bool hadError:1;        // Hit a syntax error, at start or during a
                                // token.
        bool hitOOM:1;          // Hit OOM.

        Flags()
          : isEOF(), isDirtyLine(), sawOctalEscape(), hadError(), hitOOM()
        {}
    };

    bool awaitIsKeyword = false;
    friend class AutoAwaitIsKeyword;

  public:
    typedef Token::Modifier Modifier;
    static constexpr Modifier None = Token::None;
    static constexpr Modifier Operand = Token::Operand;
    static constexpr Modifier KeywordIsName = Token::KeywordIsName;
    static constexpr Modifier TemplateTail = Token::TemplateTail;

    typedef Token::ModifierException ModifierException;
    static constexpr ModifierException NoException = Token::NoException;
    static constexpr ModifierException NoneIsOperand = Token::NoneIsOperand;
    static constexpr ModifierException OperandIsNone = Token::OperandIsNone;
    static constexpr ModifierException NoneIsKeywordIsName = Token::NoneIsKeywordIsName;

    void addModifierException(ModifierException modifierException) {
#ifdef DEBUG
        const Token& next = nextToken();
        if (next.modifierException == NoneIsOperand)
        {
            // Token after yield expression without operand already has
            // NoneIsOperand exception.
            MOZ_ASSERT(modifierException == OperandIsNone);
            MOZ_ASSERT(next.type != TOK_DIV,
                       "next token requires contextual specifier to be parsed unambiguously");

            // Do not update modifierException.
            return;
        }

        MOZ_ASSERT(next.modifierException == NoException);
        switch (modifierException) {
          case NoneIsOperand:
            MOZ_ASSERT(next.modifier == Operand);
            MOZ_ASSERT(next.type != TOK_DIV,
                       "next token requires contextual specifier to be parsed unambiguously");
            break;
          case OperandIsNone:
            MOZ_ASSERT(next.modifier == None);
            MOZ_ASSERT(next.type != TOK_DIV && next.type != TOK_REGEXP,
                       "next token requires contextual specifier to be parsed unambiguously");
            break;
          case NoneIsKeywordIsName:
            MOZ_ASSERT(next.modifier == KeywordIsName);
            MOZ_ASSERT(next.type != TOK_NAME);
            break;
          default:
            MOZ_CRASH("unexpected modifier exception");
        }
        tokens[(cursor + 1) & ntokensMask].modifierException = modifierException;
#endif
    }

    void
    verifyConsistentModifier(Modifier modifier, Token lookaheadToken) {
#ifdef DEBUG
        // Easy case: modifiers match.
        if (modifier == lookaheadToken.modifier)
            return;

        if (lookaheadToken.modifierException == OperandIsNone) {
            // getToken(Operand) permissibly following getToken().
            if (modifier == Operand && lookaheadToken.modifier == None)
                return;
        }

        if (lookaheadToken.modifierException == NoneIsOperand) {
            // getToken() permissibly following getToken(Operand).
            if (modifier == None && lookaheadToken.modifier == Operand)
                return;
        }

        if (lookaheadToken.modifierException == NoneIsKeywordIsName) {
            // getToken() permissibly following getToken(KeywordIsName).
            if (modifier == None && lookaheadToken.modifier == KeywordIsName)
                return;
        }

        MOZ_ASSERT_UNREACHABLE("this token was previously looked up with a "
                               "different modifier, potentially making "
                               "tokenization non-deterministic");
#endif
    }

    // Advance to the next token.  If the token stream encountered an error,
    // return false.  Otherwise return true and store the token kind in |*ttp|.
    MOZ_MUST_USE bool getToken(TokenKind* ttp, Modifier modifier = None) {
        // Check for a pushed-back token resulting from mismatching lookahead.
        if (lookahead != 0) {
            MOZ_ASSERT(!flags.hadError);
            lookahead--;
            cursor = (cursor + 1) & ntokensMask;
            TokenKind tt = currentToken().type;
            MOZ_ASSERT(tt != TOK_EOL);
            verifyConsistentModifier(modifier, currentToken());
            *ttp = tt;
            return true;
        }

        return getTokenInternal(ttp, modifier);
    }

    // Push the last scanned token back into the stream.
    void ungetToken() {
        MOZ_ASSERT(lookahead < maxLookahead);
        lookahead++;
        cursor = (cursor - 1) & ntokensMask;
    }

    MOZ_MUST_USE bool peekToken(TokenKind* ttp, Modifier modifier = None) {
        if (lookahead > 0) {
            MOZ_ASSERT(!flags.hadError);
            verifyConsistentModifier(modifier, nextToken());
            *ttp = nextToken().type;
            return true;
        }
        if (!getTokenInternal(ttp, modifier))
            return false;
        ungetToken();
        return true;
    }

    MOZ_MUST_USE bool peekTokenPos(TokenPos* posp, Modifier modifier = None) {
        if (lookahead == 0) {
            TokenKind tt;
            if (!getTokenInternal(&tt, modifier))
                return false;
            ungetToken();
            MOZ_ASSERT(hasLookahead());
        } else {
            MOZ_ASSERT(!flags.hadError);
            verifyConsistentModifier(modifier, nextToken());
        }
        *posp = nextToken().pos;
        return true;
    }

    MOZ_MUST_USE bool peekOffset(uint32_t* offset, Modifier modifier = None) {
        TokenPos pos;
        if (!peekTokenPos(&pos, modifier))
            return false;
        *offset = pos.begin;
        return true;
    }

    // This is like peekToken(), with one exception:  if there is an EOL
    // between the end of the current token and the start of the next token, it
    // return true and store TOK_EOL in |*ttp|.  In that case, no token with
    // TOK_EOL is actually created, just a TOK_EOL TokenKind is returned, and
    // currentToken() shouldn't be consulted.  (This is the only place TOK_EOL
    // is produced.)
    MOZ_ALWAYS_INLINE MOZ_MUST_USE bool
    peekTokenSameLine(TokenKind* ttp, Modifier modifier = None) {
        const Token& curr = currentToken();

        // If lookahead != 0, we have scanned ahead at least one token, and
        // |lineno| is the line that the furthest-scanned token ends on.  If
        // it's the same as the line that the current token ends on, that's a
        // stronger condition than what we are looking for, and we don't need
        // to return TOK_EOL.
        if (lookahead != 0) {
            bool onThisLine;
            if (!srcCoords.isOnThisLine(curr.pos.end, lineno, &onThisLine))
                return reportError(JSMSG_OUT_OF_MEMORY);
            if (onThisLine) {
                MOZ_ASSERT(!flags.hadError);
                verifyConsistentModifier(modifier, nextToken());
                *ttp = nextToken().type;
                return true;
            }
        }

        // The above check misses two cases where we don't have to return
        // TOK_EOL.
        // - The next token starts on the same line, but is a multi-line token.
        // - The next token starts on the same line, but lookahead==2 and there
        //   is a newline between the next token and the one after that.
        // The following test is somewhat expensive but gets these cases (and
        // all others) right.
        TokenKind tmp;
        if (!getToken(&tmp, modifier))
            return false;
        const Token& next = currentToken();
        ungetToken();

        *ttp = srcCoords.lineNum(curr.pos.end) == srcCoords.lineNum(next.pos.begin)
             ? next.type
             : TOK_EOL;
        return true;
    }

    // Get the next token from the stream if its kind is |tt|.
    MOZ_MUST_USE bool matchToken(bool* matchedp, TokenKind tt, Modifier modifier = None) {
        TokenKind token;
        if (!getToken(&token, modifier))
            return false;
        if (token == tt) {
            *matchedp = true;
        } else {
            ungetToken();
            *matchedp = false;
        }
        return true;
    }

    void consumeKnownToken(TokenKind tt, Modifier modifier = None) {
        bool matched;
        MOZ_ASSERT(hasLookahead());
        MOZ_ALWAYS_TRUE(matchToken(&matched, tt, modifier));
        MOZ_ALWAYS_TRUE(matched);
    }

    // Like matchToken(..., TOK_NAME) but further matching the name token only
    // if it has the given characters, without containing escape sequences.
    // If the name token has the given characters yet *does* contain an escape,
    // a syntax error will be reported.
    //
    // This latter behavior makes this method unsuitable for use in any context
    // where ASI might occur.  In such places, an escaped "contextual keyword"
    // on a new line is the start of an ExpressionStatement, not a continuation
    // of a StatementListItem (or ImportDeclaration or ExportDeclaration, in
    // modules).
    MOZ_MUST_USE bool matchContextualKeyword(bool* matchedp, Handle<PropertyName*> keyword,
                                             Modifier modifier = None)
    {
        TokenKind token;
        if (!getToken(&token, modifier))
            return false;
        if (token == TOK_NAME && currentToken().name() == keyword) {
            if (currentToken().nameContainsEscape()) {
                reportError(JSMSG_ESCAPED_KEYWORD);
                return false;
            }

            *matchedp = true;
        } else {
            *matchedp = false;
            ungetToken();
        }
        return true;
    }

    MOZ_MUST_USE bool nextTokenEndsExpr(bool* endsExpr) {
        TokenKind tt;
        if (!peekToken(&tt))
            return false;
        *endsExpr = isExprEnding[tt];
        return true;
    }

    class MOZ_STACK_CLASS Position {
      public:
        // The Token fields may contain pointers to atoms, so for correct
        // rooting we must ensure collection of atoms is disabled while objects
        // of this class are live.  Do this by requiring a dummy AutoKeepAtoms
        // reference in the constructor.
        //
        // This class is explicity ignored by the analysis, so don't add any
        // more pointers to GC things here!
        explicit Position(AutoKeepAtoms&) { }
      private:
        Position(const Position&) = delete;
        friend class TokenStream;
        const char16_t* buf;
        Flags flags;
        unsigned lineno;
        size_t linebase;
        size_t prevLinebase;
        Token currentToken;
        unsigned lookahead;
        Token lookaheadTokens[maxLookahead];
    };

    MOZ_MUST_USE bool advance(size_t position);
    void tell(Position*);
    void seek(const Position& pos);
    MOZ_MUST_USE bool seek(const Position& pos, const TokenStream& other);
#ifdef DEBUG
    inline bool debugHasNoLookahead() const {
        return lookahead == 0;
    }
#endif

    const char16_t* rawCharPtrAt(size_t offset) const {
        return userbuf.rawCharPtrAt(offset);
    }

    const char16_t* rawLimit() const {
        return userbuf.limit();
    }

    bool hasDisplayURL() const {
        return displayURL_ != nullptr;
    }

    char16_t* displayURL() {
        return displayURL_.get();
    }

    bool hasSourceMapURL() const {
        return sourceMapURL_ != nullptr;
    }

    char16_t* sourceMapURL() {
        return sourceMapURL_.get();
    }

    // If |atom| is not a keyword in this version, return true with *ttp
    // unchanged.
    //
    // If it is a reserved word in this version and strictness mode, and thus
    // can't be present in correct code, report a SyntaxError and return false.
    //
    // If it is a keyword, like "if", return true with the keyword's TokenKind
    // in *ttp.
    MOZ_MUST_USE bool checkForKeyword(JSAtom* atom, TokenKind* ttp);

    // Same semantics as above, but for the provided keyword.
    MOZ_MUST_USE bool checkForKeyword(const KeywordInfo* kw, TokenKind* ttp);

    // This class maps a userbuf offset (which is 0-indexed) to a line number
    // (which is 1-indexed) and a column index (which is 0-indexed).
    class SourceCoords
    {
        // For a given buffer holding source code, |lineStartOffsets_| has one
        // element per line of source code, plus one sentinel element.  Each
        // non-sentinel element holds the buffer offset for the start of the
        // corresponding line of source code.  For this example script:
        //
        // 1  // xyz            [line starts at offset 0]
        // 2  var x;            [line starts at offset 7]
        // 3                    [line starts at offset 14]
        // 4  var y;            [line starts at offset 15]
        //
        // |lineStartOffsets_| is:
        //
        //   [0, 7, 14, 15, MAX_PTR]
        //
        // To convert a "line number" to a "line index" (i.e. an index into
        // |lineStartOffsets_|), subtract |initialLineNum_|.  E.g. line 3's
        // line index is (3 - initialLineNum_), which is 2.  Therefore
        // lineStartOffsets_[2] holds the buffer offset for the start of line 3,
        // which is 14.  (Note that |initialLineNum_| is often 1, but not
        // always.)
        //
        // The first element is always 0, and the last element is always the
        // MAX_PTR sentinel.
        //
        // offset-to-line/column lookups are O(log n) in the worst case (binary
        // search), but in practice they're heavily clustered and we do better
        // than that by using the previous lookup's result (lastLineIndex_) as
        // a starting point.
        //
        // Checking if an offset lies within a particular line number
        // (isOnThisLine()) is O(1).
        //
        Vector<uint32_t, 128> lineStartOffsets_;
        uint32_t            initialLineNum_;

        // This is mutable because it's modified on every search, but that fact
        // isn't visible outside this class.
        mutable uint32_t    lastLineIndex_;

        uint32_t lineIndexOf(uint32_t offset) const;

        static const uint32_t MAX_PTR = UINT32_MAX;

        uint32_t lineIndexToNum(uint32_t lineIndex) const { return lineIndex + initialLineNum_; }
        uint32_t lineNumToIndex(uint32_t lineNum)   const { return lineNum   - initialLineNum_; }

      public:
        SourceCoords(ExclusiveContext* cx, uint32_t ln);

        MOZ_MUST_USE bool add(uint32_t lineNum, uint32_t lineStartOffset);
        MOZ_MUST_USE bool fill(const SourceCoords& other);

        bool isOnThisLine(uint32_t offset, uint32_t lineNum, bool* onThisLine) const {
            uint32_t lineIndex = lineNumToIndex(lineNum);
            if (lineIndex + 1 >= lineStartOffsets_.length()) // +1 due to sentinel
                return false;
            *onThisLine = lineStartOffsets_[lineIndex] <= offset &&
                          offset < lineStartOffsets_[lineIndex + 1];
            return true;
        }

        uint32_t lineNum(uint32_t offset) const;
        uint32_t columnIndex(uint32_t offset) const;
        void lineNumAndColumnIndex(uint32_t offset, uint32_t* lineNum, uint32_t* columnIndex) const;
    };

    SourceCoords srcCoords;

    JSAtomState& names() const {
        return cx->names();
    }

    ExclusiveContext* context() const {
        return cx;
    }

    const ReadOnlyCompileOptions& options() const {
        return options_;
    }

  private:
    // This is the low-level interface to the JS source code buffer.  It just
    // gets raw chars, basically.  TokenStreams functions are layered on top
    // and do some extra stuff like converting all EOL sequences to '\n',
    // tracking the line number, and setting |flags.isEOF|.  (The "raw" in "raw
    // chars" refers to the lack of EOL sequence normalization.)
    //
    // buf[0..length-1] often represents a substring of some larger source,
    // where we have only the substring in memory. The |startOffset| argument
    // indicates the offset within this larger string at which our string
    // begins, the offset of |buf[0]|.
    class TokenBuf {
      public:
        TokenBuf(ExclusiveContext* cx, const char16_t* buf, size_t length, size_t startOffset)
          : base_(buf),
            startOffset_(startOffset),
            limit_(buf + length),
            ptr(buf)
        { }

        bool hasRawChars() const {
            return ptr < limit_;
        }

        bool atStart() const {
            return offset() == 0;
        }

        size_t startOffset() const {
            return startOffset_;
        }

        size_t offset() const {
            return startOffset_ + mozilla::PointerRangeSize(base_, ptr);
        }

        const char16_t* rawCharPtrAt(size_t offset) const {
            MOZ_ASSERT(startOffset_ <= offset);
            MOZ_ASSERT(offset - startOffset_ <= mozilla::PointerRangeSize(base_, limit_));
            return base_ + (offset - startOffset_);
        }

        const char16_t* limit() const {
            return limit_;
        }

        char16_t getRawChar() {
            return *ptr++;      // this will nullptr-crash if poisoned
        }

        char16_t peekRawChar() const {
            return *ptr;        // this will nullptr-crash if poisoned
        }

        bool matchRawChar(char16_t c) {
            if (*ptr == c) {    // this will nullptr-crash if poisoned
                ptr++;
                return true;
            }
            return false;
        }

        bool matchRawCharBackwards(char16_t c) {
            MOZ_ASSERT(ptr);     // make sure it hasn't been poisoned
            if (*(ptr - 1) == c) {
                ptr--;
                return true;
            }
            return false;
        }

        void ungetRawChar() {
            MOZ_ASSERT(ptr);     // make sure it hasn't been poisoned
            ptr--;
        }

        const char16_t* addressOfNextRawChar(bool allowPoisoned = false) const {
            MOZ_ASSERT_IF(!allowPoisoned, ptr);     // make sure it hasn't been poisoned
            return ptr;
        }

        // Use this with caution!
        void setAddressOfNextRawChar(const char16_t* a, bool allowPoisoned = false) {
            MOZ_ASSERT_IF(!allowPoisoned, a);
            ptr = a;
        }

#ifdef DEBUG
        // Poison the TokenBuf so it cannot be accessed again.
        void poison() {
            ptr = nullptr;
        }
#endif

        static bool isRawEOLChar(int32_t c) {
            return c == '\n' || c == '\r' || c == LINE_SEPARATOR || c == PARA_SEPARATOR;
        }

        // Returns the offset of the next EOL, but stops once 'max' characters
        // have been scanned (*including* the char at startOffset_).
        size_t findEOLMax(size_t start, size_t max);

      private:
        const char16_t* base_;          // base of buffer
        uint32_t startOffset_;          // offset of base_[0]
        const char16_t* limit_;         // limit for quick bounds check
        const char16_t* ptr;            // next char to get
    };

    MOZ_MUST_USE bool getTokenInternal(TokenKind* ttp, Modifier modifier);

    MOZ_MUST_USE bool getBracedUnicode(uint32_t* code);
    MOZ_MUST_USE bool getStringOrTemplateToken(int untilChar, Token** tp);

    int32_t getChar();
    int32_t getCharIgnoreEOL();
    void ungetChar(int32_t c);
    void ungetCharIgnoreEOL(int32_t c);
    Token* newToken(ptrdiff_t adjust);
    uint32_t peekUnicodeEscape(uint32_t* codePoint);
    uint32_t peekExtendedUnicodeEscape(uint32_t* codePoint);
    uint32_t matchUnicodeEscapeIdStart(uint32_t* codePoint);
    bool matchUnicodeEscapeIdent(uint32_t* codePoint);
    bool matchTrailForLeadSurrogate(char16_t lead, char16_t* trail, uint32_t* codePoint);
    bool peekChars(int n, char16_t* cp);

    MOZ_MUST_USE bool getDirectives(bool isMultiline, bool shouldWarnDeprecated);
    MOZ_MUST_USE bool getDirective(bool isMultiline, bool shouldWarnDeprecated,
                                   const char* directive, int directiveLength,
                                   const char* errorMsgPragma,
                                   UniquePtr<char16_t[], JS::FreePolicy>* destination);
    MOZ_MUST_USE bool getDisplayURL(bool isMultiline, bool shouldWarnDeprecated);
    MOZ_MUST_USE bool getSourceMappingURL(bool isMultiline, bool shouldWarnDeprecated);

    // |expect| cannot be an EOL char.
    bool matchChar(int32_t expect) {
        MOZ_ASSERT(!TokenBuf::isRawEOLChar(expect));
        return MOZ_LIKELY(userbuf.hasRawChars()) &&
               userbuf.matchRawChar(expect);
    }

    void consumeKnownChar(int32_t expect) {
        mozilla::DebugOnly<int32_t> c = getChar();
        MOZ_ASSERT(c == expect);
    }

    int32_t peekChar() {
        int32_t c = getChar();
        ungetChar(c);
        return c;
    }

    void skipChars(int n) {
        while (--n >= 0)
            getChar();
    }

    void skipCharsIgnoreEOL(int n) {
        while (--n >= 0)
            getCharIgnoreEOL();
    }

    void updateLineInfoForEOL();
    void updateFlagsForEOL();

    const Token& nextToken() const {
        MOZ_ASSERT(hasLookahead());
        return tokens[(cursor + 1) & ntokensMask];
    }

    bool hasLookahead() const { return lookahead > 0; }

    // Options used for parsing/tokenizing.
    const ReadOnlyCompileOptions& options_;

    Token               tokens[ntokens];    // circular token buffer
    unsigned            cursor;             // index of last parsed token
    unsigned            lookahead;          // count of lookahead tokens
    unsigned            lineno;             // current line number
    Flags               flags;              // flags -- see above
    size_t              linebase;           // start of current line
    size_t              prevLinebase;       // start of previous line;  size_t(-1) if on the first line
    TokenBuf            userbuf;            // user input buffer
    const char*         filename;           // input filename or null
    UniqueTwoByteChars  displayURL_;        // the user's requested source URL or null
    UniqueTwoByteChars  sourceMapURL_;      // source map's filename or null
    CharBuffer          tokenbuf;           // current token string buffer
    uint8_t             isExprEnding[TOK_LIMIT];// which tokens definitely terminate exprs?
    ExclusiveContext*   const cx;
    bool                mutedErrors;
    StrictModeGetter*   strictModeGetter;  // used to test for strict mode
};

class MOZ_STACK_CLASS AutoAwaitIsKeyword
{
private:
    TokenStream* ts_;
    bool oldAwaitIsKeyword_;

public:
    AutoAwaitIsKeyword(TokenStream* ts, bool awaitIsKeyword) {
        ts_ = ts;
        oldAwaitIsKeyword_ = ts_->awaitIsKeyword;
        ts_->awaitIsKeyword = awaitIsKeyword;
    }

    ~AutoAwaitIsKeyword() {
        ts_->awaitIsKeyword = oldAwaitIsKeyword_;
        ts_ = nullptr;
    }
};

extern const char*
TokenKindToDesc(TokenKind tt);

} // namespace frontend
} // namespace js

extern JS_FRIEND_API(int)
js_fgets(char* buf, int size, FILE* file);

#ifdef DEBUG
extern const char*
TokenKindToString(js::frontend::TokenKind tt);
#endif

#endif /* frontend_TokenStream_h */