MIRROR: javascript for 🐜's, a tiny runtime with big ambitions
hyper-optimize next()
+465
-163
+1
-1
meson.build
+1
-1
meson.build
···
96
96
build_date = run_command('date', '+%Y-%m-%d', check: true).stdout().strip()
97
97
98
98
version_conf = configuration_data()
99
-
version_conf.set('ANT_VERSION', '0.3.2.18')
99
+
version_conf.set('ANT_VERSION', '0.3.2.19')
100
100
version_conf.set('ANT_GIT_HASH', git_hash)
101
101
version_conf.set('ANT_BUILD_DATE', build_date)
102
102
+464
-162
src/ant.c
+464
-162
src/ant.c
···
3576
3576
return parsekeyword(decoded, decoded_len);
3577
3577
}
3578
3578
3579
-
static uint8_t next(struct js *js) {
3580
-
if (js->consumed == 0) return js->tok;
3579
+
#define CHAR_DIGIT 0x01
3580
+
#define CHAR_XDIGIT 0x02
3581
+
#define CHAR_ALPHA 0x04
3582
+
#define CHAR_IDENT 0x08
3583
+
#define CHAR_IDENT1 0x10
3584
+
#define CHAR_WS 0x20
3585
+
#define CHAR_OCTAL 0x40
3586
+
3587
+
static const uint8_t char_type[256] = {
3588
+
['\t'] = CHAR_WS, ['\n'] = CHAR_WS, ['\r'] = CHAR_WS, [' '] = CHAR_WS,
3589
+
['0'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3590
+
['1'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3591
+
['2'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3592
+
['3'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3593
+
['4'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3594
+
['5'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3595
+
['6'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3596
+
['7'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT | CHAR_OCTAL,
3597
+
['8'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT,
3598
+
['9'] = CHAR_DIGIT | CHAR_XDIGIT | CHAR_IDENT,
3599
+
['A'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3600
+
['B'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3601
+
['C'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3602
+
['D'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3603
+
['E'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3604
+
['F'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3605
+
['a'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3606
+
['b'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3607
+
['c'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3608
+
['d'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3609
+
['e'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3610
+
['f'] = CHAR_XDIGIT | CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3611
+
['G'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['H'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3612
+
['I'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['J'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3613
+
['K'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['L'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3614
+
['M'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['N'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3615
+
['O'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['P'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3616
+
['Q'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['R'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3617
+
['S'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['T'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3618
+
['U'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['V'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3619
+
['W'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['X'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3620
+
['Y'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['Z'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3621
+
['g'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['h'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3622
+
['i'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['j'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3623
+
['k'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['l'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3624
+
['m'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['n'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3625
+
['o'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['p'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3626
+
['q'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['r'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3627
+
['s'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['t'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3628
+
['u'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['v'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3629
+
['w'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['x'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3630
+
['y'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1, ['z'] = CHAR_ALPHA | CHAR_IDENT | CHAR_IDENT1,
3631
+
['_'] = CHAR_IDENT | CHAR_IDENT1,
3632
+
['$'] = CHAR_IDENT | CHAR_IDENT1,
3633
+
};
3634
+
3635
+
#define IS_DIGIT(c) (char_type[(uint8_t)(c)] & CHAR_DIGIT)
3636
+
#define IS_XDIGIT(c) (char_type[(uint8_t)(c)] & CHAR_XDIGIT)
3637
+
#define IS_IDENT(c) (char_type[(uint8_t)(c)] & CHAR_IDENT)
3638
+
#define IS_IDENT1(c) (char_type[(uint8_t)(c)] & CHAR_IDENT1)
3639
+
#define IS_OCTAL(c) (char_type[(uint8_t)(c)] & CHAR_OCTAL)
3640
+
3641
+
#define likely(x) __builtin_expect(!!(x), 1)
3642
+
#define unlikely(x) __builtin_expect(!!(x), 0)
3643
+
3644
+
static const uint8_t single_char_tok[128] = {
3645
+
['('] = TOK_LPAREN,
3646
+
[')'] = TOK_RPAREN,
3647
+
['{'] = TOK_LBRACE,
3648
+
['}'] = TOK_RBRACE,
3649
+
['['] = TOK_LBRACKET,
3650
+
[']'] = TOK_RBRACKET,
3651
+
[';'] = TOK_SEMICOLON,
3652
+
[','] = TOK_COMMA,
3653
+
[':'] = TOK_COLON,
3654
+
['~'] = TOK_TILDA,
3655
+
['#'] = TOK_HASH,
3656
+
};
3657
+
3658
+
static inline jsoff_t parse_decimal(const char *buf, jsoff_t maxlen, double *out) {
3659
+
double val = 0.0;
3660
+
jsoff_t i = 0;
3661
+
3662
+
while (i < maxlen && (IS_DIGIT(buf[i]) || buf[i] == '_')) {
3663
+
if (buf[i] != '_') val = val * 10.0 + (buf[i] - '0');
3664
+
i++;
3665
+
}
3666
+
3667
+
if (i < maxlen && buf[i] == '.') {
3668
+
i++;
3669
+
double frac = 0.1;
3670
+
while (i < maxlen && (IS_DIGIT(buf[i]) || buf[i] == '_')) {
3671
+
if (buf[i] != '_') { val += (buf[i] - '0') * frac; frac *= 0.1; }
3672
+
i++;
3673
+
}
3674
+
}
3675
+
3676
+
if (i < maxlen && (buf[i] == 'e' || buf[i] == 'E')) {
3677
+
i++;
3678
+
int exp_sign = 1, exp_val = 0;
3679
+
if (i < maxlen && (buf[i] == '+' || buf[i] == '-')) {
3680
+
exp_sign = (buf[i] == '-') ? -1 : 1;
3681
+
i++;
3682
+
}
3683
+
while (i < maxlen && (IS_DIGIT(buf[i]) || buf[i] == '_')) {
3684
+
if (buf[i] != '_') exp_val = exp_val * 10 + (buf[i] - '0');
3685
+
i++;
3686
+
}
3687
+
static const double pow10[] = {
3688
+
1e0,1e1,1e2,1e3,1e4,1e5,1e6,1e7,1e8,1e9,1e10,
3689
+
1e11,1e12,1e13,1e14,1e15,1e16,1e17,1e18,1e19,1e20
3690
+
};
3691
+
3692
+
if (exp_val <= 20) {
3693
+
val = (exp_sign > 0) ? val * pow10[exp_val] : val / pow10[exp_val];
3694
+
} else val *= pow(10.0, exp_sign * exp_val);
3695
+
}
3696
+
3697
+
*out = val;
3698
+
return i;
3699
+
}
3700
+
3701
+
static inline jsoff_t parse_binary(const char *buf, jsoff_t maxlen, double *out) {
3702
+
double val = 0;
3703
+
jsoff_t i = 2;
3704
+
while (i < maxlen && (buf[i] == '0' || buf[i] == '1' || buf[i] == '_')) {
3705
+
if (buf[i] != '_') val = val * 2 + (buf[i] - '0');
3706
+
i++;
3707
+
}
3708
+
*out = val;
3709
+
return i;
3710
+
}
3711
+
3712
+
static inline jsoff_t parse_octal(const char *buf, jsoff_t maxlen, double *out) {
3713
+
double val = 0;
3714
+
jsoff_t i = 2;
3715
+
while (i < maxlen && (IS_OCTAL(buf[i]) || buf[i] == '_')) {
3716
+
if (buf[i] != '_') val = val * 8 + (buf[i] - '0');
3717
+
i++;
3718
+
}
3719
+
*out = val;
3720
+
return i;
3721
+
}
3722
+
3723
+
static inline jsoff_t parse_legacy_octal(const char *buf, jsoff_t maxlen, double *out) {
3724
+
double val = 0;
3725
+
jsoff_t i = 1;
3726
+
while (i < maxlen && IS_OCTAL(buf[i])) {
3727
+
val = val * 8 + (buf[i] - '0');
3728
+
i++;
3729
+
}
3730
+
*out = val;
3731
+
return i;
3732
+
}
3733
+
3734
+
static inline jsoff_t parse_hex(const char *buf, jsoff_t maxlen, double *out) {
3735
+
double val = 0;
3736
+
jsoff_t i = 2;
3737
+
while (i < maxlen && (IS_XDIGIT(buf[i]) || buf[i] == '_')) {
3738
+
if (buf[i] != '_') {
3739
+
int d =
3740
+
(buf[i] >= 'a') ? (buf[i] - 'a' + 10) :
3741
+
(buf[i] >= 'A') ? (buf[i] - 'A' + 10) : (buf[i] - '0');
3742
+
val = val * 16 + d;
3743
+
} i++;
3744
+
}
3745
+
*out = val;
3746
+
return i;
3747
+
}
3748
+
3749
+
static inline uint8_t parse_number(struct js *js, const char *buf, jsoff_t remaining) {
3750
+
double value = 0;
3751
+
jsoff_t numlen = 0;
3752
+
3753
+
if (buf[0] == '0' && remaining > 1) {
3754
+
char c1 = buf[1] | 0x20;
3755
+
if (c1 == 'b') {
3756
+
numlen = parse_binary(buf, remaining, &value);
3757
+
} else if (c1 == 'o') {
3758
+
numlen = parse_octal(buf, remaining, &value);
3759
+
} else if (c1 == 'x') {
3760
+
numlen = parse_hex(buf, remaining, &value);
3761
+
} else if (IS_OCTAL(buf[1])) {
3762
+
if (js->flags & F_STRICT) {
3763
+
js->tok = TOK_ERR;
3764
+
js->tlen = 1;
3765
+
return TOK_ERR;
3766
+
}
3767
+
numlen = parse_legacy_octal(buf, remaining, &value);
3768
+
} else numlen = parse_decimal(buf, remaining, &value);
3769
+
} else numlen = parse_decimal(buf, remaining, &value);
3581
3770
3771
+
js->tval = tov(value);
3772
+
if (numlen < remaining && buf[numlen] == 'n') {
3773
+
js->tok = TOK_BIGINT;
3774
+
js->tlen = numlen + 1;
3775
+
} else {
3776
+
js->tok = TOK_NUMBER;
3777
+
js->tlen = numlen;
3778
+
}
3779
+
3780
+
return js->tok;
3781
+
}
3782
+
3783
+
static inline uint8_t scan_string(struct js *js, const char *buf, jsoff_t rem, char quote) {
3784
+
jsoff_t i = 1;
3785
+
3786
+
while (i < rem) {
3787
+
const char *p = buf + i;
3788
+
jsoff_t search_len = rem - i;
3789
+
3790
+
const char *q = memchr(p, quote, search_len);
3791
+
const char *b = memchr(p, '\\', search_len);
3792
+
3793
+
if (q == NULL) {
3794
+
js->tok = TOK_ERR;
3795
+
js->tlen = rem;
3796
+
return TOK_ERR;
3797
+
}
3798
+
3799
+
if (b == NULL || q < b) {
3800
+
i = (q - buf) + 1;
3801
+
js->tok = TOK_STRING;
3802
+
js->tlen = i;
3803
+
return TOK_STRING;
3804
+
}
3805
+
3806
+
jsoff_t esc_pos = b - buf;
3807
+
if (esc_pos + 1 >= rem) {
3808
+
js->tok = TOK_ERR;
3809
+
js->tlen = rem;
3810
+
return TOK_ERR;
3811
+
}
3812
+
3813
+
char esc_char = buf[esc_pos + 1];
3814
+
jsoff_t skip = 2;
3815
+
3816
+
if (esc_char == 'x') { skip = 4; } else if (esc_char == 'u') {
3817
+
skip = (esc_pos + 2 < rem && buf[esc_pos + 2] == '{') ? 0 : 6;
3818
+
if (skip == 0) {
3819
+
jsoff_t j = esc_pos + 3;
3820
+
while (j < rem && buf[j] != '}') j++;
3821
+
skip = (j < rem) ? (j - esc_pos + 1) : (rem - esc_pos);
3822
+
}
3823
+
}
3824
+
3825
+
if (esc_pos + skip > rem) {
3826
+
js->tok = TOK_ERR;
3827
+
js->tlen = rem;
3828
+
return TOK_ERR;
3829
+
}
3830
+
3831
+
i = esc_pos + skip;
3832
+
}
3833
+
3834
+
js->tok = TOK_ERR;
3835
+
js->tlen = rem;
3836
+
return TOK_ERR;
3837
+
}
3838
+
3839
+
static inline uint8_t scan_template(struct js *js, const char *buf, jsoff_t rem) {
3840
+
jsoff_t i = 1;
3841
+
3842
+
while (i < rem) {
3843
+
const char *p = buf + i;
3844
+
jsoff_t search_len = rem - i;
3845
+
3846
+
const char *q = memchr(p, '`', search_len);
3847
+
const char *b = memchr(p, '\\', search_len);
3848
+
3849
+
if (q == NULL) {
3850
+
js->tok = TOK_ERR;
3851
+
js->tlen = rem;
3852
+
return TOK_ERR;
3853
+
}
3854
+
3855
+
if (b == NULL || q < b) {
3856
+
i = (q - buf) + 1;
3857
+
js->tok = TOK_TEMPLATE;
3858
+
js->tlen = i;
3859
+
return TOK_TEMPLATE;
3860
+
}
3861
+
3862
+
jsoff_t esc_pos = b - buf;
3863
+
i = esc_pos + 2;
3864
+
if (i > rem) {
3865
+
js->tok = TOK_ERR;
3866
+
js->tlen = rem;
3867
+
return TOK_ERR;
3868
+
}
3869
+
}
3870
+
3871
+
js->tok = TOK_ERR;
3872
+
js->tlen = rem;
3873
+
return TOK_ERR;
3874
+
}
3875
+
3876
+
static inline uint8_t parse_operator(struct js *js, const char *buf, jsoff_t rem) {
3877
+
#define MATCH2(c1,c2) (rem >= 2 && buf[1] == (c2))
3878
+
#define MATCH3(c1,c2,c3) (rem >= 3 && buf[1] == (c2) && buf[2] == (c3))
3879
+
#define MATCH4(c1,c2,c3,c4) (rem >= 4 && buf[1]==(c2) && buf[2]==(c3) && buf[3]==(c4))
3880
+
3881
+
switch (buf[0]) {
3882
+
case '?':
3883
+
if (MATCH3('?','?','=')) { js->tok = TOK_NULLISH_ASSIGN; js->tlen = 3; }
3884
+
else if (MATCH2('?','?')) { js->tok = TOK_NULLISH; js->tlen = 2; }
3885
+
else if (MATCH2('?','.')) { js->tok = TOK_OPTIONAL_CHAIN; js->tlen = 2; }
3886
+
else { js->tok = TOK_Q; js->tlen = 1; }
3887
+
break;
3888
+
3889
+
case '!':
3890
+
if (MATCH3('!','=','=')) { js->tok = TOK_SNE; js->tlen = 3; }
3891
+
else if (MATCH2('!','=')) { js->tok = TOK_NE; js->tlen = 2; }
3892
+
else { js->tok = TOK_NOT; js->tlen = 1; }
3893
+
break;
3894
+
3895
+
case '=':
3896
+
if (MATCH3('=','=','=')) { js->tok = TOK_SEQ; js->tlen = 3; }
3897
+
else if (MATCH2('=','=')) { js->tok = TOK_EQ; js->tlen = 2; }
3898
+
else if (MATCH2('=','>')) { js->tok = TOK_ARROW; js->tlen = 2; }
3899
+
else { js->tok = TOK_ASSIGN; js->tlen = 1; }
3900
+
break;
3901
+
3902
+
case '<':
3903
+
if (MATCH3('<','<','=')) { js->tok = TOK_SHL_ASSIGN; js->tlen = 3; }
3904
+
else if (MATCH2('<','<')) { js->tok = TOK_SHL; js->tlen = 2; }
3905
+
else if (MATCH2('<','=')) { js->tok = TOK_LE; js->tlen = 2; }
3906
+
else { js->tok = TOK_LT; js->tlen = 1; }
3907
+
break;
3908
+
3909
+
case '>':
3910
+
if (MATCH4('>','>','>','=')) { js->tok = TOK_ZSHR_ASSIGN; js->tlen = 4; }
3911
+
else if (MATCH3('>','>','>')) { js->tok = TOK_ZSHR; js->tlen = 3; }
3912
+
else if (MATCH3('>','>','=')) { js->tok = TOK_SHR_ASSIGN; js->tlen = 3; }
3913
+
else if (MATCH2('>','>')) { js->tok = TOK_SHR; js->tlen = 2; }
3914
+
else if (MATCH2('>','=')) { js->tok = TOK_GE; js->tlen = 2; }
3915
+
else { js->tok = TOK_GT; js->tlen = 1; }
3916
+
break;
3917
+
3918
+
case '&':
3919
+
if (MATCH3('&','&','=')) { js->tok = TOK_LAND_ASSIGN; js->tlen = 3; }
3920
+
else if (MATCH2('&','&')) { js->tok = TOK_LAND; js->tlen = 2; }
3921
+
else if (MATCH2('&','=')) { js->tok = TOK_AND_ASSIGN; js->tlen = 2; }
3922
+
else { js->tok = TOK_AND; js->tlen = 1; }
3923
+
break;
3924
+
3925
+
case '|':
3926
+
if (MATCH3('|','|','=')) { js->tok = TOK_LOR_ASSIGN; js->tlen = 3; }
3927
+
else if (MATCH2('|','|')) { js->tok = TOK_LOR; js->tlen = 2; }
3928
+
else if (MATCH2('|','=')) { js->tok = TOK_OR_ASSIGN; js->tlen = 2; }
3929
+
else { js->tok = TOK_OR; js->tlen = 1; }
3930
+
break;
3931
+
3932
+
case '+':
3933
+
if (MATCH2('+','+')) { js->tok = TOK_POSTINC; js->tlen = 2; }
3934
+
else if (MATCH2('+','=')) { js->tok = TOK_PLUS_ASSIGN; js->tlen = 2; }
3935
+
else { js->tok = TOK_PLUS; js->tlen = 1; }
3936
+
break;
3937
+
3938
+
case '-':
3939
+
if (MATCH2('-','-')) { js->tok = TOK_POSTDEC; js->tlen = 2; }
3940
+
else if (MATCH2('-','=')) { js->tok = TOK_MINUS_ASSIGN; js->tlen = 2; }
3941
+
else { js->tok = TOK_MINUS; js->tlen = 1; }
3942
+
break;
3943
+
3944
+
case '*':
3945
+
if (MATCH2('*','*')) { js->tok = TOK_EXP; js->tlen = 2; }
3946
+
else if (MATCH2('*','=')) { js->tok = TOK_MUL_ASSIGN; js->tlen = 2; }
3947
+
else { js->tok = TOK_MUL; js->tlen = 1; }
3948
+
break;
3949
+
3950
+
case '/':
3951
+
if (MATCH2('/','=')) { js->tok = TOK_DIV_ASSIGN; js->tlen = 2; }
3952
+
else { js->tok = TOK_DIV; js->tlen = 1; }
3953
+
break;
3954
+
3955
+
case '%':
3956
+
if (MATCH2('%','=')) { js->tok = TOK_REM_ASSIGN; js->tlen = 2; }
3957
+
else { js->tok = TOK_REM; js->tlen = 1; }
3958
+
break;
3959
+
3960
+
case '^':
3961
+
if (MATCH2('^','=')) { js->tok = TOK_XOR_ASSIGN; js->tlen = 2; }
3962
+
else { js->tok = TOK_XOR; js->tlen = 1; }
3963
+
break;
3964
+
3965
+
case '.':
3966
+
if (MATCH3('.','.', '.')) { js->tok = TOK_REST; js->tlen = 3; }
3967
+
else if (rem > 1 && IS_DIGIT(buf[1])) {
3968
+
double val;
3969
+
js->tlen = parse_decimal(buf, rem, &val);
3970
+
js->tval = tov(val);
3971
+
js->tok = TOK_NUMBER;
3972
+
}
3973
+
else { js->tok = TOK_DOT; js->tlen = 1; }
3974
+
break;
3975
+
3976
+
default:
3977
+
return 0;
3978
+
}
3979
+
3980
+
#undef MATCH2
3981
+
#undef MATCH3
3982
+
#undef MATCH4
3983
+
3984
+
return js->tok;
3985
+
}
3986
+
3987
+
static uint8_t next(struct js *js) {
3988
+
if (likely(js->consumed == 0)) return js->tok;
3989
+
3582
3990
js->consumed = 0;
3583
3991
js->tok = TOK_ERR;
3584
3992
js->toff = js->pos = skiptonext(js->code, js->clen, js->pos, &js->had_newline);
3585
3993
js->tlen = 0;
3586
-
3994
+
3995
+
if (unlikely(js->toff >= js->clen)) {
3996
+
js->tok = TOK_EOF;
3997
+
return TOK_EOF;
3998
+
}
3999
+
3587
4000
const char *buf = js->code + js->toff;
3588
-
if (js->toff >= js->clen) { js->tok = TOK_EOF; return js->tok; }
3589
-
3590
-
#define TOK(T, LEN) { js->tok = T; js->tlen = (LEN); break; }
3591
-
#define LOOK(OFS, CH) js->toff + OFS < js->clen && buf[OFS] == CH
3592
-
3593
-
switch (buf[0]) {
3594
-
case '?': if (LOOK(1, '?') && LOOK(2, '=')) TOK(TOK_NULLISH_ASSIGN, 3); if (LOOK(1, '?')) TOK(TOK_NULLISH, 2); if (LOOK(1, '.')) TOK(TOK_OPTIONAL_CHAIN, 2); TOK(TOK_Q, 1);
3595
-
case ':': TOK(TOK_COLON, 1);
3596
-
case '(': TOK(TOK_LPAREN, 1);
3597
-
case ')': TOK(TOK_RPAREN, 1);
3598
-
case '{': TOK(TOK_LBRACE, 1);
3599
-
case '}': TOK(TOK_RBRACE, 1);
3600
-
case '[': TOK(TOK_LBRACKET, 1);
3601
-
case ']': TOK(TOK_RBRACKET, 1);
3602
-
case ';': TOK(TOK_SEMICOLON, 1);
3603
-
case ',': TOK(TOK_COMMA, 1);
3604
-
case '!': if (LOOK(1, '=') && LOOK(2, '=')) TOK(TOK_SNE, 3); if (LOOK(1, '=')) TOK(TOK_NE, 2); TOK(TOK_NOT, 1);
3605
-
3606
-
case '.':
3607
-
if (LOOK(1, '.') && LOOK(2, '.')) TOK(TOK_REST, 3);
3608
-
if (js->toff + 1 < js->clen && is_digit(buf[1])) {
3609
-
char clean[64];
3610
-
jsoff_t ci = 0, numlen = 0;
3611
-
while (js->toff + numlen < js->clen && (is_digit(buf[numlen]) || buf[numlen] == '.' || buf[numlen] == 'e' || buf[numlen] == 'E' || buf[numlen] == '+' || buf[numlen] == '-' || buf[numlen] == '_') && ci < 63) {
3612
-
if (numlen > 0 && (buf[numlen] == '+' || buf[numlen] == '-') && buf[numlen-1] != 'e' && buf[numlen-1] != 'E') break;
3613
-
if (buf[numlen] != '_') clean[ci++] = buf[numlen];
3614
-
numlen++;
3615
-
}
3616
-
clean[ci] = '\0';
3617
-
js->tval = tov(strtod(clean, NULL));
3618
-
TOK(TOK_NUMBER, numlen);
3619
-
}
3620
-
TOK(TOK_DOT, 1);
3621
-
3622
-
case '~': TOK(TOK_TILDA, 1);
3623
-
case '-': if (LOOK(1, '-')) TOK(TOK_POSTDEC, 2); if (LOOK(1, '=')) TOK(TOK_MINUS_ASSIGN, 2); TOK(TOK_MINUS, 1);
3624
-
case '+': if (LOOK(1, '+')) TOK(TOK_POSTINC, 2); if (LOOK(1, '=')) TOK(TOK_PLUS_ASSIGN, 2); TOK(TOK_PLUS, 1);
3625
-
case '*': if (LOOK(1, '*')) TOK(TOK_EXP, 2); if (LOOK(1, '=')) TOK(TOK_MUL_ASSIGN, 2); TOK(TOK_MUL, 1);
3626
-
case '/': if (LOOK(1, '=')) TOK(TOK_DIV_ASSIGN, 2); TOK(TOK_DIV, 1);
3627
-
case '%': if (LOOK(1, '=')) TOK(TOK_REM_ASSIGN, 2); TOK(TOK_REM, 1);
3628
-
case '&': if (LOOK(1, '&') && LOOK(2, '=')) TOK(TOK_LAND_ASSIGN, 3); if (LOOK(1, '&')) TOK(TOK_LAND, 2); if (LOOK(1, '=')) TOK(TOK_AND_ASSIGN, 2); TOK(TOK_AND, 1);
3629
-
case '|': if (LOOK(1, '|') && LOOK(2, '=')) TOK(TOK_LOR_ASSIGN, 3); if (LOOK(1, '|')) TOK(TOK_LOR, 2); if (LOOK(1, '=')) TOK(TOK_OR_ASSIGN, 2); TOK(TOK_OR, 1);
3630
-
case '=': if (LOOK(1, '=') && LOOK(2, '=')) TOK(TOK_SEQ, 3); if (LOOK(1, '=')) TOK(TOK_EQ, 2); if (LOOK(1, '>')) TOK(TOK_ARROW, 2); TOK(TOK_ASSIGN, 1);
3631
-
case '<': if (LOOK(1, '<') && LOOK(2, '=')) TOK(TOK_SHL_ASSIGN, 3); if (LOOK(1, '<')) TOK(TOK_SHL, 2); if (LOOK(1, '=')) TOK(TOK_LE, 2); TOK(TOK_LT, 1);
3632
-
case '>': if (LOOK(1, '>') && LOOK(2, '>') && LOOK(3, '=')) TOK(TOK_ZSHR_ASSIGN, 4); if (LOOK(1, '>') && LOOK(2, '>')) TOK(TOK_ZSHR, 3); if (LOOK(1, '>') && LOOK(2, '=')) TOK(TOK_SHR_ASSIGN, 3); if (LOOK(1, '>')) TOK(TOK_SHR, 2); if (LOOK(1, '=')) TOK(TOK_GE, 2); TOK(TOK_GT, 1);
3633
-
case '^': if (LOOK(1, '=')) TOK(TOK_XOR_ASSIGN, 2); TOK(TOK_XOR, 1);
3634
-
case '#': TOK(TOK_HASH, 1);
3635
-
case '`':
3636
-
js->tlen++;
3637
-
while (js->toff + js->tlen < js->clen && buf[js->tlen] != '`') {
3638
-
uint8_t increment = 1;
3639
-
if (buf[js->tlen] == '\\') {
3640
-
if (js->toff + js->tlen + 2 > js->clen) break;
3641
-
increment = 2;
3642
-
}
3643
-
js->tlen += increment;
3644
-
}
3645
-
if (buf[js->tlen] == '`') js->tok = TOK_TEMPLATE, js->tlen++;
3646
-
break;
3647
-
case '"': case '\'':
3648
-
js->tlen++;
3649
-
while (js->toff + js->tlen < js->clen && buf[js->tlen] != buf[0]) {
3650
-
uint8_t increment = 1;
3651
-
if (buf[js->tlen] == '\\') {
3652
-
if (js->toff + js->tlen + 2 > js->clen) break;
3653
-
increment = 2;
3654
-
if (buf[js->tlen + 1] == 'x') {
3655
-
if (js->toff + js->tlen + 4 > js->clen) break;
3656
-
increment = 4;
3657
-
} else if (buf[js->tlen + 1] == 'u') {
3658
-
if (js->toff + js->tlen + 6 > js->clen) break;
3659
-
increment = 6;
3660
-
}
3661
-
}
3662
-
js->tlen += increment;
3663
-
}
3664
-
if (buf[0] == buf[js->tlen]) js->tok = TOK_STRING, js->tlen++;
3665
-
break;
3666
-
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': {
3667
-
if ((js->flags & F_STRICT) && buf[0] == '0' && js->toff + 1 < js->clen &&
3668
-
is_digit(buf[1]) && buf[1] != 'x' && buf[1] != 'X' && buf[1] != 'b' && buf[1] != 'B' && buf[1] != 'o' && buf[1] != 'O') {
3669
-
js->tok = TOK_ERR;
3670
-
js->tlen = 1;
3671
-
break;
3672
-
}
3673
-
3674
-
char *end;
3675
-
double value = 0;
3676
-
jsoff_t numlen = 0;
3677
-
3678
-
if (buf[0] == '0' && js->toff + 2 < js->clen) {
3679
-
if (buf[1] == 'b' || buf[1] == 'B') {
3680
-
numlen = 2;
3681
-
while (js->toff + numlen < js->clen && (buf[numlen] == '0' || buf[numlen] == '1' || buf[numlen] == '_')) {
3682
-
if (buf[numlen] != '_') value = value * 2 + (buf[numlen] - '0');
3683
-
numlen++;
3684
-
}
3685
-
js->tval = tov(value);
3686
-
} else if (buf[1] == 'o' || buf[1] == 'O') {
3687
-
numlen = 2;
3688
-
while (js->toff + numlen < js->clen && ((buf[numlen] >= '0' && buf[numlen] <= '7') || buf[numlen] == '_')) {
3689
-
if (buf[numlen] != '_') value = value * 8 + (buf[numlen] - '0');
3690
-
numlen++;
3691
-
}
3692
-
js->tval = tov(value);
3693
-
} else if (buf[1] == 'x' || buf[1] == 'X') {
3694
-
char clean[64];
3695
-
jsoff_t ci = 0;
3696
-
numlen = 0;
3697
-
while (js->toff + numlen < js->clen && (is_xdigit(buf[numlen]) || buf[numlen] == 'x' || buf[numlen] == 'X' || buf[numlen] == '_') && ci < 63) {
3698
-
if (buf[numlen] != '_') clean[ci++] = buf[numlen];
3699
-
numlen++;
3700
-
}
3701
-
clean[ci] = '\0';
3702
-
js->tval = tov(strtod(clean, &end));
3703
-
} else if (buf[1] >= '0' && buf[1] <= '7') {
3704
-
numlen = 1;
3705
-
while (js->toff + numlen < js->clen && buf[numlen] >= '0' && buf[numlen] <= '7') {
3706
-
value = value * 8 + (buf[numlen] - '0');
3707
-
numlen++;
3708
-
}
3709
-
js->tval = tov(value);
3710
-
} else {
3711
-
char clean[64];
3712
-
jsoff_t ci = 0;
3713
-
numlen = 0;
3714
-
while (js->toff + numlen < js->clen && (is_digit(buf[numlen]) || buf[numlen] == '.' || buf[numlen] == 'e' || buf[numlen] == 'E' || buf[numlen] == '+' || buf[numlen] == '-' || buf[numlen] == '_') && ci < 63) {
3715
-
if (numlen > 0 && (buf[numlen] == '+' || buf[numlen] == '-') && buf[numlen-1] != 'e' && buf[numlen-1] != 'E') break;
3716
-
if (buf[numlen] != '_') clean[ci++] = buf[numlen];
3717
-
numlen++;
3718
-
}
3719
-
clean[ci] = '\0';
3720
-
js->tval = tov(strtod(clean, &end));
3721
-
}
3722
-
} else {
3723
-
char clean[64];
3724
-
jsoff_t ci = 0;
3725
-
numlen = 0;
3726
-
while (js->toff + numlen < js->clen && (is_digit(buf[numlen]) || buf[numlen] == '.' || buf[numlen] == 'e' || buf[numlen] == 'E' || buf[numlen] == '+' || buf[numlen] == '-' || buf[numlen] == '_') && ci < 63) {
3727
-
if (numlen > 0 && (buf[numlen] == '+' || buf[numlen] == '-') && buf[numlen-1] != 'e' && buf[numlen-1] != 'E') break;
3728
-
if (buf[numlen] != '_') clean[ci++] = buf[numlen];
3729
-
numlen++;
3730
-
}
3731
-
clean[ci] = '\0';
3732
-
js->tval = tov(strtod(clean, &end));
3733
-
}
3734
-
3735
-
if (js->toff + numlen < js->clen && buf[numlen] == 'n') {
3736
-
js->tok = TOK_BIGINT;
3737
-
js->tlen = numlen + 1;
3738
-
} else {
3739
-
TOK(TOK_NUMBER, numlen);
3740
-
}
3741
-
break;
4001
+
jsoff_t rem = js->clen - js->toff;
4002
+
uint8_t c = (uint8_t)buf[0];
4003
+
4004
+
if (likely(c < 128)) {
4005
+
uint8_t simple_tok = single_char_tok[c];
4006
+
if (simple_tok != 0) {
4007
+
js->tok = simple_tok;
4008
+
js->tlen = 1;
4009
+
js->pos = js->toff + 1;
4010
+
return simple_tok;
3742
4011
}
3743
-
default: js->tok = parseident(buf, js->clen - js->toff, &js->tlen); break;
3744
4012
}
3745
-
3746
-
if (js->tlen == 0) js->tlen = 1;
3747
-
js->pos = js->toff + js->tlen;
3748
-
return js->tok;
4013
+
4014
+
if (likely(IS_IDENT1(c))) {
4015
+
js->tok = parseident(buf, rem, &js->tlen);
4016
+
js->pos = js->toff + js->tlen;
4017
+
return js->tok;
4018
+
}
4019
+
4020
+
if (IS_DIGIT(c)) {
4021
+
parse_number(js, buf, rem);
4022
+
if (js->tlen == 0) js->tlen = 1;
4023
+
js->pos = js->toff + js->tlen;
4024
+
return js->tok;
4025
+
}
4026
+
4027
+
if (c == '"' || c == '\'') {
4028
+
scan_string(js, buf, rem, c);
4029
+
if (js->tlen == 0) js->tlen = 1;
4030
+
js->pos = js->toff + js->tlen;
4031
+
return js->tok;
4032
+
}
4033
+
4034
+
if (c == '`') {
4035
+
scan_template(js, buf, rem);
4036
+
if (js->tlen == 0) js->tlen = 1;
4037
+
js->pos = js->toff + js->tlen;
4038
+
return js->tok;
4039
+
}
4040
+
4041
+
if (parse_operator(js, buf, rem)) {
4042
+
if (js->tlen == 0) js->tlen = 1;
4043
+
js->pos = js->toff + js->tlen;
4044
+
return js->tok;
4045
+
}
4046
+
4047
+
js->tok = TOK_ERR;
4048
+
js->tlen = 1;
4049
+
js->pos = js->toff + 1;
4050
+
return TOK_ERR;
3749
4051
}
3750
4052
3751
4053
static inline uint8_t lookahead(struct js *js) {