| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
# This code can be redistributed and modified under the terms of the GNU |
|
2
|
|
|
|
|
|
|
# General Public License as published by the Free Software Foundation, either |
|
3
|
|
|
|
|
|
|
# version 3 of the License, or (at your option) any later version. |
|
4
|
|
|
|
|
|
|
# See the "COPYING" file for details. |
|
5
|
|
|
|
|
|
|
package HTML::Blitz::Matcher 0.1001; |
|
6
|
11
|
|
|
11
|
|
84
|
use HTML::Blitz::pragma; |
|
|
11
|
|
|
|
|
23
|
|
|
|
11
|
|
|
|
|
87
|
|
|
7
|
11
|
|
|
|
|
1014
|
use HTML::Blitz::SelectorType qw( |
|
8
|
|
|
|
|
|
|
LT_DESCENDANT |
|
9
|
|
|
|
|
|
|
LT_CHILD |
|
10
|
|
|
|
|
|
|
LT_SIBLING |
|
11
|
|
|
|
|
|
|
LT_ADJACENT_SIBLING |
|
12
|
11
|
|
|
11
|
|
12707
|
); |
|
|
11
|
|
|
|
|
41
|
|
|
13
|
11
|
|
|
11
|
|
105
|
use Scalar::Util (); |
|
|
11
|
|
|
|
|
26
|
|
|
|
11
|
|
|
|
|
632
|
|
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
use constant { |
|
16
|
11
|
|
|
|
|
2285
|
INTBITS => length(sprintf '%b', ~0), |
|
17
|
11
|
|
|
11
|
|
64
|
}; |
|
|
11
|
|
|
|
|
26
|
|
|
18
|
|
|
|
|
|
|
|
|
19
|
276
|
50
|
|
276
|
0
|
823
|
method new($class: $rules) { |
|
|
276
|
50
|
|
|
|
695
|
|
|
|
276
|
|
|
|
|
552
|
|
|
|
276
|
|
|
|
|
567
|
|
|
|
276
|
|
|
|
|
427
|
|
|
20
|
276
|
|
|
|
|
11440
|
bless { |
|
21
|
|
|
|
|
|
|
slices => [ |
|
22
|
|
|
|
|
|
|
map [ $_, { cur => 0, stack => [{ extra_bits => 0 }] } ], @$rules |
|
23
|
|
|
|
|
|
|
], |
|
24
|
|
|
|
|
|
|
doc_state => [ |
|
25
|
|
|
|
|
|
|
{ |
|
26
|
|
|
|
|
|
|
nth_child => 0, |
|
27
|
|
|
|
|
|
|
nth_child_of_type => {}, |
|
28
|
|
|
|
|
|
|
on_leave => [], |
|
29
|
|
|
|
|
|
|
}, |
|
30
|
|
|
|
|
|
|
], |
|
31
|
|
|
|
|
|
|
}, $class |
|
32
|
|
|
|
|
|
|
} |
|
33
|
|
|
|
|
|
|
|
|
34
|
1546
|
|
|
1546
|
|
9094
|
fun _guniq(@values) { |
|
|
1546
|
|
|
|
|
9752
|
|
|
35
|
1546
|
|
|
|
|
5453
|
my ($seen_undef, %seen_ref, %seen_str); |
|
36
|
|
|
|
|
|
|
grep |
|
37
|
|
|
|
|
|
|
!( |
|
38
|
|
|
|
|
|
|
ref($_) ? $seen_ref{Scalar::Util::refaddr $_} : |
|
39
|
1546
|
0
|
|
|
|
9896
|
defined($_) ? $seen_str{$_} : |
|
|
|
50
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
$seen_undef |
|
41
|
|
|
|
|
|
|
)++, |
|
42
|
|
|
|
|
|
|
@values |
|
43
|
|
|
|
|
|
|
} |
|
44
|
|
|
|
|
|
|
|
|
45
|
1546
|
50
|
|
1546
|
0
|
3274
|
method enter($tag, $attributes) { |
|
|
1546
|
50
|
|
|
|
3027
|
|
|
|
1546
|
|
|
|
|
2392
|
|
|
|
1546
|
|
|
|
|
20500
|
|
|
|
1546
|
|
|
|
|
2165
|
|
|
46
|
1546
|
|
|
|
|
2689
|
my $doc_state = $self->{doc_state}; |
|
47
|
1546
|
|
|
|
|
6509
|
my $dsp = $doc_state->[-1]; |
|
48
|
1546
|
|
|
|
|
2896
|
my $nth_child = ++$dsp->{nth_child}; |
|
49
|
1546
|
|
|
|
|
7556
|
my $nth_child_of_type = ++$dsp->{nth_child_of_type}{$tag}; |
|
50
|
1546
|
|
|
|
|
9493
|
push @$doc_state, { |
|
51
|
|
|
|
|
|
|
nth_child => 0, |
|
52
|
|
|
|
|
|
|
nth_child_of_type => {}, |
|
53
|
|
|
|
|
|
|
on_leave => [], |
|
54
|
|
|
|
|
|
|
}; |
|
55
|
|
|
|
|
|
|
|
|
56
|
1546
|
|
|
|
|
2653
|
my @ret; |
|
57
|
|
|
|
|
|
|
|
|
58
|
1546
|
|
|
|
|
2076
|
for my $slice (@{$self->{slices}}) { |
|
|
1546
|
|
|
|
|
8134
|
|
|
59
|
1771
|
|
|
|
|
3628
|
my ($glass, $goop) = @$slice; |
|
60
|
1771
|
|
|
|
|
2887
|
my $cur = $goop->{cur}; |
|
61
|
1771
|
|
|
|
|
4337
|
my $stack = $goop->{stack}; |
|
62
|
1771
|
|
|
|
|
11114
|
my $sp = $stack->[-1]; |
|
63
|
1771
|
|
|
|
|
2907
|
my $extra_volatile = $sp->{extra_volatile}; |
|
64
|
1771
|
|
|
|
|
3223
|
$sp->{extra_volatile} = []; |
|
65
|
|
|
|
|
|
|
|
|
66
|
1771
|
|
|
|
|
12050
|
push @$stack, my $sp_next = { |
|
67
|
|
|
|
|
|
|
extra_bits => 0, |
|
68
|
|
|
|
|
|
|
}; |
|
69
|
1771
|
|
|
|
|
7186
|
my $cur_next; |
|
70
|
|
|
|
|
|
|
|
|
71
|
1771
|
|
|
|
|
2560
|
for my $i ($cur, @{$sp->{extra}}, @$extra_volatile) { |
|
|
1771
|
|
|
|
|
4980
|
|
|
72
|
1806
|
|
|
|
|
3021
|
my $sss = $glass->[$i]; |
|
73
|
1806
|
100
|
|
|
|
7119
|
$sss->matches($tag, $attributes, $nth_child, $nth_child_of_type) |
|
74
|
|
|
|
|
|
|
or next; |
|
75
|
|
|
|
|
|
|
|
|
76
|
521
|
|
|
|
|
1569
|
my $link = $sss->link_type; |
|
77
|
521
|
|
|
|
|
890
|
my $k = $i + 1; |
|
78
|
521
|
|
|
|
|
937
|
my $bit_shift = $k - $cur - 1; |
|
79
|
521
|
50
|
|
|
|
1141
|
$bit_shift < INTBITS |
|
80
|
|
|
|
|
|
|
or die "Internal error: Too many combinators in a single selector (" . ($bit_shift + 1) . " exceeds limit of " . INTBITS . ")"; |
|
81
|
521
|
|
|
|
|
905
|
my $bit = 1 << $bit_shift; |
|
82
|
|
|
|
|
|
|
|
|
83
|
521
|
100
|
|
|
|
1169
|
if (!defined $link) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
84
|
439
|
|
|
|
|
1234
|
push @ret, $glass->[$k]; |
|
85
|
|
|
|
|
|
|
} elsif ($link eq LT_DESCENDANT) { |
|
86
|
33
|
|
|
|
|
57
|
$cur_next = $k; |
|
87
|
|
|
|
|
|
|
} elsif ($link eq LT_CHILD) { |
|
88
|
44
|
50
|
|
|
|
72
|
if (!($sp_next->{extra_bits} & $bit)) { |
|
89
|
44
|
|
|
|
|
49
|
$sp_next->{extra_bits} |= $bit; |
|
90
|
44
|
|
|
|
|
4072
|
push @{$sp_next->{extra}}, $k; |
|
|
44
|
|
|
|
|
90
|
|
|
91
|
|
|
|
|
|
|
} |
|
92
|
|
|
|
|
|
|
} elsif ($link eq LT_SIBLING) { |
|
93
|
1
|
50
|
|
|
|
4
|
if (!($sp->{extra_bits} & $bit)) { |
|
94
|
1
|
|
|
|
|
2
|
$sp->{extra_bits} |= $bit; |
|
95
|
1
|
|
|
|
|
3
|
push @{$sp->{extra}}, $k; |
|
|
1
|
|
|
|
|
2
|
|
|
96
|
|
|
|
|
|
|
} |
|
97
|
|
|
|
|
|
|
} elsif ($link eq LT_ADJACENT_SIBLING) { |
|
98
|
4
|
|
|
|
|
5
|
push @{$sp->{extra_volatile}}, $k; |
|
|
4
|
|
|
|
|
8
|
|
|
99
|
|
|
|
|
|
|
} else { |
|
100
|
0
|
|
|
|
|
0
|
die "Internal error: unexpected selector combinator '$link'"; |
|
101
|
|
|
|
|
|
|
} |
|
102
|
|
|
|
|
|
|
} |
|
103
|
|
|
|
|
|
|
|
|
104
|
1771
|
100
|
|
|
|
5290
|
if (defined $cur_next) { |
|
105
|
33
|
|
|
|
|
70
|
$stack->[-1] = { |
|
106
|
|
|
|
|
|
|
cur => $cur, |
|
107
|
|
|
|
|
|
|
extra_bits => 0, |
|
108
|
|
|
|
|
|
|
}; |
|
109
|
33
|
|
|
|
|
99
|
$goop->{cur} = $cur_next; |
|
110
|
|
|
|
|
|
|
} |
|
111
|
|
|
|
|
|
|
} |
|
112
|
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
_guniq @ret |
|
114
|
1546
|
|
|
|
|
5342
|
} |
|
115
|
|
|
|
|
|
|
|
|
116
|
1508
|
50
|
|
1508
|
0
|
7387
|
method leave(@args) { |
|
|
1508
|
|
|
|
|
2389
|
|
|
|
1508
|
|
|
|
|
3098
|
|
|
|
1508
|
|
|
|
|
2069
|
|
|
117
|
1508
|
|
|
|
|
3555
|
my $dsp = pop @{$self->{doc_state}}; |
|
|
1508
|
|
|
|
|
3458
|
|
|
118
|
1508
|
100
|
|
|
|
4423
|
if (defined(my $marker = $dsp->{marker})) { |
|
119
|
4
|
|
|
|
|
8
|
splice @{$self->{slices}}, $marker; |
|
|
4
|
|
|
|
|
25
|
|
|
120
|
|
|
|
|
|
|
} |
|
121
|
|
|
|
|
|
|
|
|
122
|
1508
|
|
|
|
|
2235
|
for my $slice (@{$self->{slices}}) { |
|
|
1508
|
|
|
|
|
5910
|
|
|
123
|
1736
|
|
|
|
|
2946
|
my $goop = $slice->[1]; |
|
124
|
1736
|
|
|
|
|
5723
|
my $stack = $goop->{stack}; |
|
125
|
1736
|
|
|
|
|
2712
|
my $sp_prev = pop @$stack; |
|
126
|
1736
|
100
|
|
|
|
8324
|
if (defined(my $cur = $sp_prev->{cur})) { |
|
127
|
33
|
|
|
|
|
72
|
$goop->{cur} = $cur; |
|
128
|
|
|
|
|
|
|
} |
|
129
|
|
|
|
|
|
|
} |
|
130
|
|
|
|
|
|
|
|
|
131
|
1508
|
|
|
|
|
3026
|
for my $cb (reverse @{$dsp->{on_leave}}) { |
|
|
1508
|
|
|
|
|
8054
|
|
|
132
|
9
|
|
|
|
|
29
|
$cb->(@args); |
|
133
|
|
|
|
|
|
|
} |
|
134
|
|
|
|
|
|
|
} |
|
135
|
|
|
|
|
|
|
|
|
136
|
9
|
50
|
|
9
|
0
|
44
|
method on_leave($callback) { |
|
|
9
|
50
|
|
|
|
22
|
|
|
|
9
|
|
|
|
|
17
|
|
|
|
9
|
|
|
|
|
25
|
|
|
|
9
|
|
|
|
|
12
|
|
|
137
|
9
|
|
|
|
|
20
|
push @{$self->{doc_state}[-1]{on_leave}}, $callback; |
|
|
9
|
|
|
|
|
46
|
|
|
138
|
|
|
|
|
|
|
} |
|
139
|
|
|
|
|
|
|
|
|
140
|
7
|
50
|
|
7
|
0
|
23
|
method add_temp_rule(@temp_rules) { |
|
|
7
|
|
|
|
|
9
|
|
|
|
7
|
|
|
|
|
18
|
|
|
|
7
|
|
|
|
|
12
|
|
|
141
|
7
|
|
|
|
|
24
|
my $slices = $self->{slices}; |
|
142
|
7
|
|
66
|
|
|
37
|
$self->{doc_state}[-1]{marker} //= @$slices; |
|
143
|
7
|
|
|
|
|
55
|
push @$slices, map [ $_, { cur => 0, stack => [{ extra_bits => 0 }] } ], @temp_rules; |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
1 |