| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WARC::Record::Block; # -*- CPerl -*- |
|
2
|
|
|
|
|
|
|
|
|
3
|
28
|
|
|
28
|
|
59889
|
use strict; |
|
|
28
|
|
|
|
|
67
|
|
|
|
28
|
|
|
|
|
865
|
|
|
4
|
28
|
|
|
28
|
|
139
|
use warnings; |
|
|
28
|
|
|
|
|
46
|
|
|
|
28
|
|
|
|
|
1128
|
|
|
5
|
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
our @ISA = qw(); |
|
7
|
|
|
|
|
|
|
|
|
8
|
28
|
|
|
28
|
|
498
|
use WARC; *WARC::Record::Block::VERSION = \$WARC::VERSION; |
|
|
28
|
|
|
|
|
49
|
|
|
|
28
|
|
|
|
|
805
|
|
|
9
|
|
|
|
|
|
|
|
|
10
|
28
|
|
|
28
|
|
134
|
use Carp; |
|
|
28
|
|
|
|
|
55
|
|
|
|
28
|
|
|
|
|
1591
|
|
|
11
|
28
|
|
|
28
|
|
170
|
use Fcntl qw/SEEK_SET SEEK_CUR SEEK_END/; |
|
|
28
|
|
|
|
|
62
|
|
|
|
28
|
|
|
|
|
1777
|
|
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
# This implementation uses an array as the underlying object. |
|
14
|
|
|
|
|
|
|
|
|
15
|
28
|
|
|
28
|
|
171
|
use constant { BASE => 0, LENGTH => 1, HANDLE => 2, PARENT => 3, AT_EOF => 4 }; |
|
|
28
|
|
|
|
|
50
|
|
|
|
28
|
|
|
|
|
2621
|
|
|
16
|
28
|
|
|
28
|
|
201
|
use constant OBJECT_INIT => undef, undef, undef, undef, 0; |
|
|
28
|
|
|
|
|
56
|
|
|
|
28
|
|
|
|
|
26347
|
|
|
17
|
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
sub _dbg_dump { |
|
19
|
26
|
|
|
26
|
|
686
|
my $self = shift; |
|
20
|
|
|
|
|
|
|
|
|
21
|
26
|
|
|
|
|
63
|
my $out = 'record block '.$self->[BASE].' +> '.$self->[LENGTH]; |
|
22
|
26
|
|
|
|
|
80
|
$out .= ' @'.((tell $self->[HANDLE]) - $self->[BASE]); |
|
23
|
26
|
100
|
|
|
|
60
|
$out .= ' [EOF]' if $self->[AT_EOF]; |
|
24
|
26
|
|
|
|
|
34
|
$out .= "\n"; |
|
25
|
|
|
|
|
|
|
|
|
26
|
26
|
|
|
|
|
60
|
return $out; |
|
27
|
|
|
|
|
|
|
} |
|
28
|
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
sub TIEHANDLE { |
|
30
|
152
|
|
|
152
|
|
6269
|
my $class = shift; |
|
31
|
152
|
|
|
|
|
201
|
my $parent = shift; |
|
32
|
|
|
|
|
|
|
|
|
33
|
152
|
|
|
|
|
327
|
my $handle = $parent->volume->open; |
|
34
|
|
|
|
|
|
|
|
|
35
|
152
|
100
|
|
|
|
1334
|
if (defined $parent->{compression}) { |
|
36
|
1
|
50
|
|
|
|
3
|
seek $handle, $parent->offset, SEEK_SET or die "seek: $!"; |
|
37
|
|
|
|
|
|
|
|
|
38
|
1
|
|
|
|
|
7
|
my $z_reader = $parent->{compression}; |
|
39
|
1
|
50
|
|
|
|
9
|
my $zhandle = new $z_reader ($handle, MultiStream => 0, Transparent => 0, |
|
40
|
|
|
|
|
|
|
AutoClose => 1) |
|
41
|
|
|
|
|
|
|
or die "$z_reader: ".$parent->_get_compression_error; |
|
42
|
|
|
|
|
|
|
|
|
43
|
1
|
|
|
|
|
1593
|
$handle = $zhandle; |
|
44
|
|
|
|
|
|
|
} |
|
45
|
|
|
|
|
|
|
|
|
46
|
152
|
|
|
|
|
369
|
my $ob = [OBJECT_INIT]; |
|
47
|
152
|
|
|
|
|
413
|
@$ob[PARENT, HANDLE] = ($parent, $handle); |
|
48
|
|
|
|
|
|
|
@$ob[BASE, LENGTH] = |
|
49
|
152
|
|
|
|
|
456
|
($parent->{data_offset}, $parent->field('Content-Length')); |
|
50
|
|
|
|
|
|
|
|
|
51
|
152
|
|
|
|
|
605
|
bless $ob, $class; |
|
52
|
|
|
|
|
|
|
|
|
53
|
152
|
|
|
|
|
412
|
$ob->SEEK(0, SEEK_SET); |
|
54
|
|
|
|
|
|
|
|
|
55
|
152
|
|
|
|
|
775
|
return $ob; |
|
56
|
|
|
|
|
|
|
} |
|
57
|
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
sub READLINE { |
|
59
|
283
|
|
|
283
|
|
409
|
my $self = shift; |
|
60
|
|
|
|
|
|
|
|
|
61
|
283
|
100
|
|
|
|
491
|
if (wantarray) { |
|
62
|
10
|
|
|
|
|
23
|
my @data = (); |
|
63
|
10
|
|
|
|
|
80
|
while (defined(my $line = $self->READLINE())) { push @data, $line } |
|
|
13
|
|
|
|
|
37
|
|
|
64
|
|
|
|
|
|
|
return @data |
|
65
|
10
|
|
|
|
|
94
|
} |
|
66
|
|
|
|
|
|
|
|
|
67
|
273
|
100
|
|
|
|
507
|
return undef if $self->[AT_EOF]; |
|
68
|
|
|
|
|
|
|
|
|
69
|
259
|
|
|
|
|
3980
|
my $line = readline $self->[HANDLE]; |
|
70
|
259
|
100
|
|
|
|
610
|
unless (defined $line) { $self->[AT_EOF] = 1; return undef } |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
3
|
|
|
71
|
|
|
|
|
|
|
|
|
72
|
258
|
|
|
|
|
584
|
my $excess = (tell $self->[HANDLE]) - $self->[BASE] - $self->[LENGTH]; |
|
73
|
258
|
100
|
|
|
|
477
|
$self->[AT_EOF] = 1 unless $excess < 0; |
|
74
|
258
|
100
|
|
|
|
571
|
$line = substr $line, 0, -$excess if $excess > 0; |
|
75
|
|
|
|
|
|
|
|
|
76
|
258
|
|
|
|
|
1036
|
return $line; |
|
77
|
|
|
|
|
|
|
} |
|
78
|
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
# This sub must rely on the aliasing effect of @_. |
|
80
|
|
|
|
|
|
|
sub READ { |
|
81
|
34
|
|
|
34
|
|
1138
|
my $self = shift; |
|
82
|
|
|
|
|
|
|
# args now: 0: buffer 1: length 2: offset into buffer or undef |
|
83
|
34
|
|
|
|
|
41
|
my $length = $_[1]; |
|
84
|
34
|
|
100
|
|
|
90
|
my $offset = $_[2] || 0; |
|
85
|
|
|
|
|
|
|
|
|
86
|
34
|
100
|
|
|
|
71
|
return 0 if $self->[AT_EOF]; |
|
87
|
|
|
|
|
|
|
|
|
88
|
31
|
|
|
|
|
61
|
my $excess = (($length + tell $self->[HANDLE]) |
|
89
|
|
|
|
|
|
|
- $self->[BASE] - $self->[LENGTH]); |
|
90
|
31
|
100
|
|
|
|
58
|
$self->[AT_EOF] = 1 unless $excess < 0; |
|
91
|
31
|
100
|
|
|
|
62
|
$length -= $excess if $excess > 0; |
|
92
|
|
|
|
|
|
|
|
|
93
|
31
|
|
|
|
|
34
|
my $buf; my $count = read $self->[HANDLE], $buf, $length; |
|
|
31
|
|
|
|
|
88
|
|
|
94
|
31
|
50
|
|
|
|
83
|
return undef unless defined $count; |
|
95
|
|
|
|
|
|
|
|
|
96
|
31
|
100
|
|
|
|
55
|
$_[0] = '' unless defined $_[0]; |
|
97
|
31
|
100
|
|
|
|
51
|
$_[0] .= "\0" x ($offset - length($_[0])) if $offset > length $_[0]; |
|
98
|
31
|
|
|
|
|
51
|
substr $_[0], $offset, (length($_[0]) - $offset), $buf; |
|
99
|
31
|
|
|
|
|
75
|
return $count; |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
sub GETC { |
|
103
|
23
|
|
|
23
|
|
391
|
my $self = shift; |
|
104
|
|
|
|
|
|
|
|
|
105
|
23
|
|
|
|
|
27
|
my $ch; |
|
106
|
23
|
100
|
|
|
|
34
|
return undef unless $self->READ($ch, 1); |
|
107
|
21
|
|
|
|
|
58
|
return $ch; |
|
108
|
|
|
|
|
|
|
} |
|
109
|
|
|
|
|
|
|
|
|
110
|
24
|
|
|
24
|
|
4795
|
sub EOF { (shift)->[AT_EOF] } |
|
111
|
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
sub SEEK { |
|
113
|
169
|
|
|
169
|
|
242
|
my $self = shift; |
|
114
|
169
|
|
|
|
|
209
|
my $offset = shift; |
|
115
|
169
|
|
|
|
|
222
|
my $whence = shift; |
|
116
|
|
|
|
|
|
|
|
|
117
|
169
|
|
|
|
|
221
|
my $npos; |
|
118
|
169
|
|
|
|
|
256
|
$self->[AT_EOF] = 0; |
|
119
|
|
|
|
|
|
|
|
|
120
|
169
|
100
|
|
|
|
324
|
if ($whence == SEEK_SET) { $npos = $offset } |
|
|
164
|
100
|
|
|
|
212
|
|
|
|
|
100
|
|
|
|
|
|
|
121
|
2
|
|
|
|
|
6
|
elsif ($whence == SEEK_CUR) { $npos = $offset + $self->TELL } |
|
122
|
2
|
|
|
|
|
4
|
elsif ($whence == SEEK_END) { $npos = $self->[LENGTH] + $offset } |
|
123
|
1
|
|
|
|
|
194
|
else { croak "unknown WHENCE $whence in call to seek" } |
|
124
|
|
|
|
|
|
|
|
|
125
|
168
|
100
|
|
|
|
326
|
return 0 if $npos < 0; |
|
126
|
167
|
100
|
|
|
|
452
|
if ($npos >= $self->[LENGTH]) { $self->[AT_EOF] = 1; $npos = $self->[LENGTH] } |
|
|
2
|
|
|
|
|
4
|
|
|
|
2
|
|
|
|
|
4
|
|
|
127
|
|
|
|
|
|
|
|
|
128
|
167
|
|
|
|
|
1461
|
seek $self->[HANDLE], $self->[BASE] + $npos, SEEK_SET; |
|
129
|
|
|
|
|
|
|
} |
|
130
|
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
sub TELL { |
|
132
|
39
|
|
|
39
|
|
424
|
my $self = shift; |
|
133
|
|
|
|
|
|
|
|
|
134
|
39
|
100
|
|
|
|
118
|
return $self->[LENGTH] if $self->[AT_EOF]; |
|
135
|
|
|
|
|
|
|
|
|
136
|
21
|
|
|
|
|
71
|
return ((tell $self->[HANDLE]) - $self->[BASE]); |
|
137
|
|
|
|
|
|
|
} |
|
138
|
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
sub CLOSE { |
|
140
|
51
|
|
|
51
|
|
90
|
my $self = shift; |
|
141
|
51
|
|
|
|
|
118
|
@$self[BASE, LENGTH, AT_EOF] = (0, 0, 1); |
|
142
|
51
|
|
|
|
|
683
|
close $self->[HANDLE]; |
|
143
|
|
|
|
|
|
|
} |
|
144
|
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
1; |
|
146
|
|
|
|
|
|
|
__END__ |