| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WWW::AUR::Iterator; |
|
2
|
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
24606
|
use warnings 'FATAL' => 'all'; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
44
|
|
|
4
|
1
|
|
|
1
|
|
6
|
use strict; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
72
|
|
|
5
|
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
555
|
use WWW::AUR::Package qw(); |
|
|
1
|
|
|
|
|
4
|
|
|
|
1
|
|
|
|
|
29
|
|
|
7
|
1
|
|
|
1
|
|
6
|
use WWW::AUR::URI qw( pkg_uri ); |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
50
|
|
|
8
|
1
|
|
|
1
|
|
6
|
use WWW::AUR qw( _path_params _category_name _useragent ); |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
1002
|
|
|
9
|
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
sub new |
|
11
|
|
|
|
|
|
|
{ |
|
12
|
4
|
|
|
4
|
0
|
2655
|
my $class = shift; |
|
13
|
4
|
|
|
|
|
17
|
my $self = bless {}, $class; |
|
14
|
4
|
|
|
|
|
15
|
$self->init( @_ ); |
|
15
|
|
|
|
|
|
|
} |
|
16
|
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
sub init |
|
18
|
|
|
|
|
|
|
{ |
|
19
|
4
|
|
|
4
|
0
|
8
|
my $self = shift; |
|
20
|
|
|
|
|
|
|
|
|
21
|
4
|
|
|
|
|
8
|
my $startidx; |
|
22
|
4
|
50
|
|
|
|
21
|
$startidx = shift if @_ % 2 == 1; |
|
23
|
|
|
|
|
|
|
|
|
24
|
4
|
|
|
|
|
30
|
%$self = ( %$self, _path_params( @_ )); |
|
25
|
4
|
|
|
|
|
23
|
$self->reset(); |
|
26
|
4
|
50
|
|
|
|
14
|
$self->set_pos( $startidx ) if $startidx; |
|
27
|
|
|
|
|
|
|
|
|
28
|
4
|
|
|
|
|
18
|
return $self; |
|
29
|
|
|
|
|
|
|
} |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub set_pos |
|
32
|
|
|
|
|
|
|
{ |
|
33
|
1
|
|
|
1
|
1
|
6
|
my ($self, $startidx) = @_; |
|
34
|
1
|
50
|
|
|
|
9
|
Carp::croak 'Argument to set_pos() must an integer' |
|
35
|
|
|
|
|
|
|
unless $startidx =~ /\A\d+\z/; |
|
36
|
|
|
|
|
|
|
|
|
37
|
1
|
|
|
|
|
6
|
$self->{'curridx'} = $startidx; |
|
38
|
1
|
|
|
|
|
3
|
$self->{'finished'} = 0; |
|
39
|
1
|
|
|
|
|
4
|
$self->{'packages'} = []; |
|
40
|
|
|
|
|
|
|
|
|
41
|
1
|
|
|
|
|
5
|
return; |
|
42
|
|
|
|
|
|
|
} |
|
43
|
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
sub reset |
|
45
|
|
|
|
|
|
|
{ |
|
46
|
4
|
|
|
4
|
1
|
11
|
my ($self) = @_; |
|
47
|
4
|
|
|
|
|
11
|
$self->{'curridx'} = 0; |
|
48
|
4
|
|
|
|
|
9
|
$self->{'finished'} = 0; |
|
49
|
4
|
|
|
|
|
12
|
$self->{'packages'} = []; |
|
50
|
4
|
|
|
|
|
21
|
$self->{'useragent'} = _useragent(); |
|
51
|
4
|
|
|
|
|
3929
|
return; |
|
52
|
|
|
|
|
|
|
} |
|
53
|
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
#---HELPER FUNCTION--- |
|
55
|
|
|
|
|
|
|
sub _pkglist_uri |
|
56
|
|
|
|
|
|
|
{ |
|
57
|
4
|
|
|
4
|
|
9
|
my ($startidx) = @_; |
|
58
|
4
|
|
|
|
|
25
|
return pkg_uri( q{SB} => q{n}, q{O} => $startidx, |
|
59
|
|
|
|
|
|
|
q{SO} => q{a}, q{PP} => 100 ); |
|
60
|
|
|
|
|
|
|
} |
|
61
|
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
#---PRIVATE METHOD--- |
|
63
|
|
|
|
|
|
|
sub _scrape_pkglist |
|
64
|
|
|
|
|
|
|
{ |
|
65
|
4
|
|
|
4
|
|
10
|
my ($self) = @_; |
|
66
|
|
|
|
|
|
|
|
|
67
|
4
|
|
|
|
|
16
|
my $uri = _pkglist_uri( $self->{'curridx'} ); |
|
68
|
4
|
|
|
|
|
69
|
my $resp = $self->{'useragent'}->get( $uri ); |
|
69
|
|
|
|
|
|
|
|
|
70
|
4
|
50
|
|
|
|
1471079
|
Carp::croak 'Failed to GET package list webpage: ' . $resp->status_line |
|
71
|
|
|
|
|
|
|
unless $resp->is_success; |
|
72
|
|
|
|
|
|
|
|
|
73
|
4
|
|
|
|
|
52
|
my @pkginfos; |
|
74
|
4
|
|
|
|
|
28
|
my @rows = _splitrows( $resp->content ); |
|
75
|
4
|
|
|
|
|
22
|
shift @rows; # remove the header column |
|
76
|
|
|
|
|
|
|
|
|
77
|
4
|
|
|
|
|
20
|
for my $rowhtml ( @rows ) { |
|
78
|
300
|
|
|
|
|
536
|
my @cols = _splitcols( $rowhtml ); |
|
79
|
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
# cat, name, version, votes, desc, maintainer |
|
81
|
300
|
|
|
|
|
1189
|
push @pkginfos, @cols; |
|
82
|
|
|
|
|
|
|
} |
|
83
|
|
|
|
|
|
|
|
|
84
|
4
|
|
|
|
|
260
|
return \@pkginfos; |
|
85
|
|
|
|
|
|
|
} |
|
86
|
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
sub _splitrows |
|
88
|
|
|
|
|
|
|
{ |
|
89
|
4
|
|
|
4
|
|
1761
|
my ($html) = @_; |
|
90
|
4
|
|
|
|
|
1266
|
my @rows = $html =~ m{ |
]*> ( .*? )
}gxs;
|
91
|
4
|
|
|
|
|
88
|
return @rows; |
|
92
|
|
|
|
|
|
|
} |
|
93
|
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
sub _splitcols |
|
95
|
|
|
|
|
|
|
{ |
|
96
|
300
|
|
|
300
|
|
365
|
my ($rowhtml) = @_; |
|
97
|
300
|
|
|
|
|
4032
|
my @cols = $rowhtml =~ m{ | ]*> ( .*? ) | }gxs;
|
98
|
300
|
|
|
|
|
552
|
for ( @cols ) { |
|
99
|
1800
|
|
|
|
|
4235
|
s/<[^>]+>//g; # delete tags |
|
100
|
1800
|
|
|
|
|
2869
|
s/\A\s+//; s/\s+\z//; # trim whitespace |
|
|
1800
|
|
|
|
|
3474
|
|
|
101
|
|
|
|
|
|
|
} |
|
102
|
300
|
|
|
|
|
1220
|
return @cols; |
|
103
|
|
|
|
|
|
|
} |
|
104
|
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
sub next |
|
106
|
|
|
|
|
|
|
{ |
|
107
|
206
|
|
|
206
|
1
|
950
|
my ($self) = @_; |
|
108
|
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
# There are no more packages to iterate over... |
|
110
|
206
|
100
|
|
|
|
742
|
return undef if $self->{'finished'}; |
|
111
|
|
|
|
|
|
|
|
|
112
|
205
|
|
|
|
|
339
|
my @pkginfo = splice @{ $self->{'packages'} }, 0, 6; |
|
|
205
|
|
|
|
|
1045
|
|
|
113
|
205
|
100
|
|
|
|
623
|
if ( @pkginfo ) { |
|
114
|
201
|
|
|
|
|
438
|
my $pkg; |
|
115
|
201
|
|
|
|
|
736
|
my @k = qw/cat name version votes desc/; |
|
116
|
201
|
|
|
|
|
599
|
for my $i (0 .. $#k) { |
|
117
|
1005
|
|
|
|
|
2485
|
$pkg->{$k[$i]} = $pkginfo[$i]; |
|
118
|
|
|
|
|
|
|
} |
|
119
|
|
|
|
|
|
|
|
|
120
|
201
|
|
|
|
|
461
|
my $maint = $pkginfo[5]; |
|
121
|
201
|
50
|
|
|
|
898
|
$pkg->{'maint'} = ($maint eq 'orphan' ? undef : $maint); |
|
122
|
201
|
|
|
|
|
837
|
return $pkg; |
|
123
|
|
|
|
|
|
|
} |
|
124
|
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
# Load a new batch of packages if our internal list is empty... |
|
126
|
4
|
|
|
|
|
16
|
my $newpkgs = $self->_scrape_pkglist; |
|
127
|
|
|
|
|
|
|
|
|
128
|
4
|
|
|
|
|
23
|
$self->{'curridx'} += 100; |
|
129
|
4
|
|
|
|
|
16
|
$self->{'packages'} = $newpkgs; |
|
130
|
4
|
100
|
|
|
|
24
|
$self->{'finished'} = 1 if scalar @$newpkgs == 0; |
|
131
|
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
# Recurse, just avoids code copy/pasting... |
|
133
|
4
|
|
|
|
|
30
|
return $self->next(); |
|
134
|
|
|
|
|
|
|
} |
|
135
|
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
sub next_obj |
|
137
|
|
|
|
|
|
|
{ |
|
138
|
100
|
|
|
100
|
1
|
1447
|
my ($self) = @_; |
|
139
|
|
|
|
|
|
|
|
|
140
|
100
|
|
|
|
|
418
|
my $next = $self->next; |
|
141
|
100
|
50
|
|
|
|
1696
|
return ( $next |
|
142
|
|
|
|
|
|
|
? WWW::AUR::Package->new( $next->{'name'}, %$self ) |
|
143
|
|
|
|
|
|
|
: undef ); |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
1; |
|
147
|
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
__END__ |