line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Search::Scrape; |
2
|
|
|
|
|
|
|
|
3
|
2
|
|
|
2
|
|
40744
|
use warnings; |
|
2
|
|
|
|
|
6
|
|
|
2
|
|
|
|
|
70
|
|
4
|
2
|
|
|
2
|
|
13
|
use strict; |
|
2
|
|
|
|
|
3
|
|
|
2
|
|
|
|
|
65
|
|
5
|
|
|
|
|
|
|
|
6
|
2
|
|
|
2
|
|
2169
|
use Data::Dumper; |
|
2
|
|
|
|
|
20373
|
|
|
2
|
|
|
|
|
140
|
|
7
|
|
|
|
|
|
|
#use Smart::Comments; |
8
|
|
|
|
|
|
|
|
9
|
2
|
|
|
2
|
|
19
|
use Carp; |
|
2
|
|
|
|
|
3
|
|
|
2
|
|
|
|
|
118
|
|
10
|
|
|
|
|
|
|
|
11
|
2
|
|
|
2
|
|
1173
|
use WWW::Search::Scrape::Google; |
|
2
|
|
|
|
|
9
|
|
|
2
|
|
|
|
|
73
|
|
12
|
2
|
|
|
2
|
|
1440
|
use WWW::Search::Scrape::Bing; |
|
2
|
|
|
|
|
7
|
|
|
2
|
|
|
|
|
60
|
|
13
|
2
|
|
|
2
|
|
1081
|
use WWW::Search::Scrape::Yahoo; |
|
2
|
|
|
|
|
6
|
|
|
2
|
|
|
|
|
103
|
|
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
=head1 NAME |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
WWW::Search::Scrape - Scrape search engine results |
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
=head1 VERSION |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
Version 0.08 |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
=cut |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
our $VERSION = '0.08'; |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
=head1 SYNOPSIS |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
use WWW::Search::Scrape qw/:all/; |
32
|
|
|
|
|
|
|
my $result = search({engine => 'google', keyword =>'keywords', results => 10}); |
33
|
|
|
|
|
|
|
print "Google returns " . $result->{num} . " results\n"; |
34
|
|
|
|
|
|
|
print $_, "\n" foreach (@{$result->{results}}); |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
=head1 DESCRIPTION |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
Most search engines do not provide search API. |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
Google finally stop its Google search API in Sept 2009, while the registration for it had already been disabled for years. Google AJAX API is not powerful enough. |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
The purpose of this module is to provide a simple interface to extract top search results from Google search engines (as well as others), and keep this interface as simple as possible (as soon as possible as well). |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
Currently, it supports English Google and Bing only. I schedule to add more functions soon. |
45
|
|
|
|
|
|
|
|
46
|
|
|
|
|
|
|
=head1 EXPORT |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
There is only one function in WWW::Search::Scrape -- search. |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
=cut |
52
|
|
|
|
|
|
|
|
53
|
2
|
|
|
2
|
|
12
|
use vars qw($VERSION @ISA @EXPORT @EXPORT_OK); |
|
2
|
|
|
|
|
6
|
|
|
2
|
|
|
|
|
813
|
|
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
require Exporter; |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
our @ISA = qw(Exporter); |
58
|
|
|
|
|
|
|
our %EXPORT_TAGS = ('all' => [qw/search/]); |
59
|
|
|
|
|
|
|
our @EXPORT_OK = (@{$EXPORT_TAGS{all}}); |
60
|
|
|
|
|
|
|
our @EXPORT = qw(); |
61
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
=head1 FUNCTIONS |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
=head2 search |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
search is the most important function in this module. It is used as a dispatcher for corresponding search engines -- Google, Yahoo, Bing etc. |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
It accepts a config hash. Possible keys are, |
69
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
+---------+--------------------------------------------------------+ |
71
|
|
|
|
|
|
|
| engine | The name for the search engine, like 'google', 'bing' | |
72
|
|
|
|
|
|
|
+---------+--------------------------------------------------------+ |
73
|
|
|
|
|
|
|
| keyword | The keyword(s) for the searching terms | |
74
|
|
|
|
|
|
|
+---------+--------------------------------------------------------+ |
75
|
|
|
|
|
|
|
| results | How many results should be returned (default: 10) | |
76
|
|
|
|
|
|
|
+---------+--------------------------------------------------------+ |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
It returns a hash ref |
80
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
+---------+-------------------------------------------------------------------------+ |
82
|
|
|
|
|
|
|
| num | How many items the search engine are able to return? (estimated number) | |
83
|
|
|
|
|
|
|
+---------+-------------------------------------------------------------------------+ |
84
|
|
|
|
|
|
|
| results | List of returned results + |
85
|
|
|
|
|
|
|
+---------+-------------------------------------------------------------------------+ |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
=cut |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
sub search (%) { |
90
|
8
|
|
|
8
|
1
|
476
|
my ($q) = @_; |
91
|
|
|
|
|
|
|
### search got query: $q |
92
|
8
|
50
|
|
|
|
37
|
unless (ref($q) eq "HASH") |
93
|
|
|
|
|
|
|
{ |
94
|
0
|
|
|
|
|
0
|
carp 'search query should be hash'; |
95
|
0
|
|
|
|
|
0
|
return undef; |
96
|
|
|
|
|
|
|
} |
97
|
|
|
|
|
|
|
|
98
|
8
|
100
|
|
|
|
24
|
unless ($q->{engine}) |
99
|
|
|
|
|
|
|
{ |
100
|
1
|
|
|
|
|
130
|
carp 'engine not set.'; |
101
|
1
|
|
|
|
|
89
|
return undef; |
102
|
|
|
|
|
|
|
} |
103
|
|
|
|
|
|
|
|
104
|
7
|
100
|
66
|
|
|
50
|
unless ($q->{results} && $q->{results} >= 0) { |
105
|
2
|
|
|
|
|
257
|
carp 'search results number not set.'; |
106
|
2
|
|
|
|
|
189
|
return undef; |
107
|
|
|
|
|
|
|
} |
108
|
|
|
|
|
|
|
|
109
|
5
|
100
|
66
|
|
|
39
|
unless ($q->{keyword} && length($q->{keyword}) >= 1) { |
110
|
1
|
|
|
|
|
223
|
carp 'search keyword not set.'; |
111
|
1
|
|
|
|
|
193
|
return undef; |
112
|
|
|
|
|
|
|
} |
113
|
|
|
|
|
|
|
|
114
|
4
|
100
|
|
|
|
26
|
if (lc $q->{engine} eq 'google') { |
|
|
50
|
|
|
|
|
|
115
|
3
|
50
|
|
|
|
10
|
if ($q->{geo_location}) { |
116
|
0
|
|
|
|
|
0
|
$WWW::Search::Scrape::Google::geo_location = $q->{geo_location}; |
117
|
|
|
|
|
|
|
} |
118
|
3
|
100
|
|
|
|
11
|
if ($q->{frontpage}) { |
119
|
2
|
|
|
|
|
9
|
$WWW::Search::Scrape::Google::frontpage = $q->{frontpage}; |
120
|
|
|
|
|
|
|
} |
121
|
3
|
|
|
|
|
17
|
return WWW::Search::Scrape::Google::search($q->{keyword}, $q->{results}); |
122
|
|
|
|
|
|
|
} elsif (lc $q->{engine} eq 'bing') { |
123
|
1
|
|
|
|
|
8
|
return WWW::Search::Scrape::Bing::search($q->{keyword}, $q->{results}); |
124
|
|
|
|
|
|
|
} |
125
|
|
|
|
|
|
|
else { |
126
|
0
|
|
|
|
|
|
carp 'Search engine ' . $q->{engine} . ' not implemented yet.'; |
127
|
0
|
|
|
|
|
|
return undef; |
128
|
|
|
|
|
|
|
} |
129
|
|
|
|
|
|
|
} |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=head1 AUTHOR |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
Quan Sun, C<< >> |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
=head1 BUGS |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
Please report any bugs or feature requests to C, or C, or through |
138
|
|
|
|
|
|
|
the web interface at L. I will be notified, and then you'll |
139
|
|
|
|
|
|
|
automatically be notified of progress on your bug as I make changes. |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
L |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
=head1 SUPPORT |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
You can find documentation for this module with the perldoc command. |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
perldoc WWW::Search::Scrape |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
You can also look for information at: |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
=over 4 |
155
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
=item * RT: CPAN's request tracker |
157
|
|
|
|
|
|
|
|
158
|
|
|
|
|
|
|
L |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
=item * AnnoCPAN: Annotated CPAN documentation |
161
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
L |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
=item * CPAN Ratings |
165
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
L |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
=item * Search CPAN |
169
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
L |
171
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
=back |
173
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
=head1 ACKNOWLEDGEMENTS |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
=head1 COPYRIGHT & LICENSE |
179
|
|
|
|
|
|
|
|
180
|
|
|
|
|
|
|
Copyright 2009 Quan Sun. |
181
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify it |
183
|
|
|
|
|
|
|
under the terms of either: the GNU General Public License as published |
184
|
|
|
|
|
|
|
by the Free Software Foundation; or the Artistic License. |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
See http://dev.perl.org/licenses/ for more information. |
187
|
|
|
|
|
|
|
|
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
=cut |
190
|
|
|
|
|
|
|
|
191
|
|
|
|
|
|
|
1; # End of WWW::Search::Scrape |