3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
|
6 |
my $q = 'AD=Croatia'; |
our $q = 'AD=Croatia'; |
7 |
my $range_size = 500; |
my $range_size = 500; |
8 |
|
|
9 |
my $dump = @ARGV ? 1 : 0; |
my $dump = @ARGV ? 1 : 0; |
11 |
$q = 'TS=psychology AND AD=Croatia'; |
$q = 'TS=psychology AND AD=Croatia'; |
12 |
|
|
13 |
use WWW::Mechanize; |
use WWW::Mechanize; |
14 |
use Data::Dump qw/dump/; |
use Data::Dump qw(dump); |
15 |
|
use File::Path; |
16 |
|
|
17 |
our $mech = WWW::Mechanize->new( |
our $mech = WWW::Mechanize->new( |
18 |
autocheck => 1, |
autocheck => 1, |
21 |
|
|
22 |
our $step = 0; |
our $step = 0; |
23 |
|
|
24 |
|
my $dir = '/tmp/isi/'; |
25 |
|
rmtree $dir if -e $dir; |
26 |
|
mkdir $dir; |
27 |
|
|
28 |
sub save_mech { |
sub save_mech { |
29 |
my ( $mech, $path ) = @_; |
my ( $mech, $path ) = @_; |
30 |
$step++; |
$step++; |
31 |
mkdir '/tmp/isi/' unless -e '/tmp/isi'; |
my $base_path = sprintf('%s/%04d', $dir,$step); |
|
my $base_path = sprintf('/tmp/isi/%04d', $step); |
|
32 |
$path ||= $base_path . ( $mech->{ct} =~ m{html}i ? '.html' : '.txt' ); |
$path ||= $base_path . ( $mech->{ct} =~ m{html}i ? '.html' : '.txt' ); |
33 |
$mech->save_content( $path ); |
$mech->save_content( $path ); |
34 |
warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n"; |
warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n"; |
41 |
save_mech $mech; |
save_mech $mech; |
42 |
|
|
43 |
sub search { |
sub search { |
|
my $q = shift; |
|
|
|
|
44 |
warn "# advanced serach"; |
warn "# advanced serach"; |
45 |
$mech->follow_link( url_regex => qr/AdvancedSearch/ ); |
$mech->follow_link( url_regex => qr/AdvancedSearch/ ); |
46 |
save_mech $mech; |
save_mech $mech; |
60 |
} |
} |
61 |
|
|
62 |
sub get_results { |
sub get_results { |
|
my $q = shift; |
|
63 |
my $from = 1; |
my $from = 1; |
64 |
|
|
65 |
while ( 1 ) { |
while ( 1 ) { |
118 |
} |
} |
119 |
|
|
120 |
sub years { |
sub years { |
121 |
my $years_url = $mech->find_link( text_regex => qr/more options/ )->url_abs; |
my $years_url = $mech->find_link( text_regex => qr/more options/ ); |
122 |
|
if ( ! $years_url ) { |
123 |
|
warn "W: can't find years\n"; |
124 |
|
return; |
125 |
|
} |
126 |
|
$years_url = $years_url->url_abs; |
127 |
warn "## $years_url"; |
warn "## $years_url"; |
128 |
$years_url =~ s{ra_name=\w+}{ra_name=PublicationYear} || die "ra_name"; |
if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) { |
129 |
|
warn "W: no ra_name\n"; |
130 |
|
return; |
131 |
|
} |
132 |
warn "# refine years (hidden by javascript)"; |
warn "# refine years (hidden by javascript)"; |
133 |
warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n"; |
warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n"; |
134 |
$mech->get( $years_url ); |
$mech->get( $years_url ); |
135 |
save_mech $mech; |
save_mech $mech; |
136 |
|
|
137 |
my $html = $mech->content; |
my $html = $mech->content; |
138 |
my @years; |
my $years; |
139 |
while ( $html =~ s{>(\d\d\d\d)\s\((\d+)\)</label.+?value="PublicationYear_}{} ) { |
while ( $html =~ s{>(\d\d\d\d)\s\((\d+)\)</label.+?value="PublicationYear_}{} ) { |
140 |
push @years, [ $1 => $2 ]; |
$years->{$1} = $2; |
141 |
} |
} |
142 |
warn "# years ",dump @years; |
warn "# years ",dump $years; |
143 |
$mech->back; |
$mech->back; |
144 |
return @years; |
return $years; |
145 |
} |
} |
146 |
|
|
147 |
search $q; |
search; |
148 |
years; |
years; |
149 |
get_results $q; |
get_results; |
150 |
|
|
151 |
|
|
152 |
citations; |
citations; |
153 |
years; |
years; |
154 |
get_results $q . '.citing'; |
$q .= '.citing'; |
155 |
|
get_results; |
156 |
|
|