1 |
dpavlin |
1280 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
dpavlin |
1289 |
my $q = 'AD=Croatia'; |
7 |
|
|
my $range_size = 500; |
8 |
dpavlin |
1280 |
|
9 |
dpavlin |
1282 |
my $dump = @ARGV ? 1 : 0; |
10 |
|
|
|
11 |
dpavlin |
1289 |
$q = 'TS=psychology AND AD=Croatia'; |
12 |
dpavlin |
1282 |
|
13 |
dpavlin |
1280 |
use WWW::Mechanize; |
14 |
|
|
use Data::Dump qw/dump/; |
15 |
|
|
|
16 |
dpavlin |
1289 |
our $mech = WWW::Mechanize->new( |
17 |
dpavlin |
1280 |
autocheck => 1, |
18 |
|
|
cookie_jar => undef, |
19 |
|
|
); |
20 |
|
|
|
21 |
dpavlin |
1281 |
our $step = 0; |
22 |
dpavlin |
1280 |
|
23 |
|
|
sub save_mech { |
24 |
dpavlin |
1281 |
my ( $mech, $path ) = @_; |
25 |
|
|
$step++; |
26 |
|
|
$path ||= sprintf('/tmp/isi.%02d.%s', $step, $mech->{ct} =~ m{html}i ? 'html' : 'txt' ); |
27 |
|
|
$mech->save_content( $path ); |
28 |
|
|
warn "# [$step] $path ", -s $path, " ", $mech->ct; |
29 |
dpavlin |
1282 |
$mech->dump_all if $dump; |
30 |
dpavlin |
1280 |
} |
31 |
|
|
|
32 |
|
|
warn "# get session"; |
33 |
|
|
$mech->get( 'http://isiknowledge.com/?DestApp=WOS' ); |
34 |
|
|
save_mech $mech; |
35 |
|
|
|
36 |
|
|
warn "# advanced serach"; |
37 |
|
|
$mech->follow_link( url_regex => qr/AdvancedSearch/ ); |
38 |
|
|
save_mech $mech; |
39 |
|
|
|
40 |
|
|
warn "# cookie_jar ", dump $mech->cookie_jar; |
41 |
|
|
|
42 |
|
|
$mech->submit_form( |
43 |
|
|
fields => { |
44 |
|
|
'value(input1)' => $q, |
45 |
|
|
} |
46 |
|
|
); |
47 |
|
|
save_mech $mech; |
48 |
|
|
|
49 |
|
|
warn "# summary"; |
50 |
|
|
$mech->follow_link( url_regex => qr/summary/ ); |
51 |
|
|
save_mech $mech; |
52 |
|
|
|
53 |
dpavlin |
1289 |
sub get_results { |
54 |
|
|
my $q = shift; |
55 |
|
|
my $from = 1; |
56 |
dpavlin |
1280 |
|
57 |
dpavlin |
1289 |
while ( 1 ) { |
58 |
dpavlin |
1280 |
|
59 |
dpavlin |
1289 |
my $to = $from + $range_size; |
60 |
dpavlin |
1280 |
|
61 |
dpavlin |
1289 |
$mech->submit_form( |
62 |
|
|
form_name => 'summary_output_form', |
63 |
|
|
fields => { |
64 |
|
|
record_select_type => 'range', |
65 |
|
|
mark_from => $from, |
66 |
|
|
mark_to => $to, |
67 |
|
|
mark_id => 'WOS', |
68 |
dpavlin |
1280 |
|
69 |
dpavlin |
1289 |
qo_fields => 'fullrecord', |
70 |
|
|
citedref => 'citedref', |
71 |
dpavlin |
1280 |
|
72 |
dpavlin |
1289 |
save_options => 'plain_text', |
73 |
dpavlin |
1280 |
|
74 |
dpavlin |
1289 |
fields => 'Full', |
75 |
|
|
format => 'save', |
76 |
|
|
}, |
77 |
|
|
button => 'save', |
78 |
|
|
); |
79 |
|
|
save_mech $mech; |
80 |
dpavlin |
1281 |
|
81 |
dpavlin |
1289 |
if ( $mech->content =~ m{invalid API call} ) { |
82 |
|
|
$mech->back; |
83 |
|
|
return; |
84 |
|
|
} |
85 |
dpavlin |
1281 |
|
86 |
dpavlin |
1289 |
warn "# save_file $from - $to [$q]"; |
87 |
|
|
$mech->follow_link( url_regex => qr/save_file/ ); |
88 |
|
|
save_mech $mech => "/tmp/isi.$q.$from-$to.txt"; |
89 |
dpavlin |
1281 |
|
90 |
dpavlin |
1289 |
$from += $range_size; |
91 |
dpavlin |
1281 |
|
92 |
dpavlin |
1289 |
$mech->back; |
93 |
|
|
$mech->back; |
94 |
|
|
#save_mech $mech; |
95 |
dpavlin |
1281 |
|
96 |
dpavlin |
1289 |
} # while |
97 |
|
|
|
98 |
dpavlin |
1281 |
} |
99 |
dpavlin |
1289 |
|
100 |
|
|
get_results $q; |
101 |
|
|
|
102 |
|
|
save_mech $mech; |
103 |
|
|
warn "# citations"; |
104 |
|
|
$mech->follow_link( url_regex => qr/search_mode=CitationReport/ ); |
105 |
|
|
save_mech $mech; |
106 |
|
|
|
107 |
|
|
$mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ ); |
108 |
|
|
save_mech $mech; |
109 |
|
|
|
110 |
|
|
get_results $q . '.citing'; |
111 |
|
|
|