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