git.rot13.org Git - webpac2/blob - bin/isi-download-results.pl

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