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 use File::Slurp;
  11
  12 my $mech = WWW::Mechanize->new(
  13         autocheck => 1,
  14         cookie_jar => undef,
  15 );
  16
  17 my $step = 1;
  18
  19 sub save_mech {
  20         my $mech = shift;
  21         my $path = "/tmp/isi.$step.html";
  22         write_file $path, $mech->content;
  23         warn "# step $step ", -s $path;
  24         $mech->dump_all;
  25         $step++;
  26 }
  27
  28 warn "# get session";
  29 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
  30 save_mech $mech;
  31
  32 warn "# advanced serach";
  33 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
  34 save_mech $mech;
  35
  36 warn "# cookie_jar ", dump $mech->cookie_jar;
  37
  38 $mech->submit_form(
  39         fields => {
  40                 'value(input1)' => $q,
  41         }
  42 );
  43 save_mech $mech;
  44
  45 warn "# summary";
  46 $mech->follow_link( url_regex => qr/summary/ );
  47 save_mech $mech;
  48
  49 my $from = 1;
  50 my $range_size = 10;
  51
  52 $mech->submit_form(
  53         form_name => 'summary_output_form',
  54         fields => {
  55                 record_select_type => 'range',
  56                 mark_from => $from,
  57                 mark_to => $from += $range_size,
  58                 mark_id => 'WOS',
  59
  60                 qo_fields => 'fullrecord',
  61                 citedref => 'citedref',
  62
  63                 save_options => 'plain_text',
  64
  65                 fields => 'Full',
  66                 format => 'save',
  67         },
  68         button => 'save',
  69 );
  70 save_mech $mech;
  71
  72 warn "# save_file";
  73 $mech->follow_link( url_regex => qr/save_file/ );
  74 save_mech $mech;
  75