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

   1 #!/usr/bin/perl
   2
   3 use warnings;
   4 use strict;
   5
   6 our $q = 'AD=Croatia';
   7 my $range_size = 500;
   8
   9 my $dump = @ARGV ? 1 : 0;
  10
  11 $q = 'TS=psychology AND AD=Croatia';
  12
  13 use WWW::Mechanize;
  14 use Data::Dump qw(dump);
  15 use File::Path;
  16
  17 our $mech = WWW::Mechanize->new(
  18         autocheck => 1,
  19         cookie_jar => undef,
  20 );
  21
  22 our $step = 0;
  23
  24 my $dir = '/tmp/isi/';
  25 rmtree $dir if -e $dir;
  26 mkdir $dir;
  27
  28 sub save_mech {
  29         my ( $mech, $path ) = @_;
  30         $step++;
  31         my $base_path = sprintf('%s/%04d', $dir,$step);
  32         $path ||= $base_path . ( $mech->{ct} =~ m{html}i ? '.html' : '.txt' );
  33         $mech->save_content( $path );
  34         warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
  35         open(my $dump, '>', "$base_path.dump.txt");
  36         $mech->dump_all($dump);
  37 }
  38
  39 warn "# get session";
  40 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
  41 save_mech $mech;
  42
  43 sub search {
  44         warn "# advanced serach";
  45         $mech->follow_link( url_regex => qr/AdvancedSearch/ );
  46         save_mech $mech;
  47
  48         warn "# cookie_jar ", dump $mech->cookie_jar;
  49
  50         $mech->submit_form(
  51                 fields => {
  52                         'value(input1)' => $q,
  53                 }
  54         );
  55         save_mech $mech;
  56
  57         warn "# summary";
  58         $mech->follow_link( url_regex => qr/summary/ );
  59         save_mech $mech;
  60 }
  61
  62 sub get_results {
  63         my $from = 1;
  64
  65         while ( 1 ) {
  66
  67                 my $to = $from + $range_size;
  68
  69                 $mech->submit_form(
  70                         form_name => 'summary_output_form',
  71                         fields => {
  72                                 record_select_type => 'range',
  73                                 mark_from => $from,
  74                                 mark_to => $to,
  75                                 mark_id => 'WOS',
  76
  77                                 qo_fields => 'fullrecord',
  78                                 citedref => 'citedref',
  79
  80                                 save_options => 'plain_text',
  81
  82                                 fields => 'Full',
  83                                 format => 'save',
  84                         },
  85                         button => 'save',
  86                 );
  87                 save_mech $mech;
  88
  89                 if ( $mech->content =~ m{invalid API call} ) {
  90                         $mech->back;
  91                         last;
  92                 }
  93
  94                 warn "range $from - $to [$q]\n";
  95                 $mech->follow_link( url_regex => qr/save_file/ );
  96                 save_mech $mech => "/tmp/isi.$q.$from-$to.txt";
  97
  98                 $from += $range_size;
  99
 100                 $mech->back;
 101                 $mech->back;
 102                 #save_mech $mech;
 103
 104         }
 105
 106 }
 107
 108
 109 sub citations {
 110         save_mech $mech;
 111         warn "# citation report";
 112         $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
 113         save_mech $mech;
 114
 115         warn "view citing articles";
 116         $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
 117         save_mech $mech;
 118 }
 119
 120 sub years {
 121         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";
 128         if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
 129                 warn "W: no ra_name\n";
 130                 return;
 131         }
 132         warn "# refine years (hidden by javascript)";
 133         warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
 134         $mech->get( $years_url );
 135         save_mech $mech;
 136
 137         my $html = $mech->content;
 138         my $years;
 139         while ( $html =~ s{>(\d\d\d\d)\s\((\d+)\)</label.+?value="PublicationYear_}{} ) {
 140                 $years->{$1} = $2;
 141         }
 142         warn "# years ",dump $years;
 143         $mech->back;
 144         return $years;
 145 }
 146
 147 search;
 148 years;
 149 get_results;
 150
 151
 152 citations;
 153 years;
 154 $q .= '.citing';
 155 get_results;
 156