fix download for new page url
authorDobrica Pavlinusic <dpavlin@rot13.org>
Wed, 8 Aug 2012 14:20:26 +0000 (14:20 +0000)
committerDobrica Pavlinusic <dpavlin@rot13.org>
Wed, 8 Aug 2012 14:20:26 +0000 (14:20 +0000)
git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1369 07558da8-63fa-0310-ba24-9fe276d99e06

bin/isi-download-results.pl

index 5bd18be..6bb37d4 100755 (executable)
@@ -12,15 +12,16 @@ use Text::Unaccent;
 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
 
 our $q = 'AD=Croatia';
-$q = 'CA=BRATKO, D';
+$q = 'AU=BRATKO, D';
+#$q = 'AD=(croat* OR hrvat*)';
 
-my $range_size = 500;
+my $range_size = 100;
 my $overlap    = 3; # between previous and this range
 
-my $results = 0;
+my $results = 1;
 my $citations = 0;
 my $cited_reference = 0; # html tables
-my $citing_articles = 1; # as many files as cited articles
+my $citing_articles = 0; # as many files as cited articles
 
 my $cites_by_year = 0;
 
@@ -55,7 +56,7 @@ sub save_mech {
 }
 
 warn "# get session";
-$mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
+$mech->get( 'http://www.webofknowledge.com/?DestApp=WOS' );
 save_mech;
 
 sub search {
@@ -93,24 +94,28 @@ sub get_results {
                my $to = $from + $range_size;
 
                warn "# submit_form results $from - $to\n";
+               save_mech;
 
                $mech->submit_form(
-                       form_name => 'summary_output_form',
+                       form_name => 'output_form',
                        fields => {
-                               record_select_type => 'range',
+                               'value(record_select_type)' => 'range',
+                               markFrom => $from,
+                               markTo => $to,
+
                                mark_from => $from,
                                mark_to => $to,
                                mark_id => 'WOS',
 
-                               qo_fields => 'fullrecord',
-                               citedref => 'citedref',
+                               fields_selection => 'ACCESSION_NUM FUNDING SUBJECT_CATEGORY JCR_CATEGORY LANG IDS PAGEC SABBR CITREFC ISSN PUBINFO KEYWORDS CITTIMES ADDRS CONFERENCE_SPONSORS DOCTYPE ABSTRACT CONFERENCE_INFO SOURCE TITLE AUTHORS',
+                               filters => 'ACCESSION_NUM FUNDING SUBJECT_CATEGORY JCR_CATEGORY LANG IDS PAGEC SABBR CITREFC ISSN PUBINFO KEYWORDS CITTIMES ADDRS CONFERENCE_SPONSORS DOCTYPE ABSTRACT CONFERENCE_INFO SOURCE TITLE AUTHORS CITREF',
+                               fullrec_fields_option => 'CITREF',
 
-                               save_options => 'plain_text',
+                               save_options => 'fieldtagged',
+                               format => 'saveToFile',
 
-                               fields => 'Full',
-                               format => 'save',
                        },
-                       button => 'save',
+                       button => 'saveToFile',
                );
                save_mech;
 
@@ -129,14 +134,16 @@ sub get_results {
                $path .= '.' . $desc if $desc;
 
                warn "save $from - $to into $path\n";
-               $mech->follow_link( url_regex => qr/save_file/ );
+               $mech->submit_form(
+                       form_name => 'etsForm',
+               );
                save_mech $path;
 
                $from += $range_size - $overlap;
 
                $mech->back;
                $mech->back;
-               #save_mech;
+               save_mech;
        }
 }