fix download for new page url

[webpac2] / bin / isi-download-results.pl

diff --git a/bin/isi-download-results.pl b/bin/isi-download-results.pl
index 9504185..6bb37d4 100755 (executable)
--- a/bin/isi-download-results.pl
+++ b/bin/isi-download-results.pl
@@ -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;
 
@@ -120,8 +125,8 @@ sub get_results {
                        last;
                }
 
-               if ( $mech->content =~ m{Please wait while your request is processed} ) {
-                       warn "WARNING: processing request";
+               if ( $mech->content !~ m{Please wait while your request is processed} ) {
+                       warn "WARNING: expecting processing request";
                }
 
 
@@ -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;
        }
 }
 
@@ -217,6 +224,27 @@ sub years {
        return $years;
 }
 
+
+our $page = 1;
+sub next_page {
+       $page++;
+       warn "next_page $page\n";
+
+       $mech->submit_form(
+               form_name => 'summary_navigation',
+               fields => {
+                       'page' => $page,
+               },
+       );
+
+       save_mech;
+
+       $mech->form_name( 'summary_navigation' );
+       my $is_next_page = $mech->value('page') == $page;
+       warn "no next_page" unless $is_next_page;
+       return $is_next_page;
+}
+
 if ( $results ) {
        search;
        years;
@@ -241,6 +269,8 @@ if ( $citations ) {
 
 }
 
+
+
 if ( $q =~ m{CA=(.+)} && $cited_reference ) {
 
        my $CA = $1;
@@ -265,17 +295,7 @@ if ( $q =~ m{CA=(.+)} && $cited_reference ) {
        while (1) {
                save_mech "/tmp/isi.$q.citedref.$page";
 
-               $page++;
-
-               $mech->submit_form(
-                       form_name => 'summary_navigation',
-                       fields => {
-                               'page' => $page,
-                       },
-               );
-
-               $mech->form_name( 'summary_navigation' );
-               last if $mech->value('page') < $page;
+               last unless next_page();
 
                if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) {
                        warn "span: $1\n";
@@ -298,6 +318,8 @@ if ( $q =~ m{CA=(.+)} && $citing_articles ) {
        my $orig_q = $q;
        my $nr = 0;
 
+       do {
+
        foreach my $link ( $mech->find_all_links( url_regex => qr/CitingArticles.do/ ) ) {
                $nr++;
                warn "link $nr\n";
@@ -306,11 +328,13 @@ if ( $q =~ m{CA=(.+)} && $citing_articles ) {
                $q = $orig_q . '.citing_article.' . $nr;
                get_results;
                $mech->back;
-               save_mech;
                $mech->back;
-               save_mech;
+
+               #last if $nr > 3; # FIXME only for development
        }
 
+       } while next_page;
+
        $q = $orig_q;
 }