processing request page is expected

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

diff --git a/bin/isi-download-results.pl b/bin/isi-download-results.pl
index 9504185..5bd18be 100755 (executable)
--- a/bin/isi-download-results.pl
+++ b/bin/isi-download-results.pl
@@ -120,8 +120,8 @@ 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";
                }
 
 
@@ -217,6 +217,27 @@ 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 +262,8 @@ if ( $citations ) {
 
 }
 
+
+
 if ( $q =~ m{CA=(.+)} && $cited_reference ) {
 
        my $CA = $1;
@@ -265,17 +288,7 @@ 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 +311,8 @@ 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 +321,13 @@ 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;
 }