From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Tue, 29 Jun 2010 18:32:27 +0000 (+0000)
Subject: download first page of citing articles
X-Git-Url: http://git.rot13.org/?p=webpac2;a=commitdiff_plain;h=8f8575598b93c1d730e66b81379ae170b25cb1d4

download first page of citing articles


git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1336 07558da8-63fa-0310-ba24-9fe276d99e06
---

diff --git a/bin/isi-download-results.pl b/bin/isi-download-results.pl
index 598ac2c..9504185 100755
--- a/bin/isi-download-results.pl
+++ b/bin/isi-download-results.pl
@@ -19,6 +19,8 @@ my $overlap    = 3; # between previous and this range
 
 my $results = 0;
 my $citations = 0;
+my $cited_reference = 0; # html tables
+my $citing_articles = 1; # as many files as cited articles
 
 my $cites_by_year = 0;
 
@@ -118,6 +120,10 @@ sub get_results {
 			last;
 		}
 
+		if ( $mech->content =~ m{Please wait while your request is processed} ) {
+			warn "WARNING: processing request";
+		}
+
 
 		my $path = "/tmp/isi.$q.$from-$to";
 		$path .= '.' . $desc if $desc;
@@ -235,7 +241,7 @@ if ( $citations ) {
 
 }
 
-if ( $q =~ m{CA=(.+)} ) {
+if ( $q =~ m{CA=(.+)} && $cited_reference ) {
 
 	my $CA = $1;
 
@@ -285,4 +291,27 @@ if ( $q =~ m{CA=(.+)} ) {
 
 }
 
+if ( $q =~ m{CA=(.+)} && $citing_articles ) {
+
+	search;
+
+	my $orig_q = $q;
+	my $nr = 0;
+
+	foreach my $link ( $mech->find_all_links( url_regex => qr/CitingArticles.do/ ) ) {
+		$nr++;
+		warn "link $nr\n";
+		$mech->get( $link->url );
+		save_mech;
+		$q = $orig_q . '.citing_article.' . $nr;
+		get_results;
+		$mech->back;
+		save_mech;
+		$mech->back;
+		save_mech;
+	}
+
+	$q = $orig_q;
+}
+
 warn "OVER\n";