From cac2e509edf8973c6a87a598aa897535d62cfbb6 Mon Sep 17 00:00:00 2001
From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Sat, 5 Jun 2010 15:49:59 +0000
Subject: [PATCH] use cited reference search for CA= query

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1335 07558da8-63fa-0310-ba24-9fe276d99e06
---
 bin/isi-download-results.pl | 74 ++++++++++++++++++++++++++++++++-----
 1 file changed, 64 insertions(+), 10 deletions(-)

diff --git a/bin/isi-download-results.pl b/bin/isi-download-results.pl
index dfab364..598ac2c 100755
--- a/bin/isi-download-results.pl
+++ b/bin/isi-download-results.pl
@@ -12,23 +12,22 @@ use Text::Unaccent;
 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
 
 our $q = 'AD=Croatia';
+$q = 'CA=BRATKO, D';
+
 my $range_size = 500;
 my $overlap    = 3; # between previous and this range
+
 my $results = 0;
 my $citations = 0;
+
 my $cites_by_year = 0;
 
 my $max_cites = 5000; # ISI limit to get cites
 
-if ( 0 ) {
-	$q = 'TS=psychology AND AD=Croatia';
-	$range_size = 50;
-	$overlap    = 0;
-	$max_cites  = 50;
-}
-
 $q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV;
 
+warn "QUERY: $q\n";
+
 our $mech = WWW::Mechanize->new(
 	autocheck => 0, # it dies in reference download with it!
 	cookie_jar => undef,
@@ -212,13 +211,16 @@ sub years {
 	return $years;
 }
 
-search;
-years;
-get_results 'results' if $results;
+if ( $results ) {
+	search;
+	years;
+	get_results;
+}
 
 if ( $citations ) {
 
 	citations;
+	years unless @ranges;
 
 	do {
 		my $part;
@@ -232,3 +234,55 @@ if ( $citations ) {
 	} while ( @ranges );
 
 }
+
+if ( $q =~ m{CA=(.+)} ) {
+
+	my $CA = $1;
+
+	warn "# citated reference search";
+	$mech->follow_link( url_regex => qr/CitedReferenceSearch/ );
+	save_mech;
+
+
+	$mech->submit_form(
+		form_name => 'WOS_CitedReferenceSearch_input_form',
+		fields => {
+			'value(input1)' => $CA,
+		},
+	);
+
+	my $page = 1;
+	my $records = $1 if $mech->content =~ m/(\d+)\s+records/;
+	warn "# found $records records\n";
+	my $last_span = 'fake';
+
+	while (1) {
+		save_mech "/tmp/isi.$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;
+
+		if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) {
+			warn "span: $1\n";
+			last if $2 == $records;
+			last if $1 == $last_span;
+			$last_span = $1;
+		} elsif ( $page > 5 ) {
+			warn "ARTIFICALLY LIMITED TO 5 PAGES WITHOUT VALID SPAN!";
+			last;
+		}
+
+	}
+
+}
+
+warn "OVER\n";
-- 
2.20.1