return years in hash

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1291 07558da8-63fa-0310-ba24-9fe276d99e06

diff --git a/bin/isi-download-results.pl b/bin/isi-download-results.pl
index 3fa0040..8e0f918 100755 (executable)
--- a/bin/isi-download-results.pl
+++ b/bin/isi-download-results.pl
@@ -3,7 +3,7 @@
 use warnings;
 use strict;
 
-my $q = 'AD=Croatia';
+our $q = 'AD=Croatia';
 my $range_size = 500;
 
 my $dump = @ARGV ? 1 : 0;
@@ -11,7 +11,8 @@ my $dump = @ARGV ? 1 : 0;
 $q = 'TS=psychology AND AD=Croatia';
 
 use WWW::Mechanize;
-use Data::Dump qw/dump/;
+use Data::Dump qw(dump);
+use File::Path;
 
 our $mech = WWW::Mechanize->new(
        autocheck => 1,
@@ -20,11 +21,14 @@ our $mech = WWW::Mechanize->new(
 
 our $step = 0;
 
+my $dir = '/tmp/isi/';
+rmtree $dir if -e $dir;
+mkdir $dir;
+
 sub save_mech {
        my ( $mech, $path ) = @_;
        $step++;
-       mkdir '/tmp/isi/' unless -e '/tmp/isi';
-       my $base_path = sprintf('/tmp/isi/%04d', $step);
+       my $base_path = sprintf('%s/%04d', $dir,$step);
        $path ||= $base_path . ( $mech->{ct} =~ m{html}i ? '.html' : '.txt' );
        $mech->save_content( $path );
        warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
@@ -37,8 +41,6 @@ $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
 save_mech $mech;
 
 sub search {
-       my $q = shift;
-
        warn "# advanced serach";
        $mech->follow_link( url_regex => qr/AdvancedSearch/ );
        save_mech $mech;
@@ -58,7 +60,6 @@ sub search {
 }
 
 sub get_results {
-       my $q = shift;
        my $from = 1;
 
        while ( 1 ) {
@@ -117,29 +118,39 @@ sub citations {
 }
 
 sub years {
-       my $years_url = $mech->find_link( text_regex => qr/more options/ )->url_abs;
+       my $years_url = $mech->find_link( text_regex => qr/more options/ );
+       if ( ! $years_url ) {
+               warn "W: can't find years\n";
+               return;
+       }
+       $years_url = $years_url->url_abs;
        warn "## $years_url";
-       $years_url =~ s{ra_name=\w+}{ra_name=PublicationYear} || die "ra_name";
+       if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
+               warn "W: no ra_name\n";
+               return;
+       }
        warn "# refine years (hidden by javascript)";
        warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
        $mech->get( $years_url );
        save_mech $mech;
 
        my $html = $mech->content;
-       my @years;
+       my $years;
        while ( $html =~ s{>(\d\d\d\d)\s\((\d+)\)</label.+?value="PublicationYear_}{} ) {
-               push @years, [ $1 => $2 ];
+               $years->{$1} = $2;
        }
-       warn "# years ",dump @years;
+       warn "# years ",dump $years;
        $mech->back;
-       return @years;
+       return $years;
 }
 
-search $q;
+search;
 years;
-get_results $q;
+get_results;
+
 
 citations;
 years;
-get_results $q . '.citing';
+$q .= '.citing';
+get_results;