6 # Advanced search syntax:
7 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
11 my $overlap = 3; # between previous and this range
13 my $max_cites = 5000; # ISI limit to get cites
16 $q = 'TS=psychology AND AD=Croatia';
23 use Data::Dump qw(dump);
26 our $mech = WWW::Mechanize->new(
34 my $dir = '/tmp/isi/';
35 rmtree $dir if -e $dir;
41 my $base_path = sprintf('%s/%04d', $dir,$step);
43 $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
44 $mech->save_content( $path );
45 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
46 open(my $dump, '>', "$base_path.dump.txt");
47 $mech->dump_all($dump);
51 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
55 warn "# advanced serach";
56 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
59 warn "# cookie_jar ", dump $mech->cookie_jar;
64 $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
67 warn "# submit_form search: $q_this\n";
70 'value(input1)' => $q_this,
76 $mech->follow_link( url_regex => qr/summary/ );
86 my $to = $from + $range_size;
88 warn "# submit_form results $from - $to\n";
91 form_name => 'summary_output_form',
93 record_select_type => 'range',
98 qo_fields => 'fullrecord',
99 citedref => 'citedref',
101 save_options => 'plain_text',
111 if ( $mech->content =~ m{invalid API call} ) {
117 my $path = "/tmp/isi.$q.$from-$to";
118 $path .= '.' . $desc if $desc;
120 warn "save $from - $to into $path\n";
121 $mech->follow_link( url_regex => qr/save_file/ );
124 $from += $range_size - $overlap;
134 warn "# citation report";
135 $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
138 warn "view citing articles";
139 $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
144 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
145 if ( ! $years_url ) {
146 warn "W: can't find ra_name link\n";
149 $years_url = $years_url->url_abs;
150 warn "## $years_url";
151 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
152 warn "W: no ra_name in $years_url\n";
155 warn "# refine years (hidden by javascript)";
156 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
157 $mech->get( $years_url );
160 my $html = $mech->content;
162 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
163 my ( $year, $count ) = ( $1, $2 );
165 $years->{$year} = $count;
167 warn "# years ",dump $years;
170 my @y = sort keys %$years;
173 my $size = $years->{$y};
177 $cites_range = [$y] if $y;
179 foreach my $y ( @y ) {
180 if ( $size + $years->{$y} > $max_cites ) {
181 push @ranges, $cites_range;
182 warn "# cites_range $size years ",dump( $cites_range ),$/;
187 $size += $years->{$y};
188 push @$cites_range, $y;
191 if ( $cites_range ) {
192 push @ranges, $cites_range;
193 warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
196 warn '# ranges ', dump @ranges;
197 @ranges = () if $#ranges == 1; # just take all
212 $part .= $ranges[0]->[0] . '.';