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
14 my $max_cites = 5000; # ISI limit to get cites
17 $q = 'TS=psychology AND AD=Croatia';
24 use Data::Dump qw(dump);
27 our $mech = WWW::Mechanize->new(
35 my $dir = '/tmp/isi/';
36 rmtree $dir if -e $dir;
42 my $base_path = sprintf('%s/%04d', $dir,$step);
44 $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
45 $mech->save_content( $path );
46 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
47 open(my $dump, '>', "$base_path.dump.txt");
48 $mech->dump_all($dump);
52 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
56 warn "# advanced serach";
57 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
60 warn "# cookie_jar ", dump $mech->cookie_jar;
65 $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
68 warn "# submit_form search: $q_this\n";
71 'value(input1)' => $q_this,
77 $mech->follow_link( url_regex => qr/summary/ );
87 my $to = $from + $range_size;
89 warn "# submit_form results $from - $to\n";
92 form_name => 'summary_output_form',
94 record_select_type => 'range',
99 qo_fields => 'fullrecord',
100 citedref => 'citedref',
102 save_options => 'plain_text',
112 if ( $mech->content =~ m{invalid API call} ) {
118 my $path = "/tmp/isi.$q.$from-$to";
119 $path .= '.' . $desc if $desc;
121 warn "save $from - $to into $path\n";
122 $mech->follow_link( url_regex => qr/save_file/ );
125 $from += $range_size - $overlap;
135 warn "# citation report";
136 $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
139 warn "view citing articles";
140 $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
145 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
146 if ( ! $years_url ) {
147 warn "W: can't find ra_name link\n";
150 $years_url = $years_url->url_abs;
151 warn "## $years_url";
152 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
153 warn "W: no ra_name in $years_url\n";
156 warn "# refine years (hidden by javascript)";
157 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
158 $mech->get( $years_url );
161 my $html = $mech->content;
163 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
164 my ( $year, $count ) = ( $1, $2 );
166 $years->{$year} = $count;
168 warn "# years ",dump $years;
171 my @y = sort keys %$years;
174 my $size = $years->{$y};
178 $cites_range = [$y] if $y;
180 foreach my $y ( @y ) {
181 if ( $size + $years->{$y} > $max_cites ) {
182 push @ranges, $cites_range;
183 warn "# cites_range $size years ",dump( $cites_range ),$/;
188 $size += $years->{$y};
189 push @$cites_range, $y;
192 if ( $cites_range ) {
193 push @ranges, $cites_range;
194 warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
197 warn '# ranges ', dump @ranges;
198 @ranges = () if $#ranges == 1; # just take all
205 get_results unless $skip_results;
213 $part .= $ranges[0]->[0] . '.';