7 use Data::Dump qw(dump);
11 # Advanced search syntax:
12 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
14 our $q = 'AD=Croatia';
18 my $overlap = 3; # between previous and this range
22 my $cited_reference = 0; # html tables
23 my $citing_articles = 1; # as many files as cited articles
25 my $cites_by_year = 0;
27 my $max_cites = 5000; # ISI limit to get cites
29 $q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV;
33 our $mech = WWW::Mechanize->new(
34 autocheck => 0, # it dies in reference download with it!
41 my $dir = '/tmp/isi/';
42 #rmtree $dir if -e $dir;
43 mkdir $dir unless -d $dir;
48 my $base_path = sprintf('%s/%04d', $dir,$step);
50 $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
51 $mech->save_content( $path );
52 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
53 open(my $dump, '>', "$base_path.dump.txt");
54 $mech->dump_all($dump);
58 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
62 warn "# advanced serach";
63 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
66 warn "# cookie_jar ", dump $mech->cookie_jar;
71 $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
74 warn "# submit_form search: $q_this\n";
77 'value(input1)' => $q_this,
83 $mech->follow_link( url_regex => qr/summary/ );
93 my $to = $from + $range_size;
95 warn "# submit_form results $from - $to\n";
98 form_name => 'summary_output_form',
100 record_select_type => 'range',
105 qo_fields => 'fullrecord',
106 citedref => 'citedref',
108 save_options => 'plain_text',
118 if ( $mech->content =~ m{invalid API call} ) {
123 if ( $mech->content =~ m{Please wait while your request is processed} ) {
124 warn "WARNING: processing request";
128 my $path = "/tmp/isi.$q.$from-$to";
129 $path .= '.' . $desc if $desc;
131 warn "save $from - $to into $path\n";
132 $mech->follow_link( url_regex => qr/save_file/ );
135 $from += $range_size - $overlap;
145 warn "# citation report";
146 $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
149 warn "view citing articles";
150 $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
155 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
156 if ( ! $years_url ) {
157 warn "W: can't find ra_name link\n";
160 $years_url = $years_url->url_abs;
161 warn "## $years_url";
162 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
163 warn "W: no ra_name in $years_url\n";
166 warn "# refine years (hidden by javascript)";
167 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
168 $mech->get( $years_url );
171 my $html = $mech->content;
173 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
174 my ( $year, $count ) = ( $1, $2 );
176 $years->{$year} = $count;
178 warn "# years ",dump $years;
181 my @y = sort keys %$years;
185 if ( $cites_by_year ) {
186 push @ranges, [ $_ ] foreach @y;
187 warn "# cites_by_year ranges ", dump @ranges;
192 my $size = $years->{$y};
195 $cites_range = [$y] if $y;
197 foreach my $y ( @y ) {
198 if ( $size + $years->{$y} > $max_cites ) {
199 push @ranges, $cites_range;
200 warn "# cites_range $size years ",dump( $cites_range ),$/;
205 $size += $years->{$y};
206 push @$cites_range, $y;
209 if ( $cites_range ) {
210 push @ranges, $cites_range;
211 warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
214 warn '# ranges ', dump @ranges;
215 @ranges = () if $#ranges == 1; # just take all
224 warn "next_page $page\n";
227 form_name => 'summary_navigation',
235 $mech->form_name( 'summary_navigation' );
236 my $is_next_page = $mech->value('page') == $page;
237 warn "no next_page" unless $is_next_page;
238 return $is_next_page;
250 years unless @ranges;
255 $part .= $ranges[0]->[0] . '.';
267 if ( $q =~ m{CA=(.+)} && $cited_reference ) {
271 warn "# citated reference search";
272 $mech->follow_link( url_regex => qr/CitedReferenceSearch/ );
277 form_name => 'WOS_CitedReferenceSearch_input_form',
279 'value(input1)' => $CA,
284 my $records = $1 if $mech->content =~ m/(\d+)\s+records/;
285 warn "# found $records records\n";
286 my $last_span = 'fake';
289 save_mech "/tmp/isi.$q.citedref.$page";
291 last unless next_page();
293 if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) {
295 last if $2 == $records;
296 last if $1 == $last_span;
298 } elsif ( $page > 5 ) {
299 warn "ARTIFICALLY LIMITED TO 5 PAGES WITHOUT VALID SPAN!";
307 if ( $q =~ m{CA=(.+)} && $citing_articles ) {
316 foreach my $link ( $mech->find_all_links( url_regex => qr/CitingArticles.do/ ) ) {
319 $mech->get( $link->url );
321 $q = $orig_q . '.citing_article.' . $nr;
326 #last if $nr > 3; # FIXME only for development