7 use Data::Dump qw(dump);
11 # Advanced search syntax:
12 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
14 our $q = 'AD=Croatia';
16 #$q = 'AD=(croat* OR hrvat*)';
19 my $overlap = 3; # between previous and this range
23 my $cited_reference = 0; # html tables
24 my $citing_articles = 0; # as many files as cited articles
26 my $cites_by_year = 0;
28 my $max_cites = 5000; # ISI limit to get cites
30 $q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV;
34 our $mech = WWW::Mechanize->new(
35 autocheck => 0, # it dies in reference download with it!
42 my $dir = '/tmp/isi/';
43 #rmtree $dir if -e $dir;
44 mkdir $dir unless -d $dir;
49 my $base_path = sprintf('%s/%04d', $dir,$step);
51 $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
52 $mech->save_content( $path );
53 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
54 open(my $dump, '>', "$base_path.dump.txt");
55 $mech->dump_all($dump);
59 $mech->get( 'http://www.webofknowledge.com/?DestApp=WOS' );
63 warn "# advanced serach";
64 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
67 warn "# cookie_jar ", dump $mech->cookie_jar;
72 $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
75 warn "# submit_form search: $q_this\n";
78 'value(input1)' => $q_this,
84 $mech->follow_link( url_regex => qr/summary/ );
94 my $to = $from + $range_size;
96 warn "# submit_form results $from - $to\n";
100 form_name => 'output_form',
102 'value(record_select_type)' => 'range',
110 fields_selection => 'ACCESSION_NUM FUNDING SUBJECT_CATEGORY JCR_CATEGORY LANG IDS PAGEC SABBR CITREFC ISSN PUBINFO KEYWORDS CITTIMES ADDRS CONFERENCE_SPONSORS DOCTYPE ABSTRACT CONFERENCE_INFO SOURCE TITLE AUTHORS',
111 filters => 'ACCESSION_NUM FUNDING SUBJECT_CATEGORY JCR_CATEGORY LANG IDS PAGEC SABBR CITREFC ISSN PUBINFO KEYWORDS CITTIMES ADDRS CONFERENCE_SPONSORS DOCTYPE ABSTRACT CONFERENCE_INFO SOURCE TITLE AUTHORS CITREF',
112 fullrec_fields_option => 'CITREF',
114 save_options => 'fieldtagged',
115 format => 'saveToFile',
118 button => 'saveToFile',
123 if ( $mech->content =~ m{invalid API call} ) {
128 if ( $mech->content !~ m{Please wait while your request is processed} ) {
129 warn "WARNING: expecting processing request";
133 my $path = "/tmp/isi.$q.$from-$to";
134 $path .= '.' . $desc if $desc;
136 warn "save $from - $to into $path\n";
138 form_name => 'etsForm',
142 $from += $range_size - $overlap;
152 warn "# citation report";
153 $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
156 warn "view citing articles";
157 $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
162 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
163 if ( ! $years_url ) {
164 warn "W: can't find ra_name link\n";
167 $years_url = $years_url->url_abs;
168 warn "## $years_url";
169 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
170 warn "W: no ra_name in $years_url\n";
173 warn "# refine years (hidden by javascript)";
174 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
175 $mech->get( $years_url );
178 my $html = $mech->content;
180 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
181 my ( $year, $count ) = ( $1, $2 );
183 $years->{$year} = $count;
185 warn "# years ",dump $years;
188 my @y = sort keys %$years;
192 if ( $cites_by_year ) {
193 push @ranges, [ $_ ] foreach @y;
194 warn "# cites_by_year ranges ", dump @ranges;
199 my $size = $years->{$y};
202 $cites_range = [$y] if $y;
204 foreach my $y ( @y ) {
205 if ( $size + $years->{$y} > $max_cites ) {
206 push @ranges, $cites_range;
207 warn "# cites_range $size years ",dump( $cites_range ),$/;
212 $size += $years->{$y};
213 push @$cites_range, $y;
216 if ( $cites_range ) {
217 push @ranges, $cites_range;
218 warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
221 warn '# ranges ', dump @ranges;
222 @ranges = () if $#ranges == 1; # just take all
231 warn "next_page $page\n";
234 form_name => 'summary_navigation',
242 $mech->form_name( 'summary_navigation' );
243 my $is_next_page = $mech->value('page') == $page;
244 warn "no next_page" unless $is_next_page;
245 return $is_next_page;
257 years unless @ranges;
262 $part .= $ranges[0]->[0] . '.';
274 if ( $q =~ m{CA=(.+)} && $cited_reference ) {
278 warn "# citated reference search";
279 $mech->follow_link( url_regex => qr/CitedReferenceSearch/ );
284 form_name => 'WOS_CitedReferenceSearch_input_form',
286 'value(input1)' => $CA,
291 my $records = $1 if $mech->content =~ m/(\d+)\s+records/;
292 warn "# found $records records\n";
293 my $last_span = 'fake';
296 save_mech "/tmp/isi.$q.citedref.$page";
298 last unless next_page();
300 if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) {
302 last if $2 == $records;
303 last if $1 == $last_span;
305 } elsif ( $page > 5 ) {
306 warn "ARTIFICALLY LIMITED TO 5 PAGES WITHOUT VALID SPAN!";
314 if ( $q =~ m{CA=(.+)} && $citing_articles ) {
323 foreach my $link ( $mech->find_all_links( url_regex => qr/CitingArticles.do/ ) ) {
326 $mech->get( $link->url );
328 $q = $orig_q . '.citing_article.' . $nr;
333 #last if $nr > 3; # FIXME only for development