6 # Advanced search syntax:
7 # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
11 my $overlap = 10; # between previous and this range
13 $q = 'TS=psychology AND AD=Croatia' if @ARGV; # FIXME debug
15 my $max_cites = 5000; # ISI limit to get cites
19 use Data::Dump qw(dump);
22 our $mech = WWW::Mechanize->new(
29 my $dir = '/tmp/isi/';
30 rmtree $dir if -e $dir;
36 my $base_path = sprintf('%s/%04d', $dir,$step);
38 $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
39 $mech->save_content( $path );
40 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
41 open(my $dump, '>', "$base_path.dump.txt");
42 $mech->dump_all($dump);
46 $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
50 warn "# advanced serach";
51 $mech->follow_link( url_regex => qr/AdvancedSearch/ );
54 warn "# cookie_jar ", dump $mech->cookie_jar;
59 $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
62 warn "# submit_form search: $q_this\n";
65 'value(input1)' => $q_this,
71 $mech->follow_link( url_regex => qr/summary/ );
81 my $to = $from + $range_size;
83 warn "# submit_form results $from - $to\n";
86 form_name => 'summary_output_form',
88 record_select_type => 'range',
93 qo_fields => 'fullrecord',
94 citedref => 'citedref',
96 save_options => 'plain_text',
106 if ( $mech->content =~ m{invalid API call} ) {
112 my $path = "/tmp/isi.$q.$from-$to";
113 $path .= '.' . $desc if $desc;
115 warn "save $from - $to into $path\n";
116 $mech->follow_link( url_regex => qr/save_file/ );
119 $from += $range_size - $overlap;
129 warn "# citation report";
130 $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
133 warn "view citing articles";
134 $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
139 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
140 if ( ! $years_url ) {
141 warn "W: can't find ra_name link\n";
144 $years_url = $years_url->url_abs;
145 warn "## $years_url";
146 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
147 warn "W: no ra_name in $years_url\n";
150 warn "# refine years (hidden by javascript)";
151 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
152 $mech->get( $years_url );
155 my $html = $mech->content;
157 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
158 my ( $year, $count ) = ( $1, $2 );
160 $years->{$year} = $count;
162 warn "# years ",dump $years;
165 my @y = sort keys %$years;
168 my $size = $years->{$y};
172 $cites_range = [$y] if $y;
174 foreach my $y ( @y ) {
175 if ( $size + $years->{$y} > $max_cites ) {
176 push @ranges, $cites_range;
177 warn "# cites_range $size years ",dump( $cites_range ),$/;
182 $size += $years->{$y};
183 push @$cites_range, $y;
186 if ( $cites_range ) {
187 push @ranges, $cites_range;
188 warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
191 warn '# ranges ', dump @ranges;
192 @ranges = () if $#ranges == 1; # just take all
207 $part .= $ranges[0]->[0] . '.';