/[webpac2]/trunk/bin/isi-download-results.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/bin/isi-download-results.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1331 - (hide annotations)
Sun Apr 18 10:56:01 2010 UTC (14 years, 1 month ago) by dpavlin
File MIME type: text/plain
File size: 4575 byte(s)
specify query as command-line params

1 dpavlin 1280 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 1292 # Advanced search syntax:
7     # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
8    
9 dpavlin 1291 our $q = 'AD=Croatia';
10 dpavlin 1289 my $range_size = 500;
11 dpavlin 1294 my $overlap = 3; # between previous and this range
12 dpavlin 1299 my $skip_results = 1;
13 dpavlin 1300 my $cites_by_year = 1;
14 dpavlin 1280
15 dpavlin 1292 my $max_cites = 5000; # ISI limit to get cites
16 dpavlin 1282
17 dpavlin 1294 if ( 0 ) {
18     $q = 'TS=psychology AND AD=Croatia';
19     $range_size = 50;
20     $overlap = 0;
21     $max_cites = 50;
22     }
23    
24 dpavlin 1331 $q = join(' ', @ARGV) if @ARGV;
25    
26 dpavlin 1280 use WWW::Mechanize;
27 dpavlin 1291 use Data::Dump qw(dump);
28     use File::Path;
29 dpavlin 1280
30 dpavlin 1289 our $mech = WWW::Mechanize->new(
31 dpavlin 1280 autocheck => 1,
32     cookie_jar => undef,
33     );
34    
35 dpavlin 1281 our $step = 0;
36 dpavlin 1294 our @ranges;
37 dpavlin 1280
38 dpavlin 1291 my $dir = '/tmp/isi/';
39 dpavlin 1331 #rmtree $dir if -e $dir;
40     mkdir $dir unless -d $dir;
41 dpavlin 1291
42 dpavlin 1280 sub save_mech {
43 dpavlin 1292 my $path = shift;
44 dpavlin 1281 $step++;
45 dpavlin 1291 my $base_path = sprintf('%s/%04d', $dir,$step);
46 dpavlin 1292 $path ||= $base_path;
47     $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
48 dpavlin 1281 $mech->save_content( $path );
49 dpavlin 1290 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
50     open(my $dump, '>', "$base_path.dump.txt");
51     $mech->dump_all($dump);
52 dpavlin 1280 }
53    
54     warn "# get session";
55     $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
56 dpavlin 1292 save_mech;
57 dpavlin 1280
58 dpavlin 1290 sub search {
59     warn "# advanced serach";
60     $mech->follow_link( url_regex => qr/AdvancedSearch/ );
61 dpavlin 1292 save_mech;
62 dpavlin 1280
63 dpavlin 1290 warn "# cookie_jar ", dump $mech->cookie_jar;
64 dpavlin 1280
65 dpavlin 1292 my $q_this = $q;
66    
67     if ( @ranges ) {
68     $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
69     }
70    
71     warn "# submit_form search: $q_this\n";
72 dpavlin 1290 $mech->submit_form(
73     fields => {
74 dpavlin 1292 'value(input1)' => $q_this,
75     },
76 dpavlin 1290 );
77 dpavlin 1292 save_mech;
78 dpavlin 1280
79 dpavlin 1290 warn "# summary";
80     $mech->follow_link( url_regex => qr/summary/ );
81 dpavlin 1292 save_mech;
82 dpavlin 1290 }
83    
84 dpavlin 1289 sub get_results {
85 dpavlin 1292 my $desc = shift;
86 dpavlin 1289 my $from = 1;
87 dpavlin 1280
88 dpavlin 1289 while ( 1 ) {
89 dpavlin 1280
90 dpavlin 1289 my $to = $from + $range_size;
91 dpavlin 1280
92 dpavlin 1292 warn "# submit_form results $from - $to\n";
93    
94 dpavlin 1289 $mech->submit_form(
95     form_name => 'summary_output_form',
96     fields => {
97     record_select_type => 'range',
98     mark_from => $from,
99     mark_to => $to,
100     mark_id => 'WOS',
101 dpavlin 1280
102 dpavlin 1289 qo_fields => 'fullrecord',
103     citedref => 'citedref',
104 dpavlin 1280
105 dpavlin 1289 save_options => 'plain_text',
106 dpavlin 1280
107 dpavlin 1289 fields => 'Full',
108     format => 'save',
109     },
110     button => 'save',
111     );
112 dpavlin 1292 save_mech;
113 dpavlin 1281
114 dpavlin 1292
115 dpavlin 1289 if ( $mech->content =~ m{invalid API call} ) {
116     $mech->back;
117 dpavlin 1290 last;
118 dpavlin 1289 }
119 dpavlin 1281
120 dpavlin 1292
121     my $path = "/tmp/isi.$q.$from-$to";
122     $path .= '.' . $desc if $desc;
123    
124     warn "save $from - $to into $path\n";
125 dpavlin 1289 $mech->follow_link( url_regex => qr/save_file/ );
126 dpavlin 1292 save_mech $path;
127 dpavlin 1281
128 dpavlin 1292 $from += $range_size - $overlap;
129 dpavlin 1281
130 dpavlin 1289 $mech->back;
131     $mech->back;
132 dpavlin 1292 #save_mech;
133 dpavlin 1290 }
134 dpavlin 1281 }
135 dpavlin 1289
136    
137 dpavlin 1290 sub citations {
138     warn "# citation report";
139     $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
140 dpavlin 1292 save_mech;
141 dpavlin 1289
142 dpavlin 1290 warn "view citing articles";
143     $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
144 dpavlin 1292 save_mech;
145 dpavlin 1290 }
146 dpavlin 1289
147 dpavlin 1290 sub years {
148 dpavlin 1292 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
149 dpavlin 1291 if ( ! $years_url ) {
150 dpavlin 1292 warn "W: can't find ra_name link\n";
151 dpavlin 1291 return;
152     }
153     $years_url = $years_url->url_abs;
154 dpavlin 1290 warn "## $years_url";
155 dpavlin 1291 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
156 dpavlin 1292 warn "W: no ra_name in $years_url\n";
157 dpavlin 1291 return;
158     }
159 dpavlin 1290 warn "# refine years (hidden by javascript)";
160 dpavlin 1292 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
161 dpavlin 1290 $mech->get( $years_url );
162 dpavlin 1292 save_mech;
163 dpavlin 1290
164     my $html = $mech->content;
165 dpavlin 1291 my $years;
166 dpavlin 1292 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
167     my ( $year, $count ) = ( $1, $2 );
168     $count =~ s{,}{}g;
169     $years->{$year} = $count;
170 dpavlin 1290 }
171 dpavlin 1291 warn "# years ",dump $years;
172 dpavlin 1290 $mech->back;
173 dpavlin 1292
174     my @y = sort keys %$years;
175    
176 dpavlin 1300 @ranges = ();
177    
178     if ( $cites_by_year ) {
179     push @ranges, [ $_ ] foreach @y;
180     warn "# cites_by_year ranges ", dump @ranges;
181     return;
182     }
183    
184 dpavlin 1292 my $y = shift @y;
185     my $size = $years->{$y};
186    
187     my $cites_range;
188     $cites_range = [$y] if $y;
189    
190     foreach my $y ( @y ) {
191     if ( $size + $years->{$y} > $max_cites ) {
192     push @ranges, $cites_range;
193     warn "# cites_range $size years ",dump( $cites_range ),$/;
194    
195     $cites_range = [];
196     $size = 0;
197     }
198     $size += $years->{$y};
199     push @$cites_range, $y;
200     }
201    
202     if ( $cites_range ) {
203     push @ranges, $cites_range;
204     warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
205     }
206    
207     warn '# ranges ', dump @ranges;
208     @ranges = () if $#ranges == 1; # just take all
209    
210 dpavlin 1291 return $years;
211 dpavlin 1290 }
212    
213 dpavlin 1291 search;
214 dpavlin 1290 years;
215 dpavlin 1299 get_results unless $skip_results;
216 dpavlin 1290
217 dpavlin 1291
218 dpavlin 1290 citations;
219 dpavlin 1289
220 dpavlin 1292 do {
221     my $part;
222     if ( @ranges ) {
223     $part .= $ranges[0]->[0] . '.';
224     search;
225 dpavlin 1294 citations;
226 dpavlin 1292 }
227     $part .= 'citing';
228     get_results $part;
229     } while ( @ranges );
230    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26