/[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 1294 - (hide annotations)
Sat Sep 19 22:54:45 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 4337 byte(s)
really download citations
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 1280
13 dpavlin 1292 my $max_cites = 5000; # ISI limit to get cites
14 dpavlin 1282
15 dpavlin 1294 if ( 0 ) {
16     $q = 'TS=psychology AND AD=Croatia';
17     $range_size = 50;
18     $overlap = 0;
19     $max_cites = 50;
20     }
21    
22 dpavlin 1280 use WWW::Mechanize;
23 dpavlin 1291 use Data::Dump qw(dump);
24     use File::Path;
25 dpavlin 1280
26 dpavlin 1289 our $mech = WWW::Mechanize->new(
27 dpavlin 1280 autocheck => 1,
28     cookie_jar => undef,
29     );
30    
31 dpavlin 1281 our $step = 0;
32 dpavlin 1294 our @ranges;
33 dpavlin 1280
34 dpavlin 1291 my $dir = '/tmp/isi/';
35     rmtree $dir if -e $dir;
36     mkdir $dir;
37    
38 dpavlin 1280 sub save_mech {
39 dpavlin 1292 my $path = shift;
40 dpavlin 1281 $step++;
41 dpavlin 1291 my $base_path = sprintf('%s/%04d', $dir,$step);
42 dpavlin 1292 $path ||= $base_path;
43     $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
44 dpavlin 1281 $mech->save_content( $path );
45 dpavlin 1290 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
46     open(my $dump, '>', "$base_path.dump.txt");
47     $mech->dump_all($dump);
48 dpavlin 1280 }
49    
50     warn "# get session";
51     $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
52 dpavlin 1292 save_mech;
53 dpavlin 1280
54 dpavlin 1290 sub search {
55     warn "# advanced serach";
56     $mech->follow_link( url_regex => qr/AdvancedSearch/ );
57 dpavlin 1292 save_mech;
58 dpavlin 1280
59 dpavlin 1290 warn "# cookie_jar ", dump $mech->cookie_jar;
60 dpavlin 1280
61 dpavlin 1292 my $q_this = $q;
62    
63     if ( @ranges ) {
64     $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
65     }
66    
67     warn "# submit_form search: $q_this\n";
68 dpavlin 1290 $mech->submit_form(
69     fields => {
70 dpavlin 1292 'value(input1)' => $q_this,
71     },
72 dpavlin 1290 );
73 dpavlin 1292 save_mech;
74 dpavlin 1280
75 dpavlin 1290 warn "# summary";
76     $mech->follow_link( url_regex => qr/summary/ );
77 dpavlin 1292 save_mech;
78 dpavlin 1290 }
79    
80 dpavlin 1289 sub get_results {
81 dpavlin 1292 my $desc = shift;
82 dpavlin 1289 my $from = 1;
83 dpavlin 1280
84 dpavlin 1289 while ( 1 ) {
85 dpavlin 1280
86 dpavlin 1289 my $to = $from + $range_size;
87 dpavlin 1280
88 dpavlin 1292 warn "# submit_form results $from - $to\n";
89    
90 dpavlin 1289 $mech->submit_form(
91     form_name => 'summary_output_form',
92     fields => {
93     record_select_type => 'range',
94     mark_from => $from,
95     mark_to => $to,
96     mark_id => 'WOS',
97 dpavlin 1280
98 dpavlin 1289 qo_fields => 'fullrecord',
99     citedref => 'citedref',
100 dpavlin 1280
101 dpavlin 1289 save_options => 'plain_text',
102 dpavlin 1280
103 dpavlin 1289 fields => 'Full',
104     format => 'save',
105     },
106     button => 'save',
107     );
108 dpavlin 1292 save_mech;
109 dpavlin 1281
110 dpavlin 1292
111 dpavlin 1289 if ( $mech->content =~ m{invalid API call} ) {
112     $mech->back;
113 dpavlin 1290 last;
114 dpavlin 1289 }
115 dpavlin 1281
116 dpavlin 1292
117     my $path = "/tmp/isi.$q.$from-$to";
118     $path .= '.' . $desc if $desc;
119    
120     warn "save $from - $to into $path\n";
121 dpavlin 1289 $mech->follow_link( url_regex => qr/save_file/ );
122 dpavlin 1292 save_mech $path;
123 dpavlin 1281
124 dpavlin 1292 $from += $range_size - $overlap;
125 dpavlin 1281
126 dpavlin 1289 $mech->back;
127     $mech->back;
128 dpavlin 1292 #save_mech;
129 dpavlin 1290 }
130 dpavlin 1281 }
131 dpavlin 1289
132    
133 dpavlin 1290 sub citations {
134     warn "# citation report";
135     $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
136 dpavlin 1292 save_mech;
137 dpavlin 1289
138 dpavlin 1290 warn "view citing articles";
139     $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
140 dpavlin 1292 save_mech;
141 dpavlin 1290 }
142 dpavlin 1289
143 dpavlin 1290 sub years {
144 dpavlin 1292 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
145 dpavlin 1291 if ( ! $years_url ) {
146 dpavlin 1292 warn "W: can't find ra_name link\n";
147 dpavlin 1291 return;
148     }
149     $years_url = $years_url->url_abs;
150 dpavlin 1290 warn "## $years_url";
151 dpavlin 1291 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
152 dpavlin 1292 warn "W: no ra_name in $years_url\n";
153 dpavlin 1291 return;
154     }
155 dpavlin 1290 warn "# refine years (hidden by javascript)";
156 dpavlin 1292 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
157 dpavlin 1290 $mech->get( $years_url );
158 dpavlin 1292 save_mech;
159 dpavlin 1290
160     my $html = $mech->content;
161 dpavlin 1291 my $years;
162 dpavlin 1292 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
163     my ( $year, $count ) = ( $1, $2 );
164     $count =~ s{,}{}g;
165     $years->{$year} = $count;
166 dpavlin 1290 }
167 dpavlin 1291 warn "# years ",dump $years;
168 dpavlin 1290 $mech->back;
169 dpavlin 1292
170     my @y = sort keys %$years;
171    
172     my $y = shift @y;
173     my $size = $years->{$y};
174    
175     @ranges = ();
176     my $cites_range;
177     $cites_range = [$y] if $y;
178    
179     foreach my $y ( @y ) {
180     if ( $size + $years->{$y} > $max_cites ) {
181     push @ranges, $cites_range;
182     warn "# cites_range $size years ",dump( $cites_range ),$/;
183    
184     $cites_range = [];
185     $size = 0;
186     }
187     $size += $years->{$y};
188     push @$cites_range, $y;
189     }
190    
191     if ( $cites_range ) {
192     push @ranges, $cites_range;
193     warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
194     }
195    
196     warn '# ranges ', dump @ranges;
197     @ranges = () if $#ranges == 1; # just take all
198    
199 dpavlin 1291 return $years;
200 dpavlin 1290 }
201    
202 dpavlin 1291 search;
203 dpavlin 1290 years;
204 dpavlin 1291 get_results;
205 dpavlin 1290
206 dpavlin 1291
207 dpavlin 1290 citations;
208 dpavlin 1289
209 dpavlin 1292 do {
210     my $part;
211     if ( @ranges ) {
212     $part .= $ranges[0]->[0] . '.';
213     search;
214 dpavlin 1294 citations;
215 dpavlin 1292 }
216     $part .= 'citing';
217     get_results $part;
218     } while ( @ranges );
219    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26