/[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 1292 - (hide annotations)
Sat Sep 19 20:35:43 2009 UTC (14 years, 7 months ago) by dpavlin
File MIME type: text/plain
File size: 4278 byte(s)
- get citing records grouped into max_cites sized chunks using PY as criteria
- implement 10 record $overlap when getting more than $range_size (500) records
- cleanup save_mech
- fix PublictionYear extraction from html

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

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26