/[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 1299 - (hide annotations)
Sat Sep 19 23:37:55 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 4380 byte(s)
$skip_results to get just 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 1299 my $skip_results = 1;
13 dpavlin 1280
14 dpavlin 1292 my $max_cites = 5000; # ISI limit to get cites
15 dpavlin 1282
16 dpavlin 1294 if ( 0 ) {
17     $q = 'TS=psychology AND AD=Croatia';
18     $range_size = 50;
19     $overlap = 0;
20     $max_cites = 50;
21     }
22    
23 dpavlin 1280 use WWW::Mechanize;
24 dpavlin 1291 use Data::Dump qw(dump);
25     use File::Path;
26 dpavlin 1280
27 dpavlin 1289 our $mech = WWW::Mechanize->new(
28 dpavlin 1280 autocheck => 1,
29     cookie_jar => undef,
30     );
31    
32 dpavlin 1281 our $step = 0;
33 dpavlin 1294 our @ranges;
34 dpavlin 1280
35 dpavlin 1291 my $dir = '/tmp/isi/';
36     rmtree $dir if -e $dir;
37     mkdir $dir;
38    
39 dpavlin 1280 sub save_mech {
40 dpavlin 1292 my $path = shift;
41 dpavlin 1281 $step++;
42 dpavlin 1291 my $base_path = sprintf('%s/%04d', $dir,$step);
43 dpavlin 1292 $path ||= $base_path;
44     $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
45 dpavlin 1281 $mech->save_content( $path );
46 dpavlin 1290 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
47     open(my $dump, '>', "$base_path.dump.txt");
48     $mech->dump_all($dump);
49 dpavlin 1280 }
50    
51     warn "# get session";
52     $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
53 dpavlin 1292 save_mech;
54 dpavlin 1280
55 dpavlin 1290 sub search {
56     warn "# advanced serach";
57     $mech->follow_link( url_regex => qr/AdvancedSearch/ );
58 dpavlin 1292 save_mech;
59 dpavlin 1280
60 dpavlin 1290 warn "# cookie_jar ", dump $mech->cookie_jar;
61 dpavlin 1280
62 dpavlin 1292 my $q_this = $q;
63    
64     if ( @ranges ) {
65     $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
66     }
67    
68     warn "# submit_form search: $q_this\n";
69 dpavlin 1290 $mech->submit_form(
70     fields => {
71 dpavlin 1292 'value(input1)' => $q_this,
72     },
73 dpavlin 1290 );
74 dpavlin 1292 save_mech;
75 dpavlin 1280
76 dpavlin 1290 warn "# summary";
77     $mech->follow_link( url_regex => qr/summary/ );
78 dpavlin 1292 save_mech;
79 dpavlin 1290 }
80    
81 dpavlin 1289 sub get_results {
82 dpavlin 1292 my $desc = shift;
83 dpavlin 1289 my $from = 1;
84 dpavlin 1280
85 dpavlin 1289 while ( 1 ) {
86 dpavlin 1280
87 dpavlin 1289 my $to = $from + $range_size;
88 dpavlin 1280
89 dpavlin 1292 warn "# submit_form results $from - $to\n";
90    
91 dpavlin 1289 $mech->submit_form(
92     form_name => 'summary_output_form',
93     fields => {
94     record_select_type => 'range',
95     mark_from => $from,
96     mark_to => $to,
97     mark_id => 'WOS',
98 dpavlin 1280
99 dpavlin 1289 qo_fields => 'fullrecord',
100     citedref => 'citedref',
101 dpavlin 1280
102 dpavlin 1289 save_options => 'plain_text',
103 dpavlin 1280
104 dpavlin 1289 fields => 'Full',
105     format => 'save',
106     },
107     button => 'save',
108     );
109 dpavlin 1292 save_mech;
110 dpavlin 1281
111 dpavlin 1292
112 dpavlin 1289 if ( $mech->content =~ m{invalid API call} ) {
113     $mech->back;
114 dpavlin 1290 last;
115 dpavlin 1289 }
116 dpavlin 1281
117 dpavlin 1292
118     my $path = "/tmp/isi.$q.$from-$to";
119     $path .= '.' . $desc if $desc;
120    
121     warn "save $from - $to into $path\n";
122 dpavlin 1289 $mech->follow_link( url_regex => qr/save_file/ );
123 dpavlin 1292 save_mech $path;
124 dpavlin 1281
125 dpavlin 1292 $from += $range_size - $overlap;
126 dpavlin 1281
127 dpavlin 1289 $mech->back;
128     $mech->back;
129 dpavlin 1292 #save_mech;
130 dpavlin 1290 }
131 dpavlin 1281 }
132 dpavlin 1289
133    
134 dpavlin 1290 sub citations {
135     warn "# citation report";
136     $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
137 dpavlin 1292 save_mech;
138 dpavlin 1289
139 dpavlin 1290 warn "view citing articles";
140     $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
141 dpavlin 1292 save_mech;
142 dpavlin 1290 }
143 dpavlin 1289
144 dpavlin 1290 sub years {
145 dpavlin 1292 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
146 dpavlin 1291 if ( ! $years_url ) {
147 dpavlin 1292 warn "W: can't find ra_name link\n";
148 dpavlin 1291 return;
149     }
150     $years_url = $years_url->url_abs;
151 dpavlin 1290 warn "## $years_url";
152 dpavlin 1291 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
153 dpavlin 1292 warn "W: no ra_name in $years_url\n";
154 dpavlin 1291 return;
155     }
156 dpavlin 1290 warn "# refine years (hidden by javascript)";
157 dpavlin 1292 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
158 dpavlin 1290 $mech->get( $years_url );
159 dpavlin 1292 save_mech;
160 dpavlin 1290
161     my $html = $mech->content;
162 dpavlin 1291 my $years;
163 dpavlin 1292 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
164     my ( $year, $count ) = ( $1, $2 );
165     $count =~ s{,}{}g;
166     $years->{$year} = $count;
167 dpavlin 1290 }
168 dpavlin 1291 warn "# years ",dump $years;
169 dpavlin 1290 $mech->back;
170 dpavlin 1292
171     my @y = sort keys %$years;
172    
173     my $y = shift @y;
174     my $size = $years->{$y};
175    
176     @ranges = ();
177     my $cites_range;
178     $cites_range = [$y] if $y;
179    
180     foreach my $y ( @y ) {
181     if ( $size + $years->{$y} > $max_cites ) {
182     push @ranges, $cites_range;
183     warn "# cites_range $size years ",dump( $cites_range ),$/;
184    
185     $cites_range = [];
186     $size = 0;
187     }
188     $size += $years->{$y};
189     push @$cites_range, $y;
190     }
191    
192     if ( $cites_range ) {
193     push @ranges, $cites_range;
194     warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
195     }
196    
197     warn '# ranges ', dump @ranges;
198     @ranges = () if $#ranges == 1; # just take all
199    
200 dpavlin 1291 return $years;
201 dpavlin 1290 }
202    
203 dpavlin 1291 search;
204 dpavlin 1290 years;
205 dpavlin 1299 get_results unless $skip_results;
206 dpavlin 1290
207 dpavlin 1291
208 dpavlin 1290 citations;
209 dpavlin 1289
210 dpavlin 1292 do {
211     my $part;
212     if ( @ranges ) {
213     $part .= $ranges[0]->[0] . '.';
214     search;
215 dpavlin 1294 citations;
216 dpavlin 1292 }
217     $part .= 'citing';
218     get_results $part;
219     } while ( @ranges );
220    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26