/[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 1335 - (hide annotations)
Sat Jun 5 15:49:59 2010 UTC (13 years, 11 months ago) by dpavlin
File MIME type: text/plain
File size: 5619 byte(s)
use cited reference search for CA= query

1 dpavlin 1280 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 1332 use WWW::Mechanize;
7     use Data::Dump qw(dump);
8     use File::Path;
9     use Text::Unaccent;
10    
11 dpavlin 1292 # Advanced search syntax:
12     # http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html
13    
14 dpavlin 1291 our $q = 'AD=Croatia';
15 dpavlin 1335 $q = 'CA=BRATKO, D';
16    
17 dpavlin 1289 my $range_size = 500;
18 dpavlin 1294 my $overlap = 3; # between previous and this range
19 dpavlin 1335
20 dpavlin 1334 my $results = 0;
21     my $citations = 0;
22 dpavlin 1335
23 dpavlin 1332 my $cites_by_year = 0;
24 dpavlin 1280
25 dpavlin 1292 my $max_cites = 5000; # ISI limit to get cites
26 dpavlin 1282
27 dpavlin 1332 $q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV;
28 dpavlin 1331
29 dpavlin 1335 warn "QUERY: $q\n";
30    
31 dpavlin 1289 our $mech = WWW::Mechanize->new(
32 dpavlin 1333 autocheck => 0, # it dies in reference download with it!
33 dpavlin 1280 cookie_jar => undef,
34     );
35    
36 dpavlin 1281 our $step = 0;
37 dpavlin 1294 our @ranges;
38 dpavlin 1280
39 dpavlin 1291 my $dir = '/tmp/isi/';
40 dpavlin 1331 #rmtree $dir if -e $dir;
41     mkdir $dir unless -d $dir;
42 dpavlin 1291
43 dpavlin 1280 sub save_mech {
44 dpavlin 1292 my $path = shift;
45 dpavlin 1281 $step++;
46 dpavlin 1291 my $base_path = sprintf('%s/%04d', $dir,$step);
47 dpavlin 1292 $path ||= $base_path;
48     $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
49 dpavlin 1281 $mech->save_content( $path );
50 dpavlin 1290 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
51     open(my $dump, '>', "$base_path.dump.txt");
52     $mech->dump_all($dump);
53 dpavlin 1280 }
54    
55     warn "# get session";
56     $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
57 dpavlin 1292 save_mech;
58 dpavlin 1280
59 dpavlin 1290 sub search {
60     warn "# advanced serach";
61     $mech->follow_link( url_regex => qr/AdvancedSearch/ );
62 dpavlin 1292 save_mech;
63 dpavlin 1280
64 dpavlin 1290 warn "# cookie_jar ", dump $mech->cookie_jar;
65 dpavlin 1280
66 dpavlin 1292 my $q_this = $q;
67    
68     if ( @ranges ) {
69     $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
70     }
71    
72     warn "# submit_form search: $q_this\n";
73 dpavlin 1290 $mech->submit_form(
74     fields => {
75 dpavlin 1292 'value(input1)' => $q_this,
76     },
77 dpavlin 1290 );
78 dpavlin 1292 save_mech;
79 dpavlin 1280
80 dpavlin 1290 warn "# summary";
81     $mech->follow_link( url_regex => qr/summary/ );
82 dpavlin 1292 save_mech;
83 dpavlin 1290 }
84    
85 dpavlin 1289 sub get_results {
86 dpavlin 1292 my $desc = shift;
87 dpavlin 1289 my $from = 1;
88 dpavlin 1280
89 dpavlin 1289 while ( 1 ) {
90 dpavlin 1280
91 dpavlin 1289 my $to = $from + $range_size;
92 dpavlin 1280
93 dpavlin 1292 warn "# submit_form results $from - $to\n";
94    
95 dpavlin 1289 $mech->submit_form(
96     form_name => 'summary_output_form',
97     fields => {
98     record_select_type => 'range',
99     mark_from => $from,
100     mark_to => $to,
101     mark_id => 'WOS',
102 dpavlin 1280
103 dpavlin 1289 qo_fields => 'fullrecord',
104     citedref => 'citedref',
105 dpavlin 1280
106 dpavlin 1289 save_options => 'plain_text',
107 dpavlin 1280
108 dpavlin 1289 fields => 'Full',
109     format => 'save',
110     },
111     button => 'save',
112     );
113 dpavlin 1292 save_mech;
114 dpavlin 1281
115 dpavlin 1292
116 dpavlin 1289 if ( $mech->content =~ m{invalid API call} ) {
117     $mech->back;
118 dpavlin 1290 last;
119 dpavlin 1289 }
120 dpavlin 1281
121 dpavlin 1292
122     my $path = "/tmp/isi.$q.$from-$to";
123     $path .= '.' . $desc if $desc;
124    
125     warn "save $from - $to into $path\n";
126 dpavlin 1289 $mech->follow_link( url_regex => qr/save_file/ );
127 dpavlin 1292 save_mech $path;
128 dpavlin 1281
129 dpavlin 1292 $from += $range_size - $overlap;
130 dpavlin 1281
131 dpavlin 1289 $mech->back;
132     $mech->back;
133 dpavlin 1292 #save_mech;
134 dpavlin 1290 }
135 dpavlin 1281 }
136 dpavlin 1289
137    
138 dpavlin 1290 sub citations {
139     warn "# citation report";
140     $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
141 dpavlin 1292 save_mech;
142 dpavlin 1289
143 dpavlin 1290 warn "view citing articles";
144     $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
145 dpavlin 1292 save_mech;
146 dpavlin 1290 }
147 dpavlin 1289
148 dpavlin 1290 sub years {
149 dpavlin 1292 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
150 dpavlin 1291 if ( ! $years_url ) {
151 dpavlin 1292 warn "W: can't find ra_name link\n";
152 dpavlin 1291 return;
153     }
154     $years_url = $years_url->url_abs;
155 dpavlin 1290 warn "## $years_url";
156 dpavlin 1291 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
157 dpavlin 1292 warn "W: no ra_name in $years_url\n";
158 dpavlin 1291 return;
159     }
160 dpavlin 1290 warn "# refine years (hidden by javascript)";
161 dpavlin 1292 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
162 dpavlin 1290 $mech->get( $years_url );
163 dpavlin 1292 save_mech;
164 dpavlin 1290
165     my $html = $mech->content;
166 dpavlin 1291 my $years;
167 dpavlin 1292 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
168     my ( $year, $count ) = ( $1, $2 );
169     $count =~ s{,}{}g;
170     $years->{$year} = $count;
171 dpavlin 1290 }
172 dpavlin 1291 warn "# years ",dump $years;
173 dpavlin 1290 $mech->back;
174 dpavlin 1292
175     my @y = sort keys %$years;
176    
177 dpavlin 1300 @ranges = ();
178    
179     if ( $cites_by_year ) {
180     push @ranges, [ $_ ] foreach @y;
181     warn "# cites_by_year ranges ", dump @ranges;
182     return;
183     }
184    
185 dpavlin 1292 my $y = shift @y;
186     my $size = $years->{$y};
187    
188     my $cites_range;
189     $cites_range = [$y] if $y;
190    
191     foreach my $y ( @y ) {
192     if ( $size + $years->{$y} > $max_cites ) {
193     push @ranges, $cites_range;
194     warn "# cites_range $size years ",dump( $cites_range ),$/;
195    
196     $cites_range = [];
197     $size = 0;
198     }
199     $size += $years->{$y};
200     push @$cites_range, $y;
201     }
202    
203     if ( $cites_range ) {
204     push @ranges, $cites_range;
205     warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
206     }
207    
208     warn '# ranges ', dump @ranges;
209     @ranges = () if $#ranges == 1; # just take all
210    
211 dpavlin 1291 return $years;
212 dpavlin 1290 }
213    
214 dpavlin 1335 if ( $results ) {
215     search;
216     years;
217     get_results;
218     }
219 dpavlin 1290
220 dpavlin 1334 if ( $citations ) {
221 dpavlin 1291
222 dpavlin 1334 citations;
223 dpavlin 1335 years unless @ranges;
224 dpavlin 1289
225 dpavlin 1334 do {
226     my $part;
227     if ( @ranges ) {
228     $part .= $ranges[0]->[0] . '.';
229     search;
230     citations;
231     }
232     $part .= 'citing';
233     get_results $part;
234     } while ( @ranges );
235 dpavlin 1292
236 dpavlin 1334 }
237 dpavlin 1335
238     if ( $q =~ m{CA=(.+)} ) {
239    
240     my $CA = $1;
241    
242     warn "# citated reference search";
243     $mech->follow_link( url_regex => qr/CitedReferenceSearch/ );
244     save_mech;
245    
246    
247     $mech->submit_form(
248     form_name => 'WOS_CitedReferenceSearch_input_form',
249     fields => {
250     'value(input1)' => $CA,
251     },
252     );
253    
254     my $page = 1;
255     my $records = $1 if $mech->content =~ m/(\d+)\s+records/;
256     warn "# found $records records\n";
257     my $last_span = 'fake';
258    
259     while (1) {
260     save_mech "/tmp/isi.$q.citedref.$page";
261    
262     $page++;
263    
264     $mech->submit_form(
265     form_name => 'summary_navigation',
266     fields => {
267     'page' => $page,
268     },
269     );
270    
271     $mech->form_name( 'summary_navigation' );
272     last if $mech->value('page') < $page;
273    
274     if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) {
275     warn "span: $1\n";
276     last if $2 == $records;
277     last if $1 == $last_span;
278     $last_span = $1;
279     } elsif ( $page > 5 ) {
280     warn "ARTIFICALLY LIMITED TO 5 PAGES WITHOUT VALID SPAN!";
281     last;
282     }
283    
284     }
285    
286     }
287    
288     warn "OVER\n";

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26