12 |
# http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html |
# http://images.isiknowledge.com/WOK46/help/WOS/h_advanced_examples.html |
13 |
|
|
14 |
our $q = 'AD=Croatia'; |
our $q = 'AD=Croatia'; |
15 |
|
$q = 'CA=BRATKO, D'; |
16 |
|
|
17 |
my $range_size = 500; |
my $range_size = 500; |
18 |
my $overlap = 3; # between previous and this range |
my $overlap = 3; # between previous and this range |
19 |
my $skip_results = 0; |
|
20 |
|
my $results = 0; |
21 |
|
my $citations = 0; |
22 |
|
my $cited_reference = 0; # html tables |
23 |
|
my $citing_articles = 1; # as many files as cited articles |
24 |
|
|
25 |
my $cites_by_year = 0; |
my $cites_by_year = 0; |
26 |
|
|
27 |
my $max_cites = 5000; # ISI limit to get cites |
my $max_cites = 5000; # ISI limit to get cites |
28 |
|
|
|
if ( 0 ) { |
|
|
$q = 'TS=psychology AND AD=Croatia'; |
|
|
$range_size = 50; |
|
|
$overlap = 0; |
|
|
$max_cites = 50; |
|
|
} |
|
|
|
|
29 |
$q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV; |
$q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV; |
30 |
|
|
31 |
|
warn "QUERY: $q\n"; |
32 |
|
|
33 |
our $mech = WWW::Mechanize->new( |
our $mech = WWW::Mechanize->new( |
34 |
autocheck => 1, |
autocheck => 0, # it dies in reference download with it! |
35 |
cookie_jar => undef, |
cookie_jar => undef, |
36 |
); |
); |
37 |
|
|
120 |
last; |
last; |
121 |
} |
} |
122 |
|
|
123 |
|
if ( $mech->content =~ m{Please wait while your request is processed} ) { |
124 |
|
warn "WARNING: processing request"; |
125 |
|
} |
126 |
|
|
127 |
|
|
128 |
my $path = "/tmp/isi.$q.$from-$to"; |
my $path = "/tmp/isi.$q.$from-$to"; |
129 |
$path .= '.' . $desc if $desc; |
$path .= '.' . $desc if $desc; |
217 |
return $years; |
return $years; |
218 |
} |
} |
219 |
|
|
220 |
search; |
if ( $results ) { |
221 |
years; |
search; |
222 |
get_results unless $skip_results; |
years; |
223 |
|
get_results; |
224 |
|
} |
225 |
|
|
226 |
|
if ( $citations ) { |
227 |
|
|
228 |
citations; |
citations; |
229 |
|
years unless @ranges; |
230 |
|
|
231 |
do { |
do { |
232 |
my $part; |
my $part; |
233 |
if ( @ranges ) { |
if ( @ranges ) { |
234 |
$part .= $ranges[0]->[0] . '.'; |
$part .= $ranges[0]->[0] . '.'; |
235 |
search; |
search; |
236 |
citations; |
citations; |
237 |
} |
} |
238 |
$part .= 'citing'; |
$part .= 'citing'; |
239 |
get_results $part; |
get_results $part; |
240 |
} while ( @ranges ); |
} while ( @ranges ); |
241 |
|
|
242 |
|
} |
243 |
|
|
244 |
|
if ( $q =~ m{CA=(.+)} && $cited_reference ) { |
245 |
|
|
246 |
|
my $CA = $1; |
247 |
|
|
248 |
|
warn "# citated reference search"; |
249 |
|
$mech->follow_link( url_regex => qr/CitedReferenceSearch/ ); |
250 |
|
save_mech; |
251 |
|
|
252 |
|
|
253 |
|
$mech->submit_form( |
254 |
|
form_name => 'WOS_CitedReferenceSearch_input_form', |
255 |
|
fields => { |
256 |
|
'value(input1)' => $CA, |
257 |
|
}, |
258 |
|
); |
259 |
|
|
260 |
|
my $page = 1; |
261 |
|
my $records = $1 if $mech->content =~ m/(\d+)\s+records/; |
262 |
|
warn "# found $records records\n"; |
263 |
|
my $last_span = 'fake'; |
264 |
|
|
265 |
|
while (1) { |
266 |
|
save_mech "/tmp/isi.$q.citedref.$page"; |
267 |
|
|
268 |
|
$page++; |
269 |
|
|
270 |
|
$mech->submit_form( |
271 |
|
form_name => 'summary_navigation', |
272 |
|
fields => { |
273 |
|
'page' => $page, |
274 |
|
}, |
275 |
|
); |
276 |
|
|
277 |
|
$mech->form_name( 'summary_navigation' ); |
278 |
|
last if $mech->value('page') < $page; |
279 |
|
|
280 |
|
if ( $mech->content =~ m/(\d+\s*-\s*(\d+))/ ) { |
281 |
|
warn "span: $1\n"; |
282 |
|
last if $2 == $records; |
283 |
|
last if $1 == $last_span; |
284 |
|
$last_span = $1; |
285 |
|
} elsif ( $page > 5 ) { |
286 |
|
warn "ARTIFICALLY LIMITED TO 5 PAGES WITHOUT VALID SPAN!"; |
287 |
|
last; |
288 |
|
} |
289 |
|
|
290 |
|
} |
291 |
|
|
292 |
|
} |
293 |
|
|
294 |
|
if ( $q =~ m{CA=(.+)} && $citing_articles ) { |
295 |
|
|
296 |
|
search; |
297 |
|
|
298 |
|
my $orig_q = $q; |
299 |
|
my $nr = 0; |
300 |
|
|
301 |
|
foreach my $link ( $mech->find_all_links( url_regex => qr/CitingArticles.do/ ) ) { |
302 |
|
$nr++; |
303 |
|
warn "link $nr\n"; |
304 |
|
$mech->get( $link->url ); |
305 |
|
save_mech; |
306 |
|
$q = $orig_q . '.citing_article.' . $nr; |
307 |
|
get_results; |
308 |
|
$mech->back; |
309 |
|
save_mech; |
310 |
|
$mech->back; |
311 |
|
save_mech; |
312 |
|
} |
313 |
|
|
314 |
|
$q = $orig_q; |
315 |
|
} |
316 |
|
|
317 |
|
warn "OVER\n"; |