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 |
|
|
20 |
my $results = 0; |
my $results = 0; |
21 |
my $citations = 0; |
my $citations = 0; |
22 |
|
|
23 |
my $cites_by_year = 0; |
my $cites_by_year = 0; |
24 |
|
|
25 |
my $max_cites = 5000; # ISI limit to get cites |
my $max_cites = 5000; # ISI limit to get cites |
26 |
|
|
|
if ( 0 ) { |
|
|
$q = 'TS=psychology AND AD=Croatia'; |
|
|
$range_size = 50; |
|
|
$overlap = 0; |
|
|
$max_cites = 50; |
|
|
} |
|
|
|
|
27 |
$q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV; |
$q = unac_string( 'utf-8', join(' ', @ARGV) ) if @ARGV; |
28 |
|
|
29 |
|
warn "QUERY: $q\n"; |
30 |
|
|
31 |
our $mech = WWW::Mechanize->new( |
our $mech = WWW::Mechanize->new( |
32 |
autocheck => 0, # it dies in reference download with it! |
autocheck => 0, # it dies in reference download with it! |
33 |
cookie_jar => undef, |
cookie_jar => undef, |
211 |
return $years; |
return $years; |
212 |
} |
} |
213 |
|
|
214 |
search; |
if ( $results ) { |
215 |
years; |
search; |
216 |
get_results 'results' if $results; |
years; |
217 |
|
get_results; |
218 |
|
} |
219 |
|
|
220 |
if ( $citations ) { |
if ( $citations ) { |
221 |
|
|
222 |
citations; |
citations; |
223 |
|
years unless @ranges; |
224 |
|
|
225 |
do { |
do { |
226 |
my $part; |
my $part; |
234 |
} while ( @ranges ); |
} while ( @ranges ); |
235 |
|
|
236 |
} |
} |
237 |
|
|
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"; |