/[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 1300 - (hide annotations)
Sat Sep 19 23:49:04 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 4526 byte(s)
$cites_by_year as default for faster searching
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 1300 my $cites_by_year = 1;
14 dpavlin 1280
15 dpavlin 1292 my $max_cites = 5000; # ISI limit to get cites
16 dpavlin 1282
17 dpavlin 1294 if ( 0 ) {
18     $q = 'TS=psychology AND AD=Croatia';
19     $range_size = 50;
20     $overlap = 0;
21     $max_cites = 50;
22     }
23    
24 dpavlin 1280 use WWW::Mechanize;
25 dpavlin 1291 use Data::Dump qw(dump);
26     use File::Path;
27 dpavlin 1280
28 dpavlin 1289 our $mech = WWW::Mechanize->new(
29 dpavlin 1280 autocheck => 1,
30     cookie_jar => undef,
31     );
32    
33 dpavlin 1281 our $step = 0;
34 dpavlin 1294 our @ranges;
35 dpavlin 1280
36 dpavlin 1291 my $dir = '/tmp/isi/';
37     rmtree $dir if -e $dir;
38     mkdir $dir;
39    
40 dpavlin 1280 sub save_mech {
41 dpavlin 1292 my $path = shift;
42 dpavlin 1281 $step++;
43 dpavlin 1291 my $base_path = sprintf('%s/%04d', $dir,$step);
44 dpavlin 1292 $path ||= $base_path;
45     $path .= $mech->{ct} =~ m{html}i ? '.html' : '.txt';
46 dpavlin 1281 $mech->save_content( $path );
47 dpavlin 1290 warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n";
48     open(my $dump, '>', "$base_path.dump.txt");
49     $mech->dump_all($dump);
50 dpavlin 1280 }
51    
52     warn "# get session";
53     $mech->get( 'http://isiknowledge.com/?DestApp=WOS' );
54 dpavlin 1292 save_mech;
55 dpavlin 1280
56 dpavlin 1290 sub search {
57     warn "# advanced serach";
58     $mech->follow_link( url_regex => qr/AdvancedSearch/ );
59 dpavlin 1292 save_mech;
60 dpavlin 1280
61 dpavlin 1290 warn "# cookie_jar ", dump $mech->cookie_jar;
62 dpavlin 1280
63 dpavlin 1292 my $q_this = $q;
64    
65     if ( @ranges ) {
66     $q_this .= ' AND (' . join(' OR ', map { "PY=$_" } @{ shift @ranges } ) . ')';
67     }
68    
69     warn "# submit_form search: $q_this\n";
70 dpavlin 1290 $mech->submit_form(
71     fields => {
72 dpavlin 1292 'value(input1)' => $q_this,
73     },
74 dpavlin 1290 );
75 dpavlin 1292 save_mech;
76 dpavlin 1280
77 dpavlin 1290 warn "# summary";
78     $mech->follow_link( url_regex => qr/summary/ );
79 dpavlin 1292 save_mech;
80 dpavlin 1290 }
81    
82 dpavlin 1289 sub get_results {
83 dpavlin 1292 my $desc = shift;
84 dpavlin 1289 my $from = 1;
85 dpavlin 1280
86 dpavlin 1289 while ( 1 ) {
87 dpavlin 1280
88 dpavlin 1289 my $to = $from + $range_size;
89 dpavlin 1280
90 dpavlin 1292 warn "# submit_form results $from - $to\n";
91    
92 dpavlin 1289 $mech->submit_form(
93     form_name => 'summary_output_form',
94     fields => {
95     record_select_type => 'range',
96     mark_from => $from,
97     mark_to => $to,
98     mark_id => 'WOS',
99 dpavlin 1280
100 dpavlin 1289 qo_fields => 'fullrecord',
101     citedref => 'citedref',
102 dpavlin 1280
103 dpavlin 1289 save_options => 'plain_text',
104 dpavlin 1280
105 dpavlin 1289 fields => 'Full',
106     format => 'save',
107     },
108     button => 'save',
109     );
110 dpavlin 1292 save_mech;
111 dpavlin 1281
112 dpavlin 1292
113 dpavlin 1289 if ( $mech->content =~ m{invalid API call} ) {
114     $mech->back;
115 dpavlin 1290 last;
116 dpavlin 1289 }
117 dpavlin 1281
118 dpavlin 1292
119     my $path = "/tmp/isi.$q.$from-$to";
120     $path .= '.' . $desc if $desc;
121    
122     warn "save $from - $to into $path\n";
123 dpavlin 1289 $mech->follow_link( url_regex => qr/save_file/ );
124 dpavlin 1292 save_mech $path;
125 dpavlin 1281
126 dpavlin 1292 $from += $range_size - $overlap;
127 dpavlin 1281
128 dpavlin 1289 $mech->back;
129     $mech->back;
130 dpavlin 1292 #save_mech;
131 dpavlin 1290 }
132 dpavlin 1281 }
133 dpavlin 1289
134    
135 dpavlin 1290 sub citations {
136     warn "# citation report";
137     $mech->follow_link( url_regex => qr/search_mode=CitationReport/ );
138 dpavlin 1292 save_mech;
139 dpavlin 1289
140 dpavlin 1290 warn "view citing articles";
141     $mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ );
142 dpavlin 1292 save_mech;
143 dpavlin 1290 }
144 dpavlin 1289
145 dpavlin 1290 sub years {
146 dpavlin 1292 my $years_url = $mech->find_link( url_regex => qr/ra_name=/ );
147 dpavlin 1291 if ( ! $years_url ) {
148 dpavlin 1292 warn "W: can't find ra_name link\n";
149 dpavlin 1291 return;
150     }
151     $years_url = $years_url->url_abs;
152 dpavlin 1290 warn "## $years_url";
153 dpavlin 1291 if ( $years_url !~ s{ra_name=\w+}{ra_name=PublicationYear} ) {
154 dpavlin 1292 warn "W: no ra_name in $years_url\n";
155 dpavlin 1291 return;
156     }
157 dpavlin 1290 warn "# refine years (hidden by javascript)";
158 dpavlin 1292 # warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n";
159 dpavlin 1290 $mech->get( $years_url );
160 dpavlin 1292 save_mech;
161 dpavlin 1290
162     my $html = $mech->content;
163 dpavlin 1291 my $years;
164 dpavlin 1292 while ( $html =~ s{<label.+?PublicationYear.+?>(\d{4})\s\(([\d,]+)\)</label>}{} ) {
165     my ( $year, $count ) = ( $1, $2 );
166     $count =~ s{,}{}g;
167     $years->{$year} = $count;
168 dpavlin 1290 }
169 dpavlin 1291 warn "# years ",dump $years;
170 dpavlin 1290 $mech->back;
171 dpavlin 1292
172     my @y = sort keys %$years;
173    
174 dpavlin 1300 @ranges = ();
175    
176     if ( $cites_by_year ) {
177     push @ranges, [ $_ ] foreach @y;
178     warn "# cites_by_year ranges ", dump @ranges;
179     return;
180     }
181    
182 dpavlin 1292 my $y = shift @y;
183     my $size = $years->{$y};
184    
185     my $cites_range;
186     $cites_range = [$y] if $y;
187    
188     foreach my $y ( @y ) {
189     if ( $size + $years->{$y} > $max_cites ) {
190     push @ranges, $cites_range;
191     warn "# cites_range $size years ",dump( $cites_range ),$/;
192    
193     $cites_range = [];
194     $size = 0;
195     }
196     $size += $years->{$y};
197     push @$cites_range, $y;
198     }
199    
200     if ( $cites_range ) {
201     push @ranges, $cites_range;
202     warn "# cites_range $size years ",dump( $cites_range ), " FINAL\n"
203     }
204    
205     warn '# ranges ', dump @ranges;
206     @ranges = () if $#ranges == 1; # just take all
207    
208 dpavlin 1291 return $years;
209 dpavlin 1290 }
210    
211 dpavlin 1291 search;
212 dpavlin 1290 years;
213 dpavlin 1299 get_results unless $skip_results;
214 dpavlin 1290
215 dpavlin 1291
216 dpavlin 1290 citations;
217 dpavlin 1289
218 dpavlin 1292 do {
219     my $part;
220     if ( @ranges ) {
221     $part .= $ranges[0]->[0] . '.';
222     search;
223 dpavlin 1294 citations;
224 dpavlin 1292 }
225     $part .= 'citing';
226     get_results $part;
227     } while ( @ranges );
228    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26