/[webpac]/trunk2/all2all.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

Contents of /trunk2/all2all.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 501 - (show annotations)
Sun Oct 10 11:25:10 2004 UTC (19 years, 5 months ago) by dpavlin
File MIME type: text/plain
File size: 8292 byte(s)
implement CROVOC marker for tree

1 #!/usr/bin/perl -w
2
3 =head1 NAME
4
5 all2all.pl - basic script for all WebPAC needs
6
7 =cut
8
9 use strict;
10 use locale;
11 use YAML;
12 use Carp;
13 use Getopt::Long;
14
15 use lib './lib';
16 use WebPAC;
17 use WebPAC::jsFind;
18 use WebPAC::Index;
19 use WebPAC::Tree;
20
21 # options which can be changed via command line
22 #
23 my $code_page = 'ISO-8859-2';
24 my ($limit_mfn, $start_mfn, $debug, $low_mem);
25 my $index_path = './out/index';
26
27 my $result = GetOptions(
28 "code_page=s" => \$code_page,
29 "limit_mfn=i" => \$limit_mfn,
30 "start_mfn=i" => \$start_mfn,
31 "debug!" => \$debug,
32 "low_mem!" => \$low_mem,
33 );
34
35 my $filter = {
36 'CROVOC' => sub {
37 my $tmp = shift || return;
38 return undef unless ($tmp =~ s/\s+CROVOC.*$/ */);
39 return $tmp;
40 },
41 'CROVOC_tree' => sub {
42 my $tmp = shift || return;
43 $tmp =~ s/\s+CROVOC.*$/ */;
44 $tmp =~ s/\s+EUROVOC.*//;
45 return $tmp;
46 },
47 };
48
49 # create WebPAC object
50 #
51 my $webpac = new WebPAC(
52 code_page => $code_page,
53 limit_mfn => $limit_mfn,
54 start_mfn => $start_mfn,
55 debug => $debug,
56 low_mem => $low_mem,
57 filter => $filter,
58 ) || die;
59
60 my $log = $webpac->_get_logger() || die "can't get logger";
61
62 $log->debug("creating WebPAC::jsFind object");
63
64 my $index = new WebPAC::jsFind(
65 index_path => $index_path,
66 keys => 10,
67 ) || die;
68
69 my $thes;
70
71 $|=1;
72
73 my $maxmfn = $webpac->open_isis(
74 filename => shift @ARGV || '/data/hidra/THS/THS',
75 lookup => [
76 { 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' },
77 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
78 # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
79 # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
80 { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
81 { 'key' => '900_mfn:v900', 'val' => 'v000' },
82 # tree structure
83 { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
84 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
85 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
86 ],
87 );
88
89 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
90
91 $log->info("rows: $maxmfn");
92
93 $webpac->open_import_xml(type => 'isis_hidra_ths');
94
95 if(1) { # XXX
96
97 while (my $rec = $webpac->fetch_rec) {
98
99 my @ds = $webpac->data_structure($rec);
100
101 if (0 && $log->is_debug) {
102 $log->debug("rec = ",Dump($rec));
103 $log->debug("ds = ",Dump(\@ds));
104 }
105
106 next if (! @ds);
107
108 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
109
110 if ($filename) {
111 $webpac->output_file(
112 file => $filename,
113 template => 'html.tt',
114 data => \@ds,
115 headline => $webpac->{'headline'},
116 );
117 } else {
118 print $webpac->output(
119 template => 'text.tt',
120 data => \@ds,
121 headline => $webpac->{'headline'},
122 );
123 }
124
125 my $headline = $webpac->{'headline'};
126
127 my $f = $filename;
128 $f =~ s!out/!!;
129
130 # save into index
131 foreach my $ds (@ds) {
132 next if (! $ds->{'swish'});
133
134 # strip all non word characters from beginning or end
135 # of word
136 my $words = join(" ",@{$ds->{'swish'}});
137 $words =~ s/^\W+//;
138 $words =~ s/\W*\s+\W*/ /g;
139 $words =~ s/\W+$//;
140
141 $index->insert(
142 index_name => $ds->{'tag'},
143 #path => $f,
144 path => $webpac->mfn,
145 headline => $headline,
146 words => $words,
147 );
148 }
149
150 # save into sorted index (thesaurus)
151 foreach my $ds (@ds) {
152 next if (! $ds->{'index'});
153
154 $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
155
156 foreach my $h (@{$ds->{'index'}}) {
157 $thes->{$ds->{'tag'}}->insert(
158 headline => $h,
159 mfn => $webpac->mfn,
160 );
161 }
162 }
163
164 # print Dump(\@ds);
165
166 }
167
168 foreach my $t (keys %{$thes}) {
169
170 my @e = $thes->{$t}->elements;
171 if (! @e) {
172 $log->logwarn("no elements in sorted index $t?");
173 next;
174 }
175
176 my $file = "./out/bfilter/$t.txt";
177 $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
178
179 $webpac->output_file(
180 file => $file,
181 template => 'index.tt',
182 data => \@e,
183 index_name => $t,
184 );
185 }
186
187 if (0 && $log->is_debug) {
188 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
189 $log->debug("data hash: ",Dump($webpac->{'data'}));
190 foreach my $t (keys %{$thes}) {
191 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
192 }
193 }
194
195 } # XXX if(0)
196
197 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
198
199 $log->info("creating tree");
200
201 #
202 # define tree structure
203 #
204
205 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
206
207 my @tree = ({
208 # level 0
209 code_arr => sub { sort keys %{$l} },
210 filter_code => sub {
211 my $t = shift;
212 return $t if ($t =~ s/root://);
213 },
214 lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
215 lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
216 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
217 have_children => sub { return $l->{"a:".$_[0]."::"} },
218 have_children_at_level => sub {
219 return unless (defined($l->{"code:".$_[1]}));
220 my $code = shift @{$l->{"code:".$_[1]}};
221 print STDERR "## $_[1] -> $code\n";
222 return undef unless($code);
223 return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
224 },
225 style => 'display: none',
226 },{
227 # 1
228 code_arr => sub { @{$_[0]} },
229 filter_code => sub { shift }, # nop
230 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
231 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
232 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
233 have_children => sub { return $l->{"a:".$_[1].":"} },
234 style => 'display: none',
235 },{
236 # 2
237 code_arr => sub { @{$_[0]} },
238 filter_code => sub { shift },
239 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
240 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
241 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
242 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
243 #style => 'display: none',
244 },{
245 # 3 u¾i pojam
246 code_arr => sub { @{$_[0]} },
247 filter_code => sub { shift },
248 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
249 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
250 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
251 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
252 },{
253 # 4
254 code_arr => sub { @{$_[0]} },
255 filter_code => sub { shift },
256 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
257 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
258 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
259 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
260 },{
261 # 5
262 code_arr => sub { @{$_[0]} },
263 filter_code => sub { shift },
264 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
265 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
266 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
267 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
268 },{
269 # 6
270 code_arr => sub { @{$_[0]} },
271 filter_code => sub { shift },
272 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
273 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
274 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
275 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
276 },{
277 # 7
278 code_arr => sub { @{$_[0]} },
279 filter_code => sub { shift },
280 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
281 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
282 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
283 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
284 },{
285 # 8
286 code_arr => sub { @{$_[0]} },
287 filter_code => sub { shift },
288 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
289 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
290 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
291 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
292 have_children => sub { 0 },
293 },{
294 # 9 - level which is never reached except explicitly
295 code_arr => sub { @{$_[0]} },
296 filter_code => sub { shift },
297 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
298 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
299 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
300 have_children => sub { 0 },
301 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
302 },{
303 });
304
305 my $tree = new WebPAC::Tree(
306 dir => './out',
307 html => 'browse.html',
308 template => './output_template/tree.tt',
309 js => 'tree-ids.js',
310 tree => \@tree,
311 );
312
313 $log->info("closing index");
314 $index->close;
315
316 $log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));

Properties

Name Value
cvs2svn:cvs-rev 1.64
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26