/[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 466 - (show annotations)
Thu Sep 23 23:17:42 2004 UTC (14 years, 11 months ago) by dpavlin
File MIME type: text/plain
File size: 7761 byte(s)
remove non-word characters from data before inserting into jsFind index

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 YAML;
11 use Carp;
12 use Getopt::Long;
13
14 use lib './lib';
15 use WebPAC;
16 use WebPAC::jsFind;
17 use WebPAC::Index;
18 use WebPAC::Tree;
19
20 # options which can be changed via command line
21 #
22 my $code_page = 'ISO-8859-2';
23 my ($limit_mfn, $start_mfn, $debug, $low_mem);
24 my $index_path = './out/index';
25
26 my $result = GetOptions(
27 "code_page=s" => \$code_page,
28 "limit_mfn=i" => \$limit_mfn,
29 "start_mfn=i" => \$start_mfn,
30 "debug!" => \$debug,
31 "low_mem!" => \$low_mem,
32 );
33
34 # create WebPAC object
35 #
36 my $webpac = new WebPAC(
37 code_page => $code_page,
38 limit_mfn => $limit_mfn,
39 start_mfn => $start_mfn,
40 debug => $debug,
41 low_mem => $low_mem,
42 ) || die;
43
44 my $log = $webpac->_get_logger() || die "can't get logger";
45
46 $log->debug("creating WebPAC::jsFind object");
47
48 my $index = new WebPAC::jsFind(
49 index_path => $index_path,
50 keys => 10,
51 ) || die;
52
53 my $thes;
54
55 $|=1;
56
57 my $maxmfn = $webpac->open_isis(
58 filename => shift @ARGV || '/data/hidra/THS/THS',
59 lookup => [
60 { 'key' => 'd:v900', 'val' => 'v250^a' },
61 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
62 # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
63 # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
64 { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
65 { 'key' => '900_mfn:v900', 'val' => 'v000' },
66 # tree structure
67 { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
68 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
69 { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
70 ],
71 );
72
73 $log->debug("isis file ",$webpac->{'isis_filename'}," opened");
74
75 $log->info("rows: $maxmfn");
76
77 $webpac->open_import_xml(type => 'isis_hidra_ths');
78
79 if(1) { # XXX
80
81 while (my $rec = $webpac->fetch_rec) {
82
83 my @ds = $webpac->data_structure($rec);
84
85 if (0 && $log->is_debug) {
86 $log->debug("rec = ",Dump($rec));
87 $log->debug("ds = ",Dump(\@ds));
88 }
89
90 next if (! @ds);
91
92 my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
93
94 if ($filename) {
95 $webpac->output_file(
96 file => $filename,
97 template => 'html.tt',
98 data => \@ds,
99 headline => $webpac->{'headline'},
100 );
101 } else {
102 print $webpac->output(
103 template => 'text.tt',
104 data => \@ds,
105 headline => $webpac->{'headline'},
106 );
107 }
108
109 my $headline = $webpac->{'headline'};
110
111 my $f = $filename;
112 $f =~ s!out/!!;
113
114 # save into index
115 foreach my $ds (@ds) {
116 next if (! $ds->{'swish'});
117
118 # strip all non word characters from beginning or end
119 # of word
120 my $words = join(" ",@{$ds->{'swish'}});
121 $words =~ s/^\W+//;
122 $words =~ s/\W*\s+\W*/ /g;
123 $words =~ s/\W+$//;
124
125 $index->insert(
126 index_name => $ds->{'tag'},
127 #path => $f,
128 path => $webpac->mfn,
129 headline => $headline,
130 words => $words,
131 );
132 }
133
134 # save into sorted index (thesaurus)
135 foreach my $ds (@ds) {
136 next if (! $ds->{'index'});
137
138 $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
139
140 foreach my $h (@{$ds->{'index'}}) {
141 $thes->{$ds->{'tag'}}->insert(
142 headline => $h,
143 mfn => $webpac->mfn,
144 );
145 }
146 }
147
148 # print Dump(\@ds);
149
150 }
151
152 foreach my $t (keys %{$thes}) {
153
154 my @e = $thes->{$t}->elements;
155 if (! @e) {
156 $log->logwarn("no elements in sorted index $t?");
157 next;
158 }
159
160 my $file = "./out/bfilter/$t.txt";
161 $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
162
163 $webpac->output_file(
164 file => $file,
165 template => 'index.tt',
166 data => \@e,
167 index_name => $t,
168 );
169 }
170
171 if (0 && $log->is_debug) {
172 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
173 $log->debug("data hash: ",Dump($webpac->{'data'}));
174 foreach my $t (keys %{$thes}) {
175 $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
176 }
177 }
178
179 } # XXX if(0)
180
181 $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
182
183 $log->info("creating tree");
184
185 #
186 # define tree structure
187 #
188
189 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
190
191 my @tree = ({
192 # level 0
193 code_arr => sub { sort keys %{$l} },
194 filter_code => sub {
195 my $t = shift;
196 return $t if ($t =~ s/root://);
197 },
198 lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
199 lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
200 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
201 have_children => sub { return $l->{"a:".$_[0]."::"} },
202 have_children_at_level => sub {
203 my $l2 = shift;
204 my $l1 = substr($l2,0,1);
205 return(9, "a:$l1:$l2:" ) if ($l->{"a:$l1:$l2"});
206 },
207 style => 'display: none',
208 },{
209 # 1
210 code_arr => sub { @{$_[0]} },
211 filter_code => sub { shift }, # nop
212 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
213 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
214 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
215 have_children => sub { return $l->{"a:".$_[1].":"} },
216 style => 'display: none',
217 },{
218 # 2
219 code_arr => sub { @{$_[0]} },
220 filter_code => sub { shift },
221 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
222 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
223 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
224 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
225 #style => 'display: none',
226 },{
227 # 3 u¾i pojam
228 code_arr => sub { @{$_[0]} },
229 filter_code => sub { shift },
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].":".$_[0]} },
234 },{
235 # 4
236 code_arr => sub { @{$_[0]} },
237 filter_code => sub { shift },
238 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
239 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
240 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
241 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
242 },{
243 # 5
244 code_arr => sub { @{$_[0]} },
245 filter_code => sub { shift },
246 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
247 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
248 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
249 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
250 },{
251 # 6
252 code_arr => sub { @{$_[0]} },
253 filter_code => sub { shift },
254 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
255 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
256 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
257 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
258 },{
259 # 7
260 code_arr => sub { @{$_[0]} },
261 filter_code => sub { shift },
262 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
263 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
264 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
265 have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
266 },{
267 # 8
268 code_arr => sub { @{$_[0]} },
269 filter_code => sub { shift },
270 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
271 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
272 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
273 # have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
274 have_children => sub { 0 },
275 },{
276 # 9 - level which is never reached except explicitly
277 code_arr => sub { @{$l->{$_[0]}} },
278 filter_code => sub { shift },
279 lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
280 lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
281 lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
282 have_children => sub { 0 },
283 have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,"a:".$_[1].":".$_[0]) },
284 },{
285 });
286
287 my $tree = new WebPAC::Tree(
288 dir => './out',
289 html => 'browse.html',
290 template => './output_template/tree.tt',
291 js => 'tree-ids.js',
292 tree => \@tree,
293 );
294
295 $log->info("closing index");
296 $index->close;
297
298 $log->info("elapsed time: ",$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