/[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 420 - (show annotations)
Thu Sep 9 18:25:34 2004 UTC (15 years, 1 month ago) by dpavlin
File MIME type: text/plain
File size: 4925 byte(s)
warn, don't die

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 Data::Dumper;
11 use Carp;
12
13 use lib './lib';
14 use WebPAC;
15 use WebPAC::jsFind;
16 use WebPAC::Index;
17
18 my $webpac = new WebPAC(
19 code_page => 'ISO-8859-2',
20 # limit_mfn => 500,
21 # debug => 1,
22 ) || die;
23
24 my $log = $webpac->_get_logger();
25
26 my $index = new WebPAC::jsFind(
27 index_path => './out/index',
28 keys => 10,
29 ) || die;
30
31 my $thes;
32
33 $|=1;
34
35 my $maxmfn = $webpac->open_isis(
36 filename => shift @ARGV || '/data/hidra/THS/THS',
37 lookup => [
38 { 'key' => 'd:v900', 'val' => 'v250^a' },
39 # { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
40 # { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
41 # { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
42 { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
43 { 'key' => '900_mfn:v900', 'val' => 'v000' },
44 # tree structure
45 { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
46 { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
47 ],
48 );
49
50 $log->info("rows: $maxmfn");
51
52 $webpac->open_import_xml(type => 'isis_hidra_ths');
53
54 while (my $rec = $webpac->fetch_rec) {
55
56 my @ds = $webpac->data_structure($rec);
57
58 if (0 && $log->is_debug) {
59 $log->debug("rec = ",Dumper($rec));
60 $log->debug("ds = ",Dumper(\@ds));
61 }
62
63 next if (! @ds);
64
65 my $filename = $webpac->{'current_filename'};
66
67 if ($filename) {
68 $webpac->output_file(
69 file => $filename,
70 template => 'html.tt',
71 data => \@ds,
72 headline => $webpac->{'headline'},
73 );
74 } else {
75 print $webpac->output(
76 template => 'text.tt',
77 data => \@ds,
78 headline => $webpac->{'headline'},
79 );
80 }
81
82 my $headline = $webpac->{'headline'};
83
84 my $f = $filename;
85 $f =~ s!out/!!;
86
87 # save into index
88 foreach my $ds (@ds) {
89 next if (! $ds->{'swish'});
90
91 $index->insert(
92 index_name => $ds->{'tag'},
93 path => $f,
94 headline => $headline,
95 words => join(" ",@{$ds->{'swish'}})
96 );
97 }
98
99 # save into sorted index (thesaurus)
100 foreach my $ds (@ds) {
101 next if (! $ds->{'index'});
102
103 $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
104
105 foreach my $h (@{$ds->{'index'}}) {
106 $thes->{$ds->{'tag'}}->insert(
107 path => $f,
108 headline => $h,
109 );
110 }
111 }
112
113 # print Dumper(\@ds);
114
115 }
116
117 foreach my $t (keys %{$thes}) {
118
119 my @e = $thes->{$t}->elements;
120 if (! @e) {
121 $log->logwarn("no elements in sorted index $t?");
122 next;
123 }
124
125 my $file = "./out/bfilter/$t.txt";
126 $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
127
128 $webpac->output_file(
129 file => $file,
130 template => 'index.tt',
131 data => \@e,
132 index_name => $t,
133 );
134 }
135
136 if (0 && $log->is_debug) {
137 $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
138 $log->debug("data hash: ",Dumper($webpac->{'data'}));
139 foreach my $t (keys %{$thes}) {
140 $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
141 }
142 }
143
144 $log->info("lookup hash: ",Dumper($webpac->{'lookup'}));
145
146 #
147 # tree dump implementation which shouldn't be here :-)
148 #
149
150 my $file = 'out/browse.html';
151 open(HTML, "> $file") || $log->logdie("can't open '$file'");
152
153 print HTML qq{
154 <html>
155 <head>
156 <title>Browse</title>
157 <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/>
158 </head>
159 <body>
160 <ul>
161 };
162
163 foreach my $code (sort keys %{$webpac->{'lookup'}}) {
164
165 my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
166
167 if ($code =~ m/^root:/) {
168 my $v900 = shift @{ $l->{$code} } || $log->logconfess("can't lookup '$code'");
169 $code =~ s/^root:// || die;
170
171 my $term = shift @{ $l->{"d:${v900}"} } || die;
172 my $mfn = shift @{ $l->{"900_mfn:${v900}"} } || die;
173
174 $log->info("$code -> $v900 : $term [$mfn]");
175
176 print HTML qq{ <li><a href="thes/$mfn.html">$term</a></li>\n} if (-e "out/thes/$mfn.html");
177
178 unless ($l->{"a:${code}::"}) {
179 $log->warn("can't find 'a:${code}::'");
180 next;
181 }
182
183 print HTML qq{ <ul>\n};
184
185 foreach my $l2_v900 (@{ $l->{"a:${code}::"} }) {
186
187 my $l2_code = shift @{ $l->{"code:${l2_v900}"} } || die;
188 my $l2_term = shift @{ $l->{"d:$l2_v900"} } || die;
189 my $l2_mfn = shift @{ $l->{"900_mfn:${l2_v900}"} } || die;
190
191 $log->info("$l2_code -> $l2_v900 : $l2_term [$l2_mfn]");
192
193 print HTML qq{ <li><a href="thes/$l2_mfn.html">$l2_term</a></li>\n} if (-e "out/thes/$l2_mfn.html");
194
195 unless ($l->{"a:${l2_code}:"}) {
196 $log->warn("can't find 'a:${l2_code}:'");
197 next;
198 }
199
200 print HTML qq{ <ul>\n};
201
202 foreach my $l3_v900 (@{ $l->{"a:${l2_code}:"} }) {
203
204 my $l3_term = shift @{ $l->{"d:$l3_v900"} } || die;
205 my $l3_mfn = shift @{ $l->{"900_mfn:${l3_v900}"} } || die;
206
207 $log->info(" -> $l3_v900 : $l3_term [$l3_mfn]");
208
209 print HTML qq{ <li><a href="thes/$l3_mfn.html">$l3_term</a></li>\n} if (-e "out/thes/$l3_mfn.html");
210 }
211
212 print HTML qq{ </ul>\n};
213 }
214 print HTML qq{ </ul>\n};
215 }
216 }
217
218 print HTML qq{
219 </ul>
220 </body>
221 </html>
222 };
223
224 $index->close;
225

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.26