2 |
|
|
3 |
=head1 NAME |
=head1 NAME |
4 |
|
|
5 |
all2all.pl - basic script for all WebPac needs |
all2all.pl - basic script for all WebPAC needs |
6 |
|
|
7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
|
use OpenIsis; |
|
|
use Text::Iconv; |
|
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
use Carp; |
use Carp; |
12 |
|
|
13 |
use lib './lib'; |
use lib './lib'; |
14 |
use WebPac; |
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
use WebPAC::Index; |
17 |
|
|
18 |
my $webpac = new WebPac( |
my $webpac = new WebPAC( |
19 |
code_page => 'ISO-8859-2', |
code_page => 'ISO-8859-2', |
20 |
|
# limit_mfn => 100, |
21 |
|
# debug => 1, |
22 |
|
low_mem => 1, |
23 |
) || die; |
) || die; |
24 |
|
|
25 |
|
my $log = $webpac->_get_logger() || die "can't get logger"; |
26 |
|
|
27 |
|
$log->debug("creating WebPAC::jsFind object"); |
28 |
|
|
29 |
|
my $index = new WebPAC::jsFind( |
30 |
|
index_path => './out/index', |
31 |
|
keys => 10, |
32 |
|
) || die; |
33 |
|
|
34 |
|
my $thes; |
35 |
|
|
36 |
$|=1; |
$|=1; |
37 |
|
|
|
print "reading database\n"; |
|
38 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
39 |
filename => shift @ARGV || '/data/hidra/THS-500/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
40 |
lookup => [ |
lookup => [ |
41 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
42 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
43 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
44 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
45 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
46 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
47 |
|
# tree structure |
48 |
|
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
49 |
|
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
50 |
], |
], |
51 |
); |
); |
52 |
|
|
53 |
print "rows: $maxmfn\n\n"; |
$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); |
54 |
|
|
55 |
|
$log->info("rows: $maxmfn"); |
56 |
|
|
57 |
|
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
58 |
|
|
59 |
|
$log->debug("data: ",Dumper($webpac->{'data'})); |
60 |
|
|
61 |
|
while (my $rec = $webpac->fetch_rec) { |
62 |
|
|
63 |
|
my @ds = $webpac->data_structure($rec); |
64 |
|
|
65 |
|
if (0 && $log->is_debug) { |
66 |
|
$log->debug("rec = ",Dumper($rec)); |
67 |
|
$log->debug("ds = ",Dumper(\@ds)); |
68 |
|
} |
69 |
|
|
70 |
|
next if (! @ds); |
71 |
|
|
72 |
|
my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); |
73 |
|
|
74 |
|
if ($filename) { |
75 |
|
$webpac->output_file( |
76 |
|
file => $filename, |
77 |
|
template => 'html.tt', |
78 |
|
data => \@ds, |
79 |
|
headline => $webpac->{'headline'}, |
80 |
|
); |
81 |
|
} else { |
82 |
|
print $webpac->output( |
83 |
|
template => 'text.tt', |
84 |
|
data => \@ds, |
85 |
|
headline => $webpac->{'headline'}, |
86 |
|
); |
87 |
|
} |
88 |
|
|
89 |
|
my $headline = $webpac->{'headline'}; |
90 |
|
|
91 |
|
my $f = $filename; |
92 |
|
$f =~ s!out/!!; |
93 |
|
|
94 |
|
# save into index |
95 |
|
foreach my $ds (@ds) { |
96 |
|
next if (! $ds->{'swish'}); |
97 |
|
|
98 |
|
$index->insert( |
99 |
|
index_name => $ds->{'tag'}, |
100 |
|
path => $f, |
101 |
|
headline => $headline, |
102 |
|
words => join(" ",@{$ds->{'swish'}}) |
103 |
|
); |
104 |
|
} |
105 |
|
|
106 |
|
# save into sorted index (thesaurus) |
107 |
|
foreach my $ds (@ds) { |
108 |
|
next if (! $ds->{'index'}); |
109 |
|
|
110 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
111 |
|
|
112 |
|
foreach my $h (@{$ds->{'index'}}) { |
113 |
|
$thes->{$ds->{'tag'}}->insert( |
114 |
|
path => $f, |
115 |
|
headline => $h, |
116 |
|
); |
117 |
|
} |
118 |
|
} |
119 |
|
|
120 |
|
# print Dumper(\@ds); |
121 |
|
|
122 |
|
} |
123 |
|
|
124 |
|
foreach my $t (keys %{$thes}) { |
125 |
|
|
126 |
|
my @e = $thes->{$t}->elements; |
127 |
|
if (! @e) { |
128 |
|
$log->logwarn("no elements in sorted index $t?"); |
129 |
|
next; |
130 |
|
} |
131 |
|
|
132 |
|
my $file = "./out/bfilter/$t.txt"; |
133 |
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
134 |
|
|
135 |
|
$webpac->output_file( |
136 |
|
file => $file, |
137 |
|
template => 'index.tt', |
138 |
|
data => \@e, |
139 |
|
index_name => $t, |
140 |
|
); |
141 |
|
} |
142 |
|
|
143 |
|
if (0 && $log->is_debug) { |
144 |
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
145 |
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
146 |
|
foreach my $t (keys %{$thes}) { |
147 |
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
148 |
|
} |
149 |
|
} |
150 |
|
|
151 |
|
$log->info("lookup hash: ",Dumper($webpac->{'lookup'})); |
152 |
|
|
153 |
|
# |
154 |
|
# tree dump implementation which shouldn't be here :-) |
155 |
|
# |
156 |
|
|
157 |
|
my $file = 'out/browse.html'; |
158 |
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
159 |
|
|
160 |
|
print HTML qq{ |
161 |
|
<html> |
162 |
|
<head> |
163 |
|
<title>Browse</title> |
164 |
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
165 |
|
</head> |
166 |
|
<body> |
167 |
|
<ul> |
168 |
|
}; |
169 |
|
|
170 |
|
foreach my $code (sort keys %{$webpac->{'lookup'}}) { |
171 |
|
|
172 |
|
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
173 |
|
|
174 |
|
if ($code =~ m/^root:/) { |
175 |
|
my $v900 = shift @{ $l->{$code} } || $log->logconfess("can't lookup '$code'"); |
176 |
|
$code =~ s/^root:// || die; |
177 |
|
|
178 |
|
my $term = shift @{ $l->{"d:${v900}"} } || die; |
179 |
|
my $mfn = shift @{ $l->{"900_mfn:${v900}"} } || die; |
180 |
|
|
181 |
|
$log->debug("$code -> $v900 : $term [$mfn]"); |
182 |
|
|
183 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
print HTML qq{ <li><a href="thes/$mfn.html">$term</a></li>\n} if (-e "out/thes/$mfn.html"); |
|
my $rec = $webpac->{'data'}->{$mfn} || die "no record with mfn $mfn"; |
|
184 |
|
|
185 |
print "-- ",$webpac->fill_in($rec,'v250^a (v901^a)'),"\n"; |
unless ($l->{"a:${code}::"}) { |
186 |
|
$log->warn("can't find 'a:${code}::'"); |
187 |
|
next; |
188 |
|
} |
189 |
|
|
190 |
my @t = $webpac->fill_in($rec,'v553^1;;v553^a'); |
print HTML qq{ <ul>\n}; |
|
print " Uži pojam: ",join("\t\n",@t),"\n" if (@t); |
|
191 |
|
|
192 |
@t = $webpac->fill_in($rec,'[a:v251::];;[d:[a:v251::]]'); |
foreach my $l2_v900 (@{ $l->{"a:${code}::"} }) { |
193 |
@t = map { $webpac->lookup($_) => $_ } @t; |
|
194 |
print " Područje: ",join("\t\n",@t),"\n" if (@t); |
my $l2_code = shift @{ $l->{"code:${l2_v900}"} } || die; |
195 |
|
my $l2_term = shift @{ $l->{"d:$l2_v900"} } || die; |
196 |
|
my $l2_mfn = shift @{ $l->{"900_mfn:${l2_v900}"} } || die; |
197 |
|
|
198 |
@t = $webpac->fill_in($rec,'[a:v561^4:v251:];;[d:[a:v561^4:v251:]]'); |
$log->debug("$l2_code -> $l2_v900 : $l2_term [$l2_mfn]"); |
|
print " Mikrotezaurus: ",join("\t\n",@t),"\n" if (@t); |
|
199 |
|
|
200 |
@t = $webpac->fill_in($rec,'[a:v561^4:v562^4:v900];;[d:[a:v561^4:v562^4:v900]]'); |
print HTML qq{ <li><a href="thes/$l2_mfn.html">$l2_term</a></li>\n} if (-e "out/thes/$l2_mfn.html"); |
201 |
print " Deskriptor: ",join("\t\n",@t),"\n" if (@t); |
|
202 |
|
unless ($l->{"a:${l2_code}:"}) { |
203 |
|
$log->warn("can't find 'a:${l2_code}:'"); |
204 |
|
next; |
205 |
|
} |
206 |
|
|
207 |
|
print HTML qq{ <ul>\n}; |
208 |
|
|
209 |
|
foreach my $l3_v900 (@{ $l->{"a:${l2_code}:"} }) { |
210 |
|
|
211 |
|
my $l3_term = shift @{ $l->{"d:$l3_v900"} } || die; |
212 |
|
my $l3_mfn = shift @{ $l->{"900_mfn:${l3_v900}"} } || die; |
213 |
|
|
214 |
|
$log->debug(" -> $l3_v900 : $l3_term [$l3_mfn]"); |
215 |
|
|
216 |
|
print HTML qq{ <li><a href="thes/$l3_mfn.html">$l3_term</a></li>\n} if (-e "out/thes/$l3_mfn.html"); |
217 |
|
} |
218 |
|
|
219 |
|
print HTML qq{ </ul>\n}; |
220 |
|
} |
221 |
|
print HTML qq{ </ul>\n}; |
222 |
|
} |
223 |
} |
} |
224 |
|
|
225 |
print "## lookup ",Dumper($webpac->{'lookup'}); |
print HTML qq{ |
226 |
print "## data ",Dumper($webpac->{'data'}); |
</ul> |
227 |
|
</body> |
228 |
|
</html> |
229 |
|
}; |
230 |
|
|
231 |
|
$index->close; |
232 |
|
|
233 |
|
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |