1 |
dpavlin |
1 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
dpavlin |
348 |
=head1 NAME |
4 |
|
|
|
5 |
dpavlin |
354 |
all2all.pl - basic script for all WebPAC needs |
6 |
dpavlin |
348 |
|
7 |
|
|
=cut |
8 |
|
|
|
9 |
dpavlin |
1 |
use strict; |
10 |
|
|
use Data::Dumper; |
11 |
dpavlin |
352 |
use Carp; |
12 |
dpavlin |
1 |
|
13 |
dpavlin |
352 |
use lib './lib'; |
14 |
dpavlin |
354 |
use WebPAC; |
15 |
dpavlin |
390 |
use WebPAC::jsFind; |
16 |
dpavlin |
410 |
use WebPAC::Index; |
17 |
dpavlin |
348 |
|
18 |
dpavlin |
354 |
my $webpac = new WebPAC( |
19 |
dpavlin |
352 |
code_page => 'ISO-8859-2', |
20 |
dpavlin |
419 |
# limit_mfn => 500, |
21 |
dpavlin |
372 |
# debug => 1, |
22 |
dpavlin |
352 |
) || die; |
23 |
|
|
|
24 |
dpavlin |
373 |
my $log = $webpac->_get_logger(); |
25 |
|
|
|
26 |
dpavlin |
390 |
my $index = new WebPAC::jsFind( |
27 |
|
|
index_path => './out/index', |
28 |
dpavlin |
413 |
keys => 10, |
29 |
dpavlin |
390 |
) || die; |
30 |
|
|
|
31 |
dpavlin |
410 |
my $thes; |
32 |
|
|
|
33 |
dpavlin |
10 |
$|=1; |
34 |
dpavlin |
9 |
|
35 |
dpavlin |
352 |
my $maxmfn = $webpac->open_isis( |
36 |
dpavlin |
357 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
37 |
dpavlin |
352 |
lookup => [ |
38 |
dpavlin |
348 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
39 |
dpavlin |
352 |
# { '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 |
dpavlin |
374 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
44 |
dpavlin |
419 |
# tree structure |
45 |
|
|
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
46 |
|
|
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
47 |
dpavlin |
352 |
], |
48 |
dpavlin |
54 |
); |
49 |
dpavlin |
3 |
|
50 |
dpavlin |
372 |
$log->info("rows: $maxmfn"); |
51 |
dpavlin |
170 |
|
52 |
dpavlin |
363 |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
53 |
|
|
|
54 |
dpavlin |
362 |
while (my $rec = $webpac->fetch_rec) { |
55 |
dpavlin |
358 |
|
56 |
dpavlin |
370 |
my @ds = $webpac->data_structure($rec); |
57 |
dpavlin |
366 |
|
58 |
dpavlin |
411 |
if (0 && $log->is_debug) { |
59 |
dpavlin |
372 |
$log->debug("rec = ",Dumper($rec)); |
60 |
|
|
$log->debug("ds = ",Dumper(\@ds)); |
61 |
|
|
} |
62 |
|
|
|
63 |
dpavlin |
374 |
next if (! @ds); |
64 |
dpavlin |
366 |
|
65 |
dpavlin |
374 |
my $filename = $webpac->{'current_filename'}; |
66 |
|
|
|
67 |
|
|
if ($filename) { |
68 |
dpavlin |
411 |
$webpac->output_file( |
69 |
|
|
file => $filename, |
70 |
dpavlin |
374 |
template => 'html.tt', |
71 |
|
|
data => \@ds, |
72 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
73 |
dpavlin |
374 |
); |
74 |
|
|
} else { |
75 |
|
|
print $webpac->output( |
76 |
|
|
template => 'text.tt', |
77 |
|
|
data => \@ds, |
78 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
79 |
dpavlin |
374 |
); |
80 |
|
|
} |
81 |
|
|
|
82 |
dpavlin |
398 |
my $headline = $webpac->{'headline'}; |
83 |
dpavlin |
390 |
|
84 |
dpavlin |
392 |
my $f = $filename; |
85 |
|
|
$f =~ s!out/!!; |
86 |
|
|
|
87 |
dpavlin |
390 |
# save into index |
88 |
|
|
foreach my $ds (@ds) { |
89 |
|
|
next if (! $ds->{'swish'}); |
90 |
|
|
|
91 |
|
|
$index->insert( |
92 |
|
|
index_name => $ds->{'tag'}, |
93 |
dpavlin |
392 |
path => $f, |
94 |
dpavlin |
390 |
headline => $headline, |
95 |
|
|
words => join(" ",@{$ds->{'swish'}}) |
96 |
|
|
); |
97 |
|
|
} |
98 |
|
|
|
99 |
dpavlin |
410 |
# 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 |
dpavlin |
415 |
foreach my $h (@{$ds->{'index'}}) { |
106 |
|
|
$thes->{$ds->{'tag'}}->insert( |
107 |
|
|
path => $f, |
108 |
|
|
headline => $h, |
109 |
|
|
); |
110 |
|
|
} |
111 |
dpavlin |
410 |
} |
112 |
|
|
|
113 |
dpavlin |
390 |
# print Dumper(\@ds); |
114 |
|
|
|
115 |
dpavlin |
3 |
} |
116 |
|
|
|
117 |
dpavlin |
411 |
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 |
dpavlin |
415 |
my $file = "./out/bfilter/$t.txt"; |
126 |
|
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
127 |
dpavlin |
411 |
|
128 |
|
|
$webpac->output_file( |
129 |
dpavlin |
415 |
file => $file, |
130 |
dpavlin |
411 |
template => 'index.tt', |
131 |
|
|
data => \@e, |
132 |
|
|
index_name => $t, |
133 |
|
|
); |
134 |
|
|
} |
135 |
|
|
|
136 |
|
|
if (0 && $log->is_debug) { |
137 |
dpavlin |
372 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
138 |
|
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
139 |
dpavlin |
410 |
foreach my $t (keys %{$thes}) { |
140 |
|
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
141 |
|
|
} |
142 |
dpavlin |
372 |
} |
143 |
dpavlin |
390 |
|
144 |
dpavlin |
419 |
# |
145 |
|
|
# tree dump implementation which shouldn't be here :-) |
146 |
|
|
# |
147 |
|
|
|
148 |
|
|
my $file = 'out/browse.html'; |
149 |
|
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
150 |
|
|
|
151 |
|
|
print HTML qq{ |
152 |
|
|
<html> |
153 |
|
|
<head> |
154 |
|
|
<title>Browse</title> |
155 |
|
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
156 |
|
|
</head> |
157 |
|
|
<body> |
158 |
|
|
<ul> |
159 |
|
|
}; |
160 |
|
|
|
161 |
|
|
foreach my $code (sort keys %{$webpac->{'lookup'}}) { |
162 |
|
|
|
163 |
|
|
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
164 |
|
|
|
165 |
|
|
if ($code =~ m/^root:/) { |
166 |
|
|
my $v900 = shift @{ $l->{$code} } || $log->logconfess("can't lookup '$code'"); |
167 |
|
|
$code =~ s/^root:// || die; |
168 |
|
|
|
169 |
|
|
my $term = shift @{ $l->{"d:${v900}"} } || die; |
170 |
|
|
my $mfn = shift @{ $l->{"900_mfn:${v900}"} } || die; |
171 |
|
|
|
172 |
|
|
$log->info("$code -> $v900 : $term [$mfn]"); |
173 |
|
|
|
174 |
|
|
print HTML qq{ <li><a href="thes/$mfn.html">$term</a></li>\n} if (-e "out/thes/$mfn.html"); |
175 |
|
|
|
176 |
|
|
$log->logconfess("can't find 'a:${code}::'") unless ($l->{"a:${code}::"}); |
177 |
|
|
|
178 |
|
|
print HTML qq{ <ul>\n}; |
179 |
|
|
|
180 |
|
|
foreach my $l2_v900 (@{ $l->{"a:${code}::"} }) { |
181 |
|
|
|
182 |
|
|
my $l2_code = shift @{ $l->{"code:${l2_v900}"} } || die; |
183 |
|
|
my $l2_term = shift @{ $l->{"d:$l2_v900"} } || die; |
184 |
|
|
my $l2_mfn = shift @{ $l->{"900_mfn:${l2_v900}"} } || die; |
185 |
|
|
|
186 |
|
|
$log->info("$l2_code -> $l2_v900 : $l2_term [$l2_mfn]"); |
187 |
|
|
|
188 |
|
|
print HTML qq{ <li><a href="thes/$l2_mfn.html">$l2_term</a></li>\n} if (-e "out/thes/$l2_mfn.html"); |
189 |
|
|
|
190 |
|
|
print HTML qq{ <ul>\n}; |
191 |
|
|
|
192 |
|
|
foreach my $l3_v900 (@{ $l->{"a:${l2_code}:"} }) { |
193 |
|
|
|
194 |
|
|
my $l3_term = shift @{ $l->{"d:$l3_v900"} } || die; |
195 |
|
|
my $l3_mfn = shift @{ $l->{"900_mfn:${l3_v900}"} } || die; |
196 |
|
|
|
197 |
|
|
$log->info(" -> $l3_v900 : $l3_term [$l3_mfn]"); |
198 |
|
|
|
199 |
|
|
print HTML qq{ <li><a href="thes/$l3_mfn.html">$l3_term</a></li>\n} if (-e "out/thes/$l3_mfn.html"); |
200 |
|
|
} |
201 |
|
|
|
202 |
|
|
print HTML qq{ </ul>\n}; |
203 |
|
|
} |
204 |
|
|
print HTML qq{ </ul>\n}; |
205 |
|
|
} |
206 |
|
|
} |
207 |
|
|
|
208 |
|
|
print HTML qq{ |
209 |
|
|
</ul> |
210 |
|
|
</body> |
211 |
|
|
</html> |
212 |
|
|
}; |
213 |
|
|
|
214 |
dpavlin |
390 |
$index->close; |
215 |
|
|
|