7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
10 |
use Data::Dumper; |
use YAML; |
11 |
use Carp; |
use Carp; |
12 |
|
use Getopt::Long; |
13 |
|
|
14 |
use lib './lib'; |
use lib './lib'; |
15 |
use WebPAC; |
use WebPAC; |
16 |
use WebPAC::jsFind; |
use WebPAC::jsFind; |
17 |
use WebPAC::Index; |
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( |
my $webpac = new WebPAC( |
37 |
code_page => 'ISO-8859-2', |
code_page => $code_page, |
38 |
limit_mfn => 500, |
limit_mfn => $limit_mfn, |
39 |
# debug => 1, |
start_mfn => $start_mfn, |
40 |
# low_mem => 1, |
debug => $debug, |
41 |
|
low_mem => $low_mem, |
42 |
) || die; |
) || die; |
43 |
|
|
44 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
46 |
$log->debug("creating WebPAC::jsFind object"); |
$log->debug("creating WebPAC::jsFind object"); |
47 |
|
|
48 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
49 |
index_path => './out/index', |
index_path => $index_path, |
50 |
keys => 10, |
keys => 10, |
51 |
) || die; |
) || die; |
52 |
|
|
66 |
# tree structure |
# tree structure |
67 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
68 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ '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 |
|
|
83 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
84 |
|
|
85 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
86 |
$log->debug("rec = ",Dumper($rec)); |
$log->debug("rec = ",Dump($rec)); |
87 |
$log->debug("ds = ",Dumper(\@ds)); |
$log->debug("ds = ",Dump(\@ds)); |
88 |
} |
} |
89 |
|
|
90 |
next if (! @ds); |
next if (! @ds); |
115 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
116 |
next if (! $ds->{'swish'}); |
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( |
$index->insert( |
126 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
127 |
path => $f, |
#path => $f, |
128 |
|
path => $webpac->mfn, |
129 |
headline => $headline, |
headline => $headline, |
130 |
words => join(" ",@{$ds->{'swish'}}) |
words => $words, |
131 |
); |
); |
132 |
} |
} |
133 |
|
|
139 |
|
|
140 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
141 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
|
path => $f, |
|
142 |
headline => $h, |
headline => $h, |
143 |
|
mfn => $webpac->mfn, |
144 |
); |
); |
145 |
} |
} |
146 |
} |
} |
147 |
|
|
148 |
# print Dumper(\@ds); |
# print Dump(\@ds); |
149 |
|
|
150 |
} |
} |
151 |
|
|
169 |
} |
} |
170 |
|
|
171 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
172 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
173 |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
$log->debug("data hash: ",Dump($webpac->{'data'})); |
174 |
foreach my $t (keys %{$thes}) { |
foreach my $t (keys %{$thes}) { |
175 |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
$log->debug("thesaurus $t hash: ",Dump($thes->{$t})); |
176 |
} |
} |
177 |
} |
} |
178 |
|
|
179 |
} # XXX if(0) |
} # XXX if(0) |
180 |
|
|
181 |
$log->info("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
182 |
|
|
183 |
|
$log->info("creating tree"); |
184 |
|
|
185 |
# |
# |
186 |
# tree dump implementation which shouldn't be here :-) |
# define tree structure |
187 |
# |
# |
188 |
|
|
|
my $file = 'out/browse.html'; |
|
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
|
|
|
|
|
print HTML qq{ |
|
|
<html> |
|
|
<head> |
|
|
<title>Browse</title> |
|
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
|
|
<script type="text/javascript" src="bfilter/hide.js"></script> |
|
|
</head> |
|
|
<body> |
|
|
<ul> |
|
|
}; |
|
|
|
|
189 |
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
190 |
|
|
191 |
my @tree = ({ |
my @tree = ({ |
198 |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
199 |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
200 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
201 |
have_children => sub { $l->{"a:".$_[0]."::" } }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
202 |
child_code => sub { return $_[0] }, |
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:$l2"}); |
206 |
|
}, |
207 |
style => 'display: none', |
style => 'display: none', |
208 |
},{ |
},{ |
209 |
# 1 |
# 1 |
210 |
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
code_arr => sub { @{$_[0]} }, |
211 |
filter_code => sub { shift }, # nop |
filter_code => sub { shift }, # nop |
212 |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
213 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
214 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
215 |
have_children => sub { $l->{"a:".$_[1].":" } }, |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
|
child_code => sub { return $_[1] }, |
|
216 |
style => 'display: none', |
style => 'display: none', |
217 |
},{ |
},{ |
218 |
# 2 |
# 2 |
219 |
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
code_arr => sub { @{$_[0]} }, |
220 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
221 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
222 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
223 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
224 |
have_children => sub { $l->{"a:".$_[2].":".$_[1]} }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
225 |
child_code => sub { return "a:".$_[2].":".$_[1] }, |
#style => 'display: none', |
|
style => 'display: none', |
|
226 |
},{ |
},{ |
227 |
# 3 uži pojam |
# 3 uži pojam |
228 |
|
code_arr => sub { @{$_[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->{$_[0]}} }, |
code_arr => sub { @{$l->{$_[0]}} }, |
278 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
279 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
280 |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
281 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
282 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
283 |
child_code => sub { 0 }, |
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,"a:".$_[1].":".$_[0]) }, |
284 |
|
},{ |
285 |
}); |
}); |
286 |
|
|
287 |
unroll(0,''); |
my $tree = new WebPAC::Tree( |
288 |
|
dir => './out', |
289 |
$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99")); |
html => 'browse.html', |
290 |
|
template => './output_template/tree.tt', |
291 |
sub unroll { |
js => 'tree-ids.js', |
292 |
my ($level,$start_code) = @_; |
tree => \@tree, |
293 |
|
); |
|
$log->logconfess("need level") unless (defined($level)); |
|
|
|
|
|
# all levels passed? |
|
|
return if (! defined($tree[$level])); |
|
|
|
|
|
$log->debug("unroll level $level, start code $start_code"); |
|
|
|
|
|
foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) { |
|
|
|
|
|
if ($code = $tree[$level]->{'filter_code'}->($code)) { |
|
|
|
|
|
$log->debug("# $level filter passed code $code"); |
|
|
|
|
|
my $v900 = $tree[$level]->{'lookup_v900'}->($code) || $log->logdie("can't lookup_v900 '$code'"); |
|
|
$log->debug("# $level lookup_v900($code) = $v900"); |
|
|
my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->logdie("can't lookup_term '$v900'"); |
|
|
$log->debug("# $level lookup_term($code,$v900) = $term"); |
|
|
my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->logdie("can't lookup_mfn '$v900'"); |
|
|
$log->debug("# $level lookup_mfn($code,$v900) = $mfn"); |
|
|
|
|
|
$log->debug("$code -> $v900 : $term [$mfn]"); |
|
|
|
|
|
|
|
|
my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code); |
|
|
|
|
|
my ($link_start,$link_end) = ('',''); |
|
|
|
|
|
($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return hide_show('mfn$mfn');">},qq{</a>}) if ($have_children); |
|
|
|
|
|
my $mfn_link = "thes/$mfn.html"; |
|
|
if (-e "out/$mfn_link") { |
|
|
print HTML " " x $level . |
|
|
qq{<li>${link_start}${term}${link_end}}. |
|
|
qq{ <a href="$mfn_link">»</a></li>\n}; |
|
|
} else { |
|
|
$log->warn("file 'out/$mfn_link' doesn't exist, skipping"); |
|
|
} |
|
|
|
|
|
unless ($have_children) { |
|
|
$log->warn("doesn't have_children($code,$v900,$start_code)"); |
|
|
next; |
|
|
} |
|
|
|
|
|
print HTML " " x $level . |
|
|
qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}. |
|
|
($tree[$level]->{'style'} ? ' style="'.$tree[$level]->{'style'}.'"' : ''). |
|
|
qq{>\n}; |
|
|
|
|
|
unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code)); |
|
|
|
|
|
print HTML " " x $level . qq{</ul>\n}; |
|
|
|
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
print HTML qq{ |
|
|
</ul> |
|
|
</body> |
|
|
</html> |
|
|
}; |
|
|
|
|
|
close(HTML); |
|
294 |
|
|
295 |
|
$log->info("closing index"); |
296 |
$index->close; |
$index->close; |
297 |
|
|
298 |
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |