1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
use strict; |
=head1 NAME |
|
use OpenIsis; |
|
|
use Getopt::Std; |
|
|
use Data::Dumper; |
|
|
use XML::Simple; |
|
|
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
|
|
use Text::Iconv; |
|
|
use Config::IniFiles; |
|
|
|
|
|
$|=1; |
|
|
|
|
|
my $config_file = $0; |
|
|
$config_file =~ s/\.pl$/.conf/; |
|
|
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
|
4 |
|
|
5 |
my $config; |
all2all.pl - basic script for all WebPAC needs |
6 |
|
|
7 |
use index_DBI; # there is no other, right now ;-) |
=cut |
8 |
|
|
9 |
my $index = new index_DBI(); # open index |
use strict; |
10 |
|
use Data::Dumper; |
11 |
|
use Carp; |
12 |
|
|
13 |
my %opts; |
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 |
|
# low_mem => 1, |
23 |
|
) || 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 |
# usage: |
my $thes; |
|
# -d directory name |
|
|
# -m multiple directories |
|
|
# -q quiet |
|
|
# -s run swish |
|
35 |
|
|
36 |
getopts('d:m:qs', \%opts); |
$|=1; |
37 |
|
|
38 |
my $db_dir; |
my $maxmfn = $webpac->open_isis( |
39 |
|
filename => shift @ARGV || '/data/hidra/THS/THS', |
40 |
|
lookup => [ |
41 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
42 |
|
# { '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' }, |
44 |
|
# { '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' }, |
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 |
|
$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); |
54 |
|
|
55 |
|
$log->info("rows: $maxmfn"); |
56 |
|
|
57 |
|
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
58 |
|
|
59 |
|
if(1) { # XXX |
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 |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
next if (! @ds); |
71 |
|
|
72 |
#print Dumper($config->{indexer}); |
my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); |
|
#print "-" x 70,"\n"; |
|
73 |
|
|
74 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
if ($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 $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
my $headline = $webpac->{'headline'}; |
|
#my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
|
|
my $isis_codepage; |
|
|
my $index_codepage; |
|
|
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
|
|
my $xml_codepage; |
|
90 |
|
|
91 |
sub isis2xml { |
my $f = $filename; |
92 |
|
$f =~ s!out/!!; |
93 |
|
|
94 |
use xmlify; |
# 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 |
my $row = shift @_; |
# save into sorted index (thesaurus) |
107 |
my $add_xml = shift @_; |
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 |
my $xml; |
# print Dumper(\@ds); |
121 |
|
|
122 |
use parse_format; |
} |
123 |
|
|
124 |
my $html = ""; # html formatted display output |
foreach my $t (keys %{$thes}) { |
125 |
|
|
126 |
my %field_usage; # counter for usage of each field |
my @e = $thes->{$t}->elements; |
127 |
|
if (! @e) { |
128 |
|
$log->logwarn("no elements in sorted index $t?"); |
129 |
|
next; |
130 |
|
} |
131 |
|
|
132 |
foreach my $field (keys %{$config->{indexer}}) { |
my $file = "./out/bfilter/$t.txt"; |
133 |
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
134 |
|
|
135 |
$field_usage{$field}++; |
$webpac->output_file( |
136 |
|
file => $file, |
137 |
|
template => 'index.tt', |
138 |
|
data => \@e, |
139 |
|
index_name => $t, |
140 |
|
); |
141 |
|
} |
142 |
|
|
143 |
my $swish_data = ""; |
if (0 && $log->is_debug) { |
144 |
my $display_data = ""; |
$log->debug("lookup hash: ",Dumper($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 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
} # XXX if(0) |
152 |
|
|
153 |
my $format = $x->{content}; |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
|
|
$s = 0 if (lc($x->{type}) eq "display"); |
|
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
|
|
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
|
154 |
|
|
155 |
# FIX: this is ugly, UGLY, cludge string is returned |
# |
156 |
# in UTF8 encoding , but as if source charset |
# tree dump implementation which shouldn't be here :-) |
157 |
# is ISO8859-1 and not some other. This break other |
# |
158 |
# character encodings, so we convert it first |
|
159 |
# back to ISO8859-1 |
|
160 |
$format = $cludge_codepage->convert($format); |
my $file = 'out/browse.html'; |
161 |
|
my $js_url = 'tree-ids.js'; |
162 |
|
|
163 |
|
$log->info("creating '$file' for tree html"); |
164 |
|
open(HTML, "> $file") || $log->logdie("can't open '$file'"); |
165 |
|
|
166 |
|
print HTML qq{ |
167 |
|
<html> |
168 |
|
<head> |
169 |
|
<title>Browse</title> |
170 |
|
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2"/> |
171 |
|
<script type="text/javascript" src="bfilter/hide.js"></script> |
172 |
|
<script type="text/javascript" src="$js_url"></script> |
173 |
|
</head> |
174 |
|
|
175 |
|
<!-- |
176 |
|
<body onLoad="show_hide_display(show,hide);"> |
177 |
|
--> |
178 |
|
<body> |
179 |
|
|
180 |
|
<div style="float: right; width: 10em;"> |
181 |
|
Folding: |
182 |
|
<a href="#" onClick="show_hide_display(show,hide);">default</a> |
183 |
|
<a href="#" onClick="show_display(show); show_display(hide);">all</a> |
184 |
|
<a href="#" onClick="hide_display(hide); hide_display(show);">none</a> |
185 |
|
<a href="#" onClick="alert('show: '+show.length+', hide: '+hide.length);">debug</a> |
186 |
|
</div> |
187 |
|
|
188 |
|
|
189 |
|
<ul> |
190 |
|
}; |
191 |
|
|
192 |
|
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
193 |
|
|
194 |
|
my @tree = ({ |
195 |
|
# level 0 |
196 |
|
code_arr => sub { sort keys %{$l} }, |
197 |
|
filter_code => sub { |
198 |
|
my $t = shift; |
199 |
|
return $t if ($t =~ s/root://); |
200 |
|
}, |
201 |
|
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
202 |
|
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
203 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
204 |
|
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
205 |
|
child_code => sub { return $_[0] }, |
206 |
|
style => 'display: none', |
207 |
|
},{ |
208 |
|
# 1 |
209 |
|
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
210 |
|
filter_code => sub { shift }, # nop |
211 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
212 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
213 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
214 |
|
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
215 |
|
child_code => sub { return $_[1] }, |
216 |
|
style => 'display: none', |
217 |
|
},{ |
218 |
|
# 2 |
219 |
|
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
220 |
|
filter_code => sub { shift }, |
221 |
|
lookup_v900 => sub { shift }, |
222 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
223 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
224 |
|
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
225 |
|
child_code => sub { return "a:".$_[2].":".$_[1] }, |
226 |
|
style => 'display: none', |
227 |
|
},{ |
228 |
|
# 3 uľi pojam |
229 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
230 |
|
filter_code => sub { shift }, |
231 |
|
lookup_v900 => sub { |
232 |
|
my ($c,$p) = @_; |
233 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
234 |
|
return "a:".$p.":".$c; |
235 |
|
}, |
236 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
237 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
238 |
|
have_children => sub { defined($l->{$_[1]}) }, |
239 |
|
child_code => sub { return $_[1] }, |
240 |
|
},{ |
241 |
|
# 4 |
242 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
243 |
|
filter_code => sub { shift }, |
244 |
|
lookup_v900 => sub { |
245 |
|
my ($c,$p) = @_; |
246 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
247 |
|
return "a:".$p.":".$c; |
248 |
|
}, |
249 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
250 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
251 |
|
have_children => sub { defined($l->{$_[1]}) }, |
252 |
|
child_code => sub { return $_[1] }, |
253 |
|
},{ |
254 |
|
# 5 |
255 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
256 |
|
filter_code => sub { shift }, |
257 |
|
lookup_v900 => sub { |
258 |
|
my ($c,$p) = @_; |
259 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
260 |
|
return "a:".$p.":".$c; |
261 |
|
}, |
262 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
263 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
264 |
|
have_children => sub { defined($l->{$_[1]}) }, |
265 |
|
child_code => sub { return $_[1] }, |
266 |
|
},{ |
267 |
|
# 6 |
268 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
269 |
|
filter_code => sub { shift }, |
270 |
|
lookup_v900 => sub { |
271 |
|
my ($c,$p) = @_; |
272 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
273 |
|
return "a:".$p.":".$c; |
274 |
|
}, |
275 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
276 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
277 |
|
have_children => sub { defined($l->{$_[1]}) }, |
278 |
|
child_code => sub { return $_[1] }, |
279 |
|
},{ |
280 |
|
# 7 |
281 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
282 |
|
filter_code => sub { shift }, |
283 |
|
lookup_v900 => sub { |
284 |
|
my ($c,$p) = @_; |
285 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
286 |
|
return "a:".$p.":".$c; |
287 |
|
}, |
288 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
289 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
290 |
|
have_children => sub { defined($l->{$_[1]}) }, |
291 |
|
child_code => sub { return $_[1] }, |
292 |
|
},{ |
293 |
|
# 8 |
294 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
295 |
|
filter_code => sub { shift }, |
296 |
|
lookup_v900 => sub { |
297 |
|
my ($c,$p) = @_; |
298 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
299 |
|
return "a:".$p.":".$c; |
300 |
|
}, |
301 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
302 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
303 |
|
# have_children => sub { defined($l->{$_[1]}) }, |
304 |
|
# child_code => sub { return $_[1] }, |
305 |
|
have_children => sub { 0 }, |
306 |
|
child_code => sub { 0 }, |
307 |
|
}); |
308 |
|
|
309 |
|
my @show_ids; |
310 |
|
my @hide_ids; |
311 |
|
|
312 |
|
unroll(0,''); |
313 |
|
|
314 |
|
$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99")); |
315 |
|
|
316 |
|
sub unroll { |
317 |
|
my ($level,$start_code) = @_; |
318 |
|
|
319 |
|
$log->logconfess("need level") unless (defined($level)); |
320 |
|
|
321 |
|
# all levels passed? |
322 |
|
return if (! defined($tree[$level])); |
323 |
|
|
324 |
|
$log->debug("unroll level $level, start code $start_code"); |
325 |
|
|
326 |
|
foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) { |
327 |
|
|
328 |
|
if ($code = $tree[$level]->{'filter_code'}->($code)) { |
329 |
|
|
330 |
|
$log->debug("# $level filter passed code $code"); |
331 |
|
|
332 |
|
my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)"); |
333 |
|
$log->debug("# $level lookup_v900($code,$start_code) = $v900"); |
334 |
|
|
335 |
|
my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)"); |
336 |
|
$log->debug("# $level lookup_term($code,$v900) = $term"); |
337 |
|
|
338 |
my ($swish,$display) = parse_format($format,$row); |
my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)"); |
339 |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
$log->debug("# $level lookup_mfn($code,$v900) = $mfn"); |
340 |
|
|
341 |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
$log->debug("$code -> $v900 : $term [$mfn]"); |
342 |
|
|
343 |
# filter="name" ; filter this field through |
my ($link_start,$link_end) = ('',''); |
344 |
# filter/[name].pm |
|
345 |
my $filter; |
my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code); |
346 |
if ($x->{filter}) { |
if ($have_children) { |
347 |
$filter = "filter/".$x->{filter}.".pm"; |
($link_start,$link_end) = (qq{<a href="#mfn$mfn" onClick="return toggle_display('mfn$mfn');">},qq{</a>}); |
348 |
require $filter; |
} else { |
349 |
} |
$log->debug("# $level doesn't have_children($code,$v900,$start_code)"); |
|
# type="swish" ; field for swish |
|
|
if ($s && $swish) { |
|
|
if ($filter) { |
|
|
$swish_data .= join(" ",&filter($swish)); |
|
|
} else { |
|
|
$swish_data .= $swish if ($s && $swish); |
|
|
} |
|
350 |
} |
} |
351 |
|
|
352 |
# type="display" ; field for display |
my $mfn_link = "thes/$mfn.html"; |
353 |
if ($d && $display) { |
if (-e "out/$mfn_link") { |
354 |
if ($filter) { |
print HTML " " x $level . |
355 |
$display_data .= join(" ",&filter($display)); |
qq{<li>${link_start}${term}${link_end}}. |
356 |
} else { |
qq{ <a href="$mfn_link">»</a></li>\n}; |
357 |
$display_data .= $display if ($s && $display); |
} else { |
358 |
} |
$log->warn("file 'out/$mfn_link' doesn't exist, skipping"); |
|
} |
|
|
|
|
|
# type="index" ; insert into index |
|
|
if ($i && $display) { |
|
|
my $index_data = $index_codepage->convert($display) || $display; |
|
|
if ($filter) { |
|
|
foreach my $d (&filter($index_data)) { |
|
|
$index->insert($field, $d, $db_dir); |
|
|
} |
|
|
} else { |
|
|
$index->insert($field, $index_data, $db_dir); |
|
|
} |
|
359 |
} |
} |
|
} |
|
|
|
|
360 |
|
|
361 |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
unless ($have_children) { |
362 |
if ($display_data) { |
next; |
363 |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
} |
364 |
# FIX: this is removed and replaced by html tag. |
my $style = $tree[$level]->{'style'}; |
|
#$xml .= xmlify($field."_display", $display_data); |
|
|
|
|
|
if ($field eq "headline") { |
|
|
$xml .= xmlify("headline", $display_data); |
|
|
} else { |
|
365 |
|
|
366 |
# find field name (signular, plural) |
print HTML " " x $level . |
367 |
my $field_name = ""; |
qq{<a name="mfn$mfn"></a>\n <ul id="mfn$mfn"}. |
368 |
if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) { |
($style ? ' style="'.$style.'"' : ''). |
369 |
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
qq{>\n}; |
370 |
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
|
371 |
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
if ($style) { |
372 |
} elsif ($config->{indexer}->{$field}->{name}) { |
if ($style =~ m/display\s*:\s*none/i) { |
373 |
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
push @hide_ids, "mfn$mfn"; |
374 |
} else { |
} else { |
375 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
push @show_ids, "mfn$mfn"; |
|
} |
|
|
if ($field_name) { |
|
|
$html .= $xml_codepage->convert($field_name); |
|
376 |
} |
} |
377 |
$html .= $display_data."###\n"; |
} else { |
378 |
|
# default: show |
379 |
|
push @show_ids, "mfn$mfn"; |
380 |
} |
} |
|
} |
|
|
if ($swish_data) { |
|
|
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
|
|
# remove extra spaces |
|
|
$swish_data =~ s/ +/ /g; |
|
|
$swish_data =~ s/ +$//g; |
|
|
|
|
|
$swish_data = $i->convert($swish_data); |
|
|
$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data)); |
|
|
#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data; |
|
|
#$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data)); |
|
|
} |
|
|
|
|
|
|
|
|
} |
|
|
|
|
|
# dump formatted output in <html> |
|
|
if ($html) { |
|
|
$xml .= xmlify("html",$html); |
|
|
} |
|
|
|
|
|
if ($xml) { |
|
|
#print STDERR "x: $xml\n"; |
|
|
$xml .= $add_xml if ($add_xml); |
|
|
return "<xml>\n$xml</xml>\n"; |
|
|
} else { |
|
|
return; |
|
|
} |
|
|
} |
|
|
|
|
|
########################################################################## |
|
|
|
|
|
my $cfg = new Config::IniFiles( -file => $config_file ); |
|
|
|
|
|
foreach my $database ($cfg->Sections) { |
|
|
|
|
|
my $isis_db = $cfg -> val($database, 'isis_db'); |
|
|
my $type = $cfg -> val($database, 'type'); |
|
|
my $add_xml = $cfg -> val($database, 'xml'); |
|
|
|
|
|
# read configuration for this type |
|
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
|
|
$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
|
|
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
|
|
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8'); |
|
|
|
|
|
my $db = OpenIsis::open( $isis_db ); |
|
|
if (0) { |
|
|
# # FIX |
|
|
# if (! $db ) { |
|
|
print STDERR "WARNING: can't open '$isis_db'\n"; |
|
|
next ; |
|
|
} |
|
|
|
|
|
my $max_rowid = OpenIsis::maxRowid( $db ); |
|
|
|
|
|
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
|
381 |
|
|
382 |
my $path = $database; # was $isis_db |
unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code)); |
383 |
|
|
384 |
|
print HTML " " x $level . qq{</ul>\n}; |
385 |
|
|
|
my $last_p = 0; |
|
|
|
|
|
# { my $row_id = 4514; |
|
|
# FIX |
|
|
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
|
|
my $row = OpenIsis::read( $db, $row_id ); |
|
|
if ($row && $row->{mfn}) { |
|
|
#print STDERR "mfn: ",$row->{mfn},"\n"; |
|
|
# output current process indicator |
|
|
my $p = int($row->{mfn} * 100 / $max_rowid); |
|
|
if ($p != $last_p) { |
|
|
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q}); |
|
|
$last_p = $p; |
|
|
} |
|
|
|
|
|
if (my $xml = isis2xml($row,$add_xml)) { |
|
|
#print STDERR "--ret-->$xml\n"; |
|
|
print "Path-Name: $path#".int($row->{mfn})."\n"; |
|
|
print "Content-Length: ".(length($xml)+1)."\n"; |
|
|
print "Document-Type: XML\n\n$xml\n"; |
|
|
} |
|
386 |
} |
} |
387 |
} |
} |
|
print STDERR "\n"; |
|
388 |
} |
} |
389 |
|
|
390 |
# call this to commit index |
print HTML qq{ |
391 |
$index->close; |
</ul> |
392 |
|
</body> |
393 |
1; |
</html> |
394 |
__END__ |
}; |
395 |
########################################################################## |
|
396 |
|
close(HTML); |
397 |
=head1 NAME |
|
398 |
|
|
399 |
isis2xml.pl - read isis file and dump XML |
my $js_file = "out/$js_url"; |
400 |
|
$log->info("creating '$js_file' with arrays of shown and hidden ids"); |
401 |
|
open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!"); |
402 |
|
print JS "var show = ['",join("','",@show_ids),"'];\n"; |
403 |
|
print JS "var hide = ['",join("','",@hide_ids),"'];\n"; |
404 |
|
close(JS); |
405 |
|
|
406 |
=head1 DESCRIPTION |
$log->info("closing index"); |
407 |
|
$index->close; |
|
This command will read ISIS data file using OpenIsis perl module and |
|
|
create XML file for usage with I<SWISH-E> |
|
|
indexer. Dispite it's name, this script B<isn't general xml generator> |
|
|
from isis files (isis allready has something like that). Output of this |
|
|
script is tailor-made for SWISH-E. |
|
|
|
|
|
=head1 AUTHOR |
|
|
|
|
|
Dobrica Pavlinusic <dpavlin@rot13.org> |
|
|
|
|
|
=head1 COPYRIGHT |
|
|
|
|
|
GNU Public License (GPL) v2 or later |
|
|
|
|
|
=head1 SEE ALSO |
|
|
|
|
|
SWISH-E web site at http://www.swish-e.org |
|
408 |
|
|
409 |
=cut |
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |