1 |
dpavlin |
1 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
use OpenIsis; |
5 |
|
|
use Getopt::Std; |
6 |
|
|
use Data::Dumper; |
7 |
|
|
use XML::Simple; |
8 |
dpavlin |
5 |
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
9 |
dpavlin |
10 |
use Text::Iconv; |
10 |
dpavlin |
13 |
use Config::IniFiles; |
11 |
dpavlin |
40 |
use Encode; |
12 |
dpavlin |
1 |
|
13 |
dpavlin |
10 |
$|=1; |
14 |
dpavlin |
9 |
|
15 |
dpavlin |
13 |
my $config_file = $0; |
16 |
|
|
$config_file =~ s/\.pl$/.conf/; |
17 |
|
|
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
18 |
|
|
|
19 |
dpavlin |
10 |
my $config; |
20 |
|
|
|
21 |
|
|
use index_DBI; # there is no other, right now ;-) |
22 |
|
|
|
23 |
|
|
my $index = new index_DBI(); # open index |
24 |
|
|
|
25 |
dpavlin |
1 |
my %opts; |
26 |
|
|
|
27 |
dpavlin |
7 |
# usage: |
28 |
|
|
# -d directory name |
29 |
|
|
# -m multiple directories |
30 |
|
|
# -q quiet |
31 |
|
|
# -s run swish |
32 |
dpavlin |
1 |
|
33 |
dpavlin |
7 |
getopts('d:m:qs', \%opts); |
34 |
|
|
|
35 |
dpavlin |
13 |
my $db_dir; |
36 |
dpavlin |
1 |
|
37 |
dpavlin |
40 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
38 |
dpavlin |
1 |
|
39 |
dpavlin |
40 |
# this is encoding of all files on disk, including import_xml/*.xml file and |
40 |
|
|
# filter/*.pm files! It will be used to store strings in perl internally! |
41 |
|
|
my $codepage = 'ISO-8859-2'; |
42 |
dpavlin |
1 |
|
43 |
dpavlin |
40 |
my $utf2cp = Text::Iconv->new('UTF-8',$codepage); |
44 |
|
|
# this function will convert data from XML files to local encoding |
45 |
|
|
sub x { |
46 |
|
|
return $utf2cp->convert($_[0]); |
47 |
|
|
} |
48 |
dpavlin |
3 |
|
49 |
dpavlin |
40 |
# decode isis import codepage |
50 |
|
|
my $isis2cp; |
51 |
dpavlin |
10 |
|
52 |
dpavlin |
40 |
# outgoing xml must be in UTF-8 |
53 |
|
|
my $cp2utf = Text::Iconv->new($codepage,'UTF-8'); |
54 |
dpavlin |
29 |
|
55 |
dpavlin |
3 |
sub isis2xml { |
56 |
|
|
|
57 |
dpavlin |
10 |
use xmlify; |
58 |
|
|
|
59 |
dpavlin |
3 |
my $row = shift @_; |
60 |
dpavlin |
13 |
my $add_xml = shift @_; |
61 |
dpavlin |
3 |
|
62 |
|
|
my $xml; |
63 |
|
|
|
64 |
dpavlin |
10 |
use parse_format; |
65 |
dpavlin |
3 |
|
66 |
dpavlin |
13 |
my $html = ""; # html formatted display output |
67 |
dpavlin |
10 |
|
68 |
dpavlin |
13 |
my %field_usage; # counter for usage of each field |
69 |
|
|
|
70 |
dpavlin |
32 |
# sort subrouting using order="" attribute |
71 |
|
|
sub by_order { |
72 |
|
|
return 0 if (! $config->{indexer}->{$a}->{order}); |
73 |
|
|
return 0 if (! $config->{indexer}->{$b}->{order}); |
74 |
dpavlin |
29 |
|
75 |
dpavlin |
32 |
return $config->{indexer}->{$a}->{order} <=> |
76 |
|
|
$config->{indexer}->{$b}->{order} ; |
77 |
|
|
} |
78 |
dpavlin |
3 |
|
79 |
dpavlin |
32 |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
80 |
|
|
|
81 |
dpavlin |
40 |
$field=x($field); |
82 |
|
|
|
83 |
dpavlin |
13 |
$field_usage{$field}++; |
84 |
|
|
|
85 |
dpavlin |
10 |
my $swish_data = ""; |
86 |
dpavlin |
3 |
my $display_data = ""; |
87 |
dpavlin |
35 |
my $line_delimiter; |
88 |
dpavlin |
3 |
|
89 |
dpavlin |
34 |
my ($swish,$display); |
90 |
|
|
|
91 |
dpavlin |
3 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
92 |
|
|
|
93 |
dpavlin |
40 |
my $format = x($x->{content}); |
94 |
|
|
my $delimiter = x($x->{delimiter}) || ' '; |
95 |
dpavlin |
3 |
|
96 |
dpavlin |
29 |
my $isis_i = 0; # isis repeatable offset |
97 |
|
|
|
98 |
|
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
99 |
|
|
$s = 0 if (lc($x->{type}) eq "display"); |
100 |
|
|
$d = 0 if (lc($x->{type}) eq "swish"); |
101 |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
102 |
|
|
|
103 |
dpavlin |
35 |
# what will separate last line from this one? |
104 |
|
|
if ($display_data && $x->{append} && $x->{append} eq "1") { |
105 |
dpavlin |
34 |
$line_delimiter = ' '; |
106 |
dpavlin |
35 |
} elsif ($display_data) { |
107 |
dpavlin |
34 |
$line_delimiter = '<br/>'; |
108 |
|
|
} |
109 |
dpavlin |
29 |
|
110 |
dpavlin |
34 |
# init vars so that we go into while... |
111 |
|
|
($swish,$display) = (1,1); |
112 |
dpavlin |
29 |
|
113 |
|
|
while ($swish || $display) { |
114 |
dpavlin |
40 |
($swish,$display) = parse_format($format,$row,$isis_i++,$isis2cp); |
115 |
dpavlin |
3 |
|
116 |
dpavlin |
29 |
# filter="name" ; filter this field through |
117 |
|
|
# filter/[name].pm |
118 |
|
|
my $filter = $x->{filter}; |
119 |
dpavlin |
20 |
if ($filter) { |
120 |
dpavlin |
29 |
require "filter/".$filter.".pm"; |
121 |
dpavlin |
20 |
} |
122 |
dpavlin |
29 |
# type="swish" ; field for swish |
123 |
|
|
if ($s && $swish) { |
124 |
|
|
if ($filter) { |
125 |
|
|
no strict 'refs'; |
126 |
|
|
$swish_data .= join(" ",&$filter($swish)); |
127 |
|
|
} else { |
128 |
|
|
$swish_data .= $swish; |
129 |
|
|
} |
130 |
|
|
} |
131 |
dpavlin |
17 |
|
132 |
dpavlin |
29 |
# type="display" ; field for display |
133 |
|
|
if ($d && $display) { |
134 |
dpavlin |
35 |
if ($line_delimiter && $display_data) { |
135 |
|
|
$display_data .= $line_delimiter; |
136 |
|
|
undef $line_delimiter; |
137 |
|
|
} |
138 |
dpavlin |
29 |
if ($filter) { |
139 |
|
|
no strict 'refs'; |
140 |
|
|
$display_data .= join($delimiter,&$filter($display)); |
141 |
|
|
} else { |
142 |
|
|
if ($display_data) { |
143 |
|
|
$display_data .= $delimiter.$display; |
144 |
|
|
} else { |
145 |
|
|
$display_data .= $display; |
146 |
|
|
} |
147 |
|
|
} |
148 |
dpavlin |
20 |
} |
149 |
dpavlin |
29 |
|
150 |
|
|
# type="index" ; insert into index |
151 |
|
|
if ($i && $display) { |
152 |
dpavlin |
40 |
my $index_data = $display; |
153 |
dpavlin |
29 |
if ($filter) { |
154 |
|
|
no strict 'refs'; |
155 |
|
|
foreach my $d (&$filter($index_data)) { |
156 |
|
|
$index->insert($field, $d, $db_dir); |
157 |
|
|
} |
158 |
|
|
} else { |
159 |
|
|
$index->insert($field, $index_data, $db_dir); |
160 |
dpavlin |
20 |
} |
161 |
|
|
} |
162 |
dpavlin |
17 |
} |
163 |
dpavlin |
3 |
} |
164 |
dpavlin |
9 |
|
165 |
dpavlin |
13 |
|
166 |
dpavlin |
10 |
if ($display_data) { |
167 |
dpavlin |
29 |
|
168 |
dpavlin |
13 |
if ($field eq "headline") { |
169 |
|
|
$xml .= xmlify("headline", $display_data); |
170 |
|
|
} else { |
171 |
|
|
|
172 |
|
|
# find field name (signular, plural) |
173 |
|
|
my $field_name = ""; |
174 |
|
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) { |
175 |
|
|
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
176 |
|
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
177 |
|
|
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
178 |
dpavlin |
20 |
} elsif ($config->{indexer}->{$field}->{name}) { |
179 |
|
|
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
180 |
dpavlin |
13 |
} else { |
181 |
dpavlin |
20 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
182 |
dpavlin |
13 |
} |
183 |
|
|
if ($field_name) { |
184 |
dpavlin |
40 |
$html .= x($field_name); |
185 |
dpavlin |
13 |
} |
186 |
|
|
$html .= $display_data."###\n"; |
187 |
|
|
} |
188 |
dpavlin |
10 |
} |
189 |
|
|
if ($swish_data) { |
190 |
dpavlin |
20 |
# remove extra spaces |
191 |
|
|
$swish_data =~ s/ +/ /g; |
192 |
|
|
$swish_data =~ s/ +$//g; |
193 |
|
|
|
194 |
dpavlin |
40 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
195 |
dpavlin |
10 |
} |
196 |
|
|
|
197 |
dpavlin |
9 |
|
198 |
dpavlin |
3 |
} |
199 |
dpavlin |
13 |
|
200 |
|
|
# dump formatted output in <html> |
201 |
|
|
if ($html) { |
202 |
|
|
$xml .= xmlify("html",$html); |
203 |
|
|
} |
204 |
|
|
|
205 |
dpavlin |
3 |
if ($xml) { |
206 |
dpavlin |
13 |
$xml .= $add_xml if ($add_xml); |
207 |
dpavlin |
10 |
return "<xml>\n$xml</xml>\n"; |
208 |
dpavlin |
3 |
} else { |
209 |
|
|
return; |
210 |
|
|
} |
211 |
|
|
} |
212 |
|
|
|
213 |
|
|
########################################################################## |
214 |
|
|
|
215 |
dpavlin |
13 |
my $cfg = new Config::IniFiles( -file => $config_file ); |
216 |
dpavlin |
1 |
|
217 |
dpavlin |
13 |
foreach my $database ($cfg->Sections) { |
218 |
dpavlin |
1 |
|
219 |
dpavlin |
40 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
220 |
|
|
my $type = $cfg -> val($database, 'type') || die "$database doesn't have 'type' defined"; |
221 |
|
|
my $add_xml = $cfg -> val($database, 'xml'); # optional |
222 |
dpavlin |
1 |
|
223 |
dpavlin |
13 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
224 |
dpavlin |
1 |
|
225 |
dpavlin |
40 |
$isis2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
226 |
|
|
|
227 |
dpavlin |
3 |
my $db = OpenIsis::open( $isis_db ); |
228 |
|
|
if (0) { |
229 |
|
|
# # FIX |
230 |
|
|
# if (! $db ) { |
231 |
|
|
print STDERR "WARNING: can't open '$isis_db'\n"; |
232 |
|
|
next ; |
233 |
|
|
} |
234 |
|
|
|
235 |
dpavlin |
1 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
236 |
|
|
|
237 |
dpavlin |
3 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
238 |
dpavlin |
1 |
|
239 |
dpavlin |
13 |
my $path = $database; # was $isis_db |
240 |
dpavlin |
10 |
|
241 |
dpavlin |
3 |
my $last_p = 0; |
242 |
|
|
|
243 |
dpavlin |
10 |
# { my $row_id = 4514; |
244 |
dpavlin |
3 |
# FIX |
245 |
dpavlin |
1 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
246 |
|
|
my $row = OpenIsis::read( $db, $row_id ); |
247 |
dpavlin |
3 |
if ($row && $row->{mfn}) { |
248 |
|
|
# output current process indicator |
249 |
|
|
my $p = int($row->{mfn} * 100 / $max_rowid); |
250 |
|
|
if ($p != $last_p) { |
251 |
|
|
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q}); |
252 |
|
|
$last_p = $p; |
253 |
|
|
} |
254 |
|
|
|
255 |
dpavlin |
40 |
if (my $xml = $cp2utf->convert(isis2xml($row,$add_xml))) { |
256 |
dpavlin |
35 |
use bytes; # as opposed to chars |
257 |
dpavlin |
10 |
print "Path-Name: $path#".int($row->{mfn})."\n"; |
258 |
|
|
print "Content-Length: ".(length($xml)+1)."\n"; |
259 |
|
|
print "Document-Type: XML\n\n$xml\n"; |
260 |
dpavlin |
3 |
} |
261 |
dpavlin |
1 |
} |
262 |
|
|
} |
263 |
dpavlin |
3 |
print STDERR "\n"; |
264 |
dpavlin |
1 |
} |
265 |
dpavlin |
3 |
|
266 |
dpavlin |
10 |
# call this to commit index |
267 |
|
|
$index->close; |
268 |
dpavlin |
3 |
|
269 |
|
|
1; |
270 |
|
|
__END__ |
271 |
|
|
########################################################################## |
272 |
|
|
|
273 |
|
|
=head1 NAME |
274 |
|
|
|
275 |
|
|
isis2xml.pl - read isis file and dump XML |
276 |
|
|
|
277 |
|
|
=head1 DESCRIPTION |
278 |
|
|
|
279 |
|
|
This command will read ISIS data file using OpenIsis perl module and |
280 |
|
|
create XML file for usage with I<SWISH-E> |
281 |
|
|
indexer. Dispite it's name, this script B<isn't general xml generator> |
282 |
|
|
from isis files (isis allready has something like that). Output of this |
283 |
|
|
script is tailor-made for SWISH-E. |
284 |
|
|
|
285 |
|
|
=head1 AUTHOR |
286 |
|
|
|
287 |
|
|
Dobrica Pavlinusic <dpavlin@rot13.org> |
288 |
|
|
|
289 |
|
|
=head1 COPYRIGHT |
290 |
|
|
|
291 |
|
|
GNU Public License (GPL) v2 or later |
292 |
|
|
|
293 |
|
|
=head1 SEE ALSO |
294 |
|
|
|
295 |
|
|
SWISH-E web site at http://www.swish-e.org |
296 |
|
|
|
297 |
|
|
=cut |