/[webpac]/trunk/all2xml.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /trunk/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 35 - (show annotations)
Sun Feb 23 15:47:40 2003 UTC (16 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 8276 byte(s)
last changes; completly broken charsets

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 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
9 use Text::Iconv;
10 use Config::IniFiles;
11
12 $|=1;
13
14 my $config_file = $0;
15 $config_file =~ s/\.pl$/.conf/;
16 die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
17
18 my $config;
19
20 use index_DBI; # there is no other, right now ;-)
21
22 my $index = new index_DBI(); # open index
23
24 my %opts;
25
26 # usage:
27 # -d directory name
28 # -m multiple directories
29 # -q quiet
30 # -s run swish
31
32 getopts('d:m:qs', \%opts);
33
34 my $db_dir;
35
36 #die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
37
38 #print Dumper($config->{indexer});
39 #print "-" x 70,"\n";
40
41 Text::Iconv->raise_error(1); # Conversion errors raise exceptions
42
43 my $isis_codepage;
44 my $index_codepage;
45 my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
46 my $xml_codepage;
47
48 my $XML_CHARSET = 'UTF8';
49
50
51 sub isis2xml {
52
53 use xmlify;
54
55 my $row = shift @_;
56 my $add_xml = shift @_;
57
58 my $xml;
59
60 use parse_format;
61
62 my $html = ""; # html formatted display output
63
64 my %field_usage; # counter for usage of each field
65
66 # sort subrouting using order="" attribute
67 sub by_order {
68 return 0 if (! $config->{indexer}->{$a}->{order});
69 return 0 if (! $config->{indexer}->{$b}->{order});
70
71 return $config->{indexer}->{$a}->{order} <=>
72 $config->{indexer}->{$b}->{order} ;
73 }
74
75 foreach my $field (sort by_order keys %{$config->{indexer}}) {
76
77 $field_usage{$field}++;
78
79 my $swish_data = "";
80 my $display_data = "";
81 my $line_delimiter;
82
83 my ($swish,$display);
84
85 foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
86
87 my $format = $x->{content};
88 my $delimiter = $x->{delimiter} || ' ';
89
90 # FIX: this is ugly, UGLY, cludge string is returned
91 # in UTF8 encoding , but as if source charset
92 # is ISO8859-1 and not some other. This break other
93 # character encodings, so we convert it first
94 # back to ISO8859-1
95 $format = $xml_codepage->convert($format);
96 $delimiter = $xml_codepage->convert($delimiter) if ($delimiter);
97
98 my $isis_i = 0; # isis repeatable offset
99
100 my ($s,$d,$i) = (1,1,0); # swish, display default
101 $s = 0 if (lc($x->{type}) eq "display");
102 $d = 0 if (lc($x->{type}) eq "swish");
103 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
104 #print STDERR "## s: $s d: $d i: $i ## $format ##\n";
105
106 # what will separate last line from this one?
107 if ($display_data && $x->{append} && $x->{append} eq "1") {
108 $line_delimiter = ' ';
109 } elsif ($display_data) {
110 $line_delimiter = '<br/>';
111 }
112
113 # init vars so that we go into while...
114 ($swish,$display) = (1,1);
115
116 while ($swish || $display) {
117 ($swish,$display) = parse_format($format,$row,$isis_i++);
118 #print STDERR "s: $swish\nd: $display\n" if ($swish);
119
120 #print STDERR "swish: $swish<-- display: $display<--\n";
121
122 # filter="name" ; filter this field through
123 # filter/[name].pm
124 my $filter = $x->{filter};
125 if ($filter) {
126 require "filter/".$filter.".pm";
127 }
128 # type="swish" ; field for swish
129 if ($s && $swish) {
130 if ($filter) {
131 #print STDERR "using filter '$filter'\n";
132 no strict 'refs';
133 $swish_data .= join(" ",&$filter($swish));
134 } else {
135 $swish_data .= $swish;
136 }
137 }
138
139 # type="display" ; field for display
140 if ($d && $display) {
141 if ($line_delimiter && $display_data) {
142 $display_data .= $line_delimiter;
143 undef $line_delimiter;
144 }
145 if ($filter) {
146 no strict 'refs';
147 $display_data .= join($delimiter,&$filter($display));
148 } else {
149 if ($display_data) {
150 $display_data .= $delimiter.$display;
151 } else {
152 $display_data .= $display;
153 }
154 }
155 }
156
157 # type="index" ; insert into index
158 if ($i && $display) {
159 my $index_data = $index_codepage->convert($display) || $display;
160 if ($filter) {
161 no strict 'refs';
162 foreach my $d (&$filter($index_data)) {
163 $index->insert($field, $d, $db_dir);
164 }
165 } else {
166 $index->insert($field, $index_data, $db_dir);
167 }
168 }
169 }
170 }
171
172
173 #print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
174 if ($display_data) {
175
176 # $display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
177 # FIX: this is removed and replaced by html tag.
178 #$xml .= xmlify($field."_display", $display_data);
179
180 if ($field eq "headline") {
181 $xml .= xmlify("headline", $display_data);
182 } else {
183
184 # find field name (signular, plural)
185 my $field_name = "";
186 if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
187 $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
188 } elsif ($config->{indexer}->{$field}->{name_plural}) {
189 $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
190 } elsif ($config->{indexer}->{$field}->{name}) {
191 $field_name = $config->{indexer}->{$field}->{name}."#-#";
192 } else {
193 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
194 }
195 if ($field_name) {
196 $html .= $xml_codepage->convert($field_name);
197 }
198 $html .= $display_data."###\n";
199 }
200 }
201 if ($swish_data) {
202 my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
203 # remove extra spaces
204 $swish_data =~ s/ +/ /g;
205 $swish_data =~ s/ +$//g;
206
207 $swish_data = $i->convert($swish_data);
208 $xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
209 #$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
210 #$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
211 }
212
213
214 }
215
216 # dump formatted output in <html>
217 if ($html) {
218 $xml .= xmlify("html",$html);
219 }
220
221 if ($xml) {
222 #print STDERR "x: $xml\n";
223 $xml .= $add_xml if ($add_xml);
224 return "<xml>\n$xml</xml>\n";
225 } else {
226 return;
227 }
228 }
229
230 ##########################################################################
231
232 my $cfg = new Config::IniFiles( -file => $config_file );
233
234 foreach my $database ($cfg->Sections) {
235
236 my $isis_db = $cfg -> val($database, 'isis_db');
237 my $type = $cfg -> val($database, 'type');
238 my $add_xml = $cfg -> val($database, 'xml');
239
240 # read configuration for this type
241 $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
242 $isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET);
243 $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
244 $xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET);
245
246 my $db = OpenIsis::open( $isis_db );
247 if (0) {
248 # # FIX
249 # if (! $db ) {
250 print STDERR "WARNING: can't open '$isis_db'\n";
251 next ;
252 }
253
254 my $max_rowid = OpenIsis::maxRowid( $db );
255
256 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
257
258 my $path = $database; # was $isis_db
259
260 my $last_p = 0;
261
262 # { my $row_id = 4514;
263 # FIX
264 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
265 my $row = OpenIsis::read( $db, $row_id );
266 if ($row && $row->{mfn}) {
267 #print STDERR "mfn: ",$row->{mfn},"\n";
268 # output current process indicator
269 my $p = int($row->{mfn} * 100 / $max_rowid);
270 if ($p != $last_p) {
271 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
272 $last_p = $p;
273 }
274
275 if (my $xml = isis2xml($row,$add_xml)) {
276 #print STDERR "--ret-->$xml\n";
277 use bytes; # as opposed to chars
278 print "Path-Name: $path#".int($row->{mfn})."\n";
279 print "Content-Length: ".(length($xml)+1)."\n";
280 print "Document-Type: XML\n\n$xml\n";
281 }
282 }
283 }
284 print STDERR "\n";
285 }
286
287 # call this to commit index
288 $index->close;
289
290 1;
291 __END__
292 ##########################################################################
293
294 =head1 NAME
295
296 isis2xml.pl - read isis file and dump XML
297
298 =head1 DESCRIPTION
299
300 This command will read ISIS data file using OpenIsis perl module and
301 create XML file for usage with I<SWISH-E>
302 indexer. Dispite it's name, this script B<isn't general xml generator>
303 from isis files (isis allready has something like that). Output of this
304 script is tailor-made for SWISH-E.
305
306 =head1 AUTHOR
307
308 Dobrica Pavlinusic <dpavlin@rot13.org>
309
310 =head1 COPYRIGHT
311
312 GNU Public License (GPL) v2 or later
313
314 =head1 SEE ALSO
315
316 SWISH-E web site at http://www.swish-e.org
317
318 =cut

Properties

Name Value
cvs2svn:cvs-rev 1.15
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26