/[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 42 - (show annotations)
Sat Mar 15 21:48:48 2003 UTC (16 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 7188 byte(s)
filter fix && optimisation

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

Properties

Name Value
cvs2svn:cvs-rev 1.17
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26