/[webpac]/branches/ffzg/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

Annotation of /branches/ffzg/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 29 - (hide annotations)
Sun Feb 23 07:08:54 2003 UTC (21 years ago) by dpavlin
Original Path: trunk/all2xml.pl
File MIME type: text/plain
File size: 7880 byte(s)
repeatable field support, filter functions added, broken charset (again!)

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 1
12 dpavlin 10 $|=1;
13 dpavlin 9
14 dpavlin 13 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 dpavlin 10 my $config;
19    
20     use index_DBI; # there is no other, right now ;-)
21    
22     my $index = new index_DBI(); # open index
23    
24 dpavlin 1 my %opts;
25    
26 dpavlin 7 # usage:
27     # -d directory name
28     # -m multiple directories
29     # -q quiet
30     # -s run swish
31 dpavlin 1
32 dpavlin 7 getopts('d:m:qs', \%opts);
33    
34 dpavlin 13 my $db_dir;
35 dpavlin 1
36 dpavlin 3 #die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
37 dpavlin 1
38 dpavlin 3 #print Dumper($config->{indexer});
39     #print "-" x 70,"\n";
40 dpavlin 1
41 dpavlin 10 Text::Iconv->raise_error(1); # Conversion errors raise exceptions
42 dpavlin 3
43 dpavlin 13 my $isis_codepage;
44     my $index_codepage;
45 dpavlin 10 my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
46 dpavlin 13 my $xml_codepage;
47 dpavlin 10
48 dpavlin 29 my $XML_CHARSET = 'UTF8';
49    
50    
51 dpavlin 3 sub isis2xml {
52    
53 dpavlin 10 use xmlify;
54    
55 dpavlin 3 my $row = shift @_;
56 dpavlin 13 my $add_xml = shift @_;
57 dpavlin 3
58     my $xml;
59    
60 dpavlin 10 use parse_format;
61 dpavlin 3
62 dpavlin 13 my $html = ""; # html formatted display output
63 dpavlin 10
64 dpavlin 13 my %field_usage; # counter for usage of each field
65    
66 dpavlin 29
67 dpavlin 3 foreach my $field (keys %{$config->{indexer}}) {
68    
69 dpavlin 13 $field_usage{$field}++;
70    
71 dpavlin 10 my $swish_data = "";
72 dpavlin 3 my $display_data = "";
73 dpavlin 29 my $line_delimiter = "";
74 dpavlin 3
75     foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
76    
77     my $format = $x->{content};
78 dpavlin 29 my $delimiter = $x->{delimiter} || ' ';
79 dpavlin 3
80 dpavlin 20 # FIX: this is ugly, UGLY, cludge string is returned
81     # in UTF8 encoding , but as if source charset
82     # is ISO8859-1 and not some other. This break other
83     # character encodings, so we convert it first
84     # back to ISO8859-1
85 dpavlin 29 $format = $xml_codepage->convert($format);
86     $delimiter = $xml_codepage->convert($delimiter) if ($delimiter);
87 dpavlin 20
88 dpavlin 29 my $isis_i = 0; # isis repeatable offset
89    
90     my ($s,$d,$i) = (1,1,0); # swish, display default
91     $s = 0 if (lc($x->{type}) eq "display");
92     $d = 0 if (lc($x->{type}) eq "swish");
93     ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
94     #print STDERR "## s: $s d: $d i: $i ## $format ##\n";
95    
96     $display_data .= $line_delimiter if ($display_data && $display_data !~ /$line_delimiter$/);
97    
98     my ($swish,$display) = (1,1);
99    
100     while ($swish || $display) {
101     ($swish,$display) = parse_format($format,$row,$isis_i++);
102 dpavlin 10 #print STDERR "s: $swish\nd: $display\n" if ($swish);
103 dpavlin 3
104 dpavlin 10 #print STDERR "swish: $swish<-- display: $display<--\n";
105    
106 dpavlin 29 # filter="name" ; filter this field through
107     # filter/[name].pm
108     my $filter = $x->{filter};
109 dpavlin 20 if ($filter) {
110 dpavlin 29 require "filter/".$filter.".pm";
111 dpavlin 20 }
112 dpavlin 29 # type="swish" ; field for swish
113     if ($s && $swish) {
114     if ($filter) {
115     print STDERR "using filter '$filter'\n";
116     no strict 'refs';
117     $swish_data .= join(" ",&$filter($swish));
118     } else {
119     $swish_data .= $swish;
120     }
121     }
122 dpavlin 17
123 dpavlin 29 # type="display" ; field for display
124     if ($d && $display) {
125     if ($filter) {
126     no strict 'refs';
127     $display_data .= join($delimiter,&$filter($display));
128     } else {
129     if ($display_data) {
130     $display_data .= $delimiter.$display;
131     } else {
132     $display_data .= $display;
133     }
134     }
135 dpavlin 20 }
136 dpavlin 29
137     # type="index" ; insert into index
138     if ($i && $display) {
139     my $index_data = $index_codepage->convert($display) || $display;
140     if ($filter) {
141     no strict 'refs';
142     foreach my $d (&$filter($index_data)) {
143     $index->insert($field, $d, $db_dir);
144     }
145     } else {
146     $index->insert($field, $index_data, $db_dir);
147 dpavlin 20 }
148     }
149 dpavlin 17 }
150 dpavlin 29 if ($x->{append}) {
151     $line_delimiter = ' ';
152     } else {
153     $line_delimiter = '<br/>';
154     }
155 dpavlin 3 }
156 dpavlin 9
157 dpavlin 13
158 dpavlin 10 #print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
159     if ($display_data) {
160 dpavlin 29
161     # remove last <br>
162     $display_data =~ s/$line_delimiter$//;
163    
164 dpavlin 13 $display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
165     # FIX: this is removed and replaced by html tag.
166     #$xml .= xmlify($field."_display", $display_data);
167    
168     if ($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     $html .= $xml_codepage->convert($field_name);
185 dpavlin 29 # $html .= "-->".$field_name."<--";
186 dpavlin 13 }
187     $html .= $display_data."###\n";
188     }
189 dpavlin 10 }
190     if ($swish_data) {
191     my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
192 dpavlin 20 # remove extra spaces
193     $swish_data =~ s/ +/ /g;
194     $swish_data =~ s/ +$//g;
195    
196 dpavlin 10 $swish_data = $i->convert($swish_data);
197     $xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
198     #$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
199     #$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
200     }
201    
202 dpavlin 9
203 dpavlin 3 }
204 dpavlin 13
205     # dump formatted output in <html>
206     if ($html) {
207     $xml .= xmlify("html",$html);
208     }
209    
210 dpavlin 3 if ($xml) {
211 dpavlin 10 #print STDERR "x: $xml\n";
212 dpavlin 13 $xml .= $add_xml if ($add_xml);
213 dpavlin 10 return "<xml>\n$xml</xml>\n";
214 dpavlin 3 } else {
215     return;
216     }
217     }
218    
219     ##########################################################################
220    
221 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
222 dpavlin 1
223 dpavlin 13 foreach my $database ($cfg->Sections) {
224 dpavlin 1
225 dpavlin 13 my $isis_db = $cfg -> val($database, 'isis_db');
226     my $type = $cfg -> val($database, 'type');
227     my $add_xml = $cfg -> val($database, 'xml');
228 dpavlin 1
229 dpavlin 13 # read configuration for this type
230     $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
231 dpavlin 29 $isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET);
232 dpavlin 13 $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
233 dpavlin 29 $xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET);
234 dpavlin 1
235 dpavlin 3 my $db = OpenIsis::open( $isis_db );
236     if (0) {
237     # # FIX
238     # if (! $db ) {
239     print STDERR "WARNING: can't open '$isis_db'\n";
240     next ;
241     }
242    
243 dpavlin 1 my $max_rowid = OpenIsis::maxRowid( $db );
244    
245 dpavlin 3 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
246 dpavlin 1
247 dpavlin 13 my $path = $database; # was $isis_db
248 dpavlin 10
249 dpavlin 3 my $last_p = 0;
250    
251 dpavlin 10 # { my $row_id = 4514;
252 dpavlin 3 # FIX
253 dpavlin 1 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
254     my $row = OpenIsis::read( $db, $row_id );
255 dpavlin 3 if ($row && $row->{mfn}) {
256 dpavlin 10 #print STDERR "mfn: ",$row->{mfn},"\n";
257 dpavlin 3 # output current process indicator
258     my $p = int($row->{mfn} * 100 / $max_rowid);
259     if ($p != $last_p) {
260     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
261     $last_p = $p;
262     }
263    
264 dpavlin 13 if (my $xml = isis2xml($row,$add_xml)) {
265 dpavlin 10 #print STDERR "--ret-->$xml\n";
266     print "Path-Name: $path#".int($row->{mfn})."\n";
267     print "Content-Length: ".(length($xml)+1)."\n";
268     print "Document-Type: XML\n\n$xml\n";
269 dpavlin 3 }
270 dpavlin 1 }
271     }
272 dpavlin 3 print STDERR "\n";
273 dpavlin 1 }
274 dpavlin 3
275 dpavlin 10 # call this to commit index
276     $index->close;
277 dpavlin 3
278     1;
279     __END__
280     ##########################################################################
281    
282     =head1 NAME
283    
284     isis2xml.pl - read isis file and dump XML
285    
286     =head1 DESCRIPTION
287    
288     This command will read ISIS data file using OpenIsis perl module and
289     create XML file for usage with I<SWISH-E>
290     indexer. Dispite it's name, this script B<isn't general xml generator>
291     from isis files (isis allready has something like that). Output of this
292     script is tailor-made for SWISH-E.
293    
294     =head1 AUTHOR
295    
296     Dobrica Pavlinusic <dpavlin@rot13.org>
297    
298     =head1 COPYRIGHT
299    
300     GNU Public License (GPL) v2 or later
301    
302     =head1 SEE ALSO
303    
304     SWISH-E web site at http://www.swish-e.org
305    
306     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.12
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26