/[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

Annotation of /trunk/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 32 - (hide annotations)
Sun Feb 23 07:53:01 2003 UTC (16 years, 6 months ago) by dpavlin
File MIME type: text/plain
File size: 8151 byte(s)
display fields using order="" attribute

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 32 # 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 dpavlin 29
71 dpavlin 32 return $config->{indexer}->{$a}->{order} <=>
72     $config->{indexer}->{$b}->{order} ;
73     }
74 dpavlin 3
75 dpavlin 32 foreach my $field (sort by_order keys %{$config->{indexer}}) {
76    
77 dpavlin 13 $field_usage{$field}++;
78    
79 dpavlin 10 my $swish_data = "";
80 dpavlin 3 my $display_data = "";
81 dpavlin 29 my $line_delimiter = "";
82 dpavlin 3
83     foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
84    
85     my $format = $x->{content};
86 dpavlin 29 my $delimiter = $x->{delimiter} || ' ';
87 dpavlin 3
88 dpavlin 20 # FIX: this is ugly, UGLY, cludge string is returned
89     # in UTF8 encoding , but as if source charset
90     # is ISO8859-1 and not some other. This break other
91     # character encodings, so we convert it first
92     # back to ISO8859-1
93 dpavlin 29 $format = $xml_codepage->convert($format);
94     $delimiter = $xml_codepage->convert($delimiter) if ($delimiter);
95 dpavlin 20
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     #print STDERR "## s: $s d: $d i: $i ## $format ##\n";
103    
104     $display_data .= $line_delimiter if ($display_data && $display_data !~ /$line_delimiter$/);
105    
106     my ($swish,$display) = (1,1);
107    
108     while ($swish || $display) {
109     ($swish,$display) = parse_format($format,$row,$isis_i++);
110 dpavlin 10 #print STDERR "s: $swish\nd: $display\n" if ($swish);
111 dpavlin 3
112 dpavlin 10 #print STDERR "swish: $swish<-- display: $display<--\n";
113    
114 dpavlin 29 # filter="name" ; filter this field through
115     # filter/[name].pm
116     my $filter = $x->{filter};
117 dpavlin 20 if ($filter) {
118 dpavlin 29 require "filter/".$filter.".pm";
119 dpavlin 20 }
120 dpavlin 29 # type="swish" ; field for swish
121     if ($s && $swish) {
122     if ($filter) {
123 dpavlin 32 #print STDERR "using filter '$filter'\n";
124 dpavlin 29 no strict 'refs';
125     $swish_data .= join(" ",&$filter($swish));
126     } else {
127     $swish_data .= $swish;
128     }
129     }
130 dpavlin 17
131 dpavlin 29 # type="display" ; field for display
132     if ($d && $display) {
133     if ($filter) {
134     no strict 'refs';
135     $display_data .= join($delimiter,&$filter($display));
136     } else {
137     if ($display_data) {
138     $display_data .= $delimiter.$display;
139     } else {
140     $display_data .= $display;
141     }
142     }
143 dpavlin 20 }
144 dpavlin 29
145     # type="index" ; insert into index
146     if ($i && $display) {
147     my $index_data = $index_codepage->convert($display) || $display;
148     if ($filter) {
149     no strict 'refs';
150     foreach my $d (&$filter($index_data)) {
151     $index->insert($field, $d, $db_dir);
152     }
153     } else {
154     $index->insert($field, $index_data, $db_dir);
155 dpavlin 20 }
156     }
157 dpavlin 17 }
158 dpavlin 29 if ($x->{append}) {
159     $line_delimiter = ' ';
160     } else {
161     $line_delimiter = '<br/>';
162     }
163 dpavlin 3 }
164 dpavlin 9
165 dpavlin 13
166 dpavlin 10 #print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
167     if ($display_data) {
168 dpavlin 29
169     # remove last <br>
170     $display_data =~ s/$line_delimiter$//;
171    
172 dpavlin 13 $display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
173     # FIX: this is removed and replaced by html tag.
174     #$xml .= xmlify($field."_display", $display_data);
175    
176     if ($field eq "headline") {
177     $xml .= xmlify("headline", $display_data);
178     } else {
179    
180     # find field name (signular, plural)
181     my $field_name = "";
182     if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
183     $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
184     } elsif ($config->{indexer}->{$field}->{name_plural}) {
185     $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
186 dpavlin 20 } elsif ($config->{indexer}->{$field}->{name}) {
187     $field_name = $config->{indexer}->{$field}->{name}."#-#";
188 dpavlin 13 } else {
189 dpavlin 20 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
190 dpavlin 13 }
191     if ($field_name) {
192     $html .= $xml_codepage->convert($field_name);
193 dpavlin 29 # $html .= "-->".$field_name."<--";
194 dpavlin 13 }
195     $html .= $display_data."###\n";
196     }
197 dpavlin 10 }
198     if ($swish_data) {
199     my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
200 dpavlin 20 # remove extra spaces
201     $swish_data =~ s/ +/ /g;
202     $swish_data =~ s/ +$//g;
203    
204 dpavlin 10 $swish_data = $i->convert($swish_data);
205     $xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
206     #$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
207     #$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
208     }
209    
210 dpavlin 9
211 dpavlin 3 }
212 dpavlin 13
213     # dump formatted output in <html>
214     if ($html) {
215     $xml .= xmlify("html",$html);
216     }
217    
218 dpavlin 3 if ($xml) {
219 dpavlin 10 #print STDERR "x: $xml\n";
220 dpavlin 13 $xml .= $add_xml if ($add_xml);
221 dpavlin 10 return "<xml>\n$xml</xml>\n";
222 dpavlin 3 } else {
223     return;
224     }
225     }
226    
227     ##########################################################################
228    
229 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
230 dpavlin 1
231 dpavlin 13 foreach my $database ($cfg->Sections) {
232 dpavlin 1
233 dpavlin 13 my $isis_db = $cfg -> val($database, 'isis_db');
234     my $type = $cfg -> val($database, 'type');
235     my $add_xml = $cfg -> val($database, 'xml');
236 dpavlin 1
237 dpavlin 13 # read configuration for this type
238     $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
239 dpavlin 29 $isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET);
240 dpavlin 13 $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
241 dpavlin 29 $xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET);
242 dpavlin 1
243 dpavlin 3 my $db = OpenIsis::open( $isis_db );
244     if (0) {
245     # # FIX
246     # if (! $db ) {
247     print STDERR "WARNING: can't open '$isis_db'\n";
248     next ;
249     }
250    
251 dpavlin 1 my $max_rowid = OpenIsis::maxRowid( $db );
252    
253 dpavlin 3 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
254 dpavlin 1
255 dpavlin 13 my $path = $database; # was $isis_db
256 dpavlin 10
257 dpavlin 3 my $last_p = 0;
258    
259 dpavlin 10 # { my $row_id = 4514;
260 dpavlin 3 # FIX
261 dpavlin 1 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
262     my $row = OpenIsis::read( $db, $row_id );
263 dpavlin 3 if ($row && $row->{mfn}) {
264 dpavlin 10 #print STDERR "mfn: ",$row->{mfn},"\n";
265 dpavlin 3 # output current process indicator
266     my $p = int($row->{mfn} * 100 / $max_rowid);
267     if ($p != $last_p) {
268     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
269     $last_p = $p;
270     }
271    
272 dpavlin 13 if (my $xml = isis2xml($row,$add_xml)) {
273 dpavlin 10 #print STDERR "--ret-->$xml\n";
274     print "Path-Name: $path#".int($row->{mfn})."\n";
275     print "Content-Length: ".(length($xml)+1)."\n";
276     print "Document-Type: XML\n\n$xml\n";
277 dpavlin 3 }
278 dpavlin 1 }
279     }
280 dpavlin 3 print STDERR "\n";
281 dpavlin 1 }
282 dpavlin 3
283 dpavlin 10 # call this to commit index
284     $index->close;
285 dpavlin 3
286     1;
287     __END__
288     ##########################################################################
289    
290     =head1 NAME
291    
292     isis2xml.pl - read isis file and dump XML
293    
294     =head1 DESCRIPTION
295    
296     This command will read ISIS data file using OpenIsis perl module and
297     create XML file for usage with I<SWISH-E>
298     indexer. Dispite it's name, this script B<isn't general xml generator>
299     from isis files (isis allready has something like that). Output of this
300     script is tailor-made for SWISH-E.
301    
302     =head1 AUTHOR
303    
304     Dobrica Pavlinusic <dpavlin@rot13.org>
305    
306     =head1 COPYRIGHT
307    
308     GNU Public License (GPL) v2 or later
309    
310     =head1 SEE ALSO
311    
312     SWISH-E web site at http://www.swish-e.org
313    
314     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.13
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26