/[webpac]/trunk2/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 /trunk2/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 17 - (hide annotations)
Sat Feb 22 21:33:04 2003 UTC (16 years, 9 months ago) by dpavlin
Original Path: trunk/all2xml.pl
File MIME type: text/plain
File size: 6454 byte(s)
fix index insertion

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 $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
44     #my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
45     my $isis_codepage;
46     my $index_codepage;
47 dpavlin 10 my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
48 dpavlin 13 my $xml_codepage;
49 dpavlin 10
50 dpavlin 3 sub isis2xml {
51    
52 dpavlin 10 use xmlify;
53    
54 dpavlin 3 my $row = shift @_;
55 dpavlin 13 my $add_xml = shift @_;
56 dpavlin 3
57     my $xml;
58    
59 dpavlin 10 use parse_format;
60 dpavlin 3
61 dpavlin 13 my $html = ""; # html formatted display output
62 dpavlin 10
63 dpavlin 13 my %field_usage; # counter for usage of each field
64    
65 dpavlin 3 foreach my $field (keys %{$config->{indexer}}) {
66    
67 dpavlin 13 $field_usage{$field}++;
68    
69 dpavlin 10 my $swish_data = "";
70 dpavlin 3 my $display_data = "";
71    
72     foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
73    
74     my $format = $x->{content};
75 dpavlin 10 my ($s,$d,$i) = (1,1,0); # swish, display default
76 dpavlin 9 $s = 0 if (lc($x->{type}) eq "display");
77     $d = 0 if (lc($x->{type}) eq "swish");
78     ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
79     #print STDERR "## s: $s d: $d i: $i ## $format ##\n";
80 dpavlin 3
81 dpavlin 10 $format = $cludge_codepage->convert($format);
82     my ($swish,$display) = parse_format($format,$row);
83     #print STDERR "s: $swish\nd: $display\n" if ($swish);
84 dpavlin 3
85 dpavlin 10 #print STDERR "swish: $swish<-- display: $display<--\n";
86     # FIX: this is ugly, UGLY, cludge: OpenIsis return
87     # UTF8 encoding of strings, but as if source charset
88     # is ISO8859-1 and not some other. This breaks our
89     # isis character encoding, so we convert it first
90     # back to ISO8859-1 (which can actually be different
91     # encoding in isis)
92    
93     $swish_data .= $swish if ($s && $swish);
94     $display_data .= $display if ($d && $display);
95 dpavlin 17
96     # insert into index
97     if ($i && $display) {
98     my $index_data = $index_codepage->convert($display) || $display;
99     $index->insert($field, $index_data, $db_dir);
100     }
101 dpavlin 3 }
102 dpavlin 9
103 dpavlin 13
104 dpavlin 10 #print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
105     if ($display_data) {
106 dpavlin 13 $display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
107     # FIX: this is removed and replaced by html tag.
108     #$xml .= xmlify($field."_display", $display_data);
109    
110     if ($field eq "headline") {
111     $xml .= xmlify("headline", $display_data);
112     } else {
113    
114     # find field name (signular, plural)
115     my $field_name = "";
116     if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
117     $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
118     } elsif ($config->{indexer}->{$field}->{name_plural}) {
119     $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
120     } else {
121     $field_name = $config->{indexer}->{$field}->{name}."#-#";
122     }
123     if ($field_name) {
124     $html .= $xml_codepage->convert($field_name);
125     }
126     $html .= $display_data."###\n";
127     }
128 dpavlin 10 }
129     if ($swish_data) {
130     my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
131     $swish_data = $i->convert($swish_data);
132     $xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
133     #$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
134     #$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
135     }
136    
137 dpavlin 9
138 dpavlin 3 }
139 dpavlin 13
140     # dump formatted output in <html>
141     if ($html) {
142     $xml .= xmlify("html",$html);
143     }
144    
145 dpavlin 3 if ($xml) {
146 dpavlin 10 #print STDERR "x: $xml\n";
147 dpavlin 13 $xml .= $add_xml if ($add_xml);
148 dpavlin 10 return "<xml>\n$xml</xml>\n";
149 dpavlin 3 } else {
150     return;
151     }
152     }
153    
154     ##########################################################################
155    
156 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
157 dpavlin 1
158 dpavlin 13 foreach my $database ($cfg->Sections) {
159 dpavlin 1
160 dpavlin 13 my $isis_db = $cfg -> val($database, 'isis_db');
161     my $type = $cfg -> val($database, 'type');
162     my $add_xml = $cfg -> val($database, 'xml');
163 dpavlin 1
164 dpavlin 13 # read configuration for this type
165     $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
166     $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
167     $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
168     $xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8');
169 dpavlin 1
170 dpavlin 3 my $db = OpenIsis::open( $isis_db );
171     if (0) {
172     # # FIX
173     # if (! $db ) {
174     print STDERR "WARNING: can't open '$isis_db'\n";
175     next ;
176     }
177    
178 dpavlin 1 my $max_rowid = OpenIsis::maxRowid( $db );
179    
180 dpavlin 3 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
181 dpavlin 1
182 dpavlin 13 my $path = $database; # was $isis_db
183 dpavlin 10
184 dpavlin 3 my $last_p = 0;
185    
186 dpavlin 10 # { my $row_id = 4514;
187 dpavlin 3 # FIX
188 dpavlin 1 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
189     my $row = OpenIsis::read( $db, $row_id );
190 dpavlin 3 if ($row && $row->{mfn}) {
191 dpavlin 10 #print STDERR "mfn: ",$row->{mfn},"\n";
192 dpavlin 3 # output current process indicator
193     my $p = int($row->{mfn} * 100 / $max_rowid);
194     if ($p != $last_p) {
195     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
196     $last_p = $p;
197     }
198    
199 dpavlin 13 if (my $xml = isis2xml($row,$add_xml)) {
200 dpavlin 10 #print STDERR "--ret-->$xml\n";
201     print "Path-Name: $path#".int($row->{mfn})."\n";
202     print "Content-Length: ".(length($xml)+1)."\n";
203     print "Document-Type: XML\n\n$xml\n";
204 dpavlin 3 }
205 dpavlin 1 }
206     }
207 dpavlin 3 print STDERR "\n";
208 dpavlin 1 }
209 dpavlin 3
210 dpavlin 10 # call this to commit index
211     $index->close;
212 dpavlin 3
213     1;
214     __END__
215     ##########################################################################
216    
217     =head1 NAME
218    
219     isis2xml.pl - read isis file and dump XML
220    
221     =head1 DESCRIPTION
222    
223     This command will read ISIS data file using OpenIsis perl module and
224     create XML file for usage with I<SWISH-E>
225     indexer. Dispite it's name, this script B<isn't general xml generator>
226     from isis files (isis allready has something like that). Output of this
227     script is tailor-made for SWISH-E.
228    
229     =head1 AUTHOR
230    
231     Dobrica Pavlinusic <dpavlin@rot13.org>
232    
233     =head1 COPYRIGHT
234    
235     GNU Public License (GPL) v2 or later
236    
237     =head1 SEE ALSO
238    
239     SWISH-E web site at http://www.swish-e.org
240    
241     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.9
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26