8 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
16 $config_file =~ s/\.pl$/.conf/;
17 die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
21 use index_DBI; # there is no other, right now ;-)
23 my $index = new index_DBI(); # open index
29 # -m multiple directories
33 getopts('d:m:qs', \%opts);
37 Text::Iconv->raise_error(1); # Conversion errors raise exceptions
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';
43 my $utf2cp = Text::Iconv->new('UTF-8',$codepage);
44 # this function will convert data from XML files to local encoding
46 return $utf2cp->convert($_[0]);
49 # decode isis import codepage
52 # outgoing xml must be in UTF-8
53 my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
60 my $add_xml = shift @_;
66 my $html = ""; # html formatted display output
68 my %field_usage; # counter for usage of each field
70 # sort subrouting using order="" attribute
72 return 0 if (! $config->{indexer}->{$a}->{order});
73 return 0 if (! $config->{indexer}->{$b}->{order});
75 return $config->{indexer}->{$a}->{order} <=>
76 $config->{indexer}->{$b}->{order} ;
79 foreach my $field (sort by_order keys %{$config->{indexer}}) {
83 $field_usage{$field}++;
86 my $display_data = "";
91 foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
93 my $format = x($x->{content});
94 my $delimiter = x($x->{delimiter}) || ' ';
96 my $isis_i = 0; # isis repeatable offset
98 my ($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");
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 $line_delimiter = '<br/>';
110 # init vars so that we go into while...
111 ($swish,$display) = (1,1);
113 while ($swish || $display) {
114 ($swish,$display) = parse_format($format,$row,$isis_i++,$isis2cp);
116 # filter="name" ; filter this field through
118 my $filter = $x->{filter};
120 require "filter/".$filter.".pm";
122 # type="swish" ; field for swish
126 $swish_data .= join(" ",&$filter($swish));
127 print STDERR "#### $swish_data\n";
129 $swish_data .= $swish;
133 # type="display" ; field for display
134 if ($d && $display) {
135 if ($line_delimiter && $display_data) {
136 $display_data .= $line_delimiter;
137 undef $line_delimiter;
141 $display_data .= join($delimiter,&$filter($display));
144 $display_data .= $delimiter.$display;
146 $display_data .= $display;
151 # type="index" ; insert into index
152 if ($i && $display) {
153 my $index_data = $display;
156 foreach my $d (&$filter($index_data)) {
157 $index->insert($field, $d, $db_dir);
160 $index->insert($field, $index_data, $db_dir);
169 if ($field eq "headline") {
170 $xml .= xmlify("headline", $display_data);
173 # find field name (signular, plural)
175 if ($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}."#-#";
182 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
185 $html .= x($field_name);
187 $html .= $display_data."###\n";
191 # remove extra spaces
192 $swish_data =~ s/ +/ /g;
193 $swish_data =~ s/ +$//g;
195 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
201 # dump formatted output in <html>
203 $xml .= xmlify("html",$html);
207 $xml .= $add_xml if ($add_xml);
208 return "<xml>\n$xml</xml>\n";
214 ##########################################################################
216 my $cfg = new Config::IniFiles( -file => $config_file );
218 foreach my $database ($cfg->Sections) {
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
224 $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
226 $isis2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
228 my $db = OpenIsis::open( $isis_db );
232 print STDERR "WARNING: can't open '$isis_db'\n";
236 my $max_rowid = OpenIsis::maxRowid( $db );
238 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
240 my $path = $database; # was $isis_db
244 # { my $row_id = 4514;
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 / $max_rowid);
252 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
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";
267 # call this to commit index
272 ##########################################################################
276 isis2xml.pl - read isis file and dump XML
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.
288 Dobrica Pavlinusic <dpavlin@rot13.org>
292 GNU Public License (GPL) v2 or later
296 SWISH-E web site at http://www.swish-e.org