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; # default DBI module for index
22 use index_DBI_cache; # faster DBI module using memory cache
29 # -m multiple directories
33 getopts('d:m:qs', \%opts);
35 my $path; # this is name of database
37 Text::Iconv->raise_error(0); # Conversion errors don't 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/excel or other import codepage
52 # outgoing xml must be in UTF-8
53 my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
55 # mapping between data type and tag which specify
69 my $add_xml = shift @_;
70 # needed to read values from configuration file
72 my $database = shift @_;
78 my $html = ""; # html formatted display output
80 my %field_usage; # counter for usage of each field
82 # sort subrouting using order="" attribute
84 return 0 if (! $config->{indexer}->{$a}->{order});
85 return 0 if (! $config->{indexer}->{$b}->{order});
87 return $config->{indexer}->{$a}->{order} <=>
88 $config->{indexer}->{$b}->{order} ;
91 foreach my $field (sort by_order keys %{$config->{indexer}}) {
95 $field_usage{$field}++;
98 my $display_data = "";
101 my ($swish,$display);
103 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
104 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
106 my $format = x($x->{content});
107 my $delimiter = x($x->{delimiter}) || ' ';
109 my $repeat_off = 0; # repeatable offset
111 my ($s,$d,$i) = (1,1,0); # swish, display default
112 $s = 0 if (lc($x->{type}) eq "display");
113 $d = 0 if (lc($x->{type}) eq "swish");
114 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
116 # what will separate last line from this one?
117 if ($display_data && $x->{append} && $x->{append} eq "1") {
118 $line_delimiter = ' ';
119 } elsif ($display_data) {
120 $line_delimiter = '<br/>';
123 # init vars so that we go into while...
124 ($swish,$display) = (1,1);
126 if ($swish || $display) {
127 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
128 # filter="name" ; filter this field through
130 my $filter = $x->{filter};
132 require "filter/".$filter.".pm";
134 # type="swish" ; field for swish
138 $swish_data .= join(" ",&$filter($swish));
140 $swish_data .= $swish;
144 # type="display" ; field for display
145 if ($d && $display) {
146 if ($line_delimiter && $display_data) {
147 $display_data .= $line_delimiter;
148 undef $line_delimiter;
152 $display_data .= join($delimiter,&$filter($display));
155 $display_data .= $delimiter.$display;
157 $display_data .= $display;
162 # type="index" ; insert into index
163 if ($i && $display) {
164 my $index_data = $display;
167 foreach my $d (&$filter($index_data)) {
168 $index->insert($field, $d, $path);
171 $index->insert($field, $index_data, $path);
177 # now try to parse variables from configuration file
178 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
180 my $delimiter = x($x->{delimiter}) || ' ';
181 my $val = $cfg->val($database, x($x->{content}));
183 my ($s,$d,$i) = (1,1,0); # swish, display default
184 $s = 0 if (lc($x->{type}) eq "display");
185 $d = 0 if (lc($x->{type}) eq "swish");
186 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
189 $display_data .= $delimiter.$val if ($d);
190 $swish_data .= $val if ($s);
191 $index->insert($field, $val, $path) if ($i);
199 if ($field eq "headline") {
200 $xml .= xmlify("headline", $display_data);
203 # find field name (signular, plural)
205 if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
206 $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
207 } elsif ($config->{indexer}->{$field}->{name_plural}) {
208 $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
209 } elsif ($config->{indexer}->{$field}->{name}) {
210 $field_name = $config->{indexer}->{$field}->{name}."#-#";
212 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
215 $html .= x($field_name);
217 $html .= $display_data."###\n";
221 # remove extra spaces
222 $swish_data =~ s/ +/ /g;
223 $swish_data =~ s/ +$//g;
225 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
231 # dump formatted output in <html>
233 $xml .= xmlify("html",$html);
237 $xml .= $add_xml if ($add_xml);
238 return "<xml>\n$xml</xml>\n";
244 ##########################################################################
246 # read configuration for this script
247 my $cfg = new Config::IniFiles( -file => $config_file );
249 # read global.conf configuration
250 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
253 $index = new index_DBI(
254 $cfg_global->val('global', 'dbi_dbd'),
255 $cfg_global->val('global', 'dbi_dsn'),
256 $cfg_global->val('global', 'dbi_user'),
257 $cfg_global->val('global', 'dbi_passwd') || '',
260 foreach my $database ($cfg->Sections) {
262 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
263 my $add_xml = $cfg -> val($database, 'xml'); # optional
265 print STDERR "reading ./import_xml/$type.xml\n";
267 # extract just type basic
268 my $type_base = $type;
269 $type_base =~ s/_.+$//g;
271 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1);
273 # output current progress indicator
276 #return if (! $opts{q}); # FIXME
278 my $total = shift || 1;
279 my $p = int($current * 100 / $total);
281 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
287 print STDERR "using: $type...\n";
289 if ($type_base eq "isis") {
291 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
293 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
294 my $db = OpenIsis::open( $isis_db );
296 my $max_rowid = OpenIsis::maxRowid( $db );
298 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
300 my $path = $database;
302 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
303 my $row = OpenIsis::read( $db, $row_id );
304 if ($row && $row->{mfn}) {
306 progress($row->{mfn}, $max_rowid);
308 my $swishpath = $path."#".int($row->{mfn});
310 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
311 $xml = $cp2utf->convert($xml);
312 use bytes; # as opposed to chars
313 print "Path-Name: $swishpath\n";
314 print "Content-Length: ".(length($xml)+1)."\n";
315 print "Document-Type: XML\n\n$xml\n";
321 } elsif ($type_base eq "excel") {
322 use Spreadsheet::ParseExcel;
323 use Spreadsheet::ParseExcel::Utility qw(int2col);
325 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
326 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
328 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
329 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
331 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
334 foreach my $oWks (@{$oBook->{Worksheet}}) {
335 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
336 last if ($oWks->{Name} eq $sheet);
340 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
342 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
343 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
344 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
346 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
348 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
349 my $cell = $oWorksheet->{Cells}[$iR][$iC];
351 $row->{int2col($iC)} = $cell->Value;
355 progress($iR, $end_row);
357 # print "row[$iR/$end_row] ";
358 # foreach (keys %{$row}) {
359 # print "$_: ",$row->{$_},"\t";
363 my $swishpath = $database."#".$iR;
367 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
368 $xml = $cp2utf->convert($xml);
369 use bytes; # as opposed to chars
370 print "Path-Name: $swishpath\n";
371 print "Content-Length: ".(length($xml)+1)."\n";
372 print "Document-Type: XML\n\n$xml\n";
375 } elsif ($type_base eq "marc") {
379 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
380 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
382 # optional argument is format
383 my $format = x($config->{format}) || 'usmarc';
385 my %id_stored; # to aviod duplicates
387 print STDERR "Reading MARC file '$marc_file'\n";
390 my $nr = $marc->openmarc({
391 file=>$marc_file, format=>$format
392 }) || die "Can't open MARC file '$marc_file'";
394 my $i=0; # record nr.
400 while ($marc->nextmarc(1)) {
403 progress($i, $max_i);
405 $inc = -$inc if ($i > $max_i || $i < 0);
407 my $swishpath = $database."#".$i;
409 if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
410 $xml = $cp2utf->convert($xml);
411 use bytes; # as opposed to chars
412 print "Path-Name: $swishpath\n";
413 print "Content-Length: ".(length($xml)+1)."\n";
414 print "Document-Type: XML\n\n$xml\n";
420 # call this to commit index
425 ##########################################################################
429 isis2xml.pl - read isis file and dump XML
433 This command will read ISIS data file using OpenIsis perl module and
434 create XML file for usage with I<SWISH-E>
435 indexer. Dispite it's name, this script B<isn't general xml generator>
436 from isis files (isis allready has something like that). Output of this
437 script is tailor-made for SWISH-E.
441 Dobrica Pavlinusic <dpavlin@rot13.org>
445 GNU Public License (GPL) v2 or later
449 SWISH-E web site at http://www.swish-e.org