12 use Fcntl; # for O_RDWR
18 $config_file =~ s/\.pl$/.conf/;
19 $config_file = $ARGV[0] if ($ARGV[0] && -f $ARGV[0]);
20 die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
24 #use index_DBI; # default DBI module for index
25 use index_DBI_cache; # faster DBI module using memory cache
32 # -m multiple directories
36 getopts('d:m:qs', \%opts);
38 my $path; # this is name of database
40 Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions
42 # this is encoding of all files on disk, including import_xml/*.xml file and
43 # filter/*.pm files! It will be used to store strings in perl internally!
44 my $codepage = 'ISO-8859-2';
46 my $utf2cp = Text::Iconv->new('UTF-8',$codepage);
47 # this function will convert data from XML files to local encoding
49 return $utf2cp->convert($_[0]);
52 # decode isis/excel or other import codepage
55 # outgoing xml must be in UTF-8
56 my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
58 # mapping between data type and tag which specify
67 my $cache; # for cacheing
69 # lookup hash (tied to file)
71 # this option will cache all lookup entries in memory.
72 # if you are tight on memory, turn this off
73 my $use_lhash_cache = 1;
75 my $last_field_name; # cache to prevent repeated fields
83 my $add_xml = shift @_;
84 # needed to read values from configuration file
86 my $database = shift @_;
92 my $html = ""; # html formatted display output
94 my %field_usage; # counter for usage of each field
96 # sort subrouting using order="" attribute
98 my $va = $config->{indexer}->{$a}->{order} ||
99 $config->{indexer}->{$a};
100 my $vb = $config->{indexer}->{$b}->{order} ||
101 $config->{indexer}->{$b};
107 if ($cache->{tags_by_order}) {
108 @sorted_tags = @{$cache->{tags_by_order}};
110 @sorted_tags = sort by_order keys %{$config->{indexer}};
111 $cache->{tags_by_order} = \@sorted_tags;
117 # cache for field in pages
118 delete $cache->{display_data};
119 delete $cache->{swish_data};
120 delete $cache->{swish_exact_data};
121 delete $cache->{index_data};
122 delete $cache->{index_delimiter};
123 my @page_fields; # names of fields
126 # subs used to produce output
128 sub get_field_name($$$) {
129 my ($config,$field,$field_usage) = @_;
131 # find field name (signular, plural)
133 if ($config->{indexer}->{$field}->{name_singular} && $field_usage == 1) {
134 $field_name = $config->{indexer}->{$field}->{name_singular};
135 } elsif ($config->{indexer}->{$field}->{name_plural}) {
136 $field_name = $config->{indexer}->{$field}->{name_plural};
137 } elsif ($config->{indexer}->{$field}->{name}) {
138 $field_name = $config->{indexer}->{$field}->{name};
140 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
144 $field_name = x($field_name);
145 if (! $last_field_name) {
146 $last_field_name = $field_name;
147 return $last_field_name;
148 } elsif ($field_name ne $last_field_name) {
149 $last_field_name = $field_name;
150 return $last_field_name;
156 # init variables for different types
157 sub init_visible_type($) {
160 # swish, swish_exact, display, index, index_lookup
161 # swish and display defaults
162 my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
163 if (lc($type) eq "display") {
165 } elsif (lc($type) eq "swish") {
167 } elsif (lc($type) eq "index") {
168 ($s,$se,$d,$i) = (0,1,0,1);
169 } elsif (lc($type) eq "swish_exact") {
170 ($s,$se,$d,$i) = (0,1,0,0);
171 } elsif (lc($type) =~ /^lookup/) {
172 ($s,$se,$d,$i,$il) = (0,1,0,0,1);
174 print STDERR "WARNING: unknown type: $type\n";
176 return ($s,$se,$d,$i,$il);
183 # <delimiter>, </delimiter>
184 # <value>200a</value>
189 # <tag delimiter=", ">200a</tag>
191 # but without loosing spaces in delimiter (becasue
192 # new XML::Simple strips spaces in attribute values
193 # as defined in XML specification)
198 if (defined $x->{value}) {
199 my ($v,$d) = ($x->{value}->{content}, $x->{delimiter}->{content});
201 delete $x->{delimiter};
203 $x->{delimiter} = $d;
208 # begin real work: go field by field
209 foreach my $field (@sorted_tags) {
212 $field_usage{$field}++;
215 my $swish_exact_data = "";
216 my $display_data = "";
220 my ($swish,$display);
222 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
224 # is this field page-by-page?
225 my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page};
226 push @page_fields,$field if ($iterate_by_page);
228 # default line_delimiter if using
229 my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>';
230 $cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter};
232 my $format_name = $config->{indexer}->{$field}->{format_name};
233 my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter};
234 if ($format_name && $format_delimiter) {
235 $cache->{format}->{$field}->{format_name} = $format_name;
236 $cache->{format}->{$field}->{format_delimiter} = $format_delimiter;
239 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
243 my $format = x($x->{content});
244 my $delimiter = x($x->{delimiter}) || ' ';
246 my $repeat_off = 0; # init repeatable offset
248 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
250 # what will separate last line from this one?
251 if ($display_data && $x->{append}) {
252 $line_delimiter = $delimiter;
253 } elsif ($display_data) {
254 $line_delimiter = '<br/>';
257 # init vars so that we go into while...
258 ($swish,$display) = (1,1);
260 # placeholder for all repeatable entries for index
263 my $x = shift || die "mkformat needs tag reference";
264 my $data = shift || return;
265 my $format_name = x($x->{format_name}) || return $data;
266 my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
267 my $format_delimiter = x($x->{format_delimiter});
269 if ($format_delimiter) {
270 @data = split(/$format_delimiter/,$data);
276 my $nr = scalar $fmt =~ s/%s/%s/g;
277 if (($#data+1) == $nr) {
278 return sprintf($fmt,@data);
280 #print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
284 print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
288 # while because of repeatable fields
289 while ($swish || $display) {
290 my $page = $repeat_off;
291 $page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0));
292 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
293 if ($repeat_off > 1000) {
294 print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
298 # is this field is lookup?
299 if ($display && $x->{lookup}) {
300 my $null = "<!-- null -->";
301 if ($use_lhash_cache) {
302 if (!defined($cache->{lhash}->{$display})) {
303 my $new_display = $lhash{$display};
304 if (defined($new_display)) {
305 #print STDERR "lookup cache store '$display' = '$new_display'\n";
306 $display = $new_display;
307 $cache->{lhash}->{$display} = $new_display;
309 # print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
311 $cache->{lhash}->{$display} = $null;
314 $display = $cache->{lhash}->{$display};
317 $display = $lhash{$display} || $null;
321 # filter="name" ; filter this field through
323 my $filter = $x->{filter};
324 if ($filter && !$cache->{filter_loaded}->{$filter}) {
325 require "filter/".$filter.".pm";
326 $cache->{filter_loaded}->{$filter}++;
328 # type="swish" ; field for swish
331 if ($filter && ($s || $se)) {
333 $tmp = join(" ",&$filter($tmp)) if ($s || $se);
336 $swish_data .= $tmp if ($s && $tmp);
337 $swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se);
340 # type="display" ; field for display
341 if ($d && $display) {
342 my $ldel = $delimiter;
343 if ($line_delimiter && $display_data) {
344 $ldel = $line_delimiter;
349 foreach my $tmp (&$filter($display)) {
350 my $tmp2 = mkformat($x,$tmp);
351 push @arr,$tmp2 if ($tmp2);
353 $display_data .= $ldel if ($display_data && @arr);
354 $display_data .= join($delimiter,@arr);
356 $display_data .= $ldel if ($display_data);
357 my $tmp = mkformat($x,$display);
358 $display_data .= $tmp if ($tmp);
362 # type="index" ; insert into index
364 if ($i && $display) {
365 $idisplay = $display;
368 $idisplay = &$filter($idisplay);
370 push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
373 # store fields in lookup
374 if ($il && $display) {
375 if (lc($x->{type}) eq "lookup_key") {
377 print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
381 $lookup_key = &$filter($display);
383 $lookup_key = $display;
386 } elsif (lc($x->{type}) eq "lookup_val") {
390 $lhash{$lookup_key} = &$filter($display);
392 $lhash{$lookup_key} = $display;
395 print STDERR "WARNING: no lookup_key defined for '$display'?";
401 # store data for page-by-page repeatable fields
402 if ($iterate_by_page) {
403 sub iterate_fld($$$$$$) {
404 my ($cache,$what,$field,$page,$data,$append) = @_;
407 my $ldel = $page_line_delimiter;
408 $ldel = " " if ($append);
409 #print STDERR "line delimiter: ",Dumper($ldel) if ($ldel);
410 if (! $cache->{$what}->{$field}->[$page]) {
411 $cache->{$what}->{$field}->[$page] = $data;
413 $cache->{$what}->{$field}->[$page] .= $ldel.$data;
418 iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append});
422 iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append});
425 if ($swish_exact_data) {
426 iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append});
427 $swish_exact_data = "";
431 my $ldel=$page_line_delimiter;
433 if ($cache->{index_data}->{$field}->[$page]) {
435 @index_data = @{$cache->{index_data}->{$field}->[$page]};
439 $index_data[$#index_data] .= $idisplay;
441 push @index_data, $idisplay;
444 push @index_data, $idisplay;
447 @{$cache->{index_data}->{$field}->[$page]} = @index_data;
452 if (! $iterate_by_page) {
453 my $idel = $x->{index_delimiter};
455 foreach my $tmp (@index_data) {
457 if ($idel && $tmp =~ m/$idel/) {
458 ($i,$d) = split(/$idel/,$tmp);
460 $index->insert($field, $i, $d, $path);
466 # now try to parse variables from configuration file
467 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
471 my $delimiter = x($x->{delimiter}) || ' ';
472 my $val = $cfg->val($database, x($x->{content}));
474 # FIXME index_lookup is not supported!
475 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
478 $display_data .= $delimiter.$val if ($d);
479 $swish_data .= " ".$val if ($s);
480 $index->insert($field, $val, $path) if ($i);
483 if ($iterate_by_page) {
484 # FIXME data from config tag will appear just
488 $cache->{display_data}->{$field}->[$page] = $display_data;
492 $cache->{swish_data}->{$field}->[$page] = $swish_data;
495 if ($swish_exact_data) {
496 $cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data;
497 $swish_exact_data = "";
502 # save data page-by-page
503 foreach my $field (@page_fields) {
504 my $nr_pages = $page_max{$field} || next;
505 #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n";
506 #print STDERR Dumper($cache->{display_data});
507 for (my $page=0; $page <= $nr_pages; $page++) {
509 if ($cache->{format}->{$field}) {
510 my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
511 $display_data=$tmp if ($tmp);
513 $display_data = $cache->{display_data}->{$field}->[$page];
515 if ($display_data) { # default
516 if ($field eq "headline") {
517 $xml .= xmlify("headline", $display_data);
520 # fallback to empty field name if needed
521 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
522 $html .= "#-#".$display_data."###\n";
526 my $swish_data = $cache->{swish_data}->{$field}->[$page];
528 # remove extra spaces
529 $swish_data =~ s/ +/ /g;
530 $swish_data =~ s/ +$//g;
532 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
535 my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page];
536 if ($swish_exact_data) {
537 $swish_exact_data =~ s/ +/ /g;
538 $swish_exact_data =~ s/ +$//g;
540 # add delimiters before and after word.
541 # That is required to produce exact match
542 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
545 my $idel = $cache->{index_delimiter}->{$field};
546 foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) {
549 if ($idel && $tmp =~ m/$idel/) {
550 ($i,$d) = split(/$idel/,$tmp);
552 $index->insert($field, $i, $d, $path);
553 #print STDERR "index [$idel] $field: $i --> $d [$path]\n";
559 if (! $iterate_by_page) {
561 if ($field eq "headline") {
562 $xml .= xmlify("headline", $display_data);
565 # fallback to empty field name if needed
566 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
567 $html .= "#-#".$display_data."###\n";
571 # remove extra spaces
572 $swish_data =~ s/ +/ /g;
573 $swish_data =~ s/ +$//g;
575 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
578 if ($swish_exact_data) {
579 $swish_exact_data =~ s/ +/ /g;
580 $swish_exact_data =~ s/ +$//g;
582 # add delimiters before and after word.
583 # That is required to produce exact match
584 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
589 # dump formatted output in <html>
591 #$xml .= xmlify("html",$html);
592 $xml .= "<html><![CDATA[ $html ]]></html>";
596 $xml .= $add_xml if ($add_xml);
597 return "<xml>\n$xml</xml>\n";
603 ##########################################################################
605 # read configuration for this script
606 my $cfg = new Config::IniFiles( -file => $config_file );
608 # read global.conf configuration
609 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
612 $index = new index_DBI(
613 $cfg_global->val('global', 'dbi_dbd'),
614 $cfg_global->val('global', 'dbi_dsn'),
615 $cfg_global->val('global', 'dbi_user'),
616 $cfg_global->val('global', 'dbi_passwd') || '',
619 my $show_progress = $cfg_global->val('global', 'show_progress');
621 my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter');
622 if ($my_unac_filter) {
623 print STDERR "using $my_unac_filter to filter characters for search\n";
624 require $my_unac_filter;
626 print STDERR "### fallback to default my_unac_string!\n";
628 sub main::my_unac_string($$) {
629 my ($charset, $string) = (@_);
635 foreach my $database ($cfg->Sections) {
637 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
638 my $add_xml = $cfg -> val($database, 'xml'); # optional
640 # create new lookup file
641 my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
643 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
644 if (! -e $lookup_file) {
645 open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!";
648 tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
649 print STDERR "creating lookup file '$lookup_file'\n";
650 # delete memory cache for lookup file
651 delete $cache->{lhash};
654 # open existing lookup file
655 $lookup_file = $cfg -> val($database, 'lookup_open'); # optional
657 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644;
658 tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644;
659 print STDERR "opening lookup file '$lookup_file'\n";
662 print STDERR "reading ./import_xml/$type.xml\n";
664 # extract just type basic
665 my $type_base = $type;
666 $type_base =~ s/_.+$//g;
668 $config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 );
670 # helper for progress bar
675 my ($ss,$mm,$hh) = gmtime($t);
676 $out .= "${hh}h" if ($hh);
677 $out .= sprintf("%02d:%02d", $mm,$ss);
678 $out .= " " if ($hh == 0);
682 # output current progress indicator
684 my $start_t = time();
686 return if (! $show_progress);
688 my $total = shift || 1;
689 my $p = int($current * 100 / $total);
691 my $rate = ($current / (time() - $start_t || 1));
692 my $eta = ($total-$current) / ($rate || 1);
693 printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta));
700 return if (! $show_progress);
701 my $current = shift @_;
703 my @ind = ('-','\\','|','/','-','\\','|','/', '-');
705 $last_p += $fake_dir;
706 $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
707 if ($last_p % 10 == 0) {
708 printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
713 print STDERR "using: $type...\n";
715 # erase cache for tags by order in this database
716 delete $cache->{tags_by_order};
718 if ($type_base eq "isis") {
720 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
722 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
723 my $db = new IsisDB( isisdb => $isis_db );
725 my $max_rowid = $db->{'maxmfn'} || die "can't find maxmfn";
727 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
729 my $path = $database;
731 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
732 my $row = $db->to_hash( $row_id );
735 $row->{mfn} = $row_id;
736 $row->{record} = $db->{record};
738 progress($row->{mfn}, $max_rowid);
740 my $swishpath = $path."#".int($row->{mfn});
742 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
743 $xml = $cp2utf->convert($xml);
744 use bytes; # as opposed to chars
745 print "Path-Name: $swishpath\n";
746 print "Content-Length: ".(length($xml)+1)."\n";
747 print "Document-Type: XML\n\n$xml\n";
753 } elsif ($type_base eq "excel") {
754 require Spreadsheet::ParseExcel;
755 require Spreadsheet::ParseExcel::Utility;
756 import Spreadsheet::ParseExcel::Utility qw(int2col);
758 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
759 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
761 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
762 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
764 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
767 foreach my $oWks (@{$oBook->{Worksheet}}) {
768 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
769 last if ($oWks->{Name} eq $sheet);
773 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
775 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
776 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
777 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
779 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
781 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
782 my $cell = $oWorksheet->{Cells}[$iR][$iC];
784 # this conversion is a cludge.
785 # Files from Excell could have
786 # characters which don't fit into
787 # destination encoding.
788 $row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value;
792 progress($iR, $end_row);
794 # print "row[$iR/$end_row] ";
795 # foreach (keys %{$row}) {
796 # print "$_: ",$row->{$_},"\t";
800 my $swishpath = $database."#".$iR;
804 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
805 $xml = $cp2utf->convert($xml);
806 use bytes; # as opposed to chars
807 print "Path-Name: $swishpath\n";
808 print "Content-Length: ".(length($xml)+1)."\n";
809 print "Document-Type: XML\n\n$xml\n";
812 } elsif ($type_base eq "marc") {
814 require MARC::File::USMARC;
816 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
817 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
819 # optional argument is format
820 warn "marc_format is no longer used!" if ($config->{marc_format});
821 print STDERR "Reading MARC file '$marc_file'\n";
823 my $marc = MARC::File::USMARC->in( $marc_file )
824 || die "Can't open MARC file '$marc_file': ".$MARC::File::ERROR;
826 # count records in MARC file
828 my $filename = shift || die;
829 my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR;
831 while ($file->skip()) {
837 my $count = marc_count($marc_file) || warn "no records in '$marc_file'?";
841 while( my $rec = $marc->next() ) {
843 progress($i++,$count);
845 my $swishpath = $database."#".$i;
847 if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) {
848 $xml = $cp2utf->convert($xml);
849 use bytes; # as opposed to chars
850 print "Path-Name: $swishpath\n";
851 print "Content-Length: ".(length($xml)+1)."\n";
852 print "Document-Type: XML\n\n$xml\n";
858 } elsif ($type_base eq "feed") {
860 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
861 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
863 print STDERR "Reading feed from program '$prog'\n";
865 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
867 my $i=1; # record nr.
876 my $swishpath = $database."#".$i++;
878 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
879 $xml = $cp2utf->convert($xml);
880 use bytes; # as opposed to chars
881 print "Path-Name: $swishpath\n";
882 print "Content-Length: ".(length($xml)+1)."\n";
883 print "Document-Type: XML\n\n$xml\n";
890 $line = $1 if (s/^(\d+):\s*//);
891 $data->{$line++} = $_;
897 untie %lhash if (%lhash);
901 # call this to commit index
906 ##########################################################################
910 all2xml.pl - read various file formats and dump XML for SWISH-E
914 $ all2xml.pl [test.conf]
918 This command will read ISIS data file using OpenIsis perl module, MARC
919 records using MARC module and optionally Micro$oft Excel files to
920 create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
921 this script B<isn't general xml generator> from isis files (isis allready
922 has something like that). Output of this script is tailor-made for SWISH-E.
924 If no configuration file is specified, it will use default one called
929 Documentation is really lacking. However, in true Open Source spirit, source
930 is best documentation. I even made considerable effort to comment parts
931 which are not intuitively clear, so...
935 Dobrica Pavlinusic <dpavlin@rot13.org>
939 GNU Public License (GPL) v2 or later
943 SWISH-E web site at http://www.swish-e.org