8 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
13 use Fcntl; # for O_RDWR
19 $config_file =~ s/\.pl$/.conf/;
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!";
143 if (! $last_field_name) {
144 $last_field_name = x($field_name);
145 return $last_field_name;
146 } elsif ($field_name ne $last_field_name) {
147 $last_field_name = x($field_name);
148 return $last_field_name;
154 # begin real work: go field by field
155 foreach my $field (@sorted_tags) {
158 $field_usage{$field}++;
161 my $swish_exact_data = "";
162 my $display_data = "";
166 my ($swish,$display);
168 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
170 # is this field page-by-page?
171 my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page};
172 push @page_fields,$field if ($iterate_by_page);
174 # default line_delimiter if using
175 my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>';
176 $cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter};
178 my $format_name = $config->{indexer}->{$field}->{format_name};
179 my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter};
180 if ($format_name && $format_delimiter) {
181 $cache->{format}->{$field}->{format_name} = $format_name;
182 $cache->{format}->{$field}->{format_delimiter} = $format_delimiter;
185 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
187 my $format = x($x->{content});
188 my $delimiter = x($x->{delimiter}) || ' ';
190 my $repeat_off = 0; # init repeatable offset
192 # swish, swish_exact, display, index, index_lookup
193 # swish and display defaults
194 my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
195 $s = 0 if (lc($x->{type}) eq "display");
196 $d = 0 if (lc($x->{type}) eq "swish");
197 ($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index");
198 ($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact");
199 ($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/);
201 # what will separate last line from this one?
202 if ($display_data && $x->{append}) {
203 $line_delimiter = ' ';
204 } elsif ($display_data) {
205 $line_delimiter = '<br/>';
208 # init vars so that we go into while...
209 ($swish,$display) = (1,1);
211 # placeholder for all repeatable entries for index
215 return if (! defined($data));
216 if ($data && $data =~ s/\s*eval{([^}]+)}\s*//) {
228 my $x = shift || die "mkformat needs tag reference";
229 my $data = shift || return;
230 my $format_name = x($x->{format_name}) || return chk_eval($data);
231 my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
232 my $format_delimiter = x($x->{format_delimiter});
234 if ($format_delimiter) {
235 @data = split(/$format_delimiter/,$data);
241 my $nr = scalar $fmt =~ s/%s/%s/g;
242 if (($#data+1) == $nr) {
243 return chk_eval(sprintf($fmt,@data));
245 #print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
246 return chk_eval($data);
249 print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
253 # while because of repeatable fields
254 while ($swish || $display) {
255 my $page = $repeat_off;
256 $page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0));
257 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
258 if ($repeat_off > 1000) {
259 print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
263 # is this field is lookup?
264 if ($display && $x->{lookup}) {
265 my $null = "<!-- null -->";
266 if ($use_lhash_cache) {
267 if (!defined($cache->{lhash}->{$display})) {
268 my $new_display = $lhash{$display};
269 if (defined($new_display)) {
270 #print STDERR "lookup cache store '$display' = '$new_display'\n";
271 $display = $new_display;
272 $cache->{lhash}->{$display} = $new_display;
274 # print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
276 $cache->{lhash}->{$display} = $null;
279 $display = $cache->{lhash}->{$display};
282 $display = $lhash{$display} || $null;
286 # filter="name" ; filter this field through
288 my $filter = $x->{filter};
289 if ($filter && !$cache->{filter_loaded}->{$filter}) {
290 require "filter/".$filter.".pm";
291 $cache->{filter_loaded}->{$filter}++;
293 # type="swish" ; field for swish
296 if ($filter && ($s || $se)) {
298 $tmp = join(" ",&$filter($tmp)) if ($s || $se);
301 $tmp = chk_eval($tmp);
302 $swish_data .= $tmp if ($s && $tmp);
303 $swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se);
306 # type="display" ; field for display
307 if ($d && $display) {
308 my $ldel = $delimiter;
309 if ($line_delimiter && $display_data) {
310 $ldel = $line_delimiter;
315 foreach my $tmp (&$filter($display)) {
316 my $tmp2 = mkformat($x,$tmp);
317 push @arr,$tmp2 if ($tmp2);
319 $display_data .= $ldel if ($display_data && @arr);
320 $display_data .= join($delimiter,@arr);
322 $display_data .= $ldel if ($display_data);
323 my $tmp = mkformat($x,$display);
324 $display_data .= $tmp if ($tmp);
328 # type="index" ; insert into index
330 if ($i && $display) {
331 $idisplay = $display;
334 $idisplay = &$filter($idisplay);
336 $idisplay = chk_eval($idisplay);
337 push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
340 # store fields in lookup
341 if ($il && $display) {
342 if (lc($x->{type}) eq "lookup_key") {
344 print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
348 $lookup_key = &$filter($display);
350 $lookup_key = $display;
353 } elsif (lc($x->{type}) eq "lookup_val") {
357 $lhash{$lookup_key} = &$filter($display);
359 $lhash{$lookup_key} = $display;
362 print STDERR "WARNING: no lookup_key defined for '$display'?";
368 # store data for page-by-page repeatable fields
369 if ($iterate_by_page) {
370 sub iterate_fld($$$$$$) {
371 my ($cache,$what,$field,$page,$data,$append) = @_;
374 my $ldel = $page_line_delimiter;
375 $ldel = " " if ($append);
376 #print STDERR "line delimiter: ",Dumper($ldel) if ($ldel);
377 if (! $cache->{$what}->{$field}->[$page]) {
378 $cache->{$what}->{$field}->[$page] = $data;
380 $cache->{$what}->{$field}->[$page] .= $ldel.$data;
385 iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append});
389 iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append});
392 if ($swish_exact_data) {
393 iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append});
394 $swish_exact_data = "";
398 my $ldel=$page_line_delimiter;
400 if ($cache->{index_data}->{$field}->[$page]) {
402 @index_data = @{$cache->{index_data}->{$field}->[$page]};
406 $index_data[$#index_data] .= $idisplay;
408 push @index_data, $idisplay;
411 push @index_data, $idisplay;
414 @{$cache->{index_data}->{$field}->[$page]} = @index_data;
419 if (! $iterate_by_page) {
420 my $idel = $x->{index_delimiter};
422 foreach my $tmp (@index_data) {
424 if ($idel && $tmp =~ m/$idel/) {
425 ($i,$d) = split(/$idel/,$tmp);
427 $index->insert($field, $i, $d, $path);
433 # now try to parse variables from configuration file
434 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
436 my $delimiter = x($x->{delimiter}) || ' ';
437 my $val = $cfg->val($database, x($x->{content}));
439 my ($s,$d,$i) = (1,1,0); # swish, display default
440 $s = 0 if (lc($x->{type}) eq "display");
441 $d = 0 if (lc($x->{type}) eq "swish");
442 # no support for swish exact in config.
444 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
447 $display_data .= $delimiter.$val if ($d);
448 $swish_data .= $val if ($s);
449 $index->insert($field, $val, $path) if ($i);
452 if ($iterate_by_page) {
453 # FIXME data from config tag will appear just
457 $cache->{display_data}->{$field}->[$page] = $display_data;
461 $cache->{swish_data}->{$field}->[$page] = $swish_data;
464 if ($swish_exact_data) {
465 $cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data;
466 $swish_exact_data = "";
471 # save data page-by-page
472 foreach my $field (@page_fields) {
473 my $nr_pages = $page_max{$field} || next;
474 #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n";
475 #print STDERR Dumper($cache->{display_data});
476 for (my $page=0; $page <= $nr_pages; $page++) {
478 if ($cache->{format}->{$field}) {
479 my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
480 $display_data=$tmp if ($tmp);
482 $display_data = $cache->{display_data}->{$field}->[$page];
484 if ($display_data) { # default
485 if ($field eq "headline") {
486 $xml .= xmlify("headline", $display_data);
489 # fallback to empty field name if needed
490 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
491 $html .= "#-#".$display_data."###\n";
495 my $swish_data = $cache->{swish_data}->{$field}->[$page];
497 # remove extra spaces
498 $swish_data =~ s/ +/ /g;
499 $swish_data =~ s/ +$//g;
501 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
504 my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page];
505 if ($swish_exact_data) {
506 $swish_exact_data =~ s/ +/ /g;
507 $swish_exact_data =~ s/ +$//g;
509 # add delimiters before and after word.
510 # That is required to produce exact match
511 $xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
514 my $idel = $cache->{index_delimiter}->{$field};
515 foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) {
518 if ($idel && $tmp =~ m/$idel/) {
519 ($i,$d) = split(/$idel/,$tmp);
521 $index->insert($field, $i, $d, $path);
522 #print STDERR "index [$idel] $field: $i --> $d [$path]\n";
528 if (! $iterate_by_page) {
530 if ($field eq "headline") {
531 $xml .= xmlify("headline", $display_data);
534 # fallback to empty field name if needed
535 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
536 $html .= "#-#".$display_data."###\n";
540 # remove extra spaces
541 $swish_data =~ s/ +/ /g;
542 $swish_data =~ s/ +$//g;
544 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
547 if ($swish_exact_data) {
548 $swish_exact_data =~ s/ +/ /g;
549 $swish_exact_data =~ s/ +$//g;
551 # add delimiters before and after word.
552 # That is required to produce exact match
553 $xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
558 # dump formatted output in <html>
560 #$xml .= xmlify("html",$html);
561 $xml .= "<html><![CDATA[ $html ]]></html>";
565 $xml .= $add_xml if ($add_xml);
566 return "<xml>\n$xml</xml>\n";
572 ##########################################################################
574 # read configuration for this script
575 my $cfg = new Config::IniFiles( -file => $config_file );
577 # read global.conf configuration
578 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
581 $index = new index_DBI(
582 $cfg_global->val('global', 'dbi_dbd'),
583 $cfg_global->val('global', 'dbi_dsn'),
584 $cfg_global->val('global', 'dbi_user'),
585 $cfg_global->val('global', 'dbi_passwd') || '',
588 my $show_progress = $cfg_global->val('global', 'show_progress');
590 my $unac_filter = $cfg_global->val('global', 'unac_filter');
592 require $unac_filter;
595 foreach my $database ($cfg->Sections) {
597 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
598 my $add_xml = $cfg -> val($database, 'xml'); # optional
600 # create new lookup file
601 my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
603 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
604 tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
605 print STDERR "creating lookup file '$lookup_file'\n";
606 # delete memory cache for lookup file
607 delete $cache->{lhash};
610 # open existing lookup file
611 $lookup_file = $cfg -> val($database, 'lookup_open'); # optional
613 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644;
614 tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644;
615 print STDERR "opening lookup file '$lookup_file'\n";
618 print STDERR "reading ./import_xml/$type.xml\n";
620 # extract just type basic
621 my $type_base = $type;
622 $type_base =~ s/_.+$//g;
624 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1);
626 # output current progress indicator
629 return if (! $show_progress);
631 my $total = shift || 1;
632 my $p = int($current * 100 / $total);
634 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
641 return if (! $show_progress);
642 my $current = shift @_;
644 my @ind = ('-','\\','|','/','-','\\','|','/', '-');
646 $last_p += $fake_dir;
647 $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
648 if ($last_p % 10 == 0) {
649 printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
654 print STDERR "using: $type...\n";
656 # erase cache for tags by order in this database
657 delete $cache->{tags_by_order};
659 if ($type_base eq "isis") {
661 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
663 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
664 my $db = OpenIsis::open( $isis_db );
666 # check if .txt database for OpenIsis is zero length,
667 # if so, erase it and re-open database
669 my $isis_db = shift || die "need isis database name";
672 if (-e $isis_db.".TXT") {
673 print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n";
674 unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!";
677 if (-e $isis_db.".PTR") {
678 print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n";
679 unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!";
682 return OpenIsis::open( $isis_db ) if ($reopen);
687 $db = check_txt_db($isis_db);
689 print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n";
696 print STDERR "FATAL: OpenIsis can't find file $isis_db\n";
699 } elsif ($db == -5) {
700 print STDERR "FATAL: OpenIsis can't access file $isis_db\n";
703 print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n";
707 my $max_rowid = OpenIsis::maxRowid( $db );
709 # if 0 records, try to rease isis .txt database
710 if ($max_rowid == 0) {
711 # force removal of database
712 $db = check_txt_db($isis_db);
713 $max_rowid = OpenIsis::maxRowid( $db );
716 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
718 my $path = $database;
720 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
721 my $row = OpenIsis::read( $db, $row_id );
722 if ($row && $row->{mfn}) {
724 progress($row->{mfn}, $max_rowid);
726 my $swishpath = $path."#".int($row->{mfn});
728 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
729 $xml = $cp2utf->convert($xml);
730 use bytes; # as opposed to chars
731 print "Path-Name: $swishpath\n";
732 print "Content-Length: ".(length($xml)+1)."\n";
733 print "Document-Type: XML\n\n$xml\n";
737 # for this to work with current version of OpenIsis (0.9.0)
738 # you might need my patch from
739 # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff
740 OpenIsis::close($db);
743 } elsif ($type_base eq "excel") {
744 require Spreadsheet::ParseExcel;
745 require Spreadsheet::ParseExcel::Utility;
746 import Spreadsheet::ParseExcel::Utility qw(int2col);
748 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
749 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
751 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
752 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
754 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
757 foreach my $oWks (@{$oBook->{Worksheet}}) {
758 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
759 last if ($oWks->{Name} eq $sheet);
763 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
765 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
766 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
767 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
769 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
771 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
772 my $cell = $oWorksheet->{Cells}[$iR][$iC];
774 $row->{int2col($iC)} = $cell->Value;
778 progress($iR, $end_row);
780 # print "row[$iR/$end_row] ";
781 # foreach (keys %{$row}) {
782 # print "$_: ",$row->{$_},"\t";
786 my $swishpath = $database."#".$iR;
790 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
791 $xml = $cp2utf->convert($xml);
792 use bytes; # as opposed to chars
793 print "Path-Name: $swishpath\n";
794 print "Content-Length: ".(length($xml)+1)."\n";
795 print "Document-Type: XML\n\n$xml\n";
798 } elsif ($type_base eq "marc") {
802 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
803 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
805 # optional argument is format
806 my $format = x($config->{marc_format}) || 'usmarc';
808 print STDERR "Reading MARC file '$marc_file'\n";
811 my $nr = $marc->openmarc({
812 file=>$marc_file, format=>$format
813 }) || die "Can't open MARC file '$marc_file' with format '$format'";
815 # read MARC file in memory
818 my $max_rec = $marc->marc_count();
820 for(my $i=1; $i<=$max_rec; $i++) {
822 progress($i,$max_rec);
824 # store value for marc_sf.pm
825 $main::cache->{marc_record} = $i;
827 my $swishpath = $database."#".$i;
829 if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
830 $xml = $cp2utf->convert($xml);
831 use bytes; # as opposed to chars
832 print "Path-Name: $swishpath\n";
833 print "Content-Length: ".(length($xml)+1)."\n";
834 print "Document-Type: XML\n\n$xml\n";
840 } elsif ($type_base eq "feed") {
842 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
843 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
845 print STDERR "Reading feed from program '$prog'\n";
847 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
849 my $i=1; # record nr.
858 my $swishpath = $database."#".$i++;
860 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
861 $xml = $cp2utf->convert($xml);
862 use bytes; # as opposed to chars
863 print "Path-Name: $swishpath\n";
864 print "Content-Length: ".(length($xml)+1)."\n";
865 print "Document-Type: XML\n\n$xml\n";
872 $line = $1 if (s/^(\d+):\s*//);
873 $data->{$line++} = $_;
879 untie %lhash if (%lhash);
883 # call this to commit index
888 ##########################################################################
892 all2xml.pl - read various file formats and dump XML for SWISH-E
896 This command will read ISIS data file using OpenIsis perl module, MARC
897 records using MARC module and optionally Micro$oft Excel files to
898 create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
899 this script B<isn't general xml generator> from isis files (isis allready
900 has something like that). Output of this script is tailor-made for SWISH-E.
904 Documentation is really lacking. However, in true Open Source spirit, source
905 is best documentation. I even made considerable effort to comment parts
906 which are not intuitively clear, so...
910 Dobrica Pavlinusic <dpavlin@rot13.org>
914 GNU Public License (GPL) v2 or later
918 SWISH-E web site at http://www.swish-e.org