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
26 use index_DBI_filter; # filter support for indexes
33 # -m multiple directories
37 getopts('d:m:qs', \%opts);
39 my $path; # this is name of database
41 Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions
43 # this is encoding of all files on disk, including import_xml/*.xml file and
44 # filter/*.pm files! It will be used to store strings in perl internally!
45 my $codepage = 'ISO-8859-2';
47 my $utf2cp = Text::Iconv->new('UTF-8',$codepage);
48 # this function will convert data from XML files to local encoding
50 return $utf2cp->convert($_[0]);
53 # decode isis/excel or other import codepage
56 # outgoing xml must be in UTF-8
57 my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
59 # mapping between data type and tag which specify
68 my $cache; # for cacheing
70 # lookup hash (tied to file)
72 # this option will cache all lookup entries in memory.
73 # if you are tight on memory, turn this off
74 my $use_lhash_cache = 1;
76 my $last_field_name; # cache to prevent repeated fields
84 my $add_xml = shift @_;
85 # needed to read values from configuration file
87 my $database = shift @_;
93 my $html = ""; # html formatted display output
95 my %field_usage; # counter for usage of each field
97 # sort subrouting using order="" attribute
99 my $va = $config->{indexer}->{$a}->{order} ||
100 $config->{indexer}->{$a};
101 my $vb = $config->{indexer}->{$b}->{order} ||
102 $config->{indexer}->{$b};
108 if ($cache->{tags_by_order}) {
109 @sorted_tags = @{$cache->{tags_by_order}};
111 @sorted_tags = sort by_order keys %{$config->{indexer}};
112 $cache->{tags_by_order} = \@sorted_tags;
118 # cache for field in pages
119 delete $cache->{display_data};
120 delete $cache->{swish_data};
121 delete $cache->{swish_exact_data};
122 delete $cache->{index_data};
123 delete $cache->{index_delimiter};
124 my @page_fields; # names of fields
127 # subs used to produce output
129 sub get_field_name($$$) {
130 my ($config,$field,$field_usage) = @_;
132 # find field name (signular, plural)
134 if ($config->{indexer}->{$field}->{name_singular} && $field_usage == 1) {
135 $field_name = $config->{indexer}->{$field}->{name_singular};
136 } elsif ($config->{indexer}->{$field}->{name_plural}) {
137 $field_name = $config->{indexer}->{$field}->{name_plural};
138 } elsif ($config->{indexer}->{$field}->{name}) {
139 $field_name = $config->{indexer}->{$field}->{name};
141 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
145 $field_name = x($field_name);
146 if (! $last_field_name) {
147 $last_field_name = $field_name;
148 return $last_field_name;
149 } elsif ($field_name ne $last_field_name) {
150 $last_field_name = $field_name;
151 return $last_field_name;
157 # init variables for different types
158 sub init_visible_type($) {
161 # swish, swish_exact, display, index, index_lookup
162 # swish and display defaults
163 my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
164 if (lc($type) eq "display") {
166 } elsif (lc($type) eq "swish") {
168 } elsif (lc($type) eq "index") {
169 ($s,$se,$d,$i) = (0,1,0,1);
170 } elsif (lc($type) eq "swish_exact") {
171 ($s,$se,$d,$i) = (0,1,0,0);
172 } elsif (lc($type) =~ /^lookup/) {
173 ($s,$se,$d,$i,$il) = (0,1,0,0,1);
175 print STDERR "WARNING: unknown type: $type\n";
177 return ($s,$se,$d,$i,$il);
184 # <delimiter>, </delimiter>
185 # <value>200a</value>
190 # <tag delimiter=", ">200a</tag>
192 # but without loosing spaces in delimiter (becasue
193 # new XML::Simple strips spaces in attribute values
194 # as defined in XML specification)
199 if (defined $x->{value}) {
200 my ($v,$d) = ($x->{value}->{content}, $x->{delimiter}->{content});
202 delete $x->{delimiter};
204 $x->{delimiter} = $d;
209 # begin real work: go field by field
210 foreach my $field (@sorted_tags) {
213 $field_usage{$field}++;
216 my $swish_exact_data = "";
217 my $display_data = "";
221 my ($swish,$display);
223 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
225 # is this field page-by-page?
226 my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page};
227 push @page_fields,$field if ($iterate_by_page);
229 # default line_delimiter if using
230 my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>';
231 $cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter};
233 my $format_name = $config->{indexer}->{$field}->{format_name};
234 my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter};
235 if ($format_name && $format_delimiter) {
236 $cache->{format}->{$field}->{format_name} = $format_name;
237 $cache->{format}->{$field}->{format_delimiter} = $format_delimiter;
240 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
244 my $format = x($x->{content});
245 my $delimiter = x($x->{delimiter}) || ' ';
247 my $repeat_off = 0; # init repeatable offset
249 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
251 # what will separate last line from this one?
252 if ($display_data && $x->{append}) {
253 $line_delimiter = $delimiter;
254 } elsif ($display_data) {
255 $line_delimiter = '<br/>';
258 # init vars so that we go into while...
259 ($swish,$display) = (1,1);
262 my $x = shift || die "mkformat needs tag reference";
263 my $data = shift || return;
264 my $format_name = x($x->{format_name}) || return $data;
265 my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
266 my $format_delimiter = x($x->{format_delimiter});
268 if ($format_delimiter) {
269 @data = split(/$format_delimiter/,$data);
275 my $nr = scalar $fmt =~ s/%s/%s/g;
276 if (($#data+1) == $nr) {
277 return sprintf($fmt,@data);
279 #print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
283 print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
287 # while because of repeatable fields
288 while ($swish || $display) {
289 my $page = $repeat_off;
290 $page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0));
291 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
292 if ($repeat_off > 1000) {
293 print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
297 # is this field is lookup?
298 if ($display && $x->{lookup}) {
299 my $null = "<!-- null -->";
300 if ($use_lhash_cache) {
301 if (!defined($cache->{lhash}->{$display})) {
302 my $new_display = $lhash{$display};
303 if (defined($new_display)) {
304 #print STDERR "lookup cache store '$display' = '$new_display'\n";
305 $display = $new_display;
306 $cache->{lhash}->{$display} = $new_display;
308 # print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
310 $cache->{lhash}->{$display} = $null;
313 $display = $cache->{lhash}->{$display};
316 $display = $lhash{$display} || $null;
320 # filter="name" ; filter this field through
322 my $filter = $x->{filter};
323 if ($filter && !$cache->{filter_loaded}->{$filter}) {
324 require "filter/".$filter.".pm";
325 $cache->{filter_loaded}->{$filter}++;
327 # type="swish" ; field for swish
330 if ($filter && ($s || $se)) {
332 $tmp = join(" ",&$filter($tmp)) if ($s || $se);
335 $swish_data .= $tmp if ($s && $tmp);
336 $swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se);
339 # type="display" ; field for display
340 if ($d && $display) {
341 my $ldel = $delimiter;
342 if ($line_delimiter && $display_data) {
343 $ldel = $line_delimiter;
348 foreach my $tmp (&$filter($display)) {
349 my $tmp2 = mkformat($x,$tmp);
350 push @arr,$tmp2 if ($tmp2);
352 $display_data .= $ldel if ($display_data && @arr);
353 $display_data .= join($delimiter,@arr);
355 $display_data .= $ldel if ($display_data);
356 my $tmp = mkformat($x,$display);
357 $display_data .= $tmp if ($tmp);
361 # type="index" ; insert into index
363 if ($i && $display) {
364 $idisplay = $display;
367 $idisplay = &$filter($idisplay);
369 push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
372 # store fields in lookup
373 if ($il && $display) {
374 if (lc($x->{type}) eq "lookup_key") {
376 print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
380 $lookup_key = &$filter($display);
382 $lookup_key = $display;
385 } elsif (lc($x->{type}) eq "lookup_val") {
389 $lhash{$lookup_key} = &$filter($display);
391 $lhash{$lookup_key} = $display;
394 print STDERR "WARNING: no lookup_key defined for '$display'?";
400 # store data for page-by-page repeatable fields
401 if ($iterate_by_page) {
402 sub iterate_fld($$$$$$) {
403 my ($cache,$what,$field,$page,$data,$append) = @_;
406 my $ldel = $page_line_delimiter;
407 $ldel = " " if ($append);
408 #print STDERR "line delimiter: ",Dumper($ldel) if ($ldel);
409 if (! $cache->{$what}->{$field}->[$page]) {
410 $cache->{$what}->{$field}->[$page] = $data;
412 $cache->{$what}->{$field}->[$page] .= $ldel.$data;
417 iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append});
421 iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append});
424 if ($swish_exact_data) {
425 iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append});
426 $swish_exact_data = "";
430 my $ldel=$page_line_delimiter;
432 if ($cache->{index_data}->{$field}->[$page]) {
434 @index_data = @{$cache->{index_data}->{$field}->[$page]};
438 $index_data[$#index_data] .= $idisplay;
440 push @index_data, $idisplay;
443 push @index_data, $idisplay;
446 @{$cache->{index_data}->{$field}->[$page]} = @index_data;
451 if (! $iterate_by_page) {
452 my $idel = $x->{index_delimiter};
454 foreach my $tmp (@index_data) {
456 if ($idel && $tmp =~ m/$idel/) {
457 ($i,$d) = split(/$idel/,$tmp);
459 $index->insert($field, $i, $d, $path);
465 # now try to parse variables from configuration file
466 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
470 my $delimiter = x($x->{delimiter}) || ' ';
471 my $val = $cfg->val($database, x($x->{content}));
473 # FIXME index_lookup is not supported!
474 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
477 $display_data .= $delimiter.$val if ($d);
478 $swish_data .= " ".$val if ($s);
479 $index->insert($field, $val, $path) if ($i);
482 if ($iterate_by_page) {
483 # FIXME data from config tag will appear just
487 $cache->{display_data}->{$field}->[$page] = $display_data;
491 $cache->{swish_data}->{$field}->[$page] = $swish_data;
494 if ($swish_exact_data) {
495 $cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data;
496 $swish_exact_data = "";
501 # save data page-by-page
502 foreach my $field (@page_fields) {
503 my $nr_pages = $page_max{$field} || next;
504 #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n";
505 #print STDERR Dumper($cache->{display_data});
506 for (my $page=0; $page <= $nr_pages; $page++) {
508 if ($cache->{format}->{$field}) {
509 my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
510 $display_data=$tmp if ($tmp);
512 $display_data = $cache->{display_data}->{$field}->[$page];
514 if ($display_data) { # default
515 if ($field eq "headline") {
516 $xml .= xmlify("headline", $display_data);
519 # fallback to empty field name if needed
520 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
521 $html .= "#-#".$display_data."###\n";
525 my $swish_data = $cache->{swish_data}->{$field}->[$page];
527 # remove extra spaces
528 $swish_data =~ s/ +/ /g;
529 $swish_data =~ s/ +$//g;
531 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
534 my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page];
535 if ($swish_exact_data) {
536 $swish_exact_data =~ s/ +/ /g;
537 $swish_exact_data =~ s/ +$//g;
539 # add delimiters before and after word.
540 # That is required to produce exact match
541 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
544 my $idel = $cache->{index_delimiter}->{$field};
545 foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) {
548 if ($idel && $tmp =~ m/$idel/) {
549 ($i,$d) = split(/$idel/,$tmp);
551 $index->insert($field, $i, $d, $path);
552 #print STDERR "index [$idel] $field: $i --> $d [$path]\n";
558 if (! $iterate_by_page) {
560 if ($field eq "headline") {
561 $xml .= xmlify("headline", $display_data);
564 # fallback to empty field name if needed
565 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
566 $html .= "#-#".$display_data."###\n";
570 # remove extra spaces
571 $swish_data =~ s/ +/ /g;
572 $swish_data =~ s/ +$//g;
574 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
577 if ($swish_exact_data) {
578 $swish_exact_data =~ s/ +/ /g;
579 $swish_exact_data =~ s/ +$//g;
581 # add delimiters before and after word.
582 # That is required to produce exact match
583 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
588 # dump formatted output in <html>
590 #$xml .= xmlify("html",$html);
591 $xml .= "<html><![CDATA[ $html ]]></html>";
595 $xml .= $add_xml if ($add_xml);
596 return "<xml>\n$xml</xml>\n";
602 ##########################################################################
604 # read configuration for this script
605 my $cfg = new Config::IniFiles( -file => $config_file );
607 # read global.conf configuration
608 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
611 $index = new index_DBI(
612 $cfg_global->val('global', 'dbi_dbd'),
613 $cfg_global->val('global', 'dbi_dsn'),
614 $cfg_global->val('global', 'dbi_user'),
615 $cfg_global->val('global', 'dbi_passwd') || '',
618 my $show_progress = $cfg_global->val('global', 'show_progress');
620 my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter');
621 if ($my_unac_filter) {
622 print STDERR "using $my_unac_filter to filter characters for search\n";
623 require $my_unac_filter;
625 print STDERR "### fallback to default my_unac_string!\n";
627 sub main::my_unac_string($$) {
628 my ($charset, $string) = (@_);
634 foreach my $database ($cfg->Sections) {
636 # save database name in global variable path for later
637 # (need for index filter creation)
640 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
641 my $add_xml = $cfg -> val($database, 'xml'); # optional
643 # create new lookup file
644 my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
646 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
647 if (! -e $lookup_file) {
648 open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!";
651 tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
652 print STDERR "creating lookup file '$lookup_file'\n";
653 # delete memory cache for lookup file
654 delete $cache->{lhash};
657 # open existing lookup file
658 $lookup_file = $cfg -> val($database, 'lookup_open'); # optional
660 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644;
661 tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644;
662 print STDERR "opening lookup file '$lookup_file'\n";
665 print STDERR "reading ./import_xml/$type.xml\n";
667 # extract just type basic
668 my $type_base = $type;
669 $type_base =~ s/_.+$//g;
671 $config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 );
673 # helper for progress bar
678 my ($ss,$mm,$hh) = gmtime($t);
679 $out .= "${hh}h" if ($hh);
680 $out .= sprintf("%02d:%02d", $mm,$ss);
681 $out .= " " if ($hh == 0);
685 # output current progress indicator
687 my $start_t = time();
689 return if (! $show_progress);
691 my $total = shift || 1;
692 my $p = int($current * 100 / $total);
693 if ($p < $last_p || $current == 1) {
696 } elsif ($p != $last_p) {
697 my $rate = ($current / (time() - $start_t || 1));
698 my $eta = ($total-$current) / ($rate || 1);
699 printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta));
706 my $last_fake_t = time();
708 return if (! $show_progress);
709 my $current = shift @_;
711 my @ind = ('-','\\','|','/','-','\\','|','/');
713 if ($current < $fake_pos) {
720 if (time()-$last_fake_t >= 1) {
721 $last_fake_t = time();
722 $fake_pos += $fake_dir;
723 $fake_dir = -$fake_dir if ($fake_pos > 38);
726 if ($current % 10 == 0) {
727 my $rate = ($current / (time() - $start_t || 1));
728 printf STDERR ("%5d [%-38s] %0.1f/s\r",$current, " " x $fake_pos .$ind[($current / 10) % 8], $rate);
733 print STDERR "using: $type...\n";
735 # erase cache for tags by order in this database
736 delete $cache->{tags_by_order};
738 if ($type_base eq "isis") {
740 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
742 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
743 my $db = new Biblio::Isis( isisdb => $isis_db );
746 print STDERR "FATAL: can't read ISIS database: $isis_db, skipping...\n";
750 my $max_rowid = $db->count if ($db);
752 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
754 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
755 my $row = $db->to_hash( $row_id );
758 $row->{mfn} = $row_id;
759 $row->{record} = $db->{record};
761 progress($row->{mfn}, $max_rowid);
763 my $swishpath = $path."#".int($row->{mfn});
765 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
766 $xml = $cp2utf->convert($xml);
767 use bytes; # as opposed to chars
768 print "Path-Name: $swishpath\n";
769 print "Content-Length: ".(length($xml)+1)."\n";
770 print "Document-Type: XML\n\n$xml\n";
776 } elsif ($type_base eq "excel") {
777 require Spreadsheet::ParseExcel;
778 require Spreadsheet::ParseExcel::Utility;
779 import Spreadsheet::ParseExcel::Utility qw(int2col);
781 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
782 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
784 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
785 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
787 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
790 foreach my $oWks (@{$oBook->{Worksheet}}) {
791 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
792 last if ($oWks->{Name} eq $sheet);
796 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
798 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
799 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
800 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
802 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
804 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
805 my $cell = $oWorksheet->{Cells}[$iR][$iC];
807 # this conversion is a cludge.
808 # Files from Excell could have
809 # characters which don't fit into
810 # destination encoding.
811 $row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value;
815 progress($iR, $end_row);
817 # print "row[$iR/$end_row] ";
818 # foreach (keys %{$row}) {
819 # print "$_: ",$row->{$_},"\t";
823 my $swishpath = $database."#".$iR;
827 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
828 $xml = $cp2utf->convert($xml);
829 use bytes; # as opposed to chars
830 print "Path-Name: $swishpath\n";
831 print "Content-Length: ".(length($xml)+1)."\n";
832 print "Document-Type: XML\n\n$xml\n";
838 } elsif ($type_base eq "marc") {
840 require MARC::File::USMARC;
842 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
843 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
845 # optional argument is format
846 warn "marc_format is no longer used!" if ($config->{marc_format});
847 print STDERR "Reading MARC file '$marc_file'\n";
849 my $marc = MARC::File::USMARC->in( $marc_file );
852 print STDERR "FATAL: can't read MARC file: $marc_file, skipping...\n";
856 # count records in MARC file
858 my $filename = shift || die;
859 my $file = MARC::File::USMARC->in($filename) || return;
861 while ($file->skip()) {
867 my $count = marc_count($marc_file) || warn "no records in '$marc_file'?";
871 while( my $rec = $marc->next() ) {
875 my $swishpath = $database."#".$i;
877 if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) {
878 $xml = $cp2utf->convert($xml);
879 use bytes; # as opposed to chars
880 print "Path-Name: $swishpath\n";
881 print "Content-Length: ".(length($xml)+1)."\n";
882 print "Document-Type: XML\n\n$xml\n";
890 } elsif ($type_base eq "feed") {
892 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
893 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
895 print STDERR "Reading feed from program '$prog'\n";
897 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
899 my $i=1; # record nr.
908 my $swishpath = $database."#".$i++;
910 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
911 $xml = $cp2utf->convert($xml);
912 use bytes; # as opposed to chars
913 print "Path-Name: $swishpath\n";
914 print "Content-Length: ".(length($xml)+1)."\n";
915 print "Document-Type: XML\n\n$xml\n";
922 $line = $1 if (s/^(\d+):\s*//);
923 $data->{$line++} = $_;
929 untie %lhash if (%lhash);
933 # call this to commit index
938 ##########################################################################
942 all2xml.pl - read various file formats and dump XML for SWISH-E
946 $ all2xml.pl [test.conf]
950 This command will read ISIS data file using Biblio::Isis perl module, MARC
951 records using MARC::File module and optionally Micro$oft Excel files to
952 create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
953 this script B<isn't general xml generator> from isis files (isis allready
954 has something like that). Output of this script is tailor-made for SWISH-E.
956 If no configuration file is specified, it will use default one called
961 Documentation is really lacking. However, in true Open Source spirit, source
962 is best documentation. I even made considerable effort to comment parts
963 which are not intuitively clear, so...
967 Dobrica Pavlinusic <dpavlin@rot13.org>
971 GNU Public License (GPL) v2 or later
975 SWISH-E web site at http://www.swish-e.org