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!";
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 # init variables for different types
155 sub init_visible_type($) {
158 # swish, swish_exact, display, index, index_lookup
159 # swish and display defaults
160 my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
161 if (lc($type) eq "display") {
163 } elsif (lc($type) eq "swish") {
165 } elsif (lc($type) eq "index") {
166 ($s,$se,$d,$i) = (0,1,0,1);
167 } elsif (lc($type) eq "swish_exact") {
168 ($s,$se,$d,$i) = (0,1,0,0);
169 } elsif (lc($type) =~ /^lookup/) {
170 ($s,$se,$d,$i,$il) = (0,1,0,0,1);
172 print STDERR "WARNING: unknown type: $type\n";
174 return ($s,$se,$d,$i,$il);
181 # <delimiter>, </delimiter>
182 # <value>200a</value>
187 # <tag delimiter=", ">200a</tag>
189 # but without loosing spaces in delimiter (becasue
190 # new XML::Simple strips spaces in attribute values
191 # as defined in XML specification)
196 if (defined $x->{value}) {
197 my ($v,$d) = ($x->{value}->{content}, $x->{delimiter}->{content});
199 delete $x->{delimiter};
201 $x->{delimiter} = $d;
206 # begin real work: go field by field
207 foreach my $field (@sorted_tags) {
210 $field_usage{$field}++;
213 my $swish_exact_data = "";
214 my $display_data = "";
218 my ($swish,$display);
220 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
222 # is this field page-by-page?
223 my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page};
224 push @page_fields,$field if ($iterate_by_page);
226 # default line_delimiter if using
227 my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>';
228 $cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter};
230 my $format_name = $config->{indexer}->{$field}->{format_name};
231 my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter};
232 if ($format_name && $format_delimiter) {
233 $cache->{format}->{$field}->{format_name} = $format_name;
234 $cache->{format}->{$field}->{format_delimiter} = $format_delimiter;
237 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
241 my $format = x($x->{content});
242 my $delimiter = x($x->{delimiter}) || ' ';
244 my $repeat_off = 0; # init repeatable offset
246 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
248 # what will separate last line from this one?
249 if ($display_data && $x->{append}) {
250 $line_delimiter = $delimiter;
251 } elsif ($display_data) {
252 $line_delimiter = '<br/>';
255 # init vars so that we go into while...
256 ($swish,$display) = (1,1);
258 # placeholder for all repeatable entries for index
261 my $x = shift || die "mkformat needs tag reference";
262 my $data = shift || return;
263 my $format_name = x($x->{format_name}) || return $data;
264 my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
265 my $format_delimiter = x($x->{format_delimiter});
267 if ($format_delimiter) {
268 @data = split(/$format_delimiter/,$data);
274 my $nr = scalar $fmt =~ s/%s/%s/g;
275 if (($#data+1) == $nr) {
276 return sprintf($fmt,@data);
278 #print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
282 print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
286 # while because of repeatable fields
287 while ($swish || $display) {
288 my $page = $repeat_off;
289 $page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0));
290 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
291 if ($repeat_off > 1000) {
292 print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
296 # is this field is lookup?
297 if ($display && $x->{lookup}) {
298 my $null = "<!-- null -->";
299 if ($use_lhash_cache) {
300 if (!defined($cache->{lhash}->{$display})) {
301 my $new_display = $lhash{$display};
302 if (defined($new_display)) {
303 #print STDERR "lookup cache store '$display' = '$new_display'\n";
304 $display = $new_display;
305 $cache->{lhash}->{$display} = $new_display;
307 # print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
309 $cache->{lhash}->{$display} = $null;
312 $display = $cache->{lhash}->{$display};
315 $display = $lhash{$display} || $null;
319 # filter="name" ; filter this field through
321 my $filter = $x->{filter};
322 if ($filter && !$cache->{filter_loaded}->{$filter}) {
323 require "filter/".$filter.".pm";
324 $cache->{filter_loaded}->{$filter}++;
326 # type="swish" ; field for swish
329 if ($filter && ($s || $se)) {
331 $tmp = join(" ",&$filter($tmp)) if ($s || $se);
334 $swish_data .= $tmp if ($s && $tmp);
335 $swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se);
338 # type="display" ; field for display
339 if ($d && $display) {
340 my $ldel = $delimiter;
341 if ($line_delimiter && $display_data) {
342 $ldel = $line_delimiter;
347 foreach my $tmp (&$filter($display)) {
348 my $tmp2 = mkformat($x,$tmp);
349 push @arr,$tmp2 if ($tmp2);
351 $display_data .= $ldel if ($display_data && @arr);
352 $display_data .= join($delimiter,@arr);
354 $display_data .= $ldel if ($display_data);
355 my $tmp = mkformat($x,$display);
356 $display_data .= $tmp if ($tmp);
360 # type="index" ; insert into index
362 if ($i && $display) {
363 $idisplay = $display;
366 $idisplay = &$filter($idisplay);
368 push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
371 # store fields in lookup
372 if ($il && $display) {
373 if (lc($x->{type}) eq "lookup_key") {
375 print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
379 $lookup_key = &$filter($display);
381 $lookup_key = $display;
384 } elsif (lc($x->{type}) eq "lookup_val") {
388 $lhash{$lookup_key} = &$filter($display);
390 $lhash{$lookup_key} = $display;
393 print STDERR "WARNING: no lookup_key defined for '$display'?";
399 # store data for page-by-page repeatable fields
400 if ($iterate_by_page) {
401 sub iterate_fld($$$$$$) {
402 my ($cache,$what,$field,$page,$data,$append) = @_;
405 my $ldel = $page_line_delimiter;
406 $ldel = " " if ($append);
407 #print STDERR "line delimiter: ",Dumper($ldel) if ($ldel);
408 if (! $cache->{$what}->{$field}->[$page]) {
409 $cache->{$what}->{$field}->[$page] = $data;
411 $cache->{$what}->{$field}->[$page] .= $ldel.$data;
416 iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append});
420 iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append});
423 if ($swish_exact_data) {
424 iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append});
425 $swish_exact_data = "";
429 my $ldel=$page_line_delimiter;
431 if ($cache->{index_data}->{$field}->[$page]) {
433 @index_data = @{$cache->{index_data}->{$field}->[$page]};
437 $index_data[$#index_data] .= $idisplay;
439 push @index_data, $idisplay;
442 push @index_data, $idisplay;
445 @{$cache->{index_data}->{$field}->[$page]} = @index_data;
450 if (! $iterate_by_page) {
451 my $idel = $x->{index_delimiter};
453 foreach my $tmp (@index_data) {
455 if ($idel && $tmp =~ m/$idel/) {
456 ($i,$d) = split(/$idel/,$tmp);
458 $index->insert($field, $i, $d, $path);
464 # now try to parse variables from configuration file
465 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
469 my $delimiter = x($x->{delimiter}) || ' ';
470 my $val = $cfg->val($database, x($x->{content}));
472 # FIXME index_lookup is not supported!
473 my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
476 $display_data .= $delimiter.$val if ($d);
477 $swish_data .= " ".$val if ($s);
478 $index->insert($field, $val, $path) if ($i);
481 if ($iterate_by_page) {
482 # FIXME data from config tag will appear just
486 $cache->{display_data}->{$field}->[$page] = $display_data;
490 $cache->{swish_data}->{$field}->[$page] = $swish_data;
493 if ($swish_exact_data) {
494 $cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data;
495 $swish_exact_data = "";
500 # save data page-by-page
501 foreach my $field (@page_fields) {
502 my $nr_pages = $page_max{$field} || next;
503 #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n";
504 #print STDERR Dumper($cache->{display_data});
505 for (my $page=0; $page <= $nr_pages; $page++) {
507 if ($cache->{format}->{$field}) {
508 my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
509 $display_data=$tmp if ($tmp);
511 $display_data = $cache->{display_data}->{$field}->[$page];
513 if ($display_data) { # default
514 if ($field eq "headline") {
515 $xml .= xmlify("headline", $display_data);
518 # fallback to empty field name if needed
519 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
520 $html .= "#-#".$display_data."###\n";
524 my $swish_data = $cache->{swish_data}->{$field}->[$page];
526 # remove extra spaces
527 $swish_data =~ s/ +/ /g;
528 $swish_data =~ s/ +$//g;
530 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
533 my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page];
534 if ($swish_exact_data) {
535 $swish_exact_data =~ s/ +/ /g;
536 $swish_exact_data =~ s/ +$//g;
538 # add delimiters before and after word.
539 # That is required to produce exact match
540 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
543 my $idel = $cache->{index_delimiter}->{$field};
544 foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) {
547 if ($idel && $tmp =~ m/$idel/) {
548 ($i,$d) = split(/$idel/,$tmp);
550 $index->insert($field, $i, $d, $path);
551 #print STDERR "index [$idel] $field: $i --> $d [$path]\n";
557 if (! $iterate_by_page) {
559 if ($field eq "headline") {
560 $xml .= xmlify("headline", $display_data);
563 # fallback to empty field name if needed
564 $html .= get_field_name($config,$field,$field_usage{$field}) || '';
565 $html .= "#-#".$display_data."###\n";
569 # remove extra spaces
570 $swish_data =~ s/ +/ /g;
571 $swish_data =~ s/ +$//g;
573 $xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
576 if ($swish_exact_data) {
577 $swish_exact_data =~ s/ +/ /g;
578 $swish_exact_data =~ s/ +$//g;
580 # add delimiters before and after word.
581 # That is required to produce exact match
582 $xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
587 # dump formatted output in <html>
589 #$xml .= xmlify("html",$html);
590 $xml .= "<html><![CDATA[ $html ]]></html>";
594 $xml .= $add_xml if ($add_xml);
595 return "<xml>\n$xml</xml>\n";
601 ##########################################################################
603 # read configuration for this script
604 my $cfg = new Config::IniFiles( -file => $config_file );
606 # read global.conf configuration
607 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
610 $index = new index_DBI(
611 $cfg_global->val('global', 'dbi_dbd'),
612 $cfg_global->val('global', 'dbi_dsn'),
613 $cfg_global->val('global', 'dbi_user'),
614 $cfg_global->val('global', 'dbi_passwd') || '',
617 my $show_progress = $cfg_global->val('global', 'show_progress');
619 my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter');
620 if ($my_unac_filter) {
621 print STDERR "using $my_unac_filter to filter characters for search\n";
622 require $my_unac_filter;
624 print STDERR "### fallback to default my_unac_string!\n";
626 sub main::my_unac_string($$) {
627 my ($charset, $string) = (@_);
633 foreach my $database ($cfg->Sections) {
635 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
636 my $add_xml = $cfg -> val($database, 'xml'); # optional
638 # create new lookup file
639 my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
641 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
642 if (! -e $lookup_file) {
643 open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!";
646 tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
647 print STDERR "creating lookup file '$lookup_file'\n";
648 # delete memory cache for lookup file
649 delete $cache->{lhash};
652 # open existing lookup file
653 $lookup_file = $cfg -> val($database, 'lookup_open'); # optional
655 #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644;
656 tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644;
657 print STDERR "opening lookup file '$lookup_file'\n";
660 print STDERR "reading ./import_xml/$type.xml\n";
662 # extract just type basic
663 my $type_base = $type;
664 $type_base =~ s/_.+$//g;
666 $config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 );
668 # output current progress indicator
671 return if (! $show_progress);
673 my $total = shift || 1;
674 my $p = int($current * 100 / $total);
676 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
683 return if (! $show_progress);
684 my $current = shift @_;
686 my @ind = ('-','\\','|','/','-','\\','|','/', '-');
688 $last_p += $fake_dir;
689 $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
690 if ($last_p % 10 == 0) {
691 printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
696 print STDERR "using: $type...\n";
698 # erase cache for tags by order in this database
699 delete $cache->{tags_by_order};
701 if ($type_base eq "isis") {
703 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
705 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
706 my $db = OpenIsis::open( $isis_db );
708 # check if .txt database for OpenIsis is zero length,
709 # if so, erase it and re-open database
711 my $isis_db = shift || die "need isis database name";
714 if (-e $isis_db.".TXT") {
715 print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n";
716 unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!";
719 if (-e $isis_db.".PTR") {
720 print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n";
721 unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!";
724 return OpenIsis::open( $isis_db ) if ($reopen);
729 $db = check_txt_db($isis_db);
731 print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n";
738 print STDERR "FATAL: OpenIsis can't find file $isis_db\n";
741 } elsif ($db == -5) {
742 print STDERR "FATAL: OpenIsis can't access file $isis_db\n";
745 print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n";
749 my $max_rowid = OpenIsis::maxRowid( $db );
751 # if 0 records, try to rease isis .txt database
752 if ($max_rowid == 0) {
753 # force removal of database
754 $db = check_txt_db($isis_db);
755 $max_rowid = OpenIsis::maxRowid( $db );
758 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
760 my $path = $database;
762 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
763 my $row = OpenIsis::read( $db, $row_id );
764 if ($row && $row->{mfn}) {
766 progress($row->{mfn}, $max_rowid);
768 my $swishpath = $path."#".int($row->{mfn});
770 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
771 $xml = $cp2utf->convert($xml);
772 use bytes; # as opposed to chars
773 print "Path-Name: $swishpath\n";
774 print "Content-Length: ".(length($xml)+1)."\n";
775 print "Document-Type: XML\n\n$xml\n";
779 # for this to work with current version of OpenIsis (0.9.0)
780 # you might need my patch from
781 # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff
782 OpenIsis::close($db);
785 } elsif ($type_base eq "excel") {
786 require Spreadsheet::ParseExcel;
787 require Spreadsheet::ParseExcel::Utility;
788 import Spreadsheet::ParseExcel::Utility qw(int2col);
790 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
791 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
793 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
794 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
796 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
799 foreach my $oWks (@{$oBook->{Worksheet}}) {
800 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
801 last if ($oWks->{Name} eq $sheet);
805 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
807 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
808 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
809 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
811 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
813 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
814 my $cell = $oWorksheet->{Cells}[$iR][$iC];
816 # this conversion is a cludge.
817 # Files from Excell could have
818 # characters which don't fit into
819 # destination encoding.
820 $row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value;
824 progress($iR, $end_row);
826 # print "row[$iR/$end_row] ";
827 # foreach (keys %{$row}) {
828 # print "$_: ",$row->{$_},"\t";
832 my $swishpath = $database."#".$iR;
836 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
837 $xml = $cp2utf->convert($xml);
838 use bytes; # as opposed to chars
839 print "Path-Name: $swishpath\n";
840 print "Content-Length: ".(length($xml)+1)."\n";
841 print "Document-Type: XML\n\n$xml\n";
844 } elsif ($type_base eq "marc") {
848 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
849 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
851 # optional argument is format
852 my $format = x($config->{marc_format}) || 'usmarc';
854 print STDERR "Reading MARC file '$marc_file'\n";
857 my $nr = $marc->openmarc({
858 file=>$marc_file, format=>$format
859 }) || die "Can't open MARC file '$marc_file' with format '$format'";
861 # read MARC file in memory
864 my $max_rec = $marc->marc_count();
866 for(my $i=1; $i<=$max_rec; $i++) {
868 progress($i,$max_rec);
870 # store value for marc_sf.pm
871 $main::cache->{marc_record} = $i;
873 my $swishpath = $database."#".$i;
875 if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
876 $xml = $cp2utf->convert($xml);
877 use bytes; # as opposed to chars
878 print "Path-Name: $swishpath\n";
879 print "Content-Length: ".(length($xml)+1)."\n";
880 print "Document-Type: XML\n\n$xml\n";
886 } elsif ($type_base eq "feed") {
888 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
889 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
891 print STDERR "Reading feed from program '$prog'\n";
893 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
895 my $i=1; # record nr.
904 my $swishpath = $database."#".$i++;
906 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
907 $xml = $cp2utf->convert($xml);
908 use bytes; # as opposed to chars
909 print "Path-Name: $swishpath\n";
910 print "Content-Length: ".(length($xml)+1)."\n";
911 print "Document-Type: XML\n\n$xml\n";
918 $line = $1 if (s/^(\d+):\s*//);
919 $data->{$line++} = $_;
925 untie %lhash if (%lhash);
929 # call this to commit index
934 ##########################################################################
938 all2xml.pl - read various file formats and dump XML for SWISH-E
942 $ all2xml.pl [test.conf]
946 This command will read ISIS data file using OpenIsis perl module, MARC
947 records using MARC module and optionally Micro$oft Excel files to
948 create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
949 this script B<isn't general xml generator> from isis files (isis allready
950 has something like that). Output of this script is tailor-made for SWISH-E.
952 If no configuration file is specified, it will use default one called
957 Documentation is really lacking. However, in true Open Source spirit, source
958 is best documentation. I even made considerable effort to comment parts
959 which are not intuitively clear, so...
963 Dobrica Pavlinusic <dpavlin@rot13.org>
967 GNU Public License (GPL) v2 or later
971 SWISH-E web site at http://www.swish-e.org