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
70 my $add_xml = shift @_;
71 # needed to read values from configuration file
73 my $database = shift @_;
79 my $html = ""; # html formatted display output
81 my %field_usage; # counter for usage of each field
83 # sort subrouting using order="" attribute
85 my $va = $config->{indexer}->{$a}->{order} ||
86 $config->{indexer}->{$a};
87 my $vb = $config->{indexer}->{$b}->{order} ||
88 $config->{indexer}->{$b};
93 foreach my $field (sort by_order keys %{$config->{indexer}}) {
96 $field_usage{$field}++;
99 my $display_data = "";
102 my ($swish,$display);
104 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
105 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
107 my $format = x($x->{content});
108 my $delimiter = x($x->{delimiter}) || ' ';
110 my $repeat_off = 0; # repeatable offset
112 my ($s,$d,$i) = (1,1,0); # swish, display default
113 $s = 0 if (lc($x->{type}) eq "display");
114 $d = 0 if (lc($x->{type}) eq "swish");
115 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
117 # what will separate last line from this one?
118 if ($display_data && $x->{append} && $x->{append} eq "1") {
119 $line_delimiter = ' ';
120 } elsif ($display_data) {
121 $line_delimiter = '<br/>';
124 # init vars so that we go into while...
125 ($swish,$display) = (1,1);
127 # placeholder for all repeatable entries for index
132 my $x = shift || die "mkformat needs tag reference";
133 my $data = shift || return;
134 my $format_name = x($x->{format_name}) || return $data;
135 my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
136 my $format_delimiter = x($x->{format_delimiter});
138 if ($format_delimiter) {
139 @data = split(/$format_delimiter/,$data);
145 my $nr = scalar $fmt =~ s/%s/%s/g;
146 if (($#data+1) == $nr) {
147 return sprintf($fmt,@data);
149 print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
153 print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
157 # while because of repeatable fields
158 while ($swish || $display) {
159 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
160 if ($repeat_off > 1000) {
161 print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
165 # filter="name" ; filter this field through
167 my $filter = $x->{filter};
169 require "filter/".$filter.".pm";
171 # type="swish" ; field for swish
175 $swish_data .= join(" ",&$filter($swish));
177 $swish_data .= $swish;
181 # type="display" ; field for display
182 if ($d && $display) {
183 if ($line_delimiter && $display_data) {
184 $display_data .= $line_delimiter;
185 undef $line_delimiter;
190 $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display)));
192 $display_data = join($delimiter,mkformat($x,&$filter($display)));
196 $display_data .= $delimiter.mkformat($x,$display);
198 $display_data = mkformat($x,$display);
203 # type="index" ; insert into index
204 if ($i && $display) {
205 push @index_data, $display;
206 $index_filter = $filter if ($filter);
214 foreach my $d (@index_data) {
215 $index->insert($field, &$index_filter($d), $path);
218 foreach my $d (@index_data) {
219 $index->insert($field, $d, $path);
225 # now try to parse variables from configuration file
226 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
228 my $delimiter = x($x->{delimiter}) || ' ';
229 my $val = $cfg->val($database, x($x->{content}));
231 my ($s,$d,$i) = (1,1,0); # swish, display default
232 $s = 0 if (lc($x->{type}) eq "display");
233 $d = 0 if (lc($x->{type}) eq "swish");
234 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
237 $display_data .= $delimiter.$val if ($d);
238 $swish_data .= $val if ($s);
239 $index->insert($field, $val, $path) if ($i);
247 if ($field eq "headline") {
248 $xml .= xmlify("headline", $display_data);
251 # find field name (signular, plural)
253 if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
254 $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
255 } elsif ($config->{indexer}->{$field}->{name_plural}) {
256 $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
257 } elsif ($config->{indexer}->{$field}->{name}) {
258 $field_name = $config->{indexer}->{$field}->{name}."#-#";
260 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
263 $html .= x($field_name);
265 $html .= $display_data."###\n";
269 # remove extra spaces
270 $swish_data =~ s/ +/ /g;
271 $swish_data =~ s/ +$//g;
273 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
279 # dump formatted output in <html>
281 #$xml .= xmlify("html",$html);
282 $xml .= "<html><![CDATA[ $html ]]></html>";
286 $xml .= $add_xml if ($add_xml);
287 return "<xml>\n$xml</xml>\n";
293 ##########################################################################
295 # read configuration for this script
296 my $cfg = new Config::IniFiles( -file => $config_file );
298 # read global.conf configuration
299 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
302 $index = new index_DBI(
303 $cfg_global->val('global', 'dbi_dbd'),
304 $cfg_global->val('global', 'dbi_dsn'),
305 $cfg_global->val('global', 'dbi_user'),
306 $cfg_global->val('global', 'dbi_passwd') || '',
309 my $show_progress = $cfg_global->val('global', 'show_progress');
311 foreach my $database ($cfg->Sections) {
313 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
314 my $add_xml = $cfg -> val($database, 'xml'); # optional
316 print STDERR "reading ./import_xml/$type.xml\n";
318 # extract just type basic
319 my $type_base = $type;
320 $type_base =~ s/_.+$//g;
322 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1);
324 # output current progress indicator
327 return if (! $show_progress);
329 my $total = shift || 1;
330 my $p = int($current * 100 / $total);
332 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
339 return if (! $show_progress);
340 my $current = shift @_;
342 my @ind = ('-','\\','|','/','-','\\','|','/', '-');
344 $last_p += $fake_dir;
345 $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
346 if ($last_p % 10 == 0) {
347 printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
352 print STDERR "using: $type...\n";
354 if ($type_base eq "isis") {
356 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
358 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
359 my $db = OpenIsis::open( $isis_db );
361 # check if .txt database for OpenIsis is zero length,
362 # if so, erase it and re-open database
364 my $isis_db = shift || die "need isis database name";
367 if (-e $isis_db.".TXT") {
368 print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n";
369 unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!";
372 if (-e $isis_db.".PTR") {
373 print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n";
374 unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!";
377 return OpenIsis::open( $isis_db ) if ($reopen);
382 $db = check_txt_db($isis_db);
384 print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n";
391 print STDERR "FATAL: OpenIsis can't find file $isis_db\n";
394 } elsif ($db == -5) {
395 print STDERR "FATAL: OpenIsis can't access file $isis_db\n";
398 print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n";
402 my $max_rowid = OpenIsis::maxRowid( $db );
404 # if 0 records, try to rease isis .txt database
405 if ($max_rowid == 0) {
406 # force removal of database
407 $db = check_txt_db($isis_db);
408 $max_rowid = OpenIsis::maxRowid( $db );
411 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
413 my $path = $database;
415 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
416 my $row = OpenIsis::read( $db, $row_id );
417 if ($row && $row->{mfn}) {
419 progress($row->{mfn}, $max_rowid);
421 my $swishpath = $path."#".int($row->{mfn});
423 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
424 $xml = $cp2utf->convert($xml);
425 use bytes; # as opposed to chars
426 print "Path-Name: $swishpath\n";
427 print "Content-Length: ".(length($xml)+1)."\n";
428 print "Document-Type: XML\n\n$xml\n";
432 # for this to work with current version of OpenIsis (0.9.0)
433 # you might need my patch from
434 # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff
435 OpenIsis::close($db);
438 } elsif ($type_base eq "excel") {
439 use Spreadsheet::ParseExcel;
440 use Spreadsheet::ParseExcel::Utility qw(int2col);
442 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
443 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
445 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
446 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
448 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
451 foreach my $oWks (@{$oBook->{Worksheet}}) {
452 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
453 last if ($oWks->{Name} eq $sheet);
457 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
459 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
460 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
461 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
463 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
465 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
466 my $cell = $oWorksheet->{Cells}[$iR][$iC];
468 $row->{int2col($iC)} = $cell->Value;
472 progress($iR, $end_row);
474 # print "row[$iR/$end_row] ";
475 # foreach (keys %{$row}) {
476 # print "$_: ",$row->{$_},"\t";
480 my $swishpath = $database."#".$iR;
484 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
485 $xml = $cp2utf->convert($xml);
486 use bytes; # as opposed to chars
487 print "Path-Name: $swishpath\n";
488 print "Content-Length: ".(length($xml)+1)."\n";
489 print "Document-Type: XML\n\n$xml\n";
492 } elsif ($type_base eq "marc") {
496 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
497 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
499 # optional argument is format
500 my $format = x($config->{format}) || 'usmarc';
502 print STDERR "Reading MARC file '$marc_file'\n";
505 my $nr = $marc->openmarc({
506 file=>$marc_file, format=>$format
507 }) || die "Can't open MARC file '$marc_file'";
509 my $i=0; # record nr.
513 while ($marc->nextmarc(1)) {
518 my $swishpath = $database."#".$i;
520 if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
521 $xml = $cp2utf->convert($xml);
522 use bytes; # as opposed to chars
523 print "Path-Name: $swishpath\n";
524 print "Content-Length: ".(length($xml)+1)."\n";
525 print "Document-Type: XML\n\n$xml\n";
528 } elsif ($type_base eq "feed") {
530 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
531 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
533 print STDERR "Reading feed from program '$prog'\n";
535 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
537 my $i=1; # record nr.
546 my $swishpath = $database."#".$i++;
548 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
549 $xml = $cp2utf->convert($xml);
550 use bytes; # as opposed to chars
551 print "Path-Name: $swishpath\n";
552 print "Content-Length: ".(length($xml)+1)."\n";
553 print "Document-Type: XML\n\n$xml\n";
560 $line = $1 if (s/^(\d+):\s*//);
561 $data->{$line++} = $_;
569 # call this to commit index
574 ##########################################################################
578 all2xml.pl - read various file formats and dump XML for SWISH-E
582 This command will read ISIS data file using OpenIsis perl module, MARC
583 records using MARC module and optionally Micro$oft Excel files to
584 create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
585 this script B<isn't general xml generator> from isis files (isis allready
586 has something like that). Output of this script is tailor-made for SWISH-E.
590 Documentation is really lacking. However, in true Open Source spirit, source
591 is best documentation. I even made considerable effort to comment parts
592 which are not intuitively clear, so...
596 Dobrica Pavlinusic <dpavlin@rot13.org>
600 GNU Public License (GPL) v2 or later
604 SWISH-E web site at http://www.swish-e.org