--- branches/tehnika/all2xml.pl 2005/01/16 18:57:15 633 +++ branches/tehnika/all2xml.pl 2005/02/17 18:18:49 674 @@ -22,7 +22,8 @@ my $config; #use index_DBI; # default DBI module for index -use index_DBI_cache; # faster DBI module using memory cache +#use index_DBI_cache; # faster DBI module using memory cache +use index_DBI_filter; # filter support for indexes my $index; my %opts; @@ -170,6 +171,8 @@ ($s,$se,$d,$i) = (0,1,0,0); } elsif (lc($type) =~ /^lookup/) { ($s,$se,$d,$i,$il) = (0,1,0,0,1); + } elsif ($type) { + print STDERR "WARNING: unknown type: $type\n"; } return ($s,$se,$d,$i,$il); } @@ -255,8 +258,6 @@ # init vars so that we go into while... ($swish,$display) = (1,1); - # placeholder for all repeatable entries for index - sub mkformat($$) { my $x = shift || die "mkformat needs tag reference"; my $data = shift || return; @@ -632,6 +633,10 @@ foreach my $database ($cfg->Sections) { + # save database name in global variable path for later + # (need for index filter creation) + $path = $database; + my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; my $add_xml = $cfg -> val($database, 'xml'); # optional @@ -737,12 +742,15 @@ $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); my $db = new Biblio::Isis( isisdb => $isis_db ); - my $max_rowid = $db->count || die "can't find maxmfn"; + if (! $db) { + print STDERR "FATAL: can't read ISIS database: $isis_db, skipping...\n"; + next; + } + + my $max_rowid = $db->count if ($db); print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; - my $path = $database; - for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { my $row = $db->to_hash( $row_id ); if ($row) { @@ -824,6 +832,9 @@ print "Document-Type: XML\n\n$xml\n"; } } + + print STDERR "\n"; + } elsif ($type_base eq "marc") { require MARC::File::USMARC; @@ -835,13 +846,17 @@ warn "marc_format is no longer used!" if ($config->{marc_format}); print STDERR "Reading MARC file '$marc_file'\n"; - my $marc = MARC::File::USMARC->in( $marc_file ) - || die "Can't open MARC file '$marc_file': ".$MARC::File::ERROR; + my $marc = MARC::File::USMARC->in( $marc_file ); + + if (! $marc) { + print STDERR "FATAL: can't read MARC file: $marc_file, skipping...\n"; + next; + } # count records in MARC file sub marc_count { my $filename = shift || die; - my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; + my $file = MARC::File::USMARC->in($filename) || return; my $count = 0; while ($file->skip()) { $count++; @@ -851,11 +866,11 @@ my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; - my $i = 0; + my $i = 1; while( my $rec = $marc->next() ) { - progress($i++,$count); + progress($i,$count); my $swishpath = $database."#".$i; @@ -866,6 +881,8 @@ print "Content-Length: ".(length($xml)+1)."\n"; print "Document-Type: XML\n\n$xml\n"; } + + $i++; } print STDERR "\n"; @@ -930,8 +947,8 @@ =head1 DESCRIPTION -This command will read ISIS data file using IsisDB perl module, MARC -records using MARC module and optionally Micro$oft Excel files to +This command will read ISIS data file using Biblio::Isis perl module, MARC +records using MARC::File module and optionally Micro$oft Excel files to create one XML file for usage with I indexer. Dispite it's name, this script B from isis files (isis allready has something like that). Output of this script is tailor-made for SWISH-E.