--- trunk/all2xml.pl 2003/07/04 23:29:27 67 +++ trunk/all2xml.pl 2003/07/14 10:54:34 102 @@ -82,17 +82,17 @@ # sort subrouting using order="" attribute sub by_order { - return 0 if (! $config->{indexer}->{$a}->{order}); - return 0 if (! $config->{indexer}->{$b}->{order}); + my $va = $config->{indexer}->{$a}->{order} || + $config->{indexer}->{$a}; + my $vb = $config->{indexer}->{$b}->{order} || + $config->{indexer}->{$b}; - return $config->{indexer}->{$a}->{order} <=> - $config->{indexer}->{$b}->{order} ; + return $va <=> $vb; } foreach my $field (sort by_order keys %{$config->{indexer}}) { $field=x($field); - $field_usage{$field}++; my $swish_data = ""; @@ -124,8 +124,18 @@ # init vars so that we go into while... ($swish,$display) = (1,1); - if ($swish || $display) { + # placeholder for all repeatable entries for index + my @index_data; + my $index_filter; + + # while because of repeatable fields + while ($swish || $display) { ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); + if ($repeat_off > 1000) { + print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; + last; + } + # filter="name" ; filter this field through # filter/[name].pm my $filter = $x->{filter}; @@ -162,14 +172,21 @@ # type="index" ; insert into index if ($i && $display) { - my $index_data = $display; - if ($filter) { - no strict 'refs'; - foreach my $d (&$filter($index_data)) { - $index->insert($field, $d, $path); - } - } else { - $index->insert($field, $index_data, $path); + push @index_data, $display; + $index_filter = $filter if ($filter); + } + } + + # fill data in index + if (@index_data) { + if ($index_filter) { + no strict 'refs'; + foreach my $d (&$index_filter(@index_data)) { + $index->insert($field, $d, $path); + } + } else { + foreach my $d (@index_data) { + $index->insert($field, $d, $path); } } } @@ -231,7 +248,8 @@ # dump formatted output in if ($html) { - $xml .= xmlify("html",$html); + #$xml .= xmlify("html",$html); + $xml .= ""; } if ($xml) { @@ -258,6 +276,8 @@ $cfg_global->val('global', 'dbi_passwd') || '', ); +my $show_progress = $cfg_global->val('global', 'show_progress'); + foreach my $database ($cfg->Sections) { my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; @@ -274,7 +294,7 @@ # output current progress indicator my $last_p = 0; sub progress { - #return if (! $opts{q}); # FIXME + return if (! $show_progress); my $current = shift; my $total = shift || 1; my $p = int($current * 100 / $total); @@ -307,6 +327,19 @@ $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); my $db = OpenIsis::open( $isis_db ); + # OpenIsis::ERR_BADF + if ($db == -4) { + print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; + next; + # OpenIsis::ERR_IO + } elsif ($db == -5) { + print STDERR "FATAL: OpenIsis can't access file $isis_db\n"; + next; + } elsif ($db < 0) { + print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n"; + next; + } + my $max_rowid = OpenIsis::maxRowid( $db ); print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; @@ -431,7 +464,7 @@ open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; - my $i=0; # record nr. + my $i=1; # record nr. my $data; my $line=1; @@ -440,7 +473,7 @@ chomp; if (/^$/) { - my $swishpath = $database."#".$i; + my $swishpath = $database."#".$i++; if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { $xml = $cp2utf->convert($xml); @@ -454,9 +487,10 @@ next; } + $line = $1 if (s/^(\d+):\s*//); $data->{$line++} = $_; - fakeprogress($i++); + fakeprogress($i); } } @@ -471,15 +505,21 @@ =head1 NAME -isis2xml.pl - read isis file and dump XML +all2xml.pl - read various file formats and dump XML for SWISH-E =head1 DESCRIPTION -This command will read ISIS data file using OpenIsis perl module and -create XML file for usage with I -indexer. Dispite it's name, this script B -from isis files (isis allready has something like that). Output of this -script is tailor-made for SWISH-E. +This command will read ISIS data file using OpenIsis perl module, MARC +records using MARC module and optionally Micro$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. + +=head1 BUGS + +Documentation is really lacking. However, in true Open Source spirit, source +is best documentation. I even made considerable effort to comment parts +which are not intuitively clear, so... =head1 AUTHOR