--- trunk/all2xml.pl 2003/12/14 20:50:03 195 +++ trunk/all2xml.pl 2004/03/05 23:33:19 233 @@ -210,10 +210,23 @@ # placeholder for all repeatable entries for index + sub chk_eval($) { + my $data = shift; + if ($data =~ s/eval{([^}]+)}//) { + if (eval "$1") { + return $data; + } else { + return undef; + } + } else { + return $data; + } + } + sub mkformat($$) { my $x = shift || die "mkformat needs tag reference"; my $data = shift || return; - my $format_name = x($x->{format_name}) || return $data; + my $format_name = x($x->{format_name}) || return chk_eval($data); my $fmt = x($config->{format}->{$format_name}->{content}) || die " is not defined!"; my $format_delimiter = x($x->{format_delimiter}); my @data; @@ -226,10 +239,10 @@ if ($fmt) { my $nr = scalar $fmt =~ s/%s/%s/g; if (($#data+1) == $nr) { - return sprintf($fmt,@data); + return chk_eval(sprintf($fmt,@data)); } else { - print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; - return $data; + #print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; + return chk_eval($data); } } else { print STDERR "usage of link '$format_name' without defined format ( tag)\n"; @@ -282,32 +295,33 @@ no strict 'refs'; my $tmp = join(" ",&$filter($swish)) if ($s || $se); $swish_data .= $tmp if ($s); - $swish_exact_data .= $tmp if ($se); + $swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp ne ""); } else { $swish_data .= $swish if ($s); - $swish_exact_data .= $swish if ($se); + $swish_exact_data .= "xxbxx $swish xxexx " if ($se && $swish ne ""); } } # type="display" ; field for display if ($d && $display) { + my $ldel = $delimiter; if ($line_delimiter && $display_data) { - $display_data .= $line_delimiter; + $ldel = $line_delimiter; } if ($filter) { no strict 'refs'; - if ($display_data) { - $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); - } else { - $display_data = join($delimiter,mkformat($x,&$filter($display))); + my @arr; + foreach my $tmp (&$filter($display)) { + my $tmp2 = mkformat($x,$tmp); + push @arr,$tmp2 if ($tmp2); } + $display_data .= $ldel if ($display_data && @arr); + $display_data .= join($delimiter,@arr); } else { - if ($display_data) { - $display_data .= $delimiter.mkformat($x,$display); - } else { - $display_data = mkformat($x,$display); - } + $display_data .= $ldel if ($display_data); + my $tmp = mkformat($x,$display); + $display_data .= $tmp if ($tmp); } } @@ -328,11 +342,21 @@ if ($lookup_key) { print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; } else { - $lookup_key = $display; + if ($filter) { + no strict 'refs'; + $lookup_key = &$filter($display); + } else { + $lookup_key = $display; + } } } elsif (lc($x->{type}) eq "lookup_val") { if ($lookup_key) { - $lhash{$lookup_key} = $display; + if ($filter) { + no strict 'refs'; + $lhash{$lookup_key} = &$filter($display); + } else { + $lhash{$lookup_key} = $display; + } } else { print STDERR "WARNING: no lookup_key defined for '$display'?"; } @@ -451,7 +475,8 @@ for (my $page=0; $page <= $nr_pages; $page++) { my $display_data; if ($cache->{format}->{$field}) { - $display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); + my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); + $display_data=$tmp if ($tmp); } else { $display_data = $cache->{display_data}->{$field}->[$page]; } @@ -482,7 +507,7 @@ # add delimiters before and after word. # That is required to produce exact match - $xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); + $xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); } my $idel = $cache->{index_delimiter}->{$field}; @@ -524,7 +549,7 @@ # add delimiters before and after word. # That is required to produce exact match - $xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); + $xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); } } } @@ -577,6 +602,8 @@ #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; print STDERR "creating lookup file '$lookup_file'\n"; + # delete memory cache for lookup file + delete $cache->{lhash}; } # open existing lookup file @@ -774,23 +801,26 @@ my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; # optional argument is format - my $format = x($config->{format}) || 'usmarc'; + my $format = x($config->{marc_format}) || 'usmarc'; print STDERR "Reading MARC file '$marc_file'\n"; my $marc = new MARC; my $nr = $marc->openmarc({ file=>$marc_file, format=>$format - }) || die "Can't open MARC file '$marc_file'"; + }) || die "Can't open MARC file '$marc_file' with format '$format'"; - my $i=0; # record nr. + # read MARC file in memory + $marc->nextmarc(-1); - my $rec; + my $max_rec = $marc->marc_count(); - while ($marc->nextmarc(1)) { + for(my $i=1; $i<=$max_rec; $i++) { - # XXX - fakeprogress($i++); + progress($i,$max_rec); + + # store value for marc_sf.pm + $main::cache->{marc_record} = $i; my $swishpath = $database."#".$i; @@ -802,6 +832,9 @@ print "Document-Type: XML\n\n$xml\n"; } } + + print STDERR "\n"; + } elsif ($type_base eq "feed") { $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);