--- trunk/all2xml.pl 2003/10/29 22:46:49 136 +++ trunk/all2xml.pl 2003/11/22 22:04:05 164 @@ -96,6 +96,7 @@ $field_usage{$field}++; my $swish_data = ""; + my $swish_exact_data = ""; my $display_data = ""; my $line_delimiter; @@ -109,9 +110,10 @@ my $repeat_off = 0; # repeatable offset - my ($s,$d,$i) = (1,1,0); # swish, display default + my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default $s = 0 if (lc($x->{type}) eq "display"); $d = 0 if (lc($x->{type}) eq "swish"); + $se = 1 if (lc($x->{type}) eq "swish_exact"); ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); # what will separate last line from this one? @@ -128,6 +130,32 @@ my @index_data; my $index_filter; + sub mkformat { + my $x = shift || die "mkformat needs tag reference"; + my $data = shift || return; + my $format_name = x($x->{format_name}) || return $data; + my $fmt = x($config->{format}->{$format_name}->{content}) || die " is not defined!"; + my $format_delimiter = x($x->{format_delimiter}); + my @data; + if ($format_delimiter) { + @data = split(/$format_delimiter/,$data); + } else { + push @data,$data; + } + + if ($fmt) { + my $nr = scalar $fmt =~ s/%s/%s/g; + if (($#data+1) == $nr) { + return sprintf($fmt,@data); + } else { + print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; + return $data; + } + } else { + print STDERR "usage of link '$format_name' without defined format ( tag)\n"; + } + } + # while because of repeatable fields while ($swish || $display) { ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); @@ -143,12 +171,16 @@ require "filter/".$filter.".pm"; } # type="swish" ; field for swish - if ($s && $swish) { - if ($filter) { + if ($swish) { + if ($filter && ($s || $se)) { no strict 'refs'; - $swish_data .= join(" ",&$filter($swish)); + my $tmp = join(" ",&$filter($swish)) if ($s || $se); + $swish_data .= $tmp if ($s); + $swish_exact_data .= $tmp if ($se); + } else { - $swish_data .= $swish; + $swish_data .= $swish if ($s); + $swish_exact_data .= $swish if ($se); } } @@ -160,9 +192,17 @@ } if ($filter) { no strict 'refs'; - $display_data .= $delimiter.&$filter($display); + if ($display_data) { + $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); + } else { + $display_data = join($delimiter,mkformat($x,&$filter($display))); + } } else { - $display_data .= $delimiter.$display; + if ($display_data) { + $display_data .= $delimiter.mkformat($x,$display); + } else { + $display_data = mkformat($x,$display); + } } } @@ -197,6 +237,8 @@ my ($s,$d,$i) = (1,1,0); # swish, display default $s = 0 if (lc($x->{type}) eq "display"); $d = 0 if (lc($x->{type}) eq "swish"); + # no support for swish exact in config. + # IMHO, it's useless ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); if ($val) { @@ -239,6 +281,15 @@ $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); } + if ($swish_exact_data) { + $swish_exact_data =~ s/ +/ /g; + $swish_exact_data =~ s/ +$//g; + + # 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')); + } + } @@ -274,6 +325,11 @@ my $show_progress = $cfg_global->val('global', 'show_progress'); +my $unac_filter = $cfg_global->val('global', 'unac_filter'); +if ($unac_filter) { + require $unac_filter; +} + foreach my $database ($cfg->Sections) { my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; @@ -285,7 +341,7 @@ my $type_base = $type; $type_base =~ s/_.+$//g; - $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); + $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); # output current progress indicator my $last_p = 0;