--- trunk/all2xml.pl 2003/11/29 15:59:19 182 +++ trunk/all2xml.pl 2003/12/15 00:12:16 196 @@ -72,6 +72,8 @@ # if you are tight on memory, turn this off my $use_lhash_cache = 1; +my $last_field_name; # cache to prevent repeated fields + sub data2xml { use xmlify; @@ -117,6 +119,7 @@ delete $cache->{swish_data}; delete $cache->{swish_exact_data}; delete $cache->{index_data}; + delete $cache->{index_delimiter}; my @page_fields; # names of fields @@ -137,7 +140,13 @@ print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; } if ($field_name) { - return x($field_name); + if (! $last_field_name) { + $last_field_name = x($field_name); + return $last_field_name; + } elsif ($field_name ne $last_field_name) { + $last_field_name = x($field_name); + return $last_field_name; + } } } @@ -164,6 +173,14 @@ my %page_max = (); # default line_delimiter if using my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '
'; + $cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter}; + + my $format_name = $config->{indexer}->{$field}->{format_name}; + my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter}; + if ($format_name && $format_delimiter) { + $cache->{format}->{$field}->{format_name} = $format_name; + $cache->{format}->{$field}->{format_delimiter} = $format_delimiter; + } foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { @@ -177,7 +194,7 @@ my ($s,$se,$d,$i,$il) = (1,0,1,0,0); $s = 0 if (lc($x->{type}) eq "display"); $d = 0 if (lc($x->{type}) eq "swish"); - ($s,$se,$d,$i) = (0,0,0,1) if (lc($x->{type}) eq "index"); + ($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index"); ($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact"); ($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/); @@ -193,7 +210,7 @@ # placeholder for all repeatable entries for index - sub mkformat { + sub mkformat($$) { my $x = shift || die "mkformat needs tag reference"; my $data = shift || return; my $format_name = x($x->{format_name}) || return $data; @@ -265,11 +282,23 @@ no strict 'refs'; my $tmp = join(" ",&$filter($swish)) if ($s || $se); $swish_data .= $tmp if ($s); - $swish_exact_data .= $tmp if ($se); + if ($se) { + if ($swish_exact_data) { + $swish_exact_data .= "xxexx xxbxx ".$tmp; + } else { + $swish_exact_data .= $tmp; + } + } } else { $swish_data .= $swish if ($s); - $swish_exact_data .= $swish if ($se); + if ($se) { + if ($swish_exact_data) { + $swish_exact_data .= "xxexx xxbxx ".$swish; + } else { + $swish_exact_data .= $swish; + } + } } } @@ -360,7 +389,11 @@ @index_data = @{$cache->{index_data}->{$field}->[$page]}; } if ($x->{append}) { - $index_data[$#index_data] .= $idisplay; + if (@index_data) { + $index_data[$#index_data] .= $idisplay; + } else { + push @index_data, $idisplay; + } } else { push @index_data, $idisplay; } @@ -371,9 +404,14 @@ } if (! $iterate_by_page) { + my $idel = $x->{index_delimiter}; # fill data in index - foreach my $d (@index_data) { - $index->insert($field, $d, $path); + foreach my $tmp (@index_data) { + my $i = $d = $tmp; + if ($idel && $tmp =~ m/$idel/) { + ($i,$d) = split(/$idel/,$tmp); + } + $index->insert($field, $i, $d, $path); } @index_data = (); } @@ -423,8 +461,12 @@ #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; #print STDERR Dumper($cache->{display_data}); for (my $page=0; $page <= $nr_pages; $page++) { - - my $display_data = $cache->{display_data}->{$field}->[$page]; + my $display_data; + if ($cache->{format}->{$field}) { + $display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); + } else { + $display_data = $cache->{display_data}->{$field}->[$page]; + } if ($display_data) { # default if ($field eq "headline") { $xml .= xmlify("headline", $display_data); @@ -455,9 +497,15 @@ $xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); } - foreach my $d (@{$cache->{index_data}->{$field}->[$page]}) { - $index->insert($field, $d, $path); -#print STDERR "index $field: $d [$path]\n"; + my $idel = $cache->{index_delimiter}->{$field}; + foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { + my $i = $tmp; + my $d = $tmp; + if ($idel && $tmp =~ m/$idel/) { + ($i,$d) = split(/$idel/,$tmp); + } + $index->insert($field, $i, $d, $path); +#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; } }