--- trunk/all2xml.pl 2003/11/25 20:19:03 181
+++ trunk/all2xml.pl 2003/11/29 19:07:00 188
@@ -116,6 +116,8 @@
delete $cache->{display_data};
delete $cache->{swish_data};
delete $cache->{swish_exact_data};
+ delete $cache->{index_data};
+ delete $cache->{index_delimiter};
my @page_fields; # names of fields
@@ -150,6 +152,7 @@
my $swish_data = "";
my $swish_exact_data = "";
my $display_data = "";
+ my @index_data;
my $line_delimiter;
my ($swish,$display);
@@ -162,6 +165,7 @@
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};
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
@@ -175,12 +179,12 @@
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");
- $il = 1 if (lc($x->{type}) =~ /^lookup/);
+ ($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/);
# what will separate last line from this one?
- if ($display_data && $x->{append} && $x->{append} eq "1") {
+ if ($display_data && $x->{append}) {
$line_delimiter = ' ';
} elsif ($display_data) {
$line_delimiter = '
';
@@ -190,7 +194,6 @@
($swish,$display) = (1,1);
# placeholder for all repeatable entries for index
- my @index_data;
sub mkformat {
my $x = shift || die "mkformat needs tag reference";
@@ -294,16 +297,14 @@
}
# type="index" ; insert into index
+ my $idisplay;
if ($i && $display) {
+ $idisplay = $display;
if ($filter) {
no strict 'refs';
- $display = &$filter($display);
- }
- if ($x->{append} && @index_data) {
- $index_data[$#index_data].=$display;
- } else {
- push @index_data, $display;
+ $idisplay = &$filter($idisplay);
}
+ push @index_data, $idisplay if (! $iterate_by_page);
}
# store fields in lookup
@@ -329,17 +330,18 @@
sub iterate_fld($$$$$$) {
my ($cache,$what,$field,$page,$data,$append) = @_;
return if (!$data);
- my $line_delimiter = $page_line_delimiter;
- $line_delimiter = " " if ($append);
+
+ my $ldel = $page_line_delimiter;
+ $ldel = " " if ($append);
+#print STDERR "line delimiter: ",Dumper($ldel) if ($ldel);
if (! $cache->{$what}->{$field}->[$page]) {
$cache->{$what}->{$field}->[$page] = $data;
} else {
- $cache->{$what}->{$field}->[$page] .= $line_delimiter.$data;
+ $cache->{$what}->{$field}->[$page] .= $ldel.$data;
}
}
if ($display_data) {
-print STDERR "line delimiter: ",Dumper($line_delimiter) if ($line_delimiter);
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append});
}
$display_data = "";
@@ -351,12 +353,36 @@
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append});
$swish_exact_data = "";
}
+
+ if ($idisplay) {
+ my $ldel=$page_line_delimiter;
+ my @index_data;
+ if ($cache->{index_data}->{$field}->[$page]) {
+
+ @index_data = @{$cache->{index_data}->{$field}->[$page]};
+ }
+ if ($x->{append}) {
+ $index_data[$#index_data] .= $idisplay;
+ } else {
+ push @index_data, $idisplay;
+ }
+ $idisplay = "";
+ @{$cache->{index_data}->{$field}->[$page]} = @index_data;
+ }
}
}
- # fill data in index
- foreach my $d (@index_data) {
- $index->insert($field, $d, $path);
+ if (! $iterate_by_page) {
+ my $idel = $x->{index_delimiter};
+ # fill data in index
+ 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 = ();
}
}
@@ -435,6 +461,17 @@
# That is required to produce exact match
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx'));
}
+
+ 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";
+ }
}
}