--- trunk/all2xml.pl 2003/11/25 20:19:03 181
+++ trunk/all2xml.pl 2004/01/07 12:29:11 199
@@ -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;
@@ -116,6 +118,8 @@
delete $cache->{display_data};
delete $cache->{swish_data};
delete $cache->{swish_exact_data};
+ delete $cache->{index_data};
+ delete $cache->{index_delimiter};
my @page_fields; # names of fields
@@ -136,7 +140,13 @@
print STDERR "WARNING: field '$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;
+ }
}
}
@@ -150,6 +160,7 @@
my $swish_data = "";
my $swish_exact_data = "";
my $display_data = "";
+ my @index_data;
my $line_delimiter;
my ($swish,$display);
@@ -162,6 +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}}) {
@@ -175,12 +194,12 @@
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) {
$line_delimiter = '
';
@@ -190,9 +209,8 @@
($swish,$display) = (1,1);
# placeholder for all repeatable entries for index
- my @index_data;
- sub mkformat {
+ sub mkformat($$) {
my $x = shift || die "mkformat needs tag reference";
my $data = shift || return;
my $format_name = x($x->{format_name}) || return $data;
@@ -264,11 +282,11 @@
no strict 'refs';
my $tmp = join(" ",&$filter($swish)) if ($s || $se);
$swish_data .= $tmp if ($s);
- $swish_exact_data .= $tmp if ($se);
+ $swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp ne "");
} else {
$swish_data .= $swish if ($s);
- $swish_exact_data .= $swish if ($se);
+ $swish_exact_data .= "xxbxx $swish xxexx " if ($se && $swish ne "");
}
}
@@ -280,9 +298,9 @@
if ($filter) {
no strict 'refs';
if ($display_data) {
- $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display)));
+ $display_data .= $delimiter.mkformat($x,join($delimiter,&$filter($display)));
} else {
- $display_data = join($delimiter,mkformat($x,&$filter($display)));
+ $display_data = mkformat($x,join($delimiter,&$filter($display)));
}
} else {
if ($display_data) {
@@ -294,16 +312,14 @@
}
# type="index" ; insert into index
+ my $idisplay;
if ($i && $display) {
+ $idisplay = $display;
if ($filter) {
no strict 'refs';
- $display = &$filter($display);
- }
- if ($x->{append} && @index_data) {
- $index_data[$#index_data].=$display;
- } else {
- push @index_data, $display;
+ $idisplay = &$filter($idisplay);
}
+ push @index_data, $idisplay if (! $iterate_by_page);
}
# store fields in lookup
@@ -329,17 +345,18 @@
sub iterate_fld($$$$$$) {
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});
}
$display_data = "";
@@ -351,12 +368,40 @@
iterate_fld($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}) {
+ if (@index_data) {
+ $index_data[$#index_data] .= $idisplay;
+ } else {
+ push @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, $path);
+ }
+ @index_data = ();
}
}
@@ -404,8 +449,12 @@
#print STDERR "field '$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);
@@ -433,7 +482,18 @@
# 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};
+ 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";
}
}
@@ -464,7 +524,7 @@
# 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));
}
}
}