--- trunk2/all2all.pl	2004/09/14 17:07:59	441
+++ trunk2/all2all.pl	2004/10/10 17:36:43	504
@@ -7,7 +7,8 @@
 =cut
 
 use strict;
-use Data::Dumper;
+use locale;
+use YAML;
 use Carp;
 use Getopt::Long;
 
@@ -31,6 +32,20 @@
 	"low_mem!"	=> \$low_mem,
 );
 
+my $filter = {
+	'CROVOC' => sub {
+		my $tmp = shift || return;
+		return undef unless ($tmp =~ s/\s*CROVOC.*$/ */);
+		return $tmp;
+	},
+	'CROVOC_tree' => sub {
+		my $tmp = shift || return;
+		$tmp =~ s/\s*CROVOC.*$/ */;
+		$tmp =~ s/\s*EUROVOC.*//;
+		return $tmp;
+	},
+};
+
 # create WebPAC object
 #
 my $webpac = new WebPAC(
@@ -39,6 +54,7 @@
 	start_mfn => $start_mfn,
 	debug => $debug,
 	low_mem => $low_mem,
+	filter => $filter,
 ) || die;
 
 my $log = $webpac->_get_logger() || die "can't get logger";
@@ -57,15 +73,16 @@
 my $maxmfn = $webpac->open_isis(
 	filename => shift @ARGV || '/data/hidra/THS/THS',
 	lookup => [
-	{ 'key' => 'd:v900', 'val' => 'v250^a' },
+	{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' },
 #	{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
 #	{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
 #	{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
 	{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
 	{ 'key' => '900_mfn:v900', 'val' => 'v000' },
 	# tree structure
-	{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
+	{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
 	{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
+	{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
 	],
 );
 
@@ -82,8 +99,8 @@
 	my @ds = $webpac->data_structure($rec);
 
 	if (0 && $log->is_debug) {
-		$log->debug("rec = ",Dumper($rec));
-		$log->debug("ds = ",Dumper(\@ds));
+		$log->debug("rec = ",Dump($rec));
+		$log->debug("ds = ",Dump(\@ds));
 	}
 
 	next if (! @ds);
@@ -114,11 +131,26 @@
 	foreach my $ds (@ds) {
 		next if (! $ds->{'swish'});
 
+		# strip all non word characters from beginning or end
+		# of word
+		my $words = join(" ",@{$ds->{'swish'}});
+		$words =~ s/^\W+//;
+		$words =~ s/\W*\s+\W*/ /g;
+		$words =~ s/\W+$//;
+
+		# first try to generate headline for this entry from index
+		my $h = shift @{$ds->{'index'}};
+		# then, from display
+		$h ||= shift @{$ds->{'display'}};
+		# and as last resport, fallback to headline
+		$h ||= $headline;
+
 		$index->insert(
 			index_name => $ds->{'tag'},
-			path => $f,
-			headline => $headline,
-			words => join(" ",@{$ds->{'swish'}})
+			#path => $f,
+			path => $webpac->mfn,
+			headline => $h,
+			words => $words,
 		);
 	}
 
@@ -130,13 +162,13 @@
 
 		foreach my $h (@{$ds->{'index'}}) {
 			$thes->{$ds->{'tag'}}->insert(
-				path => $f,
 				headline => $h,
+				mfn => $webpac->mfn,
 			);
 		}
 	}
 
-#	print Dumper(\@ds);
+#	print Dump(\@ds);
 
 }
 
@@ -160,16 +192,16 @@
 }
 
 if (0 && $log->is_debug) {
-	$log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
-	$log->debug("data hash: ",Dumper($webpac->{'data'}));
+	$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
+	$log->debug("data hash: ",Dump($webpac->{'data'}));
 	foreach my $t (keys %{$thes}) {
-		$log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
+		$log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
 	}
 }
 
 } # XXX if(0)
 
-$log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
+$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
 
 $log->info("creating tree");
 
@@ -189,109 +221,92 @@
 	lookup_v900	=> sub { shift @{$l->{"root:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[1]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[1]}} },
-	have_children	=> sub { defined($l->{"a:".$_[0]."::"}) },
-	child_code	=> sub { return $_[0] },
+	have_children	=> sub { return $l->{"a:".$_[0]."::"} },
+	have_children_at_level => sub {
+				return unless (defined($l->{"code:".$_[1]}));
+				my $code = shift @{$l->{"code:".$_[1]}};
+				print STDERR "## $_[1] -> $code\n";
+				return undef unless($code);
+				return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
+			},
 	style		=> 'display: none',
 	},{
 	# 1
-	code_arr	=> sub { @{$l->{"a:".$_[0]."::"}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },	# nop
 	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{"a:".$_[1].":"}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":"} },
 	style		=> 'display: none',
 	},{
 	# 2
-	code_arr	=> sub { @{$l->{"a:".$_[0].":"}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub { shift },
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{"a:".$_[2].":".$_[1]}) },
-	child_code	=> sub { return "a:".$_[2].":".$_[1] },
-	style		=> 'display: none',
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
+	#style		=> 'display: none',
 	},{
 	# 3 uži pojam
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{$_[1]}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
 	},{
 	# 4
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{$_[1]}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
 	},{
 	# 5
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{$_[1]}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
 	},{
 	# 6
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{$_[1]}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
 	},{
 	# 7
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-	have_children	=> sub { defined($l->{$_[1]}) },
-	child_code	=> sub { return $_[1] },
+	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
 	},{
 	# 8
-	code_arr	=> sub { @{$l->{$_[0]}} },
+	code_arr	=> sub { @{$_[0]} },
 	filter_code	=> sub { shift },
-	lookup_v900	=> sub {
-				my ($c,$p) = @_;
-				$p =~ s/^a:(..:....):.*$/$1/;
-				return "a:".$p.":".$c;
-			},
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
+	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
+	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
+#	have_children	=> sub { return $l->{"a:".$_[1].":".$_[0]} },
+	have_children	=> sub { 0 },
+	},{
+	# 9 - level which is never reached except explicitly
+	code_arr	=> sub { @{$_[0]} },
+	filter_code	=> sub { shift },
+	lookup_v900	=> sub { shift @{$l->{"code:".$_[0]}} },
 	lookup_term	=> sub { shift @{$l->{"d:".$_[0]}} },
 	lookup_mfn	=> sub { shift @{$l->{"900_mfn:".$_[0]}} },
-#	have_children	=> sub { defined($l->{$_[1]}) },
-#	child_code	=> sub { return $_[1] },
 	have_children	=> sub { 0 },
-	child_code	=> sub { 0 },
+	have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
+	},{
 });
 
 my $tree = new WebPAC::Tree(