--- trunk/all2xml.pl	2002/11/24 20:52:11	1
+++ trunk2/all2all.pl	2004/09/08 15:30:07	415
@@ -1,66 +1,142 @@
 #!/usr/bin/perl -w
 
+=head1 NAME
+
+all2all.pl - basic script for all WebPAC needs
+
+=cut
+
 use strict;
-use OpenIsis;
-use Getopt::Std;
 use Data::Dumper;
-use XML::Simple;
+use Carp;
 
-my $config=XMLin();
+use lib './lib';
+use WebPAC;
+use WebPAC::jsFind;
+use WebPAC::Index;
 
-print Dumper($config);
+my $webpac = new WebPAC(
+	code_page => 'ISO-8859-2',
+	limit_mfn => 500,
+#	debug => 1,
+) || die;
 
-my %opts;
+my $log = $webpac->_get_logger();
 
-getopts('d:m:q', \%opts);
+my $index = new WebPAC::jsFind(
+	index_path => './out/index',
+	keys => 10,
+) || die;
 
-my $db_dir = $opts{d};
+my $thes;
 
-die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+$|=1;
 
-#--------------------------------------------------------------------
+my $maxmfn = $webpac->open_isis(
+	filename => shift @ARGV || '/data/hidra/THS/THS',
+	lookup => [
+	{ 'key' => 'd:v900', 'val' => 'v250^a' },
+#	{ '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' },
+	],
+);
 
-my $last_tell=0;
+$log->info("rows: $maxmfn");
 
-my @isis_dirs = ( '.' );	# use dirname as database name
+$webpac->open_import_xml(type => 'isis_hidra_ths');
 
-if ($opts{m}) {
-	@isis_dirs = split(/,/,$opts{m});
-}
+while (my $rec = $webpac->fetch_rec) {
 
-my @isis_dbs;
+	my @ds = $webpac->data_structure($rec);
 
-foreach (@isis_dirs) {
-	if (-e "$common::isis_data/$db_dir/$_/LIBRI") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI";
+	if (0 && $log->is_debug) {
+		$log->debug("rec = ",Dumper($rec));
+		$log->debug("ds = ",Dumper(\@ds));
 	}
-	if (-e "$common::isis_data/$db_dir/$_/PERI") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI";
+
+	next if (! @ds);
+
+	my $filename = $webpac->{'current_filename'};
+
+	if ($filename) {
+		$webpac->output_file(
+			file => $filename,
+			template => 'html.tt',
+			data => \@ds,
+			headline => $webpac->{'headline'},
+		);
+	} else {
+		print $webpac->output(
+			template => 'text.tt',
+			data => \@ds,
+			headline => $webpac->{'headline'},
+		);
 	}
-	if (-e "$common::isis_data/$db_dir/$_/AMS") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS";
+
+	my $headline = $webpac->{'headline'};
+
+	my $f = $filename;
+	$f =~ s!out/!!;
+
+	# save into index
+	foreach my $ds (@ds) {
+		next if (! $ds->{'swish'});
+
+		$index->insert(
+			index_name => $ds->{'tag'},
+			path => $f,
+			headline => $headline,
+			words => join(" ",@{$ds->{'swish'}})
+		);
 	}
-	if (-e "$common::isis_data/$db_dir/$_/ARTI") {
-#		push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI";
+
+	# save into sorted index (thesaurus)
+	foreach my $ds (@ds) {
+		next if (! $ds->{'index'});
+
+		$thes->{$ds->{'tag'}} ||= new WebPAC::Index;
+
+		foreach my $h (@{$ds->{'index'}}) {
+			$thes->{$ds->{'tag'}}->insert(
+				path => $f,
+				headline => $h,
+			);
+		}
 	}
-}
 
-foreach my $isis_db (@isis_dbs) {
+#	print Dumper(\@ds);
+
+}
 
-	my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'";
+foreach my $t (keys %{$thes}) {
 
-	my $max_rowid = OpenIsis::maxRowid( $db );
+	my @e = $thes->{$t}->elements;
+	if (! @e) {
+		$log->logwarn("no elements in sorted index $t?");
+		next;
+	}
 
-	my $last_pcnt = 0;
+	my $file = "./out/bfilter/$t.txt";
+	$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
 
-	for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
-		my $row = OpenIsis::read( $db, $row_id );
+	$webpac->output_file(
+		file => $file,
+		template => 'index.tt',
+		data => \@e,
+		index_name => $t,
+	);
+}
 
-		# output current process indicator
-		my $pcnt = int($row->{mfn} * 100 / $max_rowid);
-		if ($pcnt != $last_pcnt) {
-			printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q});
-			$last_pcnt = $pcnt;
-		}
+if (0 && $log->is_debug) {
+	$log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
+	$log->debug("data hash: ",Dumper($webpac->{'data'}));
+	foreach my $t (keys %{$thes}) {
+		$log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
 	}
 }
+
+$index->close;
+