--- trunk/run.pl	2005/11/25 00:22:58	141
+++ trunk/run.pl	2006/05/14 13:25:36	492
@@ -7,108 +7,285 @@
 use Data::Dumper;
 use lib './lib';
 
+use WebPAC::Common 0.02;
 use WebPAC::Lookup;
-use WebPAC::Input::ISIS;
-use WebPAC::DB 0.02;
+use WebPAC::Input 0.03;
+use WebPAC::Store 0.03;
 use WebPAC::Normalize::XML;
+use WebPAC::Normalize::Set;
 use WebPAC::Output::TT;
-use WebPAC::Output::Estraier;
 use YAML qw/LoadFile/;
+use Getopt::Long;
+use File::Path;
+use Time::HiRes qw/time/;
+use File::Slurp;
 
-my $limit = shift @ARGV;
+=head1 NAME
 
-my $config = LoadFile('conf/config.yml');
+run.pl - start WebPAC indexing
 
-print "config = ",Dumper($config);
+B<this command will probably go away. Don't get used to it!>
 
-my $type = lc($config->{input}->{type});
+Options:
 
-die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
+=over 4
 
-my $abs_path = abs_path($0);
-$abs_path =~ s#/[^/]*$#/#;
+=item --offset 42
 
-my $lookup = new WebPAC::Lookup(
-	lookup_file => $config->{input}->{lookup},
-);
+start loading (all) databases at offset 42
 
+=item --limit 100
 
+limit loading to 100 records
 
-my $isis = new WebPAC::Input::ISIS(
-	code_page => $config->{webpac}->{webpac_encoding},
-	limit_mfn => $config->{input}->{limit},
-);
+=item --clean
 
-my $maxmfn = $isis->open(
-	filename => $config->{input}->{path},
-	code_page => $config->{input}->{encoding},	# database encoding
-);
+remove database and Hyper Estraier index before indexing
 
-my $path = './db/';
+=item --only=database_name
 
-my $db = new WebPAC::DB(
-	path => $config->{webpac}->{db_path},
-);
+reindex just single database (legacy name is --one)
 
-my $n = new WebPAC::Normalize::XML(
-#	filter => { 'foo' => sub { shift } },
-	db => $db,
-	lookup_regex => $lookup->regex,
-	lookup => $lookup,
-);
+=item --config conf/config.yml
 
-$n->open(
-	tag => $config->{normalize}->{tag},
-	xml_file => $config->{normalize}->{path},
-);
+path to YAML configuration file
 
-my $out = new WebPAC::Output::TT(
-	include_path => $config->{webpac}->{template_path},
-	filters => { foo => sub { shift } },
-);
+=item --force-set
 
-my $est = new WebPAC::Output::Estraier(
-	%{ $config->{hyperestraier} }
+force conversion C<normalize->path> in C<config.yml> from
+C<.xml> to C<.pl>
+
+=back
+
+=cut
+
+my $offset;
+my $limit;
+
+my $clean = 0;
+my $config = 'conf/config.yml';
+my $debug = 0;
+my $only_db_name;
+my $force_set = 0;
+
+GetOptions(
+	"limit=i" => \$limit,
+	"offset=i" => \$offset,
+	"clean" => \$clean,
+	"one=s" => \$only_db_name,
+	"only=s" => \$only_db_name,
+	"config" => \$config,
+	"debug" => \$debug,
+	"force-set" => \$force_set,
 );
 
+$config = LoadFile($config);
+
+print "config = ",Dumper($config) if ($debug);
+
+die "no databases in config file!\n" unless ($config->{databases});
+
+my $log = _new WebPAC::Common()->_get_logger();
+
+my $use_indexer = $config->{use_indexer} || 'hyperestraier';
+$log->info("using $use_indexer indexing engine...");
+
 my $total_rows = 0;
+my $start_t = time();
+
+while (my ($database, $db_config) = each %{ $config->{databases} }) {
+
+	next if ($only_db_name && $database !~ m/$only_db_name/i);
+
+	my $indexer;
+
+	my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
+	$indexer_config->{database} = $database;
+	$indexer_config->{clean} = $clean;
+	$indexer_config->{label} = $db_config->{name};
 
-for ( 0 ... $isis->size ) {
+	# important: clean database just once!
+	$clean = 0;
 
-	my $row = $isis->fetch || next;
+	if ($use_indexer eq 'hyperestraier') {
 
-	my $mfn = $row->{'000'}->[0] || die "can't find MFN";
+		# open Hyper Estraier database
+		use WebPAC::Output::Estraier '0.10';
+		$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
+	
+	} elsif ($use_indexer eq 'kinosearch') {
 
-	my $ds = $n->data_structure($row);
+		# open KinoSearch
+		use WebPAC::Output::KinoSearch;
+		$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
+		$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
 
-#	print STDERR Dumper($row, $ds);
-
-#	my $html = $out->apply(
-#		template => 'html_ffzg.tt',
-#		data => $ds,
-#	);
-#
-#	# create test output
-#
-#	my $file = sprintf('out/%02d.html', $mfn );
-#	open(my $fh, '>', $file) or die "can't open $file: $!";
-#	print $fh $html;
-#	close($fh);
-#
-#	$html =~ s#\s*[\n\r]+\s*##gs;
-#
-#	print STDERR $html;
-
-	$est->add(
-		id => $mfn,
-		ds => $ds,
-		type => $config->{hyperestraier}->{type},
+	} else {
+		$log->logdie("unknown use_indexer: $use_indexer");
+	}
+
+	$log->logide("can't continue without valid indexer") unless ($indexer);
+
+	#
+	# now WebPAC::Store
+	#
+	my $abs_path = abs_path($0);
+	$abs_path =~ s#/[^/]*$#/#;
+
+	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
+
+	if ($clean) {
+		$log->info("creating new database $database in $db_path");
+		rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
+	} else {
+		$log->debug("working on $database in $db_path");
+	}
+
+	my $db = new WebPAC::Store(
+		path => $db_path,
+		database => $database,
+		debug => $debug,
 	);
 
-	$total_rows++;
 
-};
+	#
+	# now, iterate through input formats
+	#
+
+	my @inputs;
+	if (ref($db_config->{input}) eq 'ARRAY') {
+		@inputs = @{ $db_config->{input} };
+	} elsif ($db_config->{input}) {
+		push @inputs, $db_config->{input};
+	} else {
+		$log->info("database $database doesn't have inputs defined");
+	}
+
+	my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
+
+	foreach my $input (@inputs) {
+
+		my $type = lc($input->{type});
+
+		die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
+
+		my $lookup = new WebPAC::Lookup(
+			lookup_file => $input->{lookup},
+		);
+
+		my $input_module = $config->{webpac}->{inputs}->{$type};
+
+		$log->info("working on input '$input->{path}' [$input->{type}] using $input_module lookup '$input->{lookup}'");
+
+		my $input_db = new WebPAC::Input(
+			module => $input_module,
+			code_page => $config->{webpac}->{webpac_encoding},
+			limit => $limit || $input->{limit},
+			offset => $offset,
+			lookup => $lookup,
+			recode => $input->{recode},
+		);
+		$log->logdie("can't create input using $input_module") unless ($input);
+
+		my $maxmfn = $input_db->open(
+			path => $input->{path},
+			code_page => $input->{encoding},	# database encoding
+		);
+
+		my $n = new WebPAC::Normalize::XML(
+		#	filter => { 'foo' => sub { shift } },
+			db => $db,
+			lookup_regex => $lookup->regex,
+			lookup => $lookup,
+			prefix => $input->{name},
+		);
+
+		my $rules;
+		my $normalize_path = $input->{normalize}->{path};
+
+		if ($force_set) {
+			my $new_norm_path = $normalize_path;
+			$new_norm_path =~ s/\.xml$/.pl/;
+			if (-e $new_norm_path) {
+				$log->info("--force-set replaced $normalize_path with $new_norm_path");
+				$normalize_path = $new_norm_path;
+			} else {
+				$log->warn("--force-set failed on $new_norm_path, fallback to $normalize_path");
+			}
+		}
+
+		if ($normalize_path =~ m/\.xml$/i) {
+			$n->open(
+				tag => $input->{normalize}->{tag},
+				xml_file => $normalize_path,
+			);
+		} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) {
+			$n->open_yaml(
+				path => $normalize_path,
+				tag => $input->{normalize}->{tag},
+			);
+		} elsif ($normalize_path =~ m/\.(?:pl)$/i) {
+			$n = undef;
+			$log->info("using WebPAC::Normalize::Set to process $normalize_path");
+			$rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
+		}
+
+		foreach my $pos ( 0 ... $input_db->size ) {
+
+			my $row = $input_db->fetch || next;
+
+			my $mfn = $row->{'000'}->[0];
+
+			if (! $mfn || $mfn !~ m#^\d+$#) {
+				$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
+				$mfn = $pos;
+				push @{ $row->{'000'} }, $pos;
+			}
+
+			my $ds = $n ? $n->data_structure($row) :
+				WebPAC::Normalize::Set::data_structure(
+					row => $row,
+					rules => $rules,
+					lookup => $lookup->lookup_hash,
+				);
+
+			$indexer->add(
+				id => $input->{name} . "/" . $mfn,
+				ds => $ds,
+				type => $config->{$use_indexer}->{type},
+			);
+
+			$total_rows++;
+		}
+
+	};
+
+	eval { $indexer->finish } if ($indexer->can('finish'));
+
+	my $dt = time() - $start_t;
+	$log->info("$total_rows records indexed in " .
+		sprintf("%.2f sec [%.2f rec/sec]",
+			$dt, ($total_rows / $dt)
+		)
+	);
+
+	#
+	# add Hyper Estraier links to other databases
+	#
+	if (ref($db_config->{links}) eq 'ARRAY') {
+		foreach my $link (@{ $db_config->{links} }) {
+			if ($use_indexer eq 'hyperestraier') {
+				$log->info("adding link $database -> $link->{to} [$link->{credit}]");
+				$indexer->add_link(
+					from => $database,
+					to => $link->{to},
+					credit => $link->{credit},
+				);
+			} else {
+				$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
+			}
+		}
+	}
 
-my $log = $lookup->_get_logger;
+}
 
-$log->info("$total_rows records indexed");