--- trunk/run.pl	2005/12/05 17:47:51	217
+++ trunk/run.pl	2007/06/21 21:26:17	868
@@ -4,61 +4,318 @@
 
 use Cwd qw/abs_path/;
 use File::Temp qw/tempdir/;
-use Data::Dumper;
 use lib './lib';
 
-use WebPAC::Lookup;
-use WebPAC::Input::ISIS;
-use WebPAC::Store 0.03;
-use WebPAC::Normalize::XML;
+use WebPAC::Common 0.02;
+use WebPAC::Parser 0.08;
+use WebPAC::Input 0.16;
+use WebPAC::Store 0.14;
+use WebPAC::Normalize 0.22;
 use WebPAC::Output::TT;
-use WebPAC::Output::Estraier 0.02;
-use YAML qw/LoadFile/;
-use LWP::Simple;
+use WebPAC::Validate 0.11;
+use WebPAC::Output::MARC;
+use WebPAC::Config;
+use Getopt::Long;
+use File::Path;
+use Time::HiRes qw/time/;
+use File::Slurp;
+use Data::Dump qw/dump/;
+use Storable qw/dclone/;
+use Pod::Usage qw/pod2usage/;
 
-my $limit = shift @ARGV;
+use Proc::Queue size => 1;
+use POSIX ":sys_wait_h"; # imports WNOHANG
 
-my $config = LoadFile('conf/config.yml');
+=head1 NAME
 
-print "config = ",Dumper($config);
+run.pl - start WebPAC indexing
 
-die "no databases in config file!\n" unless ($config->{databases});
+B<this command will probably go away. Don't get used to it!>
+
+=head1 OPTIONS
+
+=over 4
+
+=item --offset 42
+
+start loading (all) databases at offset 42
+
+=item --limit 100
+
+limit loading to 100 records
+
+=item --clean
+
+remove database and Hyper Estraier index before indexing
+
+=item --only=database_name/input_filter
+
+reindex just single database (legacy name is --one)
+
+C</input_filter> is optional part which can be C<name>
+or C<type> from input
+
+=item --config conf/config.yml
+
+path to YAML configuration file
+
+=item --stats
+
+disable indexing, modify_* in configuration and dump statistics about field
+and subfield usage for each input
+
+=item --validate path/to/validation_file
+
+turn on extra validation of imput records, see L<WebPAC::Validation>
+
+You can use special variables C<$database> and $C<$input> in this parametar
+like C<--validate 'conf/validate/$database-$input'> to construct filename
+
+=item --validate-delimiters path/to/validate_delimiters_file
+
+this option is used with C<--validate> to turn on extra validation of
+delimiters. If file is non existant, it will be created on first run.
+
+=item --marc-generate
+
+Generate MARC file. This will automatically be on if file contains C<marc*> directives.
+You can use this option as C<--no-marc-generate> to disable MARC generation.
+
+=item --marc-lint
+
+By default turned on if normalisation file has C<marc*> directives. You can disable lint
+messages with C<--no-marc-lint>.
+
+=item --marc-dump
+
+Force dump or input and marc record for debugging.
+
+=item --parallel 4
+
+Run databases in parallel (aproximatly same as number of processors in
+machine if you want to use full load)
+
+=item --only-links
+
+Create just links
+
+=item --merge
+
+Create merged index of databases which have links
+
+=back
+
+=cut
+
+my $offset;
+my $limit;
+
+my $clean = 0;
+my $config_path;
+my $debug = 0;
+my $only_filter;
+my $stats = 0;
+my $validate_path;
+my $validate_delimiters_path;
+my $marc_generate = 1;
+my $marc_lint = 1;
+my $marc_dump = 0;
+my $parallel = 0;
+my $only_links = 0;
+my $merge = 0;
+my $help;
+
+my $log = _new WebPAC::Common()->_get_logger();
+
+GetOptions(
+	"limit=i" => \$limit,
+	"offset=i" => \$offset,
+	"clean" => \$clean,
+	"one=s" => \$only_filter,
+	"only=s" => \$only_filter,
+	"config" => \$config_path,
+	"debug+" => \$debug,
+	"stats" => \$stats,
+	"validate=s" => \$validate_path,
+	"validate-delimiters=s" => \$validate_delimiters_path,
+	"marc-generate!" => \$marc_generate,
+	"marc-lint!" => \$marc_lint,
+	"marc-dump!" => \$marc_dump,
+	"parallel=i" => \$parallel,
+	"only-links!" => \$only_links,
+	"merge" => \$merge,
+	"help" => \$help,
+);
+
+$marc_generate = 0 if ( $validate_delimiters_path );
+
+pod2usage(-verbose => 2) if ($help);
+
+my $config = new WebPAC::Config( path => $config_path );
+
+#print "config = ",dump($config) if ($debug);
+
+die "no databases in config file!\n" unless ($config->databases);
+
+$log->info( "-" x 79 );
+
+my $log_file = 'log';
+
+if (-e $log_file ) {	# && -s $log_file > 5 * 1024 * 1024) {
+	$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'");
+	rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!");
+}
+
+my $estcmd_fh;
+my $estcmd_path = './estcmd-merge.sh';
+if ($merge) {
+	open($estcmd_fh, '>', $estcmd_path) || $log->logdie("can't open $estcmd_path: $!");
+	print $estcmd_fh 'cd /data/estraier/_node/ || exit 1',$/;
+	print $estcmd_fh 'sudo /etc/init.d/hyperestraier stop',$/;
+	$log->info("created merge batch file $estcmd_path");
+}
+
+my $validate;
+$validate = new WebPAC::Validate(
+	delimiters => $config->webpac('delimiters'),
+) if ($validate_path || $validate_delimiters_path);
+
+my $use_indexer = $config->use_indexer;
+$stats ||= $validate;
+if ($stats) {
+	$log->debug("disabled indexing for stats collection");
+	$use_indexer = undef;
+} elsif ( $use_indexer ) {
+	$log->info("using $use_indexer indexing engine...");
+}
+
+# parse normalize files and create source files for lookup and normalization
+
+my $parser = new WebPAC::Parser( config => $config );
 
 my $total_rows = 0;
+my $start_t = time();
 
-while (my ($database, $db_config) = each %{ $config->{databases} }) {
+my @links;
 
-	my $type = lc($db_config->{input}->{type});
+if ($parallel) {
+	$log->info("Using $parallel processes for speedup");
+	Proc::Queue::size($parallel);
+}
 
-	die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
+sub create_ds_config {
+	my ($db_config, $database, $input, $mfn) = @_;
+	my $c = dclone( $db_config );
+	$c->{_} = $database || $log->logconfess("need database");
+	$c->{_mfn} = $mfn || $log->logconfess("need mfn");
+	$c->{input} = $input || $log->logconfess("need input");
+	return $c;
+}
 
-	my $abs_path = abs_path($0);
-	$abs_path =~ s#/[^/]*$#/#;
+foreach my $database ( sort keys %{ $config->databases } ) {
+	my $db_config = $config->databases->{$database};
 
-	my $lookup = new WebPAC::Lookup(
-		lookup_file => $db_config->{input}->{lookup},
-	);
+	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
+	next if ($only_database && $database !~ m/$only_database/i);
 
-	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
+	if ($parallel) {
+		my $f=fork;
+		if(defined ($f) and $f==0) {
+			$log->info("Created processes $$ for speedup");
+		} else {
+			next;
+		}
+	}
 
+	my $indexer;
+	if ($use_indexer && $parser->have_rules( 'search', $database )) {
 
-	my $log = $lookup->_get_logger;
-	$log->info("working on $database in $db_path");
+		my $cfg_name = $use_indexer;
+		$cfg_name =~ s/\-.*$//;
 
-	my $db = new WebPAC::Store(
-		path => $db_path,
-		database => $database,
-	);
+		my $indexer_config = $config->get( $cfg_name ) || $log->logdie("can't find '$cfg_name' part in confguration");
+		$indexer_config->{database} = $database;
+		$indexer_config->{clean} = $clean;
+		$indexer_config->{label} = $db_config->{name};
+
+		# force clean if database has links
+		$indexer_config->{clean} = 1 if ($db_config->{links});
+
+		if ($use_indexer eq 'hyperestraier') {
 
-	my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
-	$est_config->{database} = $database;
+			# open Hyper Estraier database
+			require WebPAC::Output::Estraier;
+			$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
+		
+		} elsif ($use_indexer eq 'hyperestraier-native') {
 
-	$log->info("using HyperEstraier URL $est_config->{masterurl}");
+			# open Hyper Estraier database
+			require WebPAC::Output::EstraierNative;
+			$indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
 
-	my $est = new WebPAC::Output::Estraier(
-		%{ $est_config },
+		} elsif ($use_indexer eq 'kinosearch') {
+
+			# open KinoSearch
+			require WebPAC::Output::KinoSearch;
+			$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
+			$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
+
+		} else {
+			$log->logdie("unknown use_indexer: $use_indexer");
+		}
+
+		$log->logdie("can't continue without valid indexer") unless ($indexer);
+	}
+
+
+	#
+	# store Hyper Estraier links to other databases
+	#
+	if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
+		foreach my $link (@{ $db_config->{links} }) {
+			if ($use_indexer eq 'hyperestraier') {
+				if ($merge) {
+					print $estcmd_fh 'sudo -u www-data estcmd merge ' . $database . ' ' . $link->{to},$/;
+				} else {
+					$log->info("saving link $database -> $link->{to} [$link->{credit}]");
+					push @links, sub {
+						$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}]");
+			}
+		}
+	}
+	next if ($only_links);
+
+
+	#
+	# now WebPAC::Store
+	#
+	my $abs_path = abs_path($0);
+	$abs_path =~ s#/[^/]*$#/#;
+
+	my $db_path = $config->webpac('db_path');
+
+	if ($clean) {
+		$log->info("creating new database '$database' in $db_path");
+		rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
+	} else {
+		$log->info("working on database '$database' in $db_path");
+	}
+
+	my $store = new WebPAC::Store(
+		path => $db_path,
+		debug => $debug,
 	);
 
+
 	#
 	# now, iterate through input formats
 	#
@@ -66,54 +323,319 @@
 	my @inputs;
 	if (ref($db_config->{input}) eq 'ARRAY') {
 		@inputs = @{ $db_config->{input} };
-	} else {
+	} elsif ($db_config->{input}) {
 		push @inputs, $db_config->{input};
+	} else {
+		$log->info("database $database doesn't have inputs defined");
 	}
 
 	foreach my $input (@inputs) {
-		$log->info("working on input $input->{path} [$input->{type}]");
 
-		my $isis = new WebPAC::Input::ISIS(
-			code_page => $config->{webpac}->{webpac_encoding},
-			limit_mfn => $input->{limit},
-		);
+		my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input));
 
-		my $maxmfn = $isis->open(
-			filename => $input->{path},
-			code_page => $input->{encoding},	# database encoding
+		next if ($only_input && ($input_name !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
+
+		my $type = lc($input->{type});
+
+		die "I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!\n" unless (grep(/$type/, $config->webpac('inputs')));
+
+		my $input_module = $config->webpac('inputs')->{$type};
+
+		my @lookups = $parser->have_lookup_create($database, $input);
+
+		$log->info("working on input '$input_name' in $input->{path} [type: $input->{type}] using $input_module",
+			@lookups ? " creating lookups: ".join(", ", @lookups) : ""
 		);
 
-		my $n = new WebPAC::Normalize::XML(
-		#	filter => { 'foo' => sub { shift } },
-			db => $db,
-			lookup_regex => $lookup->regex,
-			lookup => $lookup,
+		if ($stats) {
+			# disable modification of records if --stats is in use
+			delete($input->{modify_records});
+			delete($input->{modify_file});
+		}
+
+		my $input_db = new WebPAC::Input(
+			module => $input_module,
+			encoding => $config->webpac('webpac_encoding'),
+			limit => $limit || $input->{limit},
+			offset => $offset,
+			recode => $input->{recode},
+			stats => $stats,
+			modify_records => $input->{modify_records},
+			modify_file => $input->{modify_file},
+			input_config => $input,
 		);
+		$log->logdie("can't create input using $input_module") unless ($input);
+
+		if (defined( $input->{lookup} )) {
+			$log->warn("$database/$input_name has depriciated lookup definition, removing it...");
+			delete( $input->{lookup} );
+		}
+
+		my $lookup_coderef;
+
+		if (@lookups) {
+
+			my $rules = $parser->lookup_create_rules($database, $input) || $log->logdie("no rules found for $database/$input");
+
+			$lookup_coderef = sub {
+				my $rec = shift || die "need rec!";
+				my $mfn = $rec->{'000'}->[0] || die "need mfn in 000";
+
+				WebPAC::Normalize::data_structure(
+					row => $rec,
+					rules => $rules,
+					config => create_ds_config( $db_config, $database, $input, $mfn ),
+				);
+
+				#warn "current lookup: ", dump(WebPAC::Normalize::_get_lookup());
+			};
+
+			WebPAC::Normalize::_set_lookup( undef );
+
+			$log->debug("created lookup_coderef using:\n$rules");
+
+		};
+
+		my $lookup_jar;
+
+		my $maxmfn = $input_db->open(
+			path => $input->{path},
+			code_page => $input->{encoding},	# database encoding
+			lookup_coderef => $lookup_coderef,
+			lookup => $lookup_jar,
+			%{ $input },
+			load_row => sub {
+				my $a = shift;
+				return $store->load_row(
+					database => $database,
+					input => $input_name,
+					id => $a->{id},
+				);
+			},
+			save_row => sub {
+				my $a = shift;
+				return $store->save_row(
+					database => $database,
+					input => $input_name,
+					id => $a->{id},
+					row => $a->{row},
+				);
+			},
 
-		$n->open(
-			tag => $input->{normalize}->{tag},
-			xml_file => $input->{normalize}->{path},
 		);
 
-		for ( 0 ... $isis->size ) {
+		my $lookup_data = WebPAC::Normalize::_get_lookup();
 
-			my $row = $isis->fetch || next;
+		if (defined( $lookup_data->{$database}->{$input_name} )) {
+			$log->debug("created following lookups: ", sub { dump( $lookup_data ) } );
 
-			my $mfn = $row->{'000'}->[0] || die "can't find MFN";
+			foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) {
+				$store->save_lookup(
+					database => $database,
+					input => $input_name,
+					key => $key,
+					data => $lookup_data->{$database}->{$input_name}->{$key},
+				);
+			}
+		}
 
-			my $ds = $n->data_structure($row);
+		my $report_fh;
+		if ($stats || $validate) {
+			my $path = "out/report/${database}-${input_name}.txt";
+			open($report_fh, '>', $path) || $log->logdie("can't open $path: $!");
+
+			print $report_fh "Report for database '$database' input '$input_name' records ",
+				$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n";
+			$log->info("Generating report file $path");
+
+			if ( $validate ) {
+				$validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ) if ( $validate_path );
+				$validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ) if ( $validate_delimiters_path );
+			}
+		}
 
-			$est->add(
-				id => $input->{name} . "#" . $mfn,
-				ds => $ds,
-				type => $config->{hyperestraier}->{type},
+		my $marc;
+		if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) {
+			$marc = new WebPAC::Output::MARC(
+				path => "out/marc/${database}-${input_name}.marc",
+				lint => $marc_lint,
+				dump => $marc_dump,
 			);
+		}
+
+		my $rules = $parser->normalize_rules($database,$input_name) || $log->logdie("no normalize rules found for $database/$input_name");
+		$log->debug("parsed normalize rules:\n$rules");
+
+		# reset position in database
+		$input_db->seek(1);
+
+		# generate name of config key for indexer (strip everything after -)
+		my $indexer_config = $use_indexer;
+		$indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config);
+
+		my $lookup_hash;
+		my $depends = $parser->depends($database,$input_name);
+	
+		if ($depends) {
+			$log->debug("$database/$input_name depends on: ", dump($depends)) if ($depends);
+			$log->logdie("parser->depends didn't return HASH") unless (ref($depends) eq 'HASH');
+
+			foreach my $db (keys %$depends) {
+				foreach my $i (keys %{$depends->{$db}}) {
+					foreach my $k (keys %{$depends->{$db}->{$i}}) {
+						my $t = time();
+						$log->debug("loading lookup $db/$i");
+						$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup(
+							database => $db,
+							input => $i,
+							key => $k,
+						);
+						$log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t));
+					}
+				}
+			}
+
+			$log->debug("lookup_hash = ", sub { dump( $lookup_hash ) });
+		}
+
+
+		foreach my $pos ( 0 ... $input_db->size ) {
+
+			my $row = $input_db->fetch || next;
 
 			$total_rows++;
+
+			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'} }, $pos;
+			}
+
+
+			if ($validate) {
+				if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) {
+					$log->error( "MFN $mfn validation error:\n",
+						$validate->report_error( $errors )
+					);
+				}
+				next;	# validation doesn't create any output
+			}
+
+			my $ds = WebPAC::Normalize::data_structure(
+				row => $row,
+				rules => $rules,
+				lookup => $lookup_hash,
+				config => create_ds_config( $db_config, $database, $input, $mfn ),
+				marc_encoding => 'utf-8',
+				load_row_coderef => sub {
+					my ($database,$input,$mfn) = @_;
+					return $store->load_row(
+						database => $database,
+						input => $input,
+						id => $mfn,
+					);
+				},
+			);
+
+			$log->debug("ds = ", sub { dump($ds) }) if ($ds);
+
+			$store->save_ds(
+				database => $database,
+				input => $input_name,
+				id => $mfn,
+				ds => $ds,
+			) if ($ds && !$stats);
+
+			$indexer->add(
+				id => "${input_name}/${mfn}",
+				ds => $ds,
+				type => $config->get($indexer_config)->{type},
+			) if ($indexer && $ds);
+
+			if ($marc) {
+				my $i = 0;
+
+				while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) {
+					$marc->add(
+						id => $mfn . ( $i ? "/$i" : '' ),
+						fields => $fields,
+						leader => WebPAC::Normalize::_get_marc_leader(),
+						row => $row,
+					);
+					$i++;
+				}
+
+				$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
+			}
 		}
 
-	};
+		if ($validate) {
+			my $errors = $validate->report;
+			if ($errors) {
+				$log->info("validation errors:\n$errors\n" );
+				print $report_fh "$errors\n" if ($report_fh);
+			}
+
+			print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1, current_input => 1 ), "\n\n";
+
+			# must be last thing that touches $validate for this input
+			$validate->reset;
+		}
+
+		if ($stats) {
+			my $s = $input_db->stats;
+			$log->info("statistics of fields usage:\n$s");
+			print $report_fh "Statistics of fields usage:\n$s" if ($report_fh);
+		}
+
+		# close MARC file
+		$marc->finish if ($marc);
+
+		# close report
+		close($report_fh) if ($report_fh);
+	}
+
+	eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
+
+	my $dt = time() - $start_t;
+	$log->info("$total_rows records ", $indexer ? "indexed " : "",
+		sprintf("in %.2f sec [%.2f rec/sec]",
+			$dt, ($total_rows / $dt)
+		)
+	);
+
+
+	# end forked process
+	if ($parallel) {
+		$log->info("parallel process $$ finished");
+		exit(0);
+	}
 
-	$log->info("$total_rows records indexed");
 }
 
+if ($parallel) {
+	# wait all children to finish
+	sleep(1) while wait != -1;
+	$log->info("all parallel processes finished");
+}
+
+# save new delimiters if needed
+$validate->save_delimiters_templates if ( $validate_delimiters_path );
+
+#
+# handle links or merge after indexing
+#
+
+if ($merge) {
+	print $estcmd_fh 'sudo /etc/init.d/hyperestraier start',$/;
+	close($estcmd_fh);
+	chmod 0700, $estcmd_path || $log->warn("can't chmod 0700 $estcmd_path: $!");
+	system $estcmd_path;
+} else {
+	foreach my $link (@links) {
+		$log->logdie("coderef in link ", Dumper($link), " is ", ref($link), " and not CODE") unless (ref($link) eq 'CODE');
+		$link->();
+	}
+}