--- trunk/all2xml.pl	2003/02/16 22:41:37	13
+++ trunk2/all2all.pl	2004/09/08 15:30:07	415
@@ -1,242 +1,142 @@
 #!/usr/bin/perl -w
 
-use strict;
-use OpenIsis;
-use Getopt::Std;
-use Data::Dumper;
-use XML::Simple;
-use Text::Unaccent 1.02;	# 1.01 won't compile on my platform,
-use Text::Iconv;
-use Config::IniFiles;
-
-$|=1;
-
-my $config_file = $0;
-$config_file =~ s/\.pl$/.conf/;
-die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
-
-my $config;
-
-use index_DBI;	# there is no other, right now ;-)
-
-my $index = new index_DBI();	# open index
-
-my %opts;
-
-# usage:
-# 	-d directory name
-# 	-m multiple directories
-# 	-q quiet
-# 	-s run swish
-
-getopts('d:m:qs', \%opts);
-
-my $db_dir;
-
-#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
-
-#print Dumper($config->{indexer});
-#print "-" x 70,"\n";
-
-Text::Iconv->raise_error(1);     # Conversion errors raise exceptions
-
-#my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
-#my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
-my $isis_codepage;
-my $index_codepage;
-my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
-my $xml_codepage;
-
-sub isis2xml {
+=head1 NAME
 
-	use xmlify;
+all2all.pl - basic script for all WebPAC needs
 
-	my $row = shift @_;
-	my $add_xml = shift @_;
+=cut
 
-	my $xml;
+use strict;
+use Data::Dumper;
+use Carp;
 
-	use parse_format;
+use lib './lib';
+use WebPAC;
+use WebPAC::jsFind;
+use WebPAC::Index;
+
+my $webpac = new WebPAC(
+	code_page => 'ISO-8859-2',
+	limit_mfn => 500,
+#	debug => 1,
+) || die;
+
+my $log = $webpac->_get_logger();
+
+my $index = new WebPAC::jsFind(
+	index_path => './out/index',
+	keys => 10,
+) || die;
 
-	my $html = "";		# html formatted display output
+my $thes;
 
-	my %field_usage;	# counter for usage of each field
+$|=1;
 
-	foreach my $field (keys %{$config->{indexer}}) {
+my $maxmfn = $webpac->open_isis(
+	filename => shift @ARGV || '/data/hidra/THS/THS',
+	lookup => [
+	{ 'key' => 'd:v900', 'val' => 'v250^a' },
+#	{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
+#	{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
+#	{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
+	{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
+	{ 'key' => '900_mfn:v900', 'val' => 'v000' },
+	],
+);
+
+$log->info("rows: $maxmfn");
+
+$webpac->open_import_xml(type => 'isis_hidra_ths');
+
+while (my $rec = $webpac->fetch_rec) {
+
+	my @ds = $webpac->data_structure($rec);
+
+	if (0 && $log->is_debug) {
+		$log->debug("rec = ",Dumper($rec));
+		$log->debug("ds = ",Dumper(\@ds));
+	}
 
-		$field_usage{$field}++;
+	next if (! @ds);
 
-		my $swish_data = "";
-		my $display_data = "";
-		my $index_data = "";
+	my $filename = $webpac->{'current_filename'};
 
-		foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
+	if ($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'},
+		);
+	}
 
-			my $format = $x->{content};
-			my ($s,$d,$i) = (1,1,0);	# swish, display default
-			$s = 0 if (lc($x->{type}) eq "display");
-			$d = 0 if (lc($x->{type}) eq "swish");
-			($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
-#print STDERR "## s: $s d: $d i: $i ## $format ##\n";	
+	my $headline = $webpac->{'headline'};
 
-			$format = $cludge_codepage->convert($format);
-			my ($swish,$display) = parse_format($format,$row);
-#print STDERR "s: $swish\nd: $display\n" if ($swish);
+	my $f = $filename;
+	$f =~ s!out/!!;
 
-#print STDERR "swish: $swish<-- display: $display<--\n";
-			# FIX: this is ugly, UGLY, cludge: OpenIsis return
-			# UTF8 encoding of strings, but as if source charset
-			# is ISO8859-1 and not some other. This breaks our
-			# isis character encoding, so we convert it first
-			# back to ISO8859-1 (which can actually be different
-			# encoding in isis)
+	# 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'}})
+		);
+	}
 
-			$swish_data .= $swish if ($s && $swish);
-			$display_data .= $display if ($d && $display);
-			$index_data .= $display if ($i && $display);
+	# 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,
+			);
 		}
+	}
 
+#	print Dumper(\@ds);
 
-#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
-		if ($display_data) {
-			$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
-			# FIX: this is removed and replaced by html tag.
-			#$xml .= xmlify($field."_display", $display_data);
-
-			if ($field eq "headline") {
-				$xml .= xmlify("headline", $display_data);
-			} else {
-
-				# find field name (signular, plural)
-				my $field_name = "";
-				if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
-					$field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
-				} elsif ($config->{indexer}->{$field}->{name_plural}) {
-					$field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
-				} else {
-					$field_name = $config->{indexer}->{$field}->{name}."#-#";
-				}
-				if ($field_name) {
-					$html .= $xml_codepage->convert($field_name);
-				}
-				$html .= $display_data."###\n";
-			}
-		}
-		if ($swish_data) {
-			my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
-			$swish_data = $i->convert($swish_data);
-			$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
-			#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
-			#$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
-		}
+}
 
-		# index
-		if ($index_data && $index_data ne "") {
-			$index_data = $index_codepage->convert($index_data) || $index_data;
-			$index->insert($field, $index_data, $db_dir);
-		}
+foreach my $t (keys %{$thes}) {
 
+	my @e = $thes->{$t}->elements;
+	if (! @e) {
+		$log->logwarn("no elements in sorted index $t?");
+		next;
 	}
 
-	# dump formatted output in <html>
-	if ($html) {
-		$xml .= xmlify("html",$html);
-	}
-	
-	if ($xml) {
-#print STDERR "x: $xml\n";
-		$xml .= $add_xml if ($add_xml);
-		return "<xml>\n$xml</xml>\n";
-	} else {
-		return;
-	}
-}
-
-##########################################################################
-
-my $cfg = new Config::IniFiles( -file => $config_file );
+	my $file = "./out/bfilter/$t.txt";
+	$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
 
-foreach my $database ($cfg->Sections) {
+	$webpac->output_file(
+		file => $file,
+		template => 'index.tt',
+		data => \@e,
+		index_name => $t,
+	);
+}
 
-	my $isis_db = $cfg -> val($database, 'isis_db');
-	my $type = $cfg -> val($database, 'type');
-	my $add_xml = $cfg -> val($database, 'xml');
-
-	# read configuration for this type
-	$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
-	$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
-	$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
-	$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8');
-
-	my $db = OpenIsis::open( $isis_db );
-	if (0) {
-#	# FIX
-#	if (! $db ) {
-		print STDERR "WARNING: can't open '$isis_db'\n";
-		next ;
-	}
-
-	my $max_rowid = OpenIsis::maxRowid( $db );
-
-	print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
-
-	my $path = $database;			# was $isis_db
-
-	my $last_p = 0;
-
-#	{ my $row_id = 4514;
-# FIX
-	for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
-		my $row = OpenIsis::read( $db, $row_id );
-		if ($row && $row->{mfn}) {
-#print STDERR "mfn: ",$row->{mfn},"\n";
-			# output current process indicator
-			my $p = int($row->{mfn} * 100 / $max_rowid);
-			if ($p != $last_p) {
-				printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
-				$last_p = $p;
-			}
-
-			if (my $xml = isis2xml($row,$add_xml)) {
-#print STDERR "--ret-->$xml\n";
-				print "Path-Name: $path#".int($row->{mfn})."\n";
-				print "Content-Length: ".(length($xml)+1)."\n";
-				print "Document-Type: XML\n\n$xml\n";
-			}
-		}
+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}));
 	}
-	print STDERR "\n";
 }
 
-# call this to commit index
 $index->close;
 
-1;
-__END__
-##########################################################################
-
-=head1 NAME
-
-isis2xml.pl - read isis file and dump XML
-
-=head1 DESCRIPTION
-
-This command will read ISIS data file using OpenIsis perl module and
-create XML file for usage with I<SWISH-E>
-indexer. Dispite it's name, this script B<isn't general xml generator>
-from isis files (isis allready has something like that). Output of this
-script is tailor-made for SWISH-E.
-
-=head1 AUTHOR
-
-Dobrica Pavlinusic <dpavlin@rot13.org>
-
-=head1 COPYRIGHT
-
-GNU Public License (GPL) v2 or later
-
-=head1 SEE ALSO
-
-SWISH-E web site at http://www.swish-e.org
-
-=cut