--- Webpacus/lib/Webpacus/Model/WebPAC.pm	2005/11/22 12:57:20	94
+++ Webpacus/lib/Webpacus/Model/WebPAC.pm	2005/12/20 00:00:57	305
@@ -5,9 +5,13 @@
 use lib '/data/webpac2/lib';
 use base qw/
 	Catalyst::Model
-	WebPAC::Search::Estraier
 /;
 use Data::Dumper;
+use WebPAC::Store 0.08;
+use WebPAC::Output::TT 0.05;
+use WebPAC::Search::Estraier 0.05;
+use File::Slurp;
+use Time::HiRes;
 
 =head1 NAME
 
@@ -30,9 +34,22 @@
 
  # configuration for hyper estraier full text search engine
  hyperestraier:
-    url: 'http://localhost:1978/node/webpac2'
-    user: 'admin'
-    passwd: 'admin'
+  masterurl: 'http://localhost:1978/node/webpac2'
+  defaultnode: 'webpac2'
+  defaultdepth: 1
+  user: 'admin'
+  passwd: 'admin'
+  hits_on_page: 100
+  hits_for_pager: 1000
+
+ webpac:
+  db_path: '/data/webpac2/db'
+  template_path: '/data/webpac2/conf/output/tt'
+  template: 'html_ffzg_results_short.tt'
+  # encoding comming from webpac
+  webpac_encoding: 'iso-8859-2'
+  # encoding expected by Catalyst
+  out_encoding: 'UTF-8'
 
 =cut
 
@@ -48,30 +65,300 @@
 	my $est_cfg = $c->config->{hyperestraier};
 	$est_cfg->{'log'} = $log;
 
+	$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
+
 	$log->debug("using config:" . Dumper($est_cfg) );
 
+	if (! $est_cfg->{database}) {
+		my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration");
+		$log->info("using default node $defaultnode");
+		$est_cfg->{database} = $defaultnode;
+	}
+
 	$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
 
+	# save config parametars in object
+	foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding defaultdepth/) {
+		$self->{$f} = $c->config->{hyperestraier}->{$f} ||
+			$c->config->{webpac}->{$f};
+		$log->debug("self->{$f} = " . $self->{$f});
+	}
+	my $db_path = $self->{db_path};
+	my $template_path = $self->{template_path};
+
+	$log->debug("using db path '$db_path', template path '$template_path'");
+
+	$self->{db} = new WebPAC::Store(
+		path => $db_path,
+		read_only => 1,
+		database => $est_cfg->{database},
+	);
+
+	$self->{out} = new WebPAC::Output::TT(
+		include_path => $template_path,
+		filters => { foo => sub { shift } },
+	);
+
+	# default template from config.yaml
+	$self->{template} ||= $c->config->{webpac}->{template};
+
+	$self->{iconv} = new Text::Iconv(
+		$c->config->{webpac}->{webpac_encoding},
+		$c->config->{webpac}->{out_encoding}
+	);
+
+	$log->debug("converting encoding from webpac_encoding '" .
+		$c->config->{webpac}->{webpac_encoding} .
+		"' to '" .
+		$c->config->{webpac}->{out_encoding} .
+		"'"
+	);
+
+	$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config");
+
 	return $self;
 
 }
 
+
+=head2 search
+
+  my $m->search(
+  	phrase => 'query phrase',
+	add_attr => \@add_attr
+	get_attr => [ '@uri' ],
+	max => 42,
+	template => 'result_template.tt',
+	depth => 1,
+  );
+
+All fields are standard C<WebPAC::Search::Estraier> parametars except
+C<template> which will (if specified) return results in HTML using
+selected template.
+
+=cut
+
 sub search {
-	my ( $self, $query ) = @_;
+	my $self = shift;
+
+	my $args = {@_};
+
+	my $log = $self->{log};
+
+	$log->debug("args: " . Dumper( $args ));
+
+	my $query = $args->{phrase} || $log->warn("no query phrase") && return;
+
+	$log->debug("search model query: '$query'");
+	if ($args->{add_attr}) {
+		$log->debug(" + add_attr: " .
+			join("','", @{ $args->{add_attr} })
+		);
+	}
+
+	my $template_filename = $args->{template} || $self->{template};
+
+	$args->{max} ||= $self->{'hits_for_pager'};
+	if (! $args->{max}) {
+		$args->{max} = 100;
+		$log->warn("max not set when calling model. Using default of $args->{max}");
+	}
+
+	my $times;	# store some times for benchmarking
+
+	my $t = time();
+
+	# transfer depth of search
+	if (! $args->{depth}) {
+		my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
+		$args->{depth} = $default;
+		$log->warn("using default search depth $default");
+	}
+
+	my @results = $self->{est}->search( %{ $args } );
+
+	$times->{est} += time() - $t;
+
+	my $hits = $#results + 1;
+
+        $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );
+
+	# just return results?
+	return @results unless ($args->{'template'});
+
+	#
+	# construct HTML results
+	#
+
+	my @html_results;
+
+	for my $i ( 0 .. $#results ) {
+
+		my ($database, $prefix, $id);
+		if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
+			($database, $prefix,$id) = ($1,$2,$3);
+		} else {
+			$log->warn("can't decode database/prefix/id from " .  $results[$i]->{'@uri'});
+			next;
+		}
+
+		#$log->debug("load_ds( id => $id, prefix => '$prefix' )");
+
+		$t = time();
+
+		my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
+		if (! $ds) {
+			$log->error("can't load_ds( ${database}/${prefix}/${id} )");
+			next;
+		}
+
+		$times->{db} += time() - $t;
+
+		#$log->debug( "ds = " . Dumper( \@html_results ) );
 
-	$self->{log}->debug("search got query: $query<--");
+		$t = time();
 
-	my @results = $self->{est}->search(
-		query => $query,
-		attr => [ '@uri' ],
-		max => 100,
+		my $html = $self->{out}->apply(
+			template => $template_filename,
+			data => $ds,
+			record_uri => "${database}/${prefix}/${id}",
+			config => $self->{databases}->{$database},
+		);
+
+		$times->{out} += time() - $t;
+
+		$t = time();
+
+		$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
+
+		$times->{iconv} += time() - $t;
+
+		push @html_results, $html;
+
+	}
+
+	#$log->debug( '@html_results = ' . Dumper( \@html_results ) );
+
+	$log->debug( sprintf(
+		"time spent: db = %.2f, out = %.2f, iconv = %.2f",
+		$times->{db}, $times->{out}, $times->{iconv},
+	) );
+
+	return \@html_results;
+}
+
+=head2 record
+
+  my $html = $m->record(
+  	mfn => 42,
+	template => 'foo.tt',
+  );
+
+This will load one record, convert it to html using C<template> and return
+it.
+
+=cut
+
+sub record {
+	my $self = shift;
+
+	my $args = {@_};
+	my $log = $self->{log};
+	$log->debug("args: " . Dumper( $args ));
+
+	foreach my $f (qw/record_uri template/) {
+		$log->fatal("need $f") unless ($args->{$f});
+	}
+
+	my ($database, $prefix, $id);
+
+	if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
+		($database, $prefix, $id) = ($1,$2,$3);
+	} else {
+		$log->error("can't parse $args->{record_uri} into prefix, database and uri");
+		return;
+	}
+
+	my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
+	if (! $ds) {
+		$log->error("can't load_ds( $database/$prefix/$id )");
+		return;
+	}
+
+	my $html = $self->{out}->apply(
+		template => $args->{template},
+		data => $ds,
+		record_uri => $args->{record_uri},
+		config => $self->{databases}->{$database},
+	);
+
+	$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
+
+	return $html;
+}
+
+
+=head2 save_html
+
+  $m->save_html( '/full/path/to/file', $content );
+
+It will use C<Text::Iconv> to convert content encoding back to
+Webpac codepage, recode JavaScript Unicode entities (%u1234),
+strip extra newlines at beginning and end, and save to 
+C</full/path/to/file.new> and if that succeeds, just rename
+it over original file which should be atomic on filesystem level.
+
+=cut
+
+sub save_html {
+	my ($self, $path, $content) = @_;
+
+	sub _conv_js {
+		my $t = shift || return;
+		return $self->{iconv}->convert(chr(hex($t)));
+	}
+	$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
+	$content =~ s/^[\n\r]+//s;
+	$content =~ s/[\n\r]+$/\n/s;
+	$content =~ s/\n\r/\n/gs;
+
+	my ($from, $to) = (
+		$self->{out_encoding},
+		$self->{webpac_encoding},
 	);
 
-	return @results;
+	$self->{log}->debug("using iconv to convert from $from to $to encoding");
+
+	my $iconv_on_save = new Text::Iconv($from, $to)
+		|| $self->{log}->fatal("can't create iconv for saving");
+
+	$content = $iconv_on_save->convert( $content ) || die "no content?";
+
+	write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
+	rename $path . '.new', $path || die "can't rename to $path: $!";
 }
 
+=head2 load_html
+
+  my $html = $m->load_html('/full/path/to/file');
+
+This will convert file from Webpac encoding to Catalyst and
+convert that data to escaped HTML (for sending into
+C<< <textarea/> >> tags in html.
+
+=cut
+
+sub load_html {
+	my ($self, $path) = @_;
+
+	die "no path?" unless ($path);
 
-	
+	my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
+	#$content = $q->escapeHTML($iconv_utf8->convert($content));
+	$content = $self->{iconv}->convert($content);
+
+	return $content;
+}
 
 =head1 AUTHOR