--- trunk/WebPac.pm	2003/01/11 19:55:30	9
+++ trunk/WebPac.pm	2003/07/08 08:24:13	80
@@ -6,18 +6,31 @@
 use HTML::Pager;
 use HTML::FillInForm;
 use SWISH;
-use Unicode::String qw(utf8 utf16);
-require Unicode::Map8;
+use Text::Iconv;
 use DBI;
+use Config::IniFiles;
+use Text::Unaccent;
 
-# configuration options
-# FIX: they really should go in configuration file!
-my $TEMPLATE_PATH = '/data/webpac/template_html';
-my $CHARSET = 'ISO-8859-2';
-my $SWISH = '/data/swish/swish-e';
-my $INDEX = '/data/webpac/index/isis.index';
-my $MAX_HITS = 500;
-my $ON_PAGE = 10;
+use lib '..';
+use index_DBI;
+use back2html;
+
+
+# read global.conf configuration
+my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'";
+
+# configuration options from global.conf
+my $TEMPLATE_PATH = $cfg_global->val('webpac', 'template_html') || die "need template_html in global.conf, section webpac";
+my $CHARSET = $cfg_global->val('webpac', 'charset') || 'ISO-8859-1';
+my $SWISH = $cfg_global->val('webpac', 'swish') || '/usr/bin/swish-e';
+my $INDEX = $cfg_global->val('webpac', 'index') || die "need index in global.conf, section webpac";
+my $MAX_HITS = $cfg_global->val('webpac', 'max_hits') || 0;
+my $ON_PAGE =$cfg_global->val('webpac', 'on_page') || 10;
+
+
+Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
+
+my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
 
 
 sub setup {
@@ -61,68 +74,93 @@
 
 	my @s_arr;	# all queries are located here
 
-	for(my $i = 1; $i <=10; $i++) {
+	my @path_arr = $q->param('path');
+	my $full = $q->param('full');
+
+	for(my $i = 1; $i <=30; $i++) {
 
 		return show_index($self, $i) if ($q->param("f".$i."_index"));
-		next if (! $q->param("f$i"));
 		next if (! $q->param("v$i"));
+		next if (! $q->param("f$i"));
 
 		# re-write query from +/- to and/and not
-		my $s;
-		my $search = $q->param("v$i");
-		while ($search =~ s/\s*("[^"]+")\s*/ /) {
-			$s .= "$1 ";
-		}
-		$search =~ s/^\s+//;
-		$search =~ s/\s+$//;
+		my @param_vals = $q->param("v$i");
+		my @swish_q;
+		while (my $search = shift @param_vals) {
+			my $s;
+			# remove accents
+			$search = unac_string($CHARSET,$search);
+			while ($search =~ s/\s*("[^"]+")\s*/ /) {
+				$s .= "$1 ";
+			}
+			$search =~ s/^\s+//;
+			$search =~ s/\s+$//;
 
-		foreach (split(/\s+/,$search)) {
-			if (m/^([+-])(\S+)/) {
-				$s.= ($s) ? "and " : "";
-				$s.="not " if ($1 eq "-");
-				$s.="$2* ";
-			} else {
-				$s.="$_* ";
+			foreach (split(/\s+/,$search)) {
+				if (m/^([+-])(\S+)/) {
+					$s.= ($s) ? "and " : "";
+					$s.="not " if ($1 eq "-");
+					$s.="$2* ";
+				} elsif (m/(and|or|not)/i) {
+					$s.="$_ ";
+				} else {
+					$s.="$_* ";
+				}
 			}
+			$s =~ s/\*+/*/g;
+			push @swish_q,$s;
 		}
-
-		push @s_arr,$q->param("f$i")."_swish=($s)";
+		# FIXME default operator for multi-value fields is or. There is
+		# no way to change it, except here for now. Is there need?
+		push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")";
 	}
 
 	my $tmpl = $self->load_tmpl('results.html');
 
-	my $l2_map = Unicode::Map8->new($CHARSET) || die;
-	my $us = Unicode::String->new();
+	sub esc_html {
+		my $html = shift;
+		$html =~ s/</&lt;/g;
+		$html =~ s/>/&gt;/g;
+		return $html;
+	}
 
 	# call swish
 	my $sh = SWISH->connect('Fork',
 		prog     => $SWISH,
 		indexes  => $INDEX,
-		#properties  => [qw/god br nr/],
+		properties  => [qw/swishdocpath swishrank swishtitle headline html/],
 		results  => sub {
 			my ($sh,$hit) = @_;
 
-			$us->utf8($hit->swishtitle);
-
 			push @swish_results, {
 				nr => ($#swish_results + 2),
 				path => $hit->swishdocpath,
-				title => $l2_map->to8($us->utf16),
+				headline => esc_html($from_utf8->convert($hit->headline)),
+				html => back2html($from_utf8->convert($hit->html)),
 				rank => $hit->swishrank };
 
-#			my @fields = $hit->field_names;
-#			print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields;
 		},
 		#startnum => 0,
-		maxhits => $MAX_HITS,
+		maxhits => $MAX_HITS
 	);
 
 	die $SWISH::errstr unless $sh;
+	# construct swish query
+	my $sw_q = join(" and ",@s_arr);
+	if (@path_arr) {
+		$sw_q .= "and (swishdocpath=\"";
+		$sw_q .= join("\" or swishdocpath=\"",@path_arr);
+		$sw_q .= "\")";
+		$tmpl->param('full',1);	# show full records
+	}
 
-	my $hits = $sh->query(join(" and ",@s_arr)) || 0;	# FIX: and/or
+	my $hits = $sh->query($sw_q);
 
 	$tmpl->param('hits',$hits);
-	$tmpl->param('search',join(" and ",@s_arr));
+	$tmpl->param('search',$sw_q);
+
+	$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
+	$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 
 	# create a Pager object
 	my $pager = HTML::Pager->new(
@@ -171,12 +209,56 @@
 
 	my $q = $self->query();
 
+	my $field = $q->param("f$i");
+	my $limit = $q->param("v$i");
+
 	my $html;
 
-	$html .= "show index of ".$q->param("f$i")." for ".$q->param("v$i");
+	my $index = new index_DBI(
+		$cfg_global->val('global', 'dbi_dbd'),
+		$cfg_global->val('global', 'dbi_dsn'),
+		$cfg_global->val('global', 'dbi_user'),
+		$cfg_global->val('global', 'dbi_passwd') || ''
+	);
 
+	my $total = $index->check($field);
+	if (! $total) {
+		my $tmpl = $self->load_tmpl('no_index.html');
+		$tmpl->param('field',$field);
+		$html = $tmpl->output;
+		return $html;
+	}
 
-	return $html;
+	my $tmpl = $self->load_tmpl('index_res.html');
+	$tmpl->param('field',$field);
+	$tmpl->param('limit',$limit);
+	$tmpl->param('total',$total);
+
+# FIXME I should set offset and leave out limit from fetch!!
+#	if (! $q->param("PAGER_offset") {
+#		$q->param("Pager_offet)
+#	}
+
+	my $pager = HTML::Pager->new(
+		query => $q,
+		get_data_callback => sub {
+			my ($offset, $rows) = @_;
+
+			my @result = $index->fetch($field,'item',$limit, $offset, $rows);
+			return \@result;
+		},
+		rows => $total,
+		page_size => $ON_PAGE,
+		persist_vars => [
+			'rm', 
+			"f$i", "v$i", "f".$i."_index", 
+			'offset',
+			],
+		debug => 1,
+		template => $tmpl,
+	);
+
+	return $pager->output;
 }
 
 1;