--- trunk/html/swish.cgi	2004/04/06 19:21:07	73
+++ trunk/html/swish.cgi	2004/08/30 11:16:39	86
@@ -9,6 +9,53 @@
 use Text::Iconv;
 use Data::Pageset;
 
+
+sub get_snippet {
+	my $context_chars = 100;
+
+	my $desc = shift || return '';
+	$desc = e($desc);
+
+	# test if $desc contains any of our query words
+	my @snips;
+
+	my @colors = qw{#ffff66 #a0ffff #99ff99 #ff9999 #ff66ff};
+
+	# construct regex
+	my $re = qq/^.*?(.{$context_chars}?)(\Q/ . join("|",@_) .
+		qq/\E)(.{$context_chars})/;
+
+	while ($desc =~ s/$re//si) {
+		my ($bef,$qm,$af) = ($1, $2, $3);
+			
+		# no partial words...
+		$bef =~ s,^\S+\s+|\s+\S+$,,gs;
+		$af =~ s,^\S+\s+|\s+\S+$,,gs;
+
+		push @snips, "$bef $qm $af";
+	}
+
+	my $ellip = ' ... ';
+	my $snippet;
+	
+	if (@snips) {
+		$snippet = $ellip. join($ellip, @snips) . $ellip;
+	} else {
+		return '';
+	}
+
+	# color offset
+	my $i = 0;
+
+ 	foreach my $w (@_) {
+		$snippet =~ s,(\Q$w\E),<span style="background: $colors[$i]; color:black;">$1</span>,gsi;
+		$i++;
+		$i = 0 if ($i > $#colors);
+	}
+
+	return $snippet;
+}
+
 # for pager
 my $pages_per_set = 20;
 
@@ -16,6 +63,7 @@
 my $config=XMLin(undef,
 #		keyattr => { label => "value" },
 		forcecontent => 0,
+		ForceArray => [ 'path' ],
 	);
 
 my $from_utf8 = Text::Iconv->new('UTF8', $config->{charset});
@@ -40,6 +88,7 @@
 # FIX: doesn't work very well
 if ($config->{findaffix}) {
 	foreach my $findaffix (split(/[, ]+/,x($config->{findaffix}))) {
+		next if (! -f $findaffix);
 		my $spelling_alt = new Lingua::Spelling::Alternative;
 		$spelling_alt->load_findaffix($findaffix);
 		push @spellings,$spelling_alt;
@@ -47,6 +96,7 @@
 }
 if ($config->{affix}) {
 	foreach my $affix (split(/[, ]+/,x($config->{affix}))) {
+		next if (! -f $affix);
 		my $spelling_alt = new Lingua::Spelling::Alternative;
 		$spelling_alt->load_affix($affix);
 		push @spellings,$spelling_alt;
@@ -62,12 +112,12 @@
 	$labels{$_->{value}} = x($_->{content});
 }
 
-my $path = param('path');	# limit to this path
+my $path;
+# limit to this path
+$path .= '"'.join('*" or "',param('path')).'*"' if (param('path'));
 my %path_label;
 my @path_name;
 foreach (@{$config->{paths}->{path}}) {
-
-print STDERR "##: $_->{limit}",x($_->{content}),"\n";
 	push @path_name,x($_->{limit});
 	$path_label{$_->{limit}} = x($_->{content});
 }
@@ -90,6 +140,8 @@
 if (@path_name) {
 	print br,x($config->{text}->{limit});
 	print popup_menu(-name=>'path',-values=>\@path_name,-labels=>\%path_label,-default=>$path);
+} elsif (param('path')) {
+	print hidden(-name=>'path',-values=>param('path'));
 }
 print end_form,hr;
 
@@ -98,13 +150,22 @@
 	my $s;
 	# re-write query from +/- to and/and not
 
+	my @s_elem;
+
 	my $search = param('search');
-	my $s_phrase = "";
+
+	# strip spaces
+	$search =~ s/^\s+//;
+	$search =~ s/\s+$//;
+	# fixup search string
+	$search=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/;	# 1250 -> iso8859-2
+	$search=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/;
+
+	# extract phrases and put them first
 	while ($search =~ s/\s*("[^"]+")\s*/ /) {
 		$s .= "$1 ";
+		push @s_elem, $1;
 	}
-	$search =~ s/^\s+//;
-	$search =~ s/\s+$//;
 
 	my %words;
 
@@ -123,10 +184,10 @@
 			} else {
 				$s.="$2* ";
 			}
+			push @s_elem, $2;
 		} else {
 			if (@spellings && !param('no_affix')) {
 				my $w = $_; $w =~ s/[\*\s]+//g;
-				#$s.="(".join("* or ",$spelling_alt->alternatives($w))."*) ";
 				my $or="";
 				foreach my $spelling_alt (@spellings) {
 					$s.="$or(".join("* or ",$spelling_alt->alternatives($w))."*) ";
@@ -135,34 +196,21 @@
 			} else {
 				$s.="$_* ";
 			}
+			push @s_elem, $_;
 		}
 	}
 
-	# fixup search string
-	$s=~tr/šðžèæŠÐŽÈÆ/¹ð¾èæ©Ð®ÈÆ/;	# 1250 -> iso8859-2
-	$s=~tr/¹©ðÐèÈæƾ®/sSdDcCcCzZ/;
+	# fix multiple stars
 	$s=~s/\*\*+/*/g;
 
 	# limit to some path
-	$s = "swishdocpath=(\"*$path*\") and $s" if ($path);
+	$s = "swishdocpath=($path) and $s" if ($path);
 
 	my %params;	# optional parametars for swish
 
 	# default format for output
 	my $hit_fmt = "<a href=\"%s\">%s</a> [%s]<br>\n";
 
-	# output start of table
-	print qq{
-<table border="0">
-	};
-	# html before and after each hit
-	my $tr_pre = qq{
-<tr><td>
-	};
-	my $tr_post = qq{
-</td></tr>
-	};
-
 	if (@properties) {
 		$hit_fmt = x($config->{hit}) if (! param('no_properties'));
 		$params{properties} = \@properties;
@@ -170,56 +218,12 @@
 		$hit_fmt = x($config->{hit}) if (x($config->{hit}));
 	}
 
-#	my $sh = SWISH->connect('Fork',
-#		prog     => x($config->{prog}),
-#		indexes  => x($config->{index}),
-#		results  => sub {
-#			my ($sh,$hit) = @_;
-#
-#			if ($config->{url}) {
-#				printf ($hit_fmt ,"http://".virtual_host().x($config->{url}).$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties));
-#			} else {
-#				printf ($hit_fmt ,$hit->swishdocpath,e($hit->swishtitle) || 'untitled',$hit->swishrank, map($hit->$_, @properties) );
-#
-#			}
-#
-##			print $_[1]->as_string,"<br>\n";
-##			my @fields = $hit->field_names;
-##			print "Field '$_' = '", $hit->$_, "'<br>\n" for sort @fields;
-#		},
-#		maxhits => param('max_hits') || $max_hits,
-#		\%params,
-#	);
-#
-#	die $SWISH::errstr unless $sh;
-#
-#	$hits = $sh->query($s);
-#
-#	if ($hits && $hits > 0) {
-#		print p,hr;
-#		printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s);
-#	} else {
-#		print p;
-#		printf (x($config->{text}->{no_hits}),$s,$sh->errstr);
-#	}
-#	if ($hits && $hits > 0) {
-#		print p,hr;
-#		printf (x($config->{text}->{hits}),$hits,param('max_hits') || $max_hits,$s);
-#	} else {
-#		print p;
-#		printf (x($config->{text}->{no_hits}),$s,$sh->errstr);
-#	}
-
 	my $swish = SWISH::API->new($config->{index});
-
 	$swish->AbortLastError if $swish->Error;
-
 	my $results = $swish->Query($s);
-
 	my $hits = $results->Hits;
 
 
-
 	# build pager
 	my $current_page = param('page') || 1;
 
@@ -242,6 +246,22 @@
 		printf (x($config->{text}->{hits}),$i,$results->Hits,$s);
 	}
 
+	my %path2title;
+	foreach my $p (@{$config->{path2title}->{path}}) {
+		$path2title{$p->{dir}} = $p->{content};
+	}
+
+	# output start of table
+	print qq{
+<table border="0">
+	};
+	# html before and after each hit
+	my $tr_pre = qq{
+<tr><td>
+	};
+	my $tr_post = qq{
+</td></tr>
+	};
 
 	for(my $i=$pager->first; $i<=$pager->last; $i++) {
 
@@ -249,10 +269,14 @@
 		last if (! $result);
 
 		my @arr;
+
 		foreach my $prop (@properties) {
 			if ($prop =~ m/swishdescription/) {
-				my $tmp = $result->Property($prop);
-				$tmp =~ s/<[^>]+>//g;
+				my $tmp = get_snippet(
+					$result->Property($prop),
+					@s_elem,
+				);
+				
 				push @arr, $tmp;
 			} else {
 				push @arr, $result->Property($prop);
@@ -263,11 +287,21 @@
 		my $rank = $result->Property("swishrank");
 		my $host = $result->Property("swishdocpath");
 		$host = "http://".virtual_host().x($config->{url}).$result->Property("swishdocpath") if ($config->{url});
+
+		foreach my $p (keys %path2title) {
+			if ($host =~ m/$p/i) {
+				$title =~ s/$path2title{$p}\s*[:-]+\s*//;
+				$title = $path2title{$p}." :: ".$title;
+				last;
+			}
+		}
+
 		print $tr_pre,$i,". ";
 		# print collection name which is not link
 		if ($title =~ s/^(.+? :: )//) {
 			print $1;
 		}
+
 		printf($hit_fmt, $host, $title || 'untitled', $rank, @arr);
 		print $tr_post;
 
@@ -278,9 +312,14 @@
 
 	my $nav_fmt=qq{ <a href="%s">%s</a> };
 
+	if ($pager->current_page() > $pager->first_page) {
+		param('page', $pager->current_page - 1);
+		$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'&lt;&lt;');
+	}
+
 	if ($pager->previous_set) {
 		param('page', $pager->previous_set);
-		$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'&lt;&lt;');
+		$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..');
 	}
 
 
@@ -297,13 +336,22 @@
 
 	if ($pager->next_set) {
 		param('page', $pager->next_set);
+		$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'..');
+	}
+
+	if ($pager->current_page() < $pager->last_page) {
+		param('page', $pager->current_page + 1);
 		$nav_html .= sprintf($nav_fmt,url(-relative=>1, -query=>1),'&gt;&gt;');
 	}
 
+	if ($config->{text}->{pages}) {
+		$nav_html = x($config->{text}->{pages})." ".$nav_html;
+	}
+
 	# end html table
 	print qq{
 <tr><td>
-Pages: $nav_html
+$nav_html
 </td></tr>
 </table>
 	};