X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=vhost%2Fwebpac2.cgi;h=fd76b657b4f95a9c989e24be91b19ed7cc75a209;hb=270b0142ec5a98e84ecb5beaf1a020130ab9e013;hp=d1047daa87866f9388c76626da0a34f20b25eb9b;hpb=9dbb36a204e7ad8eb1a5d65967506ca4e416e924;p=webpac2

diff --git a/vhost/webpac2.cgi b/vhost/webpac2.cgi
index d1047da..fd76b65 100755
--- a/vhost/webpac2.cgi
+++ b/vhost/webpac2.cgi
@@ -7,22 +7,29 @@ use CGI qw/:standard/;
 use CGI::Carp qw/fatalsToBrowser/;
 use File::Slurp;
 use YAML;
-use Search::Estraier;
 use Data::Page;
 use Data::Dump qw/dump/;
+use SWISH::API;
+use JSON;
+use Text::Unaccent::PurePerl qw/unac_string/;
+use HTML::FillInForm::Lite;
+use Encode;
 
 my $range_around = 5;
-my $entries_per_page = 30;
+my @entries_per_page = ( 30, 50, 100, 500 );
+my $debug = param('debug');
 
-print header;
+print header(
+	-charset => 'utf-8',
+);
 
 sub dump_yaml {
 	my $name = shift;
-	print qq|<pre># $name\n|, YAML::Dump( @_ ), qq|</pre>|;
+	print qq|<div class=dump><tt>$name</tt><pre>|, YAML::Dump( @_ ), qq|</pre></div>| if $debug;
 }
 
 sub show_pager {
-	my ($pager,$coderef) = @_;
+	my ($pager) = @_;
 
 	my @show_pages;
 	my $after_current = 0;
@@ -46,26 +53,39 @@ sub show_pager {
 
 	my ( $prev, $next ) = ( '&lt;&lt;', '&gt;&gt;' );
 
+	sub li_a_href {
+		my ( $page, $label, $attr ) = @_;
+		param( 'current_page', $page );
+		my $url = self_url( -query => 1 );
+		$attr ||= '';
+		$label ||= $page;
+		qq|<li$attr><a href="$url" title="$page">$label</a></li>|;
+	}
+
 	return
-		  $pager->previous_page ? $coderef->( $pager->previous_page, $prev ) : qq|<span class=skip>$prev</span>|
-		, join( ' ', map {
-			if ( $_ == $pager->current_page ) {
-				qq|<span class=current_page>$_</span>|;
+		  $pager->previous_page ? li_a_href( $pager->previous_page, $prev ) : qq|<li class=skip>$prev</li>|
+		, ( map {
+			if ( $_ eq $pager->current_page ) {
+				qq|<li class=current_page>$_</li>|;
 			} elsif ( $_ eq '' ) {
-				qq|<span class=skip>...</span>|;
+				qq|<li class=skip>...</li>|;
 			} else {
-				$coderef->( $_ );
+				li_a_href( $_ );
 			}
 		} @show_pages )
-		, $pager->next_page ? $coderef->( $pager->next_page, $next ) : qq|<span class=skip>$next</span>|
+		, $pager->next_page ? li_a_href( $pager->next_page, $next ) : qq|<li class=skip>$next</li>|
 		;
 		
 }
 
 my $path = $ENV{PATH_INFO} || 'ecas';
+$path =~ s{^/+}{};
+$path =~ s{/+$}{};
 my $dir = $0;
 $dir =~ s{/[^/]+.cgi}{};
 
+dump_yaml( 'dir', $dir );
+
 my $config = YAML::LoadFile( "$dir/$path/config.yml" );
 
 my $database = (keys %{ $config->{databases} })[0];
@@ -80,104 +100,243 @@ if ( -e $html_markup ) {
 	undef $html_markup;
 }
 
-my $estraier = YAML::LoadFile( "$dir/../var/estraier/$database.yaml" );
+my $stats;
+{
+	my $path = "$dir/../var/swish/$database.yaml";
+	$stats = YAML::LoadFile( $path );
+	dump_yaml( "stats $path", $stats );
+}
 
 my $db = $config->{databases}->{$database};
 
-my @attr = keys %{ $estraier->{attr} }; # FIXME replace with real gnerated lookup
+sub read_config_txt {
+	my ( $file ) = @_;
+	my $input;
+	my $path ="$dir/$path/$path-$file.txt";
+	if ( ! -e $path ) {
+		warn "missing $path";
+		return;
+	}
+	foreach ( split(/[\n\r]+/, read_file( $path ) ) ) {
+		my ( $val,$label ) = split(/\s*\t\s*/,$_,2);
+		push @{ $input->{ '-values' } }, $val;
+				$input->{ '-labels' }->{$val} = $label;
+	}
+	return $input;
+}
+
+my $attr_labels    = read_config_txt 'labels';
+my $attr_operators = read_config_txt 'operators';
+
+my @attr = @{ $attr_labels->{'-values'} } if $attr_labels;
+@attr = keys %{ $stats->{attr} } unless @attr;
+
+
+warn dump( $attr_labels, $attr_operators );
+
+my $only_input;
+my $inputs_available = 0;
+
+foreach ( @{ $db->{input} } ) {
+	my $input = $_->{name} || die "no name in ",dump( $_ );
+	next unless defined $stats->{input}->{$input}; # skip inputs without data
+	if ( ! $only_input->{'-labels'}->{$input} ) {
+		push @{ $only_input->{'-values'} }, $input;
+		        $only_input->{'-labels'}->{$input} = $_->{description} || $input;
+		$inputs_available++;
+	}
+}
+
+warn "## only_input = ", dump( $only_input );
+
+my @style = ( '../../style.css' );
+push @style, "../../$path/$path.css" if -e "$dir/$path/$path.css";
+dump_yaml( 'style', \@style );
+
+sub search_form {
+
+	my $form_html = "$dir/$path/$path-search.html";
+	if ( -e $form_html ) {
+		my $html = read_file( $form_html );
+		my $q = CGI->new();
+		my $h = HTML::FillInForm::Lite->new();
+		return $h->fill(\$html, $q);
+	}
+
+	qq|<a name="form"></a>|,
+	start_form( -action => self_url( query => 0 ) ),
+		checkbox_group(
+			-name => 'attr',
+			%$attr_labels,
+#			-linebreak => 0,
+		),
+		textfield( -name => 'search' ),
+		$attr_operators ? popup_menu( -name => 'attr_operator', %$attr_operators ) : '',
+		submit( -value => 'Search' ),
+#		hidden( -name => 'entries_per_page', -default => $entries_per_page ),
+		popup_menu( -name => 'entries_per_page', -values => [ @entries_per_page ], -title => 'entries per page' ),
+		# we need current_page fixed at 1 so that every submit through form will reset it
+		qq|<input type=hidden name=current_page value=1 >|,
+		checkbox( -name => 'debug', -default => 0 ), # FIXME hidden?
+		qq|<div id=inputs>|,
+		$inputs_available > 1 ?
+		h2( 'Select input' ) .
+		checkbox_group(
+			-name => 'only_input',
+			%$only_input,
+			-linebreak=> 'true',
+		) : '',
+		qq|</div>|,
+		end_form,
+	;
+}
+
 
 print
 	start_html(
 		-title => $db->{name},
-		-style => '../../style.css',
+		-style => [ @style ],
 	),
 	h1( $db->{name} ),
 	qq|<div id=description>|, $db->{description}, qq|</div>|,
-	start_form( -action => self_url( query => 0 ) ),
-		radio_group(
-			-name => 'attr',
-			-values => [ @attr ],
-#			-linebreak => 0,
-		),
-		textfield( -name => 'search' ),
-		submit,
-		textfield( -name => 'entries_per_page', -default => $entries_per_page ),
-		textfield( -name => 'current_page', -default => 1 ),
 ;
 
-print	end_form;
-
 if ( my $search = param('search') ) {
 
-	print qq|<div id="results">|;
+	$search = unac_string( Encode::decode('utf-8',$search) );
+
+	print qq|
+		<a href="#form" class="skip" title="skip to search form">#</a>
+		<div id="results">
+	|;
+
+	my $swish = SWISH::API->new( "$dir/../var/swish/$database" );
+	$swish->abort_last_error if $swish->Error;
+
+	my @search = (); 
+	my @attrs = param('attr');
+	my $op = param('attr_operator');
+
+	if ( $search =~ m{(=|"|\bAND\b|\bOR\b)} ) {
+		push @search, $search;
+	} elsif ( @attrs ) {
+
+		$op ||= 'Q*';
+		my @or;
+		foreach my $attr ( @attrs ) {
+			my $v = $search;
+			$v =~ s/^\s+//;
+			warn "-- v: $v\n";
+			sub rewrite {
+				my ( $attr, $whitespace, $v ) = @_;
+				warn "## filter $op $whitespace $v\n";
+				my $template = $op;
+				   $template =~ s{Q}{$v};
+				$whitespace = " AND " if $whitespace;
+
+				# don't return -* &* and other non-word characters
+				return '' if $template =~ m/^\W\*$/ || $template =~ m/\band\b/i;
+
+				return
+					$whitespace .
+					$attr . '="' . $template . '"';
+					;
+			};
+			if ( $op =~ m{\s} ) {
+				my $template = $op;
+				   $template =~ s{Q}{$v};
+				$v = $attr . '="' . $template . '"';
+			} else {
+				$v =~ s{(\s*)(\S+)}{rewrite($attr,$1,$2)}ge;
+			}
+
+			push @or, $v;
+		
+		}
+		push @search, '(' . join(') OR (', @or) . ')';
+
+	} else {
+		push @search, "all=\"$search\"";
+	}
+
+	my $q = '(' . join(') AND (', @search) . ')';
+
+	my @only_input = param('only_input');
+	$q .= ' AND ((' . join(') OR (', map { "input=\"$_\"" } @only_input) . '))' if @only_input;
+
+	warn "# query: $q\n";
+	my $search_obj = $swish->new_search_object;
+	if ( my $sort = param('sort') ) {
+		$search_obj->set_sort( $sort );
+	}
+	my $swish_results = $search_obj->execute( $q );
 
-	my $node = Search::Estraier::Node->new(
-		url => $config->{hyperestraier}->{masterurl} . '/node/' . $database,
-		croak_on_error => 1,
-	);
+	dump_yaml( 'swish_results', $swish_results );
 
-	param( 'entries_per_page', $entries_per_page ) unless param('entries_per_page'); # FIXME not needed?
 	my $pager = Data::Page->new;
-	$pager->total_entries( param('current_page') * param('entries_per_page') );
 	$pager->$_( param($_) ) foreach ( qw/entries_per_page current_page/ );
+	$pager->total_entries( $swish_results->hits );
 
 	dump_yaml( 'pager', $pager );
 
-	my $cond = Search::Estraier::Condition->new;
-	$cond->set_phrase( $search );
-	$cond->set_skip( $pager->skipped );
-	$cond->set_max(  $pager->entries_per_page );
-	my $nres = $node->search( $cond, 0 );
-	$pager->total_entries( $nres->hits );
+	$swish_results->seek_result( $pager->first - 1 );
 
+	if ( ! $pager->total_entries ) {
+		my $no_results = 'No results for search <b>%s</b>';
+		$no_results = $swish->error_string . '<br><b>%s</b>' if $swish->error;
+		printf qq|<div class="error">$no_results</div>\n\n|, $q;
+	} else {
 
-	dump_yaml( 'cond', $cond );
+		my $results = "<b>%d</b> results for search <b>%s</b> showing results %d - %d";
+		printf qq|<div class="message">$results</div>\n\n|, $pager->total_entries, $q, $pager->first, $pager->last;
 
-	if ( ! $nres ) {
-		my $no_results = "No results for search '%s'";
-		printf qq|<div class="error">$no_results</div>|, $search;
-	} else {
+		my $pager_html = join("\n", show_pager( $pager ));
 
-		my $results = "%d results for search '%s' showing results %d - %d on page %d";
-		printf qq|<div class="message">$results</div>|, $nres->hits, $search, $pager->first, $pager->last, $pager->current_page;
-
-		my $pager_html =
-			join(' ', show_pager( $pager,
-				sub {
-					my ($page,$label) = @_;
-					param( 'current_page', $page );
-					my $url = self_url( -query => 1 );
-					$label = $page unless defined $label;
-					qq|<a href="$url">$label</a>|;
-				}
-			))
-		;
+		print qq|<ul class="pager">$pager_html</ul>\n\n| if $pager_html;
+
+		my $nr = $pager->first;
+		print qq|<ol start=$nr>\n|;
 
-		print qq|<div class="pager top">$pager_html</div>| if $pager_html;
+		my $limit = $pager->entries_on_this_page;
 
-		my $start = $pager->first;
-		print qq|<ol start=$start>|;
+		my $nr = 1;
 
-		foreach my $i ( 1 .. $nres->doc_num ) {
-			my $rdoc = $nres->get_doc( $i - 1 );
-			print qq|<li>|;
+		while ( my $result = $swish_results->next_result ) {
+
+			my $data = $result->property('data');
+			dump_yaml( 'data', $data );
+			# FIXME if we produce valid json we shouldn't need eval here!
+			eval { $data = from_json( $data, {utf8 => 1} ); };
+			if ( $@ ) {
+				warn "ERROR: $@ from ",dump( $data );
+				next;
+			}
+
+			my $li_class = '';
+			$li_class = qq| class="z"| if $nr % 2 == 0;
+			print qq|<li$li_class>|;
 			foreach my $attr ( @attr ) {
-				my $v = $rdoc->attr( $attr );
-				if ( defined $v && $html_markup && ! $html_markup_skip->{$attr} ) {
-					eval "\$v = $html_markup->$attr( \$v );";
+				next unless defined $data->{$attr};
+				my $v = $data->{$attr};
+				if ( $html_markup && ! $html_markup_skip->{$attr} ) {
+					eval "\$v = $html_markup->$attr( \$v, \$data );";
 					if ( $@ ) {
 						warn "disable html markup for $attr: $@";
 						$html_markup_skip->{$attr} = $@;
 					}
+				} else {
+					$v =~ s{(http://\S+)}{<a href="$1">$1</a>};
 				}
-				next unless defined $v;
-				print qq|<div><label>$attr</label><span class=$attr>$v</span></div>\n|;
+				my $label = $attr_labels->{'-labels'}->{$attr} || $attr;
+				print qq|<div><label>$label</label><span class=$attr>$v</span></div>\n|;
 			}
 			print qq|</li>\n|;
+
+			last if $nr++ == $pager->last;
 		}
-		print qq|</ol>|;
+		print qq|</ol>\n\n|;
 
-		print qq|<div class="pager bottom">$pager_html</div>| if $pager_html;
+		print qq|<ul class="pager bottom">$pager_html</ul>\n\n| if $pager_html;
 	}
 	print qq|</div>|;
 
@@ -185,8 +344,9 @@ if ( my $search = param('search') ) {
 
 }
 
-dump_yaml( 'estraier', $estraier );
-dump_yaml( 'db', $db );
+print search_form;
+
+dump_yaml( "config databases $database", $db );
 dump_yaml( 'html_markup_skip', $html_markup_skip );
 
 print	end_html;