X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=lib%2FMojoFacets%2FData.pm;h=70d4d039707b3cd444a8e47e1a2807d6814e4a34;hb=251b2138da0b45b4f28faa2675e45eb1d641f1ef;hp=2cd2d5ac1cfdffb43a9650a47502c95ce4de81e9;hpb=38b4bb0047fd845f1f7c078501b948a29f3b2cbc;p=MojoFacets.git

diff --git a/lib/MojoFacets/Data.pm b/lib/MojoFacets/Data.pm
index 2cd2d5a..70d4d03 100644
--- a/lib/MojoFacets/Data.pm
+++ b/lib/MojoFacets/Data.pm
@@ -12,6 +12,9 @@ use Encode;
 use locale;
 use File::Find;
 
+our $loaded;
+our $filters;
+
 sub index {
 	my $self = shift;
 
@@ -29,37 +32,64 @@ sub index {
 		}
 	}, $path);
 
-	$self->render( files => [ @files ] );
-}
+	@files = sort { lc $a cmp lc $b } @files;
+	my $size;
+	$size->{$_} = -s "$path/$_" foreach @files;
 
-our $data;
-our $stats;
+	$self->render(
+		files => [ @files ],
+		size => $size,
+		loaded => $loaded,
+		filters => $filters,
+	);
+}
 
-sub load {
-	my $self = shift;
+sub _load_path {
+	my ( $self, $path ) = @_;
 
-	my $path = $self->app->home->rel_file( 'data/' . $self->param('path') );
-	die "$path $!" unless -r $path;
+	return if defined $loaded->{$path}->{data};
 
-	$self->session('path' => $self->param('path'));
+	my $full_path = $self->app->home->rel_file( 'data/' . $path );
+	die "$full_path $!" unless -r $full_path;
 
 	# we could use Mojo::JSON here, but it's too slow
 #	$data = from_json read_file $path;
-	$data = read_file $path;
-	Encode::_utf8_on($data);
+	my $data = read_file $full_path;
 	warn "# data snippet: ", substr($data,0,200);
+	my @header;
 	if ( $path =~ m/\.js/ ) {
+		Encode::_utf8_on($data);
 		$data = from_json $data;
 	} elsif ( $path =~ m/\.txt/ ) {
-		my @lines = split(/\n/, $data);
+		my @lines = split(/\r?\n/, $data);
 		$data = { items => [] };
 
-		my @header = split(/\|/, shift @lines);
+		my $header_line = shift @lines;
+		my $multiline = $header_line =~ s/\^//g;
+		@header = split(/\|/, $header_line );
 		warn "# header ", dump( @header );
-		$self->session( 'header' => [ @header ] );
-		$self->session( 'columns' => [ @header ] );
-		foreach my $line ( @lines ) {
+		while ( my $line = shift @lines ) {
+			$line =~ s/\^//g;
+			chomp $line;
 			my @v = split(/\|/, $line);
+			while ( @lines && $#v < $#header ) {
+				$line = $lines[0];
+				$line =~ s/\^//g;
+				chomp $line;
+				my @more_v = split(/\|/, $line);
+				if ( $#v + $#more_v > $#header ) {
+					warn "short line: ",dump( @v );
+					last;
+				}
+				shift @lines;
+				$v[ $#v ] .= shift @more_v if @more_v;
+				push @v, @more_v if @more_v;
+
+				if ( $#v > $#header ) {
+					die "# splice $#header ", dump( @v );
+					@v = splice @v, 0, $#header;
+				}
+			}
 			my $item;
 			$item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
 			push @{ $data->{items} }, $item;
@@ -68,24 +98,26 @@ sub load {
 		warn "file format unknown $path";
 	}
 
-	$stats = {};
+	my $stats;
 
 	foreach my $e ( @{ $data->{items} } ) {
 		foreach my $n ( keys %$e ) {
 			$stats->{$n}->{count}++;
+			my @v;
 			if ( ref $e->{$n} eq 'ARRAY' ) {
-
 				$stats->{$n}->{array} += $#{ $e->{$n} } + 1;
-
-				foreach my $x ( @{$e->{$n}} ) {
-					$stats->{$n}->{numeric}++
-						if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
-				}
-
+				@v = @{ $e->{$n} };
 			} else {
+				@v = ( $e->{$n} );
+			}
+
+			foreach my $x ( @v ) {
 				$stats->{$n}->{numeric}++
-					if $e->{$n} =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
+					if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
+				$stats->{$n}->{empty}++
+					if $x =~ m/^\s*$/;
 			}
+
 		}
 	}
 
@@ -95,21 +127,75 @@ sub load {
 			if $stats->{$n}->{array} == $stats->{$n}->{count};
 	}
 
-	$self->session( 'header' => [
+	@header =
 		sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
 		grep { defined $stats->{$_}->{count} } keys %$stats
-	] ) unless $self->session( 'header' );
+		unless @header;
 
 	warn dump($stats);
 
-	$self->redirect_to( '/data/columns' );
+	$loaded->{ $path } = {
+		header => [ @header ],
+		stats  => $stats,
+		full_path => $full_path,
+		size => -s $full_path,
+		data => $data,
+	};
+
+}
+
+
+sub load {
+	my $self = shift;
+
+	my @paths = $self->param('paths');
+	warn "# paths ", dump @paths;
+	$self->_load_path( $_ ) foreach @paths;
+
+ 	my $path = $self->param('path') || $self->redirect_to( '/data/index' );
+	warn "# path $path\n";
+	$self->session('path' => $path);
+	$self->_load_path( $path );
+
+	$self->session( 'header' => $loaded->{$path}->{header} );
+	if ( ! defined $loaded->{$path}->{columns} ) {
+		$self->session( 'columns' => $loaded->{$path}->{header} );
+		$self->redirect_to( '/data/columns' );
+	} else {
+		$self->session( 'columns' => $loaded->{$path}->{columns} );
+		$self->redirect_to( '/data/items' );
+	}
+		
+}
+
+
+sub _loaded {
+	my ( $self, $name ) = @_;
+	my $path = $self->session('path');
+	die "$path $name doesn't exist in loaded ",dump( $loaded )
+		unless defined $loaded->{$path}->{$name};
+	return $loaded->{$path}->{$name};
+}
+
+
+sub _checked {
+	my $self = shift;
+	my $checked;
+	$checked->{$_}++ foreach @_;
+#	warn "# _checked ",dump($checked);
+	return $checked;
 }
 
 
 sub columns {
     my $self = shift;
 
-	$self->redirect_to( '/data/index' ) unless $self->session('header');
+	if ( $self->param('columns') ) {
+		$self->_perm_array('columns');
+		$self->redirect_to('/data/items');
+	}
+
+	my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
 
 	my @columns;
 	@columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
@@ -130,9 +216,11 @@ sub _perm_array {
     my ($self,$name) = @_;
 
 	my @array = $self->param($name);
+	my $path  = $self->session('path');
 
 	if ( @array ) {
 		$self->session($name => [ @array ]);
+		$loaded->{$path}->{$name} = [ @array ];
 	} elsif ( my $session = $self->session($name) ) {
 		if ( ref $session eq 'ARRAY' ) {
 			@array = @$session;
@@ -161,7 +249,7 @@ sub _perm_scalar {
 		$self->session($name => $scalar);
 	}
 
-	warn "# $name ",dump $scalar;
+	warn "# _perm_scalar $name ",dump $scalar;
 	return $scalar;
 }
 
@@ -173,61 +261,115 @@ sub filter {
 
 	warn "# filter $name vals ",dump(@vals);
 
-	my $filters = $self->session('filters');
 	if ( @vals ) {
 		$filters->{$name} = [ @vals ];
+		warn "# filter + $name $#vals\n";
 	} else {
+		warn "# filter - $name\n";
 		delete $filters->{$name};
 	}
-	$self->session( 'filters' => $filters );
 
-	warn "# filters ",dump($self->session('filters'));
+	warn "# filters ",dump($filters);
 
 	$self->session( 'offset' => 0 );
 
 	$self->redirect_to('/data/items');
 }
 
-sub _filter_item {
-	my ( $self, $filters, $i ) = @_;
-	my $pass = 1;
-	foreach my $n ( keys %$filters ) {
-		my @filter_values = @{ $filters->{$n} };
-		my $include_missing = grep { /^_missing/ } @filter_values;
-		if ( ! exists $i->{$n} ) {
-			if ( $include_missing ) {
-				$pass = 1;
-				next;
-			} else {
+
+sub _data_items {
+	my ( $self, $all ) = @_;
+ 	my $data = $self->_loaded( 'data' );
+
+	return @{ $data->{items} } if $all == 1;
+
+	my $filters = $self->_current_filters;
+	my $filter_value;
+	foreach my $f ( keys %$filters ) {
+		foreach my $n ( @{ $filters->{$f} } ) {
+			$filter_value->{$f}->{$n} = 1;
+		}
+	}
+ 	my @items = @{ $data->{items} };
+	@items = grep {
+		my $i = $_;
+		my $pass = 1;
+		foreach my $n ( keys %$filter_value ) {
+			if ( ! exists $i->{$n} ) {
+				if ( defined $filter_value->{$n}->{_missing} ) {
+					$pass = 1;
+					next;
+				} else {
+					$pass = 0;
+					last;
+				}
+			}
+			# and match any of values in element
+			my $have_values = 0;
+			foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
+				$have_values ||= 1 if defined $filter_value->{$n}->{$v};
+			}
+			if ( ! $have_values ) {
 				$pass = 0;
 				last;
 			}
 		}
-		# and match any of values in element
-		my $have_values = 0;
-		foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
-			$have_values ||= 1 if grep { m/^\Q$v\E$/ } @filter_values;
-		}
-		if ( ! $have_values ) {
-			$pass = 0;
-			last;
-		}
-	}
-	return $pass;
+		$pass;
+	} @items if $filter_value;
+	return @items;
 }
 
-sub _data_items {
+
+sub _current_filters {
 	my $self = shift;
-	my $filters = $self->session('filters');
-	grep {
-		$filters ? $self->_filter_item( $filters, $_ ) : 1;
- 	} @{ $data->{items} };
+	my $current_filters;
+	$current_filters->{ $_ } = $filters->{ $_ }
+		foreach (
+			grep { defined $filters->{ $_ } }
+			@{ $self->_loaded('header') }
+		);
+	warn "# current_filters ",dump($current_filters);
+	return $current_filters;
+}
+
+sub _data_sorted_by {
+	my ( $self, $order ) = @_;
+
+	my $path = $self->session('path');
+
+	if ( defined $loaded->{$path}->{sorted}->{$order} ) {
+		return $loaded->{$path}->{sorted}->{$order};
+	}
+
+ 	my $data = $self->_loaded( 'data' );
+	my $numeric = $self->_is_numeric($order);
+	my $missing = $numeric ? 0 : '';
+	no warnings qw(numeric);
+	my $nr = 0;
+	my @sorted = map {
+		$_->[0]
+	} sort {
+		if ( $numeric ) {
+			$a->[1] <=> $b->[1]
+		} else {
+			$a->[1] cmp $b->[1]
+		}
+	} map {
+		[ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
+	} @{ $data->{items} }
+	;
+
+	warn "sorted $order"; # ,dump( @sorted );
+
+	$loaded->{$path}->{sorted}->{$order} = [ @sorted ];
 }
 
+
 sub items {
-    my $self = shift;
+	my $self = shift;
 
-	$self->redirect_to('/data/index') unless $data->{items};
+	my $path = $self->session('path');
+	$self->redirect_to('/data/index') unless defined $loaded->{ $path };
 
 	my @columns = $self->_perm_array('columns');
 	$self->redirect_to('/data/columns') unless @columns;
@@ -240,29 +382,28 @@ sub items {
 	# fix offset when changing limit
 	$offset = int( $offset / $limit ) * $limit;
 
-	# FIXME - multi-level sort
-	my $numeric = $self->_is_numeric($order);
-	my $missing = $numeric ? 0 : '';
-	no warnings qw(numeric);
-	my @sorted = sort {
-		my $v1 = exists $a->{$order} ? join('', @{$a->{$order}}) : $missing;
-		my $v2 = exists $b->{$order} ? join('', @{$b->{$order}}) : $missing;
-		($v1,$v2) = ($v2,$v1) if $sort eq 'd';
-		$numeric ? $v1 <=> $v2 : $v1 cmp $v2 ;
-	} $self->_data_items;
+	my $sorted = $self->_data_sorted_by( $order );
+
+	my $rows = $#$sorted + 1;
 
-#	warn "# sorted ", dump @sorted;
+	my @sorted;
+	foreach my $i ( 0 .. $limit ) {
+		my $pos = $sort eq 'a' ? ( $i + $offset ) : ( $#$sorted - $i - $offset );
+		last unless defined $sorted->[$pos];
+		push @sorted, $loaded->{$path}->{data}->{items}->[ $sorted->[$pos] ];
+	}
 
-	my $rows = $#sorted + 1;
+#	warn "# sorted ", dump $sorted;
 
 	$self->render(
 		order => $order,
 		offset => $offset,
 		limit => $limit,
-		sorted => [ splice @sorted, $offset, $limit ],
+		sorted => [ @sorted ],
 		columns => [ @columns ],
 		rows => $rows,
 		numeric => { map { $_, $self->_is_numeric($_) } @columns },
+		filters => $self->_current_filters,
 	);
 
 }
@@ -278,25 +419,31 @@ sub order {
 sub _is_numeric {
 	my ( $self, $name ) = @_;
 
+	my $stats = $self->_loaded( 'stats' );
+
 	# sort facet numerically if more >50% elements are numeric
+	my $count = $stats->{$name}->{count};
+	$count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
 	defined $stats->{$name}->{numeric} &&
-		$stats->{$name}->{numeric} > $stats->{$name}->{count} / 2;
+		$stats->{$name}->{numeric} > $count / 2;
 }
 
 sub facet {
 	my $self = shift;
 
+	my $path = $self->session('path') || $self->redirect_to( '/data/index' );
+
 	if ( my $remove = $self->param('remove') ) {
-		my $f = $self->session('filters');
-		delete $f->{$remove};
-		$self->session( 'filters' => $f );
+		delete $filters->{$remove};
 		$self->redirect_to( '/data/items' );
 	}
 
 	my $facet;
 	my $name = $self->param('name') || die "no name";
 
-	foreach my $i ( $self->_data_items ) {
+	my $all = $self->_perm_scalar('all', 1);
+
+	foreach my $i ( $self->_data_items($all) ) {
 		if ( ! exists $i->{$name} ) {
 			$facet->{ _missing }++;
 		} elsif ( ref $i->{$name} eq 'ARRAY' ) {
@@ -309,18 +456,19 @@ sub facet {
 #	warn "# facet $name ",dump $facet;
 
 	my $checked;
-	if ( my $f = $self->session('filters') ) {
-		if ( defined $f->{$name} ) {
-			$checked = $self->_checked( @{ $f->{$name} } );
-		}
-	}
+	my @facet_names =
+		  $all                      ? keys %$facet
+		: defined $filters->{$name} ? @{ $filters->{$name} }
+		: keys %$facet;
+
+	$checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
 
 	my $sort = $self->param('sort') || 'c';
 
 	# sort facet numerically if more >50% elements are numeric
 	my $numeric = $self->_is_numeric($name);
 
-	my @facet_names = sort {
+	@facet_names = sort {
 		if ( $sort =~ m/a/i ) {
 			$numeric ? $a <=> $b : lc $a cmp lc $b;
 		} elsif ( $sort =~ m/d/i ) {
@@ -331,19 +479,20 @@ sub facet {
 			warn "unknown sort: $sort";
 			$a cmp $b;
 		}
-	} keys %$facet;
+	} @facet_names;
 
 	$self->render( name => $name, facet => $facet, checked => $checked,
 		facet_names => \@facet_names, sort => $sort, numeric => $numeric,
 	);
 }
 
-sub _checked {
+sub edit {
 	my $self = shift;
-	my $checked;
-	$checked->{$_}++ foreach @_;
-	warn "# _checked ",dump($checked);
-	return $checked;
+	my $content = $self->param('content');
+
+	$self->render(
+		content => $content
+	);
 }
 
 1;