unpack numeric values separated by commas

[MojoFacets.git] / lib / MojoFacets / Data.pm

diff --git a/lib/MojoFacets/Data.pm b/lib/MojoFacets/Data.pm
index b81816b..e69c16e 100644 (file)
--- a/lib/MojoFacets/Data.pm
+++ b/lib/MojoFacets/Data.pm
@@ -47,11 +47,15 @@ sub index {
 sub _load_path {
        my ( $self, $path ) = @_;
 
-       return if defined $loaded->{$path}->{data};
-
        my $full_path = $self->app->home->rel_file( 'data/' . $path );
        die "$full_path $!" unless -r $full_path;
 
+       if ( defined $loaded->{$path}->{data} ) {
+               my $mtime = (stat($full_path))[9];
+               return if $loaded->{$path}->{mtime} == $mtime;
+               warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
+       }
+
        # we could use Mojo::JSON here, but it's too slow
 #      $data = from_json read_file $path;
        my $data = read_file $full_path;
@@ -91,7 +95,12 @@ sub _load_path {
                                }
                        }
                        my $item;
-                       $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
+                       foreach my $i ( 0 .. $#v ) {
+                               my $v = $v[$i];
+                               # unpack numeric values separated by commas
+                               my $a = $v =~ m/\d+\s*,\s*\d+/ ? [ split(/\,\s*/,$v) ] : [ $v ];
+                               $item->{ $header[$i] || "f_$i" } = $a;
+                       }
                        push @{ $data->{items} }, $item;
                }
        } else {
@@ -127,6 +136,16 @@ sub _load_path {
                        if $stats->{$n}->{array} == $stats->{$n}->{count};
        }
 
+       if ( ! @header ) {
+               if ( defined $data->{header} ) {
+                       if ( ref $data->{header} eq 'ARRAY' ) {
+                               @header = @{ $data->{header} };
+                       } else {
+                               warn "header not array ", dump( $data->{header} );
+                       }
+               }
+       }
+
        @header =
                sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
                grep { defined $stats->{$_}->{count} } keys %$stats
@@ -139,6 +158,7 @@ sub _load_path {
                stats  => $stats,
                full_path => $full_path,
                size => -s $full_path,
+               mtime => (stat($full_path))[9],
                data => $data,
        };
 
@@ -370,6 +390,8 @@ sub _data_sorted_by {
 
        my $path = $self->session('path');
 
+       warn "_data_sorted_by $order from $path";
+
        if ( defined $loaded->{$path}->{sorted}->{$order} ) {
                return $loaded->{$path}->{sorted}->{$order};
        }
@@ -392,7 +414,8 @@ sub _data_sorted_by {
        } @{ $data->{items} }
        ;
 
-       warn "sorted $order"; # ,dump( @sorted );
+       warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
+       #warn "# sorted ",dump( @sorted );
 
        $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
 }
@@ -433,15 +456,15 @@ sub items {
                }
        }
 
-       my $filtered_names = join(' ',sort @filter_names);
+       my $all_filters = join(' ',sort @filter_names,'order:',$order);
 
-#      warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
+#      warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
 
-       if ( ! defined $loaded->{$path}->{filtered}->{$filtered_names} ) {
+       if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
 
                my $path_filters = $loaded->{$path}->{filters};
 
-               warn "create combined filter for $filtered_names\n";
+               warn "create combined filter for $all_filters\n";
 
                my @filtered;
                foreach my $i ( 0 .. $#$sorted ) {
@@ -461,13 +484,13 @@ sub items {
                        push @filtered, $pos;
                }
 
-               $loaded->{$path}->{filtered}->{$filtered_names} = [ @filtered ];
+               $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
        }
 
-       my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
-               if defined $loaded->{$path}->{filtered}->{$filtered_names};
+       my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
+               if defined $loaded->{$path}->{filtered}->{$all_filters};
 
-       warn "filtered_names $filtered_names",dump( $filtered ) if $filtered;
+       warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
 
        my $sorted_items;
        my $data = $self->_loaded('data');
@@ -550,16 +573,15 @@ sub facet {
        my $data = $self->_loaded('data');
 
        my $filters = $self->_current_filters;
-       my $filtered_names = join(' ',sort keys %$filters);
-#      warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
-       my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
-               if defined $loaded->{$path}->{filtered}->{$filtered_names};
+       my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
+       my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
+               if defined $loaded->{$path}->{filtered}->{$all_filters};
 
        if ( ! $filtered || $all ) {
                $filtered = [ 0 .. $#{ $data->{items} } ];
                warn "filter all values\n";
        } else {
-               warn "filter using $filtered_names\n";
+               warn "filter using $all_filters\n";
        }
 
        foreach my $i ( @$filtered ) {
@@ -583,22 +605,26 @@ sub 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 $sort = $self->param('sort');
+       # sort numeric facets with more than 5 values ascending
+       $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
+
        @facet_names = sort {
-               if ( $sort =~ m/a/i ) {
-                       $numeric ? $a <=> $b : lc $a cmp lc $b;
-               } elsif ( $sort =~ m/d/i ) {
-                       $numeric ? $b <=> $a : lc $b cmp lc $a;
-               } elsif ( $sort =~ m/c/i ) {
-                       ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
+               my $result;
+               if ( $sort eq 'a' ) {
+                       $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
+               } elsif ( $sort eq 'd' ) {
+                       $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
+               } elsif ( $sort eq 'c' ) {
+                       $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
                } else {
                        warn "unknown sort: $sort";
-                       $a cmp $b;
+                       $result = $a cmp $b;
                }
+               $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
+               $result;
        } @facet_names;
 
        $self->render( name => $name, facet => $facet, checked => $checked,