load multiple files at once

This is quote sweeping change, introducing single global $loaded
variable to cache all metadata about loaded data and data itself.

diff --git a/lib/MojoFacets/Data.pm b/lib/MojoFacets/Data.pm
index eebc114..f1d7a60 100644 (file)
--- a/lib/MojoFacets/Data.pm
+++ b/lib/MojoFacets/Data.pm
@@ -12,6 +12,8 @@ use Encode;
 use locale;
 use File::Find;
 
+our $loaded;
+
 sub index {
        my $self = shift;
 
@@ -29,38 +31,41 @@ 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,
+       );
+}
 
-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;
+       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;
-               $self->session( 'header' => 0 ); # generate later
        } elsif ( $path =~ m/\.txt/ ) {
                my @lines = split(/\r?\n/, $data);
                $data = { items => [] };
 
-               my $headers = shift @lines;
-               my $multiline = $headers =~ s/\^//g;
-               my @header = split(/\|/, $headers );
+               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 ] );
                while ( my $line = shift @lines ) {
                        chomp $line;
                        my @v = split(/\|/, $line);
@@ -90,7 +95,7 @@ sub load {
                warn "file format unknown $path";
        }
 
-       $stats = {};
+       my $stats;
 
        foreach my $e ( @{ $data->{items} } ) {
                foreach my $n ( keys %$e ) {
@@ -117,23 +122,60 @@ 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);
 
+       $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} );
+       $self->session( 'columns' => $loaded->{$path}->{header} );
+
        $self->session( 'filters' => {} );
 
        $self->redirect_to( '/data/columns' );
 }
 
 
+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 columns {
     my $self = shift;
 
-       $self->redirect_to( '/data/index' ) unless $self->session('header');
+       my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
 
        my @columns;
        @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
@@ -243,6 +285,7 @@ sub _filter_item {
 sub _data_items {
        my $self = shift;
        my $filters = $self->session('filters');
+       my $data = $self->_loaded( 'data' );
        grep {
                $filters ? $self->_filter_item( $filters, $_ ) : 1;
        } @{ $data->{items} };
@@ -251,7 +294,7 @@ sub _data_items {
 sub items {
     my $self = shift;
 
-       $self->redirect_to('/data/index') unless $data->{items};
+       $self->redirect_to('/data/index') unless defined $loaded->{ $self->session('path') };
 
        my @columns = $self->_perm_array('columns');
        $self->redirect_to('/data/columns') unless @columns;
@@ -302,6 +345,8 @@ sub order {
 sub _is_numeric {
        my ( $self, $name ) = @_;
 
+       my $stats = $self->_loaded( 'stats' );
+
        # sort facet numerically if more >50% elements are numeric
        defined $stats->{$name}->{numeric} &&
                $stats->{$name}->{numeric} > $stats->{$name}->{count} / 2;

diff --git a/public/mojo_facets.css b/public/mojo_facets.css
index 79486d8..8b72f35 100644 (file)
--- a/public/mojo_facets.css
+++ b/public/mojo_facets.css
@@ -22,7 +22,10 @@
        color: #000;
 }
 
-li label:hover {
+li label:hover, td label:hover {
        background: #ff8;
 }      
 
+td label {
+       white-space: nowrap;
+}

diff --git a/templates/data/index.html.ep b/templates/data/index.html.ep
index 10e202b..28707aa 100644 (file)
--- a/templates/data/index.html.ep
+++ b/templates/data/index.html.ep
@@ -3,20 +3,25 @@
 
 <form method=post action=/data/load >
 
-<ul>
-% my $current = session('path');
+<table>
+<tr><th>name</th><th>bytes</th></tr>
 % foreach my $n ( @$files ) {
-<li>
+<tr><td>
 <label>
-%  if ( $n eq $current ) {
-<input type=radio disabled> <b><%= $n %></b>
-%  } else {
-<input type=radio name=path value=<%= $n %> > <%= $n %>
-%  }
+<input name=path  type=radio    value="<%= $n %>" <%= session('path') eq $n ? 'disabled' : '' %>>
+<%= $n %>
 </label>
+</td><td align=right><%= $size->{$n} %>
+</td><td><input name=paths type=checkbox value="<%= $n %>" <%= defined $loaded->{$n}->{stats} ? 'checked' : '' %>>
+</td><td>
+%  if ( defined $loaded->{$n}->{header} ) {
+<%= join("\n",@{ $loaded->{$n}->{header} }) %>
+%  }
+</td></tr>
 % }
-</ul>
+</table>
 
 <input type=submit value="Load">
 
+<pre class=debug><%= dumper $loaded %></pre>
 </form>