1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
21 my $path = $self->app->home->rel_dir('data');
22 die "no data dir $path" unless -d $path;
26 my $file = $File::Find::name;
27 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
28 $file =~ s/$path\/*//;
31 warn "IGNORE: $file\n";
35 @files = sort { lc $a cmp lc $b } @files;
37 $size->{$_} = -s "$path/$_" foreach @files;
48 my ( $self, $path ) = @_;
50 return if defined $loaded->{$path}->{data};
52 my $full_path = $self->app->home->rel_file( 'data/' . $path );
53 die "$full_path $!" unless -r $full_path;
55 # we could use Mojo::JSON here, but it's too slow
56 # $data = from_json read_file $path;
57 my $data = read_file $full_path;
58 warn "# data snippet: ", substr($data,0,200);
60 if ( $path =~ m/\.js/ ) {
61 Encode::_utf8_on($data);
62 $data = from_json $data;
63 } elsif ( $path =~ m/\.txt/ ) {
64 my @lines = split(/\r?\n/, $data);
65 $data = { items => [] };
67 my $header_line = shift @lines;
68 my $multiline = $header_line =~ s/\^//g;
69 @header = split(/\|/, $header_line );
70 warn "# header ", dump( @header );
71 while ( my $line = shift @lines ) {
74 my @v = split(/\|/, $line);
75 while ( @lines && $#v < $#header ) {
79 my @more_v = split(/\|/, $line);
80 if ( $#v + $#more_v > $#header ) {
81 warn "short line: ",dump( @v );
85 $v[ $#v ] .= shift @more_v if @more_v;
86 push @v, @more_v if @more_v;
88 if ( $#v > $#header ) {
89 die "# splice $#header ", dump( @v );
90 @v = splice @v, 0, $#header;
94 $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
95 push @{ $data->{items} }, $item;
98 warn "file format unknown $path";
103 foreach my $e ( @{ $data->{items} } ) {
104 foreach my $n ( keys %$e ) {
105 $stats->{$n}->{count}++;
107 if ( ref $e->{$n} eq 'ARRAY' ) {
108 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
114 foreach my $x ( @v ) {
115 $stats->{$n}->{numeric}++
116 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
117 $stats->{$n}->{empty}++
118 if length $x == 0; # faster than $x =~ m/^\s*$/;
124 foreach my $n ( keys %$stats ) {
125 next unless defined $stats->{$n}->{array};
126 delete $stats->{$n}->{array}
127 if $stats->{$n}->{array} == $stats->{$n}->{count};
131 if ( defined $data->{header} ) {
132 if ( ref $data->{header} eq 'ARRAY' ) {
133 @header = @{ $data->{header} };
135 warn "header not array ", dump( $data->{header} );
141 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
142 grep { defined $stats->{$_}->{count} } keys %$stats
147 $loaded->{ $path } = {
148 header => [ @header ],
150 full_path => $full_path,
151 size => -s $full_path,
161 my @paths = $self->param('paths');
162 warn "# paths ", dump @paths;
163 $self->_load_path( $_ ) foreach @paths;
165 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
166 warn "# path $path\n";
167 $self->session('path' => $path);
168 $self->_load_path( $path );
170 $self->session( 'header' => $loaded->{$path}->{header} );
171 if ( ! defined $loaded->{$path}->{columns} ) {
172 $self->session( 'columns' => $loaded->{$path}->{header} );
173 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
174 $self->redirect_to( '/data/columns' );
176 $self->session( 'columns' => $loaded->{$path}->{columns} );
177 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
178 $self->redirect_to( '/data/items' );
184 my ( $self, $name ) = @_;
185 my $path = $self->session('path');
186 die "$path $name doesn't exist in loaded ",dump( $loaded )
187 unless defined $loaded->{$path}->{$name};
188 return $loaded->{$path}->{$name};
195 $checked->{$_}++ foreach @_;
196 # warn "# _checked ",dump($checked);
204 if ( $self->param('columns') ) {
205 $self->_perm_array('columns');
206 $self->redirect_to('/data/items');
209 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
212 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
214 foreach my $c ( @{ $self->session( 'header' ) } ) {
215 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
219 message => 'Select columns to display',
221 columns => \@columns,
222 checked => $self->_checked( $self->_perm_array('columns') ),
227 my ($self,$name) = @_;
229 my @array = $self->param($name);
230 my $path = $self->session('path');
233 $self->session($name => [ @array ]);
234 $loaded->{$path}->{$name} = [ @array ];
235 } elsif ( my $session = $self->session($name) ) {
236 if ( ref $session eq 'ARRAY' ) {
239 die "$name not array ",dump($session);
242 #warn "# $name ",dump @array;
247 my ($self,$name,$default) = @_;
249 my $scalar = $self->param($name);
251 if ( defined $scalar ) {
252 $self->session($name => $scalar);
254 $scalar = $self->session($name);
257 if ( ! defined $scalar ) {
259 die "no default for $name" unless defined $scalar;
260 $self->session($name => $scalar);
263 warn "# _perm_scalar $name ",dump $scalar;
270 my $name = $self->param('filter_name') || die "name?";
271 my @vals = $self->param('filter_vals');
273 $self->_remove_filter( $name );
274 $self->_filter_on_data( $name, @vals ) if @vals;
276 $self->session( 'offset' => 0 );
278 $self->redirect_to('/data/items');
281 sub _filter_on_data {
282 my ( $self, $name, @vals ) = @_;
284 my $path = $self->session('path');
286 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
287 warn "filter $name not found in data set";
291 $filters->{$name} = [ @vals ];
292 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
295 $filter_hash->{$_}++ foreach @vals;
297 #warn "# filter_hash ",dump( $filter_hash );
299 my $items = $self->_loaded('data')->{items};
301 my $include_missing = defined $filter_hash->{_missing};
304 foreach my $i ( 0 .. $#$items ) {
306 if ( defined $items->[$i]->{$name} ) {
307 foreach my $v ( @{ $items->[$i]->{$name} } ) {
308 if ( defined $filter_hash->{ $v } ) {
309 $filtered_items->{$i}++;
312 } elsif ( $include_missing ) {
313 $filtered_items->{$i}++;
317 #warn "# filter $name ",dump($filtered_items);
319 $loaded->{$path}->{filters}->{$name} = $filtered_items;
320 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
324 my ( $self, $all ) = @_;
325 my $data = $self->_loaded( 'data' );
327 return @{ $data->{items} } if $all == 1;
329 my $filters = $self->_current_filters;
331 foreach my $f ( keys %$filters ) {
332 foreach my $n ( @{ $filters->{$f} } ) {
333 $filter_value->{$f}->{$n} = 1;
336 my @items = @{ $data->{items} };
340 foreach my $n ( keys %$filter_value ) {
341 if ( ! exists $i->{$n} ) {
342 if ( defined $filter_value->{$n}->{_missing} ) {
350 # and match any of values in element
352 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
353 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
355 if ( ! $have_values ) {
361 } @items if $filter_value;
366 sub _current_filters {
369 $current_filters->{ $_ } = $filters->{ $_ }
371 grep { defined $filters->{ $_ } }
372 @{ $self->_loaded('header') }
374 #warn "# current_filters ",dump($current_filters);
375 return $current_filters;
378 sub _data_sorted_by {
379 my ( $self, $order ) = @_;
381 my $path = $self->session('path');
383 warn "_data_sorted_by $order from $path";
385 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
386 return $loaded->{$path}->{sorted}->{$order};
389 my $data = $self->_loaded( 'data' );
390 my $numeric = $self->_is_numeric($order);
391 my $missing = $numeric ? 0 : '';
392 no warnings qw(numeric);
403 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
404 } @{ $data->{items} }
407 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
408 #warn "# sorted ",dump( @sorted );
410 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
417 my $path = $self->session('path');
418 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
420 my @columns = $self->_perm_array('columns');
421 $self->redirect_to('/data/columns') unless @columns;
422 my $order = $self->_perm_scalar('order', $columns[0]);
423 my $sort = $self->_perm_scalar('sort', 'a');
424 my $offset = $self->_perm_scalar('offset', 0);
425 my $limit = $self->_perm_scalar('limit', 20);
426 $self->_perm_scalar('show', 'table');
428 # fix offset when changing limit
429 $offset = int( $offset / $limit ) * $limit;
431 my $sorted = $self->_data_sorted_by( $order );
435 foreach my $name ( keys %$filters ) {
436 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
437 warn "skip filter $name not found in $path\n";
440 push @filter_names, $name;
442 warn "filter_names ",dump( @filter_names );
443 foreach my $name ( @filter_names ) {
444 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
445 $self->_filter_on_data( $name, @{ $filters->{$name} } );
449 my $all_filters = join(' ',sort @filter_names,'order:',$order);
451 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
453 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
455 my $path_filters = $loaded->{$path}->{filters};
457 warn "create combined filter for $all_filters\n";
460 foreach my $i ( 0 .. $#$sorted ) {
461 my $pos = $sorted->[$i];
463 if ( $#filter_names == -1 ) {
464 push @filtered, $pos;
469 foreach ( @filter_names ) {
470 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
474 push @filtered, $pos;
477 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
480 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
481 if defined $loaded->{$path}->{filtered}->{$all_filters};
483 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
486 my $data = $self->_loaded('data');
487 my $from_end = $sort eq 'd' ? $#$filtered : 0;
488 foreach ( 0 .. $limit ) {
489 my $i = $_ + $offset;
490 last unless defined $filtered->[$i];
491 $i = $from_end - $i if $from_end;
492 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
495 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
501 sorted => $sorted_items,
502 columns => [ @columns ],
503 rows => $#$filtered + 1,
504 numeric => { map { $_, $self->_is_numeric($_) } @columns },
505 filters => $self->_current_filters,
513 $self->session('order', $self->param('order'));
514 $self->session('sort', $self->param('sort'));
515 $self->redirect_to('/data/items');
519 my ( $self, $name ) = @_;
521 my $stats = $self->_loaded( 'stats' );
523 # sort facet numerically if more >50% elements are numeric
524 my $count = $stats->{$name}->{count};
525 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
526 defined $stats->{$name}->{numeric} &&
527 $stats->{$name}->{numeric} > $count / 2;
531 my ($self,$name) = @_;
532 warn "_remove_filter $name\n";
534 my $path = $self->session('path');
536 delete $filters->{$name};
537 delete $loaded->{$path}->{filters}->{$name};
538 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
542 keys %{ $loaded->{$path}->{filtered} }
544 delete $loaded->{$path}->{filtered}->{$_};
545 warn "remove filtered cache $_";
552 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
554 if ( my $name = $self->param('remove') ) {
555 $self->_remove_filter( $name );
556 $self->redirect_to( '/data/items' );
560 my $name = $self->param('name') || die "no name";
562 my $all = $self->_perm_scalar('all', 1);
563 my $data = $self->_loaded('data');
565 my $filters = $self->_current_filters;
566 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
567 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
568 if defined $loaded->{$path}->{filtered}->{$all_filters};
570 if ( ! $filtered || $all ) {
571 $filtered = [ 0 .. $#{ $data->{items} } ];
572 warn "filter all values\n";
574 warn "filter using $all_filters\n";
577 foreach my $i ( @$filtered ) {
578 my $item = $data->{items}->[$i];
579 if ( ! exists $item->{$name} ) {
580 $facet->{ _missing }++;
581 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
582 $facet->{$_}++ foreach @{ $item->{$name} };
584 $facet->{ $item->{$name} }++;
588 # warn "# facet $name ",dump $facet;
593 : defined $filters->{$name} ? @{ $filters->{$name} }
596 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
598 my $numeric = $self->_is_numeric($name);
600 my $sort = $self->param('sort');
601 $sort ||= $numeric ? 'a' : 'c';
603 @facet_names = sort {
604 if ( $sort =~ m/a/i ) {
605 $numeric ? $a <=> $b : lc $a cmp lc $b;
606 } elsif ( $sort =~ m/d/i ) {
607 $numeric ? $b <=> $a : lc $b cmp lc $a;
608 } elsif ( $sort =~ m/c/i ) {
609 ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
611 warn "unknown sort: $sort";
616 $self->render( name => $name, facet => $facet, checked => $checked,
617 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
623 my $content = $self->param('content');