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 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
132 grep { defined $stats->{$_}->{count} } keys %$stats
137 $loaded->{ $path } = {
138 header => [ @header ],
140 full_path => $full_path,
141 size => -s $full_path,
151 my @paths = $self->param('paths');
152 warn "# paths ", dump @paths;
153 $self->_load_path( $_ ) foreach @paths;
155 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
156 warn "# path $path\n";
157 $self->session('path' => $path);
158 $self->_load_path( $path );
160 $self->session( 'header' => $loaded->{$path}->{header} );
161 if ( ! defined $loaded->{$path}->{columns} ) {
162 $self->session( 'columns' => $loaded->{$path}->{header} );
163 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
164 $self->redirect_to( '/data/columns' );
166 $self->session( 'columns' => $loaded->{$path}->{columns} );
167 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
168 $self->redirect_to( '/data/items' );
174 my ( $self, $name ) = @_;
175 my $path = $self->session('path');
176 die "$path $name doesn't exist in loaded ",dump( $loaded )
177 unless defined $loaded->{$path}->{$name};
178 return $loaded->{$path}->{$name};
185 $checked->{$_}++ foreach @_;
186 # warn "# _checked ",dump($checked);
194 if ( $self->param('columns') ) {
195 $self->_perm_array('columns');
196 $self->redirect_to('/data/items');
199 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
202 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
204 foreach my $c ( @{ $self->session( 'header' ) } ) {
205 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
209 message => 'Select columns to display',
211 columns => \@columns,
212 checked => $self->_checked( $self->_perm_array('columns') ),
217 my ($self,$name) = @_;
219 my @array = $self->param($name);
220 my $path = $self->session('path');
223 $self->session($name => [ @array ]);
224 $loaded->{$path}->{$name} = [ @array ];
225 } elsif ( my $session = $self->session($name) ) {
226 if ( ref $session eq 'ARRAY' ) {
229 die "$name not array ",dump($session);
232 #warn "# $name ",dump @array;
237 my ($self,$name,$default) = @_;
239 my $scalar = $self->param($name);
241 if ( defined $scalar ) {
242 $self->session($name => $scalar);
244 $scalar = $self->session($name);
247 if ( ! defined $scalar ) {
249 die "no default for $name" unless defined $scalar;
250 $self->session($name => $scalar);
253 warn "# _perm_scalar $name ",dump $scalar;
260 my $name = $self->param('filter_name') || die "name?";
261 my @vals = $self->param('filter_vals');
263 $self->_remove_filter( $name );
264 $self->_filter_on_data( $name, @vals ) if @vals;
266 $self->session( 'offset' => 0 );
268 $self->redirect_to('/data/items');
271 sub _filter_on_data {
272 my ( $self, $name, @vals ) = @_;
274 my $path = $self->session('path');
276 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
277 warn "filter $name not found in data set";
281 $filters->{$name} = [ @vals ];
282 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
285 $filter_hash->{$_}++ foreach @vals;
287 #warn "# filter_hash ",dump( $filter_hash );
289 my $items = $self->_loaded('data')->{items};
291 my $include_missing = defined $filter_hash->{_missing};
294 foreach my $i ( 0 .. $#$items ) {
296 if ( defined $items->[$i]->{$name} ) {
297 foreach my $v ( @{ $items->[$i]->{$name} } ) {
298 if ( defined $filter_hash->{ $v } ) {
299 $filtered_items->{$i}++;
302 } elsif ( $include_missing ) {
303 $filtered_items->{$i}++;
307 #warn "# filter $name ",dump($filtered_items);
309 $loaded->{$path}->{filters}->{$name} = $filtered_items;
310 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
314 my ( $self, $all ) = @_;
315 my $data = $self->_loaded( 'data' );
317 return @{ $data->{items} } if $all == 1;
319 my $filters = $self->_current_filters;
321 foreach my $f ( keys %$filters ) {
322 foreach my $n ( @{ $filters->{$f} } ) {
323 $filter_value->{$f}->{$n} = 1;
326 my @items = @{ $data->{items} };
330 foreach my $n ( keys %$filter_value ) {
331 if ( ! exists $i->{$n} ) {
332 if ( defined $filter_value->{$n}->{_missing} ) {
340 # and match any of values in element
342 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
343 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
345 if ( ! $have_values ) {
351 } @items if $filter_value;
356 sub _current_filters {
359 $current_filters->{ $_ } = $filters->{ $_ }
361 grep { defined $filters->{ $_ } }
362 @{ $self->_loaded('header') }
364 #warn "# current_filters ",dump($current_filters);
365 return $current_filters;
368 sub _data_sorted_by {
369 my ( $self, $order ) = @_;
371 my $path = $self->session('path');
373 warn "_data_sorted_by $order from $path";
375 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
376 return $loaded->{$path}->{sorted}->{$order};
379 my $data = $self->_loaded( 'data' );
380 my $numeric = $self->_is_numeric($order);
381 my $missing = $numeric ? 0 : '';
382 no warnings qw(numeric);
393 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
394 } @{ $data->{items} }
397 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
398 warn "# sorted ",dump( @sorted );
400 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
407 my $path = $self->session('path');
408 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
410 my @columns = $self->_perm_array('columns');
411 $self->redirect_to('/data/columns') unless @columns;
412 my $order = $self->_perm_scalar('order', $columns[0]);
413 my $sort = $self->_perm_scalar('sort', 'a');
414 my $offset = $self->_perm_scalar('offset', 0);
415 my $limit = $self->_perm_scalar('limit', 20);
416 $self->_perm_scalar('show', 'table');
418 # fix offset when changing limit
419 $offset = int( $offset / $limit ) * $limit;
421 my $sorted = $self->_data_sorted_by( $order );
425 foreach my $name ( keys %$filters ) {
426 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
427 warn "skip filter $name not found in $path\n";
430 push @filter_names, $name;
432 warn "filter_names ",dump( @filter_names );
433 foreach my $name ( @filter_names ) {
434 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
435 $self->_filter_on_data( $name, @{ $filters->{$name} } );
439 my $all_filters = join(' ',sort @filter_names);
441 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
443 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
445 my $path_filters = $loaded->{$path}->{filters};
447 warn "create combined filter for $all_filters\n";
450 foreach my $i ( 0 .. $#$sorted ) {
451 my $pos = $sorted->[$i];
453 if ( $#filter_names == -1 ) {
454 push @filtered, $pos;
459 foreach ( @filter_names ) {
460 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
464 push @filtered, $pos;
467 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
470 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
471 if defined $loaded->{$path}->{filtered}->{$all_filters};
473 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
476 my $data = $self->_loaded('data');
477 my $from_end = $sort eq 'd' ? $#$filtered : 0;
478 foreach ( 0 .. $limit ) {
479 my $i = $_ + $offset;
480 last unless defined $filtered->[$i];
481 $i = $from_end - $i if $from_end;
482 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
485 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
491 sorted => $sorted_items,
492 columns => [ @columns ],
493 rows => $#$filtered + 1,
494 numeric => { map { $_, $self->_is_numeric($_) } @columns },
495 filters => $self->_current_filters,
503 $self->session('order', $self->param('order'));
504 $self->session('sort', $self->param('sort'));
505 $self->redirect_to('/data/items');
509 my ( $self, $name ) = @_;
511 my $stats = $self->_loaded( 'stats' );
513 # sort facet numerically if more >50% elements are numeric
514 my $count = $stats->{$name}->{count};
515 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
516 defined $stats->{$name}->{numeric} &&
517 $stats->{$name}->{numeric} > $count / 2;
521 my ($self,$name) = @_;
522 warn "_remove_filter $name\n";
524 my $path = $self->session('path');
526 delete $filters->{$name};
527 delete $loaded->{$path}->{filters}->{$name};
528 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
532 keys %{ $loaded->{$path}->{filtered} }
534 delete $loaded->{$path}->{filtered}->{$_};
535 warn "remove filtered cache $_";
542 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
544 if ( my $name = $self->param('remove') ) {
545 $self->_remove_filter( $name );
546 $self->redirect_to( '/data/items' );
550 my $name = $self->param('name') || die "no name";
552 my $all = $self->_perm_scalar('all', 1);
553 my $data = $self->_loaded('data');
555 my $filters = $self->_current_filters;
556 my $all_filters = join(' ',sort keys %$filters);
557 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
558 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
559 if defined $loaded->{$path}->{filtered}->{$all_filters};
561 if ( ! $filtered || $all ) {
562 $filtered = [ 0 .. $#{ $data->{items} } ];
563 warn "filter all values\n";
565 warn "filter using $all_filters\n";
568 foreach my $i ( @$filtered ) {
569 my $item = $data->{items}->[$i];
570 if ( ! exists $item->{$name} ) {
571 $facet->{ _missing }++;
572 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
573 $facet->{$_}++ foreach @{ $item->{$name} };
575 $facet->{ $item->{$name} }++;
579 # warn "# facet $name ",dump $facet;
584 : defined $filters->{$name} ? @{ $filters->{$name} }
587 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
589 my $sort = $self->param('sort') || 'c';
591 # sort facet numerically if more >50% elements are numeric
592 my $numeric = $self->_is_numeric($name);
594 @facet_names = sort {
595 if ( $sort =~ m/a/i ) {
596 $numeric ? $a <=> $b : lc $a cmp lc $b;
597 } elsif ( $sort =~ m/d/i ) {
598 $numeric ? $b <=> $a : lc $b cmp lc $a;
599 } elsif ( $sort =~ m/c/i ) {
600 ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
602 warn "unknown sort: $sort";
607 $self->render( name => $name, facet => $facet, checked => $checked,
608 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
614 my $content = $self->param('content');