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} = $filtered_items;
310 warn "generated ", scalar keys %$filtered_items, " filtered items\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 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
374 return $loaded->{$path}->{sorted}->{$order};
377 my $data = $self->_loaded( 'data' );
378 my $numeric = $self->_is_numeric($order);
379 my $missing = $numeric ? 0 : '';
380 no warnings qw(numeric);
391 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
392 } @{ $data->{items} }
395 warn "sorted $order"; # ,dump( @sorted );
397 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
404 my $path = $self->session('path');
405 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
407 my @columns = $self->_perm_array('columns');
408 $self->redirect_to('/data/columns') unless @columns;
409 my $order = $self->_perm_scalar('order', $columns[0]);
410 my $sort = $self->_perm_scalar('sort', 'a');
411 my $offset = $self->_perm_scalar('offset', 0);
412 my $limit = $self->_perm_scalar('limit', 20);
413 $self->_perm_scalar('show', 'table');
415 # fix offset when changing limit
416 $offset = int( $offset / $limit ) * $limit;
418 my $sorted = $self->_data_sorted_by( $order );
422 foreach my $name ( keys %$filters ) {
423 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
424 warn "skip filter $name not found in $path\n";
427 push @filter_names, $name;
429 warn "filter_names ",dump( @filter_names );
430 foreach my $name ( @filter_names ) {
431 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
432 $self->_filter_on_data( $name, @{ $filters->{$name} } );
436 my $filtered_names = join(' ',sort @filter_names);
438 # warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
440 if ( ! defined $loaded->{$path}->{filtered}->{$filtered_names} ) {
442 my $path_filters = $loaded->{$path}->{filters};
444 warn "create combined filter for $filtered_names\n";
447 foreach my $i ( 0 .. $#$sorted ) {
448 my $pos = $sorted->[$i];
450 if ( $#filter_names == -1 ) {
451 push @filtered, $pos;
456 foreach ( @filter_names ) {
457 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
461 push @filtered, $pos;
464 $loaded->{$path}->{filtered}->{$filtered_names} = [ @filtered ];
467 my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
468 if defined $loaded->{$path}->{filtered}->{$filtered_names};
470 warn "filters $filtered_names" if $filtered;
473 my $data = $self->_loaded('data');
474 my $sort_start = $sort eq 'd' ? $#$filtered : 0;
475 foreach ( 0 .. $limit ) {
476 my $i = $sort_start - ( $_ + $offset );
477 last unless defined $filtered->[$i];
478 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
481 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit";
487 sorted => $sorted_items,
488 columns => [ @columns ],
489 rows => $#$filtered + 1,
490 numeric => { map { $_, $self->_is_numeric($_) } @columns },
491 filters => $self->_current_filters,
499 $self->session('order', $self->param('order'));
500 $self->session('sort', $self->param('sort'));
501 $self->redirect_to('/data/items');
505 my ( $self, $name ) = @_;
507 my $stats = $self->_loaded( 'stats' );
509 # sort facet numerically if more >50% elements are numeric
510 my $count = $stats->{$name}->{count};
511 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
512 defined $stats->{$name}->{numeric} &&
513 $stats->{$name}->{numeric} > $count / 2;
517 my ($self,$name) = @_;
518 warn "_remove_filter $name\n";
520 my $path = $self->session('path');
522 delete $filters->{$name};
523 delete $loaded->{$path}->{filters}->{$name};
524 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
528 keys %{ $loaded->{$path}->{filtered} }
530 delete $loaded->{$path}->{filtered}->{$_};
531 warn "remove filtered cache $_";
538 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
540 if ( my $name = $self->param('remove') ) {
541 $self->_remove_filter( $name );
542 $self->redirect_to( '/data/items' );
546 my $name = $self->param('name') || die "no name";
548 my $all = $self->_perm_scalar('all', 1);
549 my $data = $self->_loaded('data');
551 my $filters = $self->_current_filters;
552 my $filtered_names = join(' ',sort keys %$filters);
553 # warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
554 my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
555 if defined $loaded->{$path}->{filtered}->{$filtered_names};
557 if ( ! $filtered || $all ) {
558 $filtered = [ 0 .. $#{ $data->{items} } ];
559 warn "filter all values\n";
561 warn "filter using $filtered_names\n";
564 foreach my $i ( @$filtered ) {
565 my $item = $data->{items}->[$i];
566 if ( ! exists $item->{$name} ) {
567 $facet->{ _missing }++;
568 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
569 $facet->{$_}++ foreach @{ $item->{$name} };
571 $facet->{ $item->{$name} }++;
575 # warn "# facet $name ",dump $facet;
580 : defined $filters->{$name} ? @{ $filters->{$name} }
583 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
585 my $sort = $self->param('sort') || 'c';
587 # sort facet numerically if more >50% elements are numeric
588 my $numeric = $self->_is_numeric($name);
590 @facet_names = sort {
591 if ( $sort =~ m/a/i ) {
592 $numeric ? $a <=> $b : lc $a cmp lc $b;
593 } elsif ( $sort =~ m/d/i ) {
594 $numeric ? $b <=> $a : lc $b cmp lc $a;
595 } elsif ( $sort =~ m/c/i ) {
596 ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
598 warn "unknown sort: $sort";
603 $self->render( name => $name, facet => $facet, checked => $checked,
604 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
610 my $content = $self->param('content');