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}++
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->redirect_to( '/data/columns' );
165 $self->session( 'columns' => $loaded->{$path}->{columns} );
166 $self->redirect_to( '/data/items' );
173 my ( $self, $name ) = @_;
174 my $path = $self->session('path');
175 die "$path $name doesn't exist in loaded ",dump( $loaded )
176 unless defined $loaded->{$path}->{$name};
177 return $loaded->{$path}->{$name};
184 $checked->{$_}++ foreach @_;
185 # warn "# _checked ",dump($checked);
193 if ( $self->param('columns') ) {
194 $self->_perm_array('columns');
195 $self->redirect_to('/data/items');
198 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
201 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
203 foreach my $c ( @{ $self->session( 'header' ) } ) {
204 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
208 message => 'Select columns to display',
210 columns => \@columns,
211 checked => $self->_checked( $self->_perm_array('columns') ),
216 my ($self,$name) = @_;
218 my @array = $self->param($name);
219 my $path = $self->session('path');
222 $self->session($name => [ @array ]);
223 $loaded->{$path}->{$name} = [ @array ];
224 } elsif ( my $session = $self->session($name) ) {
225 if ( ref $session eq 'ARRAY' ) {
228 die "$name not array ",dump($session);
231 #warn "# $name ",dump @array;
236 my ($self,$name,$default) = @_;
238 my $scalar = $self->param($name);
240 if ( defined $scalar ) {
241 $self->session($name => $scalar);
243 $scalar = $self->session($name);
246 if ( ! defined $scalar ) {
248 die "no default for $name" unless defined $scalar;
249 $self->session($name => $scalar);
252 warn "# _perm_scalar $name ",dump $scalar;
259 my $name = $self->param('filter_name') || die "name?";
260 my @vals = $self->param('filter_vals');
262 my $path = $self->session('path');
265 $self->_filter_on_data( $name, @vals );
267 warn "# filter - $name\n";
268 delete $filters->{$name};
269 delete $loaded->{$path}->{filters}->{$name};
272 #warn "# filters ",dump($filters);
274 $self->session( 'offset' => 0 );
276 $self->redirect_to('/data/items');
279 sub _filter_on_data {
280 my ( $self, $name, @vals ) = @_;
282 my $path = $self->session('path');
284 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
285 warn "filter $name not found in data set";
289 $filters->{$name} = [ @vals ];
290 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
293 $filter_hash->{$_}++ foreach @vals;
295 #warn "# filter_hash ",dump( $filter_hash );
297 my $items = $self->_loaded('data')->{items};
299 my $include_missing = defined $filter_hash->{_missing};
302 foreach my $i ( 0 .. $#$items ) {
304 if ( defined $items->[$i]->{$name} ) {
305 foreach my $v ( @{ $items->[$i]->{$name} } ) {
306 if ( defined $filter_hash->{ $v } ) {
307 $filtered_items->{$i}++;
310 } elsif ( $include_missing ) {
311 $filtered_items->{$i}++;
315 #warn "# filter $name ",dump($filtered_items);
317 $loaded->{$path}->{filters}->{$name} = $filtered_items;
318 warn "generated ", scalar keys %$filtered_items, " filtered items\n";
322 my ( $self, $all ) = @_;
323 my $data = $self->_loaded( 'data' );
325 return @{ $data->{items} } if $all == 1;
327 my $filters = $self->_current_filters;
329 foreach my $f ( keys %$filters ) {
330 foreach my $n ( @{ $filters->{$f} } ) {
331 $filter_value->{$f}->{$n} = 1;
334 my @items = @{ $data->{items} };
338 foreach my $n ( keys %$filter_value ) {
339 if ( ! exists $i->{$n} ) {
340 if ( defined $filter_value->{$n}->{_missing} ) {
348 # and match any of values in element
350 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
351 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
353 if ( ! $have_values ) {
359 } @items if $filter_value;
364 sub _current_filters {
367 $current_filters->{ $_ } = $filters->{ $_ }
369 grep { defined $filters->{ $_ } }
370 @{ $self->_loaded('header') }
372 #warn "# current_filters ",dump($current_filters);
373 return $current_filters;
376 sub _data_sorted_by {
377 my ( $self, $order ) = @_;
379 my $path = $self->session('path');
381 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
382 return $loaded->{$path}->{sorted}->{$order};
385 my $data = $self->_loaded( 'data' );
386 my $numeric = $self->_is_numeric($order);
387 my $missing = $numeric ? 0 : '';
388 no warnings qw(numeric);
399 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
400 } @{ $data->{items} }
403 warn "sorted $order"; # ,dump( @sorted );
405 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
412 my $path = $self->session('path');
413 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
415 my @columns = $self->_perm_array('columns');
416 $self->redirect_to('/data/columns') unless @columns;
417 my $order = $self->_perm_scalar('order', $columns[0]);
418 my $sort = $self->_perm_scalar('sort', 'a');
419 my $offset = $self->_perm_scalar('offset', 0);
420 my $limit = $self->_perm_scalar('limit', 20);
421 $self->_perm_scalar('show', 'table');
423 # fix offset when changing limit
424 $offset = int( $offset / $limit ) * $limit;
426 my $sorted = $self->_data_sorted_by( $order );
430 foreach my $name ( keys %$filters ) {
431 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
432 warn "skip filter $name not found in $path\n";
435 push @filter_names, $name;
437 warn "filter_names ",dump( @filter_names );
438 foreach my $name ( @filter_names ) {
439 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
440 $self->_filter_on_data( $name, @{ $filters->{$name} } );
444 my $path_filters = $loaded->{$path}->{filters};
447 foreach my $i ( 0 .. $#$sorted ) {
448 my $pos = $sort eq 'd' ? $sorted->[$i] : $sorted->[ $#$sorted - $i ];
451 foreach ( @filter_names ) {
452 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
456 push @filtered, $pos;
460 my $data = $self->_loaded('data');
461 foreach ( $offset .. $offset + $limit ) {
462 last unless defined $filtered[$_];
463 push @$sorted_items, $data->{items}->[ $filtered[$_] ];
467 # warn "# sorted ", dump $sorted;
473 sorted => $sorted_items,
474 columns => [ @columns ],
475 rows => $#filtered + 1,
476 numeric => { map { $_, $self->_is_numeric($_) } @columns },
477 filters => $self->_current_filters,
485 $self->session('order', $self->param('order'));
486 $self->session('sort', $self->param('sort'));
487 $self->redirect_to('/data/items');
491 my ( $self, $name ) = @_;
493 my $stats = $self->_loaded( 'stats' );
495 # sort facet numerically if more >50% elements are numeric
496 my $count = $stats->{$name}->{count};
497 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
498 defined $stats->{$name}->{numeric} &&
499 $stats->{$name}->{numeric} > $count / 2;
505 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
507 if ( my $remove = $self->param('remove') ) {
508 delete $filters->{$remove};
509 delete $loaded->{$path}->{filters}->{$remove};
510 $self->redirect_to( '/data/items' );
514 my $name = $self->param('name') || die "no name";
516 my $all = $self->_perm_scalar('all', 1);
518 foreach my $i ( $self->_data_items($all) ) {
519 if ( ! exists $i->{$name} ) {
520 $facet->{ _missing }++;
521 } elsif ( ref $i->{$name} eq 'ARRAY' ) {
522 $facet->{$_}++ foreach @{ $i->{$name} };
524 $facet->{ $i->{$name} }++;
528 # warn "# facet $name ",dump $facet;
533 : defined $filters->{$name} ? @{ $filters->{$name} }
536 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
538 my $sort = $self->param('sort') || 'c';
540 # sort facet numerically if more >50% elements are numeric
541 my $numeric = $self->_is_numeric($name);
543 @facet_names = sort {
544 if ( $sort =~ m/a/i ) {
545 $numeric ? $a <=> $b : lc $a cmp lc $b;
546 } elsif ( $sort =~ m/d/i ) {
547 $numeric ? $b <=> $a : lc $b cmp lc $a;
548 } elsif ( $sort =~ m/c/i ) {
549 $facet->{$b} <=> $facet->{$a};
551 warn "unknown sort: $sort";
556 $self->render( name => $name, facet => $facet, checked => $checked,
557 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
563 my $content = $self->param('content');