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 my $full_path = $self->app->home->rel_file( 'data/' . $path );
51 die "$full_path $!" unless -r $full_path;
53 if ( defined $loaded->{$path}->{data} ) {
54 my $mtime = (stat($full_path))[9];
55 return if $loaded->{$path}->{mtime} == $mtime;
56 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
59 # we could use Mojo::JSON here, but it's too slow
60 # $data = from_json read_file $path;
61 my $data = read_file $full_path;
62 warn "# data snippet: ", substr($data,0,200);
64 if ( $path =~ m/\.js/ ) {
65 Encode::_utf8_on($data);
66 $data = from_json $data;
67 } elsif ( $path =~ m/\.txt/ ) {
68 my @lines = split(/\r?\n/, $data);
69 $data = { items => [] };
71 my $header_line = shift @lines;
72 my $multiline = $header_line =~ s/\^//g;
73 @header = split(/\|/, $header_line );
74 warn "# header ", dump( @header );
75 while ( my $line = shift @lines ) {
78 my @v = split(/\|/, $line);
79 while ( @lines && $#v < $#header ) {
83 my @more_v = split(/\|/, $line);
84 if ( $#v + $#more_v > $#header ) {
85 warn "short line: ",dump( @v );
89 $v[ $#v ] .= shift @more_v if @more_v;
90 push @v, @more_v if @more_v;
92 if ( $#v > $#header ) {
93 die "# splice $#header ", dump( @v );
94 @v = splice @v, 0, $#header;
98 foreach my $i ( 0 .. $#v ) {
100 # unpack numeric values separated by commas
101 my $a = $v =~ m/\d+\s*,\s*\d+/ ? [ split(/\,\s*/,$v) ] : [ $v ];
102 $item->{ $header[$i] || "f_$i" } = $a;
104 push @{ $data->{items} }, $item;
107 warn "file format unknown $path";
112 foreach my $e ( @{ $data->{items} } ) {
113 foreach my $n ( keys %$e ) {
114 $stats->{$n}->{count}++;
116 if ( ref $e->{$n} eq 'ARRAY' ) {
117 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
123 foreach my $x ( @v ) {
124 $stats->{$n}->{numeric}++
125 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
126 $stats->{$n}->{empty}++
127 if length $x == 0; # faster than $x =~ m/^\s*$/;
133 foreach my $n ( keys %$stats ) {
134 next unless defined $stats->{$n}->{array};
135 delete $stats->{$n}->{array}
136 if $stats->{$n}->{array} == $stats->{$n}->{count};
140 if ( defined $data->{header} ) {
141 if ( ref $data->{header} eq 'ARRAY' ) {
142 @header = @{ $data->{header} };
144 warn "header not array ", dump( $data->{header} );
150 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
151 grep { defined $stats->{$_}->{count} } keys %$stats
156 $loaded->{ $path } = {
157 header => [ @header ],
159 full_path => $full_path,
160 size => -s $full_path,
161 mtime => (stat($full_path))[9],
171 my @paths = $self->param('paths');
172 warn "# paths ", dump @paths;
173 $self->_load_path( $_ ) foreach @paths;
175 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
176 warn "# path $path\n";
177 $self->session('path' => $path);
178 $self->_load_path( $path );
180 $self->session( 'header' => $loaded->{$path}->{header} );
181 if ( ! defined $loaded->{$path}->{columns} ) {
182 $self->session( 'columns' => $loaded->{$path}->{header} );
183 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
184 $self->redirect_to( '/data/columns' );
186 $self->session( 'columns' => $loaded->{$path}->{columns} );
187 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
188 $self->redirect_to( '/data/items' );
194 my ( $self, $name ) = @_;
195 my $path = $self->session('path');
196 die "$path $name doesn't exist in loaded ",dump( $loaded )
197 unless defined $loaded->{$path}->{$name};
198 return $loaded->{$path}->{$name};
205 $checked->{$_}++ foreach @_;
206 # warn "# _checked ",dump($checked);
214 if ( $self->param('columns') ) {
215 $self->_perm_array('columns');
216 $self->redirect_to('/data/items');
219 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
222 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
224 foreach my $c ( @{ $self->session( 'header' ) } ) {
225 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
229 message => 'Select columns to display',
231 columns => \@columns,
232 checked => $self->_checked( $self->_perm_array('columns') ),
237 my ($self,$name) = @_;
239 my @array = $self->param($name);
240 my $path = $self->session('path');
243 $self->session($name => [ @array ]);
244 $loaded->{$path}->{$name} = [ @array ];
245 } elsif ( my $session = $self->session($name) ) {
246 if ( ref $session eq 'ARRAY' ) {
249 die "$name not array ",dump($session);
252 #warn "# $name ",dump @array;
257 my ($self,$name,$default) = @_;
259 my $scalar = $self->param($name);
261 if ( defined $scalar ) {
262 $self->session($name => $scalar);
264 $scalar = $self->session($name);
267 if ( ! defined $scalar ) {
269 die "no default for $name" unless defined $scalar;
270 $self->session($name => $scalar);
273 warn "# _perm_scalar $name ",dump $scalar;
280 my $name = $self->param('filter_name') || die "name?";
281 my @vals = $self->param('filter_vals');
283 $self->_remove_filter( $name );
284 $self->_filter_on_data( $name, @vals ) if @vals;
286 $self->session( 'offset' => 0 );
288 $self->redirect_to('/data/items');
291 sub _filter_on_data {
292 my ( $self, $name, @vals ) = @_;
294 my $path = $self->session('path');
296 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
297 warn "filter $name not found in data set";
301 $filters->{$name} = [ @vals ];
302 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
305 $filter_hash->{$_}++ foreach @vals;
307 #warn "# filter_hash ",dump( $filter_hash );
309 my $items = $self->_loaded('data')->{items};
311 my $include_missing = defined $filter_hash->{_missing};
314 foreach my $i ( 0 .. $#$items ) {
316 if ( defined $items->[$i]->{$name} ) {
317 foreach my $v ( @{ $items->[$i]->{$name} } ) {
318 if ( defined $filter_hash->{ $v } ) {
319 $filtered_items->{$i}++;
322 } elsif ( $include_missing ) {
323 $filtered_items->{$i}++;
327 #warn "# filter $name ",dump($filtered_items);
329 $loaded->{$path}->{filters}->{$name} = $filtered_items;
330 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
334 my ( $self, $all ) = @_;
335 my $data = $self->_loaded( 'data' );
337 return @{ $data->{items} } if $all == 1;
339 my $filters = $self->_current_filters;
341 foreach my $f ( keys %$filters ) {
342 foreach my $n ( @{ $filters->{$f} } ) {
343 $filter_value->{$f}->{$n} = 1;
346 my @items = @{ $data->{items} };
350 foreach my $n ( keys %$filter_value ) {
351 if ( ! exists $i->{$n} ) {
352 if ( defined $filter_value->{$n}->{_missing} ) {
360 # and match any of values in element
362 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
363 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
365 if ( ! $have_values ) {
371 } @items if $filter_value;
376 sub _current_filters {
379 $current_filters->{ $_ } = $filters->{ $_ }
381 grep { defined $filters->{ $_ } }
382 @{ $self->_loaded('header') }
384 #warn "# current_filters ",dump($current_filters);
385 return $current_filters;
388 sub _data_sorted_by {
389 my ( $self, $order ) = @_;
391 my $path = $self->session('path');
393 warn "_data_sorted_by $order from $path";
395 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
396 return $loaded->{$path}->{sorted}->{$order};
399 my $data = $self->_loaded( 'data' );
400 my $numeric = $self->_is_numeric($order);
401 my $missing = $numeric ? 0 : '';
402 no warnings qw(numeric);
413 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
414 } @{ $data->{items} }
417 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
418 #warn "# sorted ",dump( @sorted );
420 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
427 my $path = $self->session('path');
428 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
430 my @columns = $self->_perm_array('columns');
431 $self->redirect_to('/data/columns') unless @columns;
432 my $order = $self->_perm_scalar('order', $columns[0]);
433 my $sort = $self->_perm_scalar('sort', 'a');
434 my $offset = $self->_perm_scalar('offset', 0);
435 my $limit = $self->_perm_scalar('limit', 20);
436 $self->_perm_scalar('show', 'table');
438 # fix offset when changing limit
439 $offset = int( $offset / $limit ) * $limit;
441 my $sorted = $self->_data_sorted_by( $order );
445 foreach my $name ( keys %$filters ) {
446 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
447 warn "skip filter $name not found in $path\n";
450 push @filter_names, $name;
452 warn "filter_names ",dump( @filter_names );
453 foreach my $name ( @filter_names ) {
454 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
455 $self->_filter_on_data( $name, @{ $filters->{$name} } );
459 my $all_filters = join(' ',sort @filter_names,'order:',$order);
461 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
463 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
465 my $path_filters = $loaded->{$path}->{filters};
467 warn "create combined filter for $all_filters\n";
470 foreach my $i ( 0 .. $#$sorted ) {
471 my $pos = $sorted->[$i];
473 if ( $#filter_names == -1 ) {
474 push @filtered, $pos;
479 foreach ( @filter_names ) {
480 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
484 push @filtered, $pos;
487 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
490 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
491 if defined $loaded->{$path}->{filtered}->{$all_filters};
493 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
496 my $data = $self->_loaded('data');
497 my $from_end = $sort eq 'd' ? $#$filtered : 0;
498 foreach ( 0 .. $limit ) {
499 my $i = $_ + $offset;
500 last unless defined $filtered->[$i];
501 $i = $from_end - $i if $from_end;
502 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
505 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
511 sorted => $sorted_items,
512 columns => [ @columns ],
513 rows => $#$filtered + 1,
514 numeric => { map { $_, $self->_is_numeric($_) } @columns },
515 filters => $self->_current_filters,
523 $self->session('order', $self->param('order'));
524 $self->session('sort', $self->param('sort'));
525 $self->redirect_to('/data/items');
529 my ( $self, $name ) = @_;
531 my $stats = $self->_loaded( 'stats' );
533 # sort facet numerically if more >50% elements are numeric
534 my $count = $stats->{$name}->{count};
535 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
536 defined $stats->{$name}->{numeric} &&
537 $stats->{$name}->{numeric} > $count / 2;
541 my ($self,$name) = @_;
542 warn "_remove_filter $name\n";
544 my $path = $self->session('path');
546 delete $filters->{$name};
547 delete $loaded->{$path}->{filters}->{$name};
548 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
552 keys %{ $loaded->{$path}->{filtered} }
554 delete $loaded->{$path}->{filtered}->{$_};
555 warn "remove filtered cache $_";
562 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
564 if ( my $name = $self->param('remove') ) {
565 $self->_remove_filter( $name );
566 $self->redirect_to( '/data/items' );
570 my $name = $self->param('name') || die "no name";
572 my $all = $self->_perm_scalar('all', 1);
573 my $data = $self->_loaded('data');
575 my $filters = $self->_current_filters;
576 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
577 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
578 if defined $loaded->{$path}->{filtered}->{$all_filters};
580 if ( ! $filtered || $all ) {
581 $filtered = [ 0 .. $#{ $data->{items} } ];
582 warn "filter all values\n";
584 warn "filter using $all_filters\n";
587 foreach my $i ( @$filtered ) {
588 my $item = $data->{items}->[$i];
589 if ( ! exists $item->{$name} ) {
590 $facet->{ _missing }++;
591 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
592 $facet->{$_}++ foreach @{ $item->{$name} };
594 $facet->{ $item->{$name} }++;
598 # warn "# facet $name ",dump $facet;
603 : defined $filters->{$name} ? @{ $filters->{$name} }
606 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
608 my $numeric = $self->_is_numeric($name);
610 my $sort = $self->param('sort');
611 # sort numeric facets with more than 5 values ascending
612 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
614 @facet_names = sort {
616 if ( $sort eq 'a' ) {
617 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
618 } elsif ( $sort eq 'd' ) {
619 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
620 } elsif ( $sort eq 'c' ) {
621 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
623 warn "unknown sort: $sort";
626 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
630 $self->render( name => $name, facet => $facet, checked => $checked,
631 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
637 my $content = $self->param('content');