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 $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
99 push @{ $data->{items} }, $item;
102 warn "file format unknown $path";
107 foreach my $e ( @{ $data->{items} } ) {
108 foreach my $n ( keys %$e ) {
109 $stats->{$n}->{count}++;
111 if ( ref $e->{$n} eq 'ARRAY' ) {
112 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
118 foreach my $x ( @v ) {
119 $stats->{$n}->{numeric}++
120 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
121 $stats->{$n}->{empty}++
122 if length $x == 0; # faster than $x =~ m/^\s*$/;
128 foreach my $n ( keys %$stats ) {
129 next unless defined $stats->{$n}->{array};
130 delete $stats->{$n}->{array}
131 if $stats->{$n}->{array} == $stats->{$n}->{count};
135 if ( defined $data->{header} ) {
136 if ( ref $data->{header} eq 'ARRAY' ) {
137 @header = @{ $data->{header} };
139 warn "header not array ", dump( $data->{header} );
145 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
146 grep { defined $stats->{$_}->{count} } keys %$stats
151 $loaded->{ $path } = {
152 header => [ @header ],
154 full_path => $full_path,
155 size => -s $full_path,
156 mtime => (stat($full_path))[9],
166 my @paths = $self->param('paths');
167 warn "# paths ", dump @paths;
168 $self->_load_path( $_ ) foreach @paths;
170 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
171 warn "# path $path\n";
172 $self->session('path' => $path);
173 $self->_load_path( $path );
175 $self->session( 'header' => $loaded->{$path}->{header} );
176 if ( ! defined $loaded->{$path}->{columns} ) {
177 $self->session( 'columns' => $loaded->{$path}->{header} );
178 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
179 $self->redirect_to( '/data/columns' );
181 $self->session( 'columns' => $loaded->{$path}->{columns} );
182 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
183 $self->redirect_to( '/data/items' );
189 my ( $self, $name ) = @_;
190 my $path = $self->session('path');
191 die "$path $name doesn't exist in loaded ",dump( $loaded )
192 unless defined $loaded->{$path}->{$name};
193 return $loaded->{$path}->{$name};
200 $checked->{$_}++ foreach @_;
201 # warn "# _checked ",dump($checked);
209 if ( $self->param('columns') ) {
210 $self->_perm_array('columns');
211 $self->redirect_to('/data/items');
214 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
217 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
219 foreach my $c ( @{ $self->session( 'header' ) } ) {
220 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
224 message => 'Select columns to display',
226 columns => \@columns,
227 checked => $self->_checked( $self->_perm_array('columns') ),
232 my ($self,$name) = @_;
234 my @array = $self->param($name);
235 my $path = $self->session('path');
238 $self->session($name => [ @array ]);
239 $loaded->{$path}->{$name} = [ @array ];
240 } elsif ( my $session = $self->session($name) ) {
241 if ( ref $session eq 'ARRAY' ) {
244 die "$name not array ",dump($session);
247 #warn "# $name ",dump @array;
252 my ($self,$name,$default) = @_;
254 my $scalar = $self->param($name);
256 if ( defined $scalar ) {
257 $self->session($name => $scalar);
259 $scalar = $self->session($name);
262 if ( ! defined $scalar ) {
264 die "no default for $name" unless defined $scalar;
265 $self->session($name => $scalar);
268 warn "# _perm_scalar $name ",dump $scalar;
275 my $name = $self->param('filter_name') || die "name?";
276 my @vals = $self->param('filter_vals');
278 $self->_remove_filter( $name );
279 $self->_filter_on_data( $name, @vals ) if @vals;
281 $self->session( 'offset' => 0 );
283 $self->redirect_to('/data/items');
286 sub _filter_on_data {
287 my ( $self, $name, @vals ) = @_;
289 my $path = $self->session('path');
291 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
292 warn "filter $name not found in data set";
296 $filters->{$name} = [ @vals ];
297 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
300 $filter_hash->{$_}++ foreach @vals;
302 #warn "# filter_hash ",dump( $filter_hash );
304 my $items = $self->_loaded('data')->{items};
306 my $include_missing = defined $filter_hash->{_missing};
309 foreach my $i ( 0 .. $#$items ) {
311 if ( defined $items->[$i]->{$name} ) {
312 foreach my $v ( @{ $items->[$i]->{$name} } ) {
313 if ( defined $filter_hash->{ $v } ) {
314 $filtered_items->{$i}++;
317 } elsif ( $include_missing ) {
318 $filtered_items->{$i}++;
322 #warn "# filter $name ",dump($filtered_items);
324 $loaded->{$path}->{filters}->{$name} = $filtered_items;
325 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
329 my ( $self, $all ) = @_;
330 my $data = $self->_loaded( 'data' );
332 return @{ $data->{items} } if $all == 1;
334 my $filters = $self->_current_filters;
336 foreach my $f ( keys %$filters ) {
337 foreach my $n ( @{ $filters->{$f} } ) {
338 $filter_value->{$f}->{$n} = 1;
341 my @items = @{ $data->{items} };
345 foreach my $n ( keys %$filter_value ) {
346 if ( ! exists $i->{$n} ) {
347 if ( defined $filter_value->{$n}->{_missing} ) {
355 # and match any of values in element
357 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
358 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
360 if ( ! $have_values ) {
366 } @items if $filter_value;
371 sub _current_filters {
374 $current_filters->{ $_ } = $filters->{ $_ }
376 grep { defined $filters->{ $_ } }
377 @{ $self->_loaded('header') }
379 #warn "# current_filters ",dump($current_filters);
380 return $current_filters;
383 sub _data_sorted_by {
384 my ( $self, $order ) = @_;
386 my $path = $self->session('path');
388 warn "_data_sorted_by $order from $path";
390 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
391 return $loaded->{$path}->{sorted}->{$order};
394 my $data = $self->_loaded( 'data' );
395 my $numeric = $self->_is_numeric($order);
396 my $missing = $numeric ? 0 : '';
397 no warnings qw(numeric);
408 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
409 } @{ $data->{items} }
412 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
413 #warn "# sorted ",dump( @sorted );
415 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
422 my $path = $self->session('path');
423 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
425 my @columns = $self->_perm_array('columns');
426 $self->redirect_to('/data/columns') unless @columns;
427 my $order = $self->_perm_scalar('order', $columns[0]);
428 my $sort = $self->_perm_scalar('sort', 'a');
429 my $offset = $self->_perm_scalar('offset', 0);
430 my $limit = $self->_perm_scalar('limit', 20);
431 $self->_perm_scalar('show', 'table');
433 # fix offset when changing limit
434 $offset = int( $offset / $limit ) * $limit;
436 my $sorted = $self->_data_sorted_by( $order );
440 foreach my $name ( keys %$filters ) {
441 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
442 warn "skip filter $name not found in $path\n";
445 push @filter_names, $name;
447 warn "filter_names ",dump( @filter_names );
448 foreach my $name ( @filter_names ) {
449 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
450 $self->_filter_on_data( $name, @{ $filters->{$name} } );
454 my $all_filters = join(' ',sort @filter_names,'order:',$order);
456 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
458 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
460 my $path_filters = $loaded->{$path}->{filters};
462 warn "create combined filter for $all_filters\n";
465 foreach my $i ( 0 .. $#$sorted ) {
466 my $pos = $sorted->[$i];
468 if ( $#filter_names == -1 ) {
469 push @filtered, $pos;
474 foreach ( @filter_names ) {
475 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
479 push @filtered, $pos;
482 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
485 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
486 if defined $loaded->{$path}->{filtered}->{$all_filters};
488 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
491 my $data = $self->_loaded('data');
492 my $from_end = $sort eq 'd' ? $#$filtered : 0;
493 foreach ( 0 .. $limit ) {
494 my $i = $_ + $offset;
495 last unless defined $filtered->[$i];
496 $i = $from_end - $i if $from_end;
497 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
500 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
506 sorted => $sorted_items,
507 columns => [ @columns ],
508 rows => $#$filtered + 1,
509 numeric => { map { $_, $self->_is_numeric($_) } @columns },
510 filters => $self->_current_filters,
518 $self->session('order', $self->param('order'));
519 $self->session('sort', $self->param('sort'));
520 $self->redirect_to('/data/items');
524 my ( $self, $name ) = @_;
526 my $stats = $self->_loaded( 'stats' );
528 # sort facet numerically if more >50% elements are numeric
529 my $count = $stats->{$name}->{count};
530 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
531 defined $stats->{$name}->{numeric} &&
532 $stats->{$name}->{numeric} > $count / 2;
536 my ($self,$name) = @_;
537 warn "_remove_filter $name\n";
539 my $path = $self->session('path');
541 delete $filters->{$name};
542 delete $loaded->{$path}->{filters}->{$name};
543 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
547 keys %{ $loaded->{$path}->{filtered} }
549 delete $loaded->{$path}->{filtered}->{$_};
550 warn "remove filtered cache $_";
557 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
559 if ( my $name = $self->param('remove') ) {
560 $self->_remove_filter( $name );
561 $self->redirect_to( '/data/items' );
565 my $name = $self->param('name') || die "no name";
567 my $all = $self->_perm_scalar('all', 1);
568 my $data = $self->_loaded('data');
570 my $filters = $self->_current_filters;
571 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
572 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
573 if defined $loaded->{$path}->{filtered}->{$all_filters};
575 if ( ! $filtered || $all ) {
576 $filtered = [ 0 .. $#{ $data->{items} } ];
577 warn "filter all values\n";
579 warn "filter using $all_filters\n";
582 foreach my $i ( @$filtered ) {
583 my $item = $data->{items}->[$i];
584 if ( ! exists $item->{$name} ) {
585 $facet->{ _missing }++;
586 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
587 $facet->{$_}++ foreach @{ $item->{$name} };
589 $facet->{ $item->{$name} }++;
593 # warn "# facet $name ",dump $facet;
598 : defined $filters->{$name} ? @{ $filters->{$name} }
601 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
603 my $numeric = $self->_is_numeric($name);
605 my $sort = $self->param('sort');
606 # sort numeric facets with more than 5 values ascending
607 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
609 @facet_names = sort {
611 if ( $sort eq 'a' ) {
612 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
613 } elsif ( $sort eq 'd' ) {
614 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
615 } elsif ( $sort eq 'c' ) {
616 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
618 warn "unknown sort: $sort";
621 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
625 $self->render( name => $name, facet => $facet, checked => $checked,
626 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
632 my $content = $self->param('content');