1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
22 my $path = $self->app->home->rel_dir('data');
23 die "no data dir $path" unless -d $path;
27 my $file = $File::Find::name;
28 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
29 $file =~ s/$path\/*//;
32 warn "IGNORE: $file\n";
36 @files = sort { lc $a cmp lc $b } @files;
38 $size->{$_} = -s "$path/$_" foreach @files;
49 my ( $self, $name ) = @_;
50 my $dir = $self->app->home->rel_dir('data');
53 return '/tmp/mojo_facets.' . $name . '.storable';
57 my ( $self, $path ) = @_;
59 my $dump_path = $self->_dump_path( $path );
60 warn "save loaded to $dump_path";
61 my $info = $loaded->{$path};
62 store $info, $dump_path;
65 my $mtime = $loaded->{$path}->{mtime};
66 utime $mtime, $mtime, $dump_path;
68 warn $dump_path, ' ', -s $dump_path, " bytes\n";
73 my ( $self, $path ) = @_;
75 my $full_path = $self->app->home->rel_file( 'data/' . $path );
76 die "$full_path $!" unless -r $full_path;
78 my $dump_path = $self->_dump_path( $path );
80 if ( defined $loaded->{$path}->{data} ) {
81 my $mtime = (stat($full_path))[9];
82 return if $loaded->{$path}->{mtime} == $mtime;
83 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
84 } elsif ( -e $dump_path ) {
85 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
86 my $info = retrieve $dump_path;
87 $loaded->{ $path } = $info;
91 # we could use Mojo::JSON here, but it's too slow
92 # $data = from_json read_file $path;
93 my $data = read_file $full_path;
94 warn "# data snippet: ", substr($data,0,200);
96 if ( $path =~ m/\.js/ ) {
97 Encode::_utf8_on($data);
98 $data = from_json $data;
99 } elsif ( $path =~ m/\.txt/ ) {
100 my @lines = split(/\r?\n/, $data);
101 $data = { items => [] };
103 my $header_line = shift @lines;
104 my $multiline = $header_line =~ s/\^//g;
105 @header = split(/\|/, $header_line );
106 warn "# header ", dump( @header );
107 while ( my $line = shift @lines ) {
110 my @v = split(/\|/, $line);
111 while ( @lines && $#v < $#header ) {
115 my @more_v = split(/\|/, $line);
116 if ( $#v + $#more_v > $#header ) {
117 warn "short line: ",dump( @v );
121 $v[ $#v ] .= shift @more_v if @more_v;
122 push @v, @more_v if @more_v;
124 if ( $#v > $#header ) {
125 die "# splice $#header ", dump( @v );
126 @v = splice @v, 0, $#header;
130 foreach my $i ( 0 .. $#v ) {
131 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
133 push @{ $data->{items} }, $item;
136 warn "file format unknown $path";
141 foreach my $e ( @{ $data->{items} } ) {
142 foreach my $n ( keys %$e ) {
143 $stats->{$n}->{count}++;
145 if ( ref $e->{$n} eq 'ARRAY' ) {
146 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
152 foreach my $x ( @v ) {
153 $stats->{$n}->{numeric}++
154 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
155 $stats->{$n}->{empty}++
156 if length $x == 0; # faster than $x =~ m/^\s*$/;
162 foreach my $n ( keys %$stats ) {
163 next unless defined $stats->{$n}->{array};
164 delete $stats->{$n}->{array}
165 if $stats->{$n}->{array} == $stats->{$n}->{count};
169 if ( defined $data->{header} ) {
170 if ( ref $data->{header} eq 'ARRAY' ) {
171 @header = @{ $data->{header} };
173 warn "header not array ", dump( $data->{header} );
179 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
180 grep { defined $stats->{$_}->{count} } keys %$stats
186 header => [ @header ],
188 full_path => $full_path,
189 size => -s $full_path,
190 mtime => (stat($full_path))[9],
194 $loaded->{ $path } = $info;
195 $self->_save( $path );
203 my @paths = $self->param('paths');
204 warn "# paths ", dump @paths;
205 $self->_load_path( $_ ) foreach @paths;
207 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
208 warn "# path $path\n";
209 $self->session('path' => $path);
210 $self->_load_path( $path );
212 $self->session( 'header' => $loaded->{$path}->{header} );
213 if ( ! defined $loaded->{$path}->{columns} ) {
214 $self->session( 'columns' => $loaded->{$path}->{header} );
215 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
216 $self->redirect_to( '/data/columns' );
218 $self->session( 'columns' => $loaded->{$path}->{columns} );
219 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
220 $self->redirect_to( '/data/items' );
226 my ( $self, $name ) = @_;
227 my $path = $self->session('path');
228 $self->redirect_to('/data/index') unless $path;
229 if ( ! defined $loaded->{$path}->{$name} ) {
230 warn "$path $name doesn't exist in loaded ",dump( $loaded );
231 $self->redirect_to('/data/index');
233 return $loaded->{$path}->{$name};
240 $checked->{$_}++ foreach @_;
241 # warn "# _checked ",dump($checked);
249 if ( $self->param('columns') ) {
250 $self->_perm_array('columns');
251 $self->redirect_to('/data/items');
254 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
257 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
259 foreach my $c ( @{ $self->session( 'header' ) } ) {
260 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
264 message => 'Select columns to display',
266 columns => \@columns,
267 checked => $self->_checked( $self->_perm_array('columns') ),
272 my ($self,$name) = @_;
274 my @array = $self->param($name);
275 my $path = $self->session('path');
278 $self->session($name => [ @array ]);
279 $loaded->{$path}->{$name} = [ @array ];
280 } elsif ( my $session = $self->session($name) ) {
281 if ( ref $session eq 'ARRAY' ) {
284 die "$name not array ",dump($session);
287 #warn "# $name ",dump @array;
292 my ($self,$name,$default) = @_;
294 my $scalar = $self->param($name);
296 if ( defined $scalar ) {
297 $self->session($name => $scalar);
299 $scalar = $self->session($name);
302 if ( ! defined $scalar ) {
304 die "no default for $name" unless defined $scalar;
305 $self->session($name => $scalar);
308 warn "# _perm_scalar $name ",dump $scalar;
315 my $name = $self->param('filter_name') || die "name?";
316 my @vals = $self->param('filter_vals');
318 $self->_remove_filter( $name );
319 $self->_filter_on_data( $name, @vals ) if @vals;
321 $self->session( 'offset' => 0 );
323 $self->redirect_to('/data/items');
326 sub _filter_on_data {
327 my ( $self, $name, @vals ) = @_;
329 my $path = $self->session('path');
331 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
332 warn "filter $name not found in data set";
336 $filters->{$name} = [ @vals ];
337 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
340 $filter_hash->{$_}++ foreach @vals;
342 #warn "# filter_hash ",dump( $filter_hash );
344 my $items = $self->_loaded('data')->{items};
346 my $include_missing = defined $filter_hash->{_missing};
349 foreach my $i ( 0 .. $#$items ) {
351 if ( defined $items->[$i]->{$name} ) {
352 foreach my $v ( @{ $items->[$i]->{$name} } ) {
353 if ( defined $filter_hash->{ $v } ) {
354 $filtered_items->{$i}++;
357 } elsif ( $include_missing ) {
358 $filtered_items->{$i}++;
362 #warn "# filter $name ",dump($filtered_items);
364 $loaded->{$path}->{filters}->{$name} = $filtered_items;
365 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
369 my ( $self, $all ) = @_;
370 my $data = $self->_loaded( 'data' );
372 return @{ $data->{items} } if $all == 1;
374 my $filters = $self->_current_filters;
376 foreach my $f ( keys %$filters ) {
377 foreach my $n ( @{ $filters->{$f} } ) {
378 $filter_value->{$f}->{$n} = 1;
381 my @items = @{ $data->{items} };
385 foreach my $n ( keys %$filter_value ) {
386 if ( ! exists $i->{$n} ) {
387 if ( defined $filter_value->{$n}->{_missing} ) {
395 # and match any of values in element
397 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
398 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
400 if ( ! $have_values ) {
406 } @items if $filter_value;
411 sub _current_filters {
414 $current_filters->{ $_ } = $filters->{ $_ }
416 grep { defined $filters->{ $_ } }
417 @{ $self->_loaded('header') }
419 #warn "# current_filters ",dump($current_filters);
420 return $current_filters;
423 sub _data_sorted_by {
424 my ( $self, $order ) = @_;
426 my $path = $self->session('path');
428 warn "_data_sorted_by $order from $path";
430 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
431 return $loaded->{$path}->{sorted}->{$order};
434 my $data = $self->_loaded( 'data' );
435 my $numeric = $self->_is_numeric($order);
436 my $missing = $numeric ? 0 : '';
437 no warnings qw(numeric);
448 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
449 } @{ $data->{items} }
452 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
453 #warn "# sorted ",dump( @sorted );
455 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
462 my $path = $self->session('path');
463 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
465 my @columns = $self->_perm_array('columns');
466 $self->redirect_to('/data/columns') unless @columns;
467 my $order = $self->_perm_scalar('order', $columns[0]);
468 my $sort = $self->_perm_scalar('sort', 'a');
469 my $offset = $self->_perm_scalar('offset', 0);
470 my $limit = $self->_perm_scalar('limit', 20);
471 $self->_perm_scalar('show', 'table');
473 # fix offset when changing limit
474 $offset = int( $offset / $limit ) * $limit;
476 my $sorted = $self->_data_sorted_by( $order );
480 foreach my $name ( keys %$filters ) {
481 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
482 warn "skip filter $name not found in $path\n";
485 push @filter_names, $name;
487 warn "filter_names ",dump( @filter_names );
488 foreach my $name ( @filter_names ) {
489 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
490 $self->_filter_on_data( $name, @{ $filters->{$name} } );
494 my $all_filters = join(' ',sort @filter_names,'order:',$order);
496 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
498 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
500 my $path_filters = $loaded->{$path}->{filters};
502 warn "create combined filter for $all_filters\n";
505 foreach my $i ( 0 .. $#$sorted ) {
506 my $pos = $sorted->[$i];
508 if ( $#filter_names == -1 ) {
509 push @filtered, $pos;
514 foreach ( @filter_names ) {
515 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
519 push @filtered, $pos;
522 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
525 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
526 if defined $loaded->{$path}->{filtered}->{$all_filters};
528 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
531 my $data = $self->_loaded('data');
532 my $from_end = $sort eq 'd' ? $#$filtered : 0;
533 foreach ( 0 .. $limit ) {
534 my $i = $_ + $offset;
535 last unless defined $filtered->[$i];
536 $i = $from_end - $i if $from_end;
537 my $id = $filtered->[$i];
539 my $item = $data->{items}->[ $id ];
540 $item->{_row_id} ||= $id;
543 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
549 sorted => $sorted_items,
550 columns => [ @columns ],
551 rows => $#$filtered + 1,
552 numeric => { map { $_, $self->_is_numeric($_) } @columns },
553 filters => $self->_current_filters,
561 $self->session('order', $self->param('order'));
562 $self->session('sort', $self->param('sort'));
563 $self->redirect_to('/data/items');
567 my ( $self, $name ) = @_;
569 my $stats = $self->_loaded( 'stats' );
571 # sort facet numerically if more >50% elements are numeric
572 my $count = $stats->{$name}->{count};
573 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
574 defined $stats->{$name}->{numeric} &&
575 $stats->{$name}->{numeric} > $count / 2;
579 my ($self,$name) = @_;
580 warn "_remove_filter $name\n";
582 my $path = $self->session('path');
584 delete $filters->{$name};
585 delete $loaded->{$path}->{filters}->{$name};
586 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
590 keys %{ $loaded->{$path}->{filtered} }
592 delete $loaded->{$path}->{filtered}->{$_};
593 warn "remove filtered cache $_";
600 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
602 if ( my $name = $self->param('remove') ) {
603 $self->_remove_filter( $name );
604 $self->redirect_to( '/data/items' );
608 my $name = $self->param('name') || die "no name";
610 my $all = $self->_perm_scalar('all', 1);
611 my $data = $self->_loaded('data');
613 my $filters = $self->_current_filters;
614 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
615 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
616 if defined $loaded->{$path}->{filtered}->{$all_filters};
618 if ( ! $filtered || $all ) {
619 $filtered = [ 0 .. $#{ $data->{items} } ];
620 warn "filter all values\n";
622 warn "filter using $all_filters\n";
625 foreach my $i ( @$filtered ) {
626 my $item = $data->{items}->[$i];
627 if ( ! exists $item->{$name} ) {
628 $facet->{ _missing }++;
629 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
630 $facet->{$_}++ foreach @{ $item->{$name} };
632 $facet->{ $item->{$name} }++;
636 # warn "# facet $name ",dump $facet;
641 : defined $filters->{$name} ? @{ $filters->{$name} }
644 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
646 my $numeric = $self->_is_numeric($name);
648 my $sort = $self->param('sort');
649 # sort numeric facets with more than 5 values ascending
650 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
652 @facet_names = sort {
654 if ( $sort eq 'a' ) {
655 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
656 } elsif ( $sort eq 'd' ) {
657 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
658 } elsif ( $sort eq 'c' ) {
659 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
661 warn "unknown sort: $sort";
664 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
668 $self->render( name => $name, facet => $facet, checked => $checked,
669 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
675 my $content = $self->param('content');
677 my $i = $self->param('_row_id');
678 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
679 my $path = $self->param('path') || die "no path";
680 my $name = $self->param('name') || die "no name";
681 my $status = 200; # 200 = OK, 201 = Created
683 if ( defined $loaded->{$path}->{data}->{items}->[$i]->{$name} ) {
684 $content =~ s/^\s+//s;
685 $content =~ s/\s+$//s;
687 if ( $content =~ /\xB6/ ) { # para
688 $v = [ split(/\s*\xB6\s*/, $content) ];
693 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
695 if ( $old ne $new ) {
696 warn "# update $path $i $old -> $new\n";
697 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
699 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
700 delete $loaded->{$path}->{sorted}->{$name};
701 warn "# invalidate $path sorted $name\n";
704 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
705 delete $loaded->{$path}->{filtered}->{$_};
706 warn "# invalidate $path filtered $_\n";
709 $status = 201; # created
710 $self->session('save_path' => $path);
713 warn "# unchanged $path $i $old\n";
717 $content = "$path $i $name doesn't exist\n";
731 my $path = $self->param('path');
732 $path ||= $self->session('path');
734 my $dump_path = $self->_save( $path );
735 $self->session('save_path' => 0);
737 $self->redirect_to( '/data/items' );