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;
41 $dump_path->{$_} = $self->_dump_path( $_ ) foreach @files;
48 dump_path => $dump_path,
53 my ( $self, $name ) = @_;
54 my $dir = $self->app->home->rel_dir('data');
57 return '/tmp/mojo_facets.' . $name . '.storable';
61 my ( $self, $path ) = @_;
63 my $dump_path = $self->_dump_path( $path );
64 warn "save loaded to $dump_path";
65 my $info = $loaded->{$path};
66 store $info, $dump_path;
69 my $mtime = $loaded->{$path}->{mtime};
70 utime $mtime, $mtime, $dump_path;
72 warn $dump_path, ' ', -s $dump_path, " bytes\n";
77 my ( $self, $path ) = @_;
79 my $full_path = $self->app->home->rel_file( 'data/' . $path );
80 die "$full_path $!" unless -r $full_path;
82 my $dump_path = $self->_dump_path( $path );
84 if ( defined $loaded->{$path}->{data} ) {
85 my $mtime = (stat($full_path))[9];
86 return if $loaded->{$path}->{mtime} == $mtime;
87 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
88 } elsif ( -e $dump_path ) {
89 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
90 my $info = retrieve $dump_path;
91 $loaded->{ $path } = $info;
95 # we could use Mojo::JSON here, but it's too slow
96 # $data = from_json read_file $path;
97 my $data = read_file $full_path;
98 warn "# data snippet: ", substr($data,0,200);
100 if ( $path =~ m/\.js/ ) {
101 Encode::_utf8_on($data);
102 $data = from_json $data;
103 } elsif ( $path =~ m/\.txt/ ) {
104 my @lines = split(/\r?\n/, $data);
105 $data = { items => [] };
107 my $header_line = shift @lines;
108 my $multiline = $header_line =~ s/\^//g;
109 @header = split(/\|/, $header_line );
110 warn "# header ", dump( @header );
111 while ( my $line = shift @lines ) {
114 my @v = split(/\|/, $line);
115 while ( @lines && $#v < $#header ) {
119 my @more_v = split(/\|/, $line);
120 if ( $#v + $#more_v > $#header ) {
121 warn "short line: ",dump( @v );
125 $v[ $#v ] .= shift @more_v if @more_v;
126 push @v, @more_v if @more_v;
128 if ( $#v > $#header ) {
129 die "# splice $#header ", dump( @v );
130 @v = splice @v, 0, $#header;
134 foreach my $i ( 0 .. $#v ) {
135 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
137 push @{ $data->{items} }, $item;
140 warn "file format unknown $path";
145 foreach my $e ( @{ $data->{items} } ) {
146 foreach my $n ( keys %$e ) {
147 $stats->{$n}->{count}++;
149 if ( ref $e->{$n} eq 'ARRAY' ) {
150 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
156 foreach my $x ( @v ) {
157 $stats->{$n}->{numeric}++
158 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
159 $stats->{$n}->{empty}++
160 if length $x == 0; # faster than $x =~ m/^\s*$/;
166 foreach my $n ( keys %$stats ) {
167 next unless defined $stats->{$n}->{array};
168 delete $stats->{$n}->{array}
169 if $stats->{$n}->{array} == $stats->{$n}->{count};
173 if ( defined $data->{header} ) {
174 if ( ref $data->{header} eq 'ARRAY' ) {
175 @header = @{ $data->{header} };
177 warn "header not array ", dump( $data->{header} );
183 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
184 grep { defined $stats->{$_}->{count} } keys %$stats
190 header => [ @header ],
192 full_path => $full_path,
193 size => -s $full_path,
194 mtime => (stat($full_path))[9],
198 $loaded->{ $path } = $info;
199 $self->_save( $path );
207 my @paths = $self->param('paths');
208 warn "# paths ", dump @paths;
209 $self->_load_path( $_ ) foreach @paths;
211 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
212 warn "# path $path\n";
213 $self->session('path' => $path);
214 $self->_load_path( $path );
216 $self->session( 'header' => $loaded->{$path}->{header} );
217 if ( ! defined $loaded->{$path}->{columns} ) {
218 $self->session( 'columns' => $loaded->{$path}->{header} );
219 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
220 $self->redirect_to( '/data/columns' );
222 $self->session( 'columns' => $loaded->{$path}->{columns} );
223 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
224 $self->redirect_to( '/data/items' );
230 my ( $self, $name ) = @_;
231 my $path = $self->session('path');
232 $self->redirect_to('/data/index') unless $path;
233 if ( ! defined $loaded->{$path}->{$name} ) {
234 warn "$path $name doesn't exist in loaded ",dump( $loaded );
235 $self->redirect_to('/data/index');
237 return $loaded->{$path}->{$name};
244 $checked->{$_}++ foreach @_;
245 # warn "# _checked ",dump($checked);
253 if ( $self->param('columns') ) {
254 $self->_perm_array('columns');
255 $self->redirect_to('/data/items');
258 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
261 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
263 foreach my $c ( @{ $self->session( 'header' ) } ) {
264 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
268 message => 'Select columns to display',
270 columns => \@columns,
271 checked => $self->_checked( $self->_perm_array('columns') ),
276 my ($self,$name) = @_;
278 my @array = $self->param($name);
279 my $path = $self->session('path');
282 $self->session($name => [ @array ]);
283 $loaded->{$path}->{$name} = [ @array ];
284 } elsif ( my $session = $self->session($name) ) {
285 if ( ref $session eq 'ARRAY' ) {
288 die "$name not array ",dump($session);
291 #warn "# $name ",dump @array;
296 my ($self,$name,$default) = @_;
298 my $scalar = $self->param($name);
300 if ( defined $scalar ) {
301 $self->session($name => $scalar);
303 $scalar = $self->session($name);
306 if ( ! defined $scalar ) {
308 die "no default for $name" unless defined $scalar;
309 $self->session($name => $scalar);
312 warn "# _perm_scalar $name ",dump $scalar;
319 my $name = $self->param('filter_name') || die "name?";
320 my @vals = $self->param('filter_vals');
322 $self->_remove_filter( $name );
323 $self->_filter_on_data( $name, @vals ) if @vals;
325 $self->session( 'offset' => 0 );
327 $self->redirect_to('/data/items');
330 sub _filter_on_data {
331 my ( $self, $name, @vals ) = @_;
333 my $path = $self->session('path');
335 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
336 warn "filter $name not found in data set";
340 $filters->{$name} = [ @vals ];
341 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
344 $filter_hash->{$_}++ foreach @vals;
346 #warn "# filter_hash ",dump( $filter_hash );
348 my $items = $self->_loaded('data')->{items};
350 my $include_missing = defined $filter_hash->{_missing};
353 foreach my $i ( 0 .. $#$items ) {
355 if ( defined $items->[$i]->{$name} ) {
356 foreach my $v ( @{ $items->[$i]->{$name} } ) {
357 if ( defined $filter_hash->{ $v } ) {
358 $filtered_items->{$i}++;
361 } elsif ( $include_missing ) {
362 $filtered_items->{$i}++;
366 #warn "# filter $name ",dump($filtered_items);
368 $loaded->{$path}->{filters}->{$name} = $filtered_items;
369 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
373 my ( $self, $all ) = @_;
374 my $data = $self->_loaded( 'data' );
376 return @{ $data->{items} } if $all == 1;
378 my $filters = $self->_current_filters;
380 foreach my $f ( keys %$filters ) {
381 foreach my $n ( @{ $filters->{$f} } ) {
382 $filter_value->{$f}->{$n} = 1;
385 my @items = @{ $data->{items} };
389 foreach my $n ( keys %$filter_value ) {
390 if ( ! exists $i->{$n} ) {
391 if ( defined $filter_value->{$n}->{_missing} ) {
399 # and match any of values in element
401 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
402 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
404 if ( ! $have_values ) {
410 } @items if $filter_value;
415 sub _current_filters {
418 $current_filters->{ $_ } = $filters->{ $_ }
420 grep { defined $filters->{ $_ } }
421 @{ $self->_loaded('header') }
423 #warn "# current_filters ",dump($current_filters);
424 return $current_filters;
427 sub _data_sorted_by {
428 my ( $self, $order ) = @_;
430 my $path = $self->session('path');
432 warn "_data_sorted_by $order from $path";
434 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
435 return $loaded->{$path}->{sorted}->{$order};
438 my $data = $self->_loaded( 'data' );
439 my $numeric = $self->_is_numeric($order);
440 my $missing = $numeric ? 0 : '';
441 no warnings qw(numeric);
452 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
453 } @{ $data->{items} }
456 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
457 #warn "# sorted ",dump( @sorted );
459 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
466 my $path = $self->session('path');
467 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
469 my @columns = $self->_perm_array('columns');
470 $self->redirect_to('/data/columns') unless @columns;
471 my $order = $self->_perm_scalar('order', $columns[0]);
472 my $sort = $self->_perm_scalar('sort', 'a');
473 my $offset = $self->_perm_scalar('offset', 0);
474 my $limit = $self->_perm_scalar('limit', 20);
475 $self->_perm_scalar('show', 'table');
477 # fix offset when changing limit
478 $offset = int( $offset / $limit ) * $limit;
480 my $sorted = $self->_data_sorted_by( $order );
484 foreach my $name ( keys %$filters ) {
485 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
486 warn "skip filter $name not found in $path\n";
489 push @filter_names, $name;
491 warn "filter_names ",dump( @filter_names );
492 foreach my $name ( @filter_names ) {
493 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
494 $self->_filter_on_data( $name, @{ $filters->{$name} } );
498 my $all_filters = join(' ',sort @filter_names,'order:',$order);
500 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
502 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
504 my $path_filters = $loaded->{$path}->{filters};
506 warn "create combined filter for $all_filters\n";
509 foreach my $i ( 0 .. $#$sorted ) {
510 my $pos = $sorted->[$i];
512 if ( $#filter_names == -1 ) {
513 push @filtered, $pos;
518 foreach ( @filter_names ) {
519 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
523 push @filtered, $pos;
526 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
529 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
530 if defined $loaded->{$path}->{filtered}->{$all_filters};
532 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
535 my $data = $self->_loaded('data');
536 my $from_end = $sort eq 'd' ? $#$filtered : 0;
537 foreach ( 0 .. $limit ) {
538 my $i = $_ + $offset;
539 last unless defined $filtered->[$i];
540 $i = $from_end - $i if $from_end;
541 my $id = $filtered->[$i];
543 my $item = $data->{items}->[ $id ];
544 $item->{_row_id} ||= $id;
547 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
553 sorted => $sorted_items,
554 columns => [ @columns ],
555 rows => $#$filtered + 1,
556 numeric => { map { $_, $self->_is_numeric($_) } @columns },
557 filters => $self->_current_filters,
565 $self->session('order', $self->param('order'));
566 $self->session('sort', $self->param('sort'));
567 $self->redirect_to('/data/items');
571 my ( $self, $name ) = @_;
573 my $stats = $self->_loaded( 'stats' );
575 # sort facet numerically if more >50% elements are numeric
576 my $count = $stats->{$name}->{count};
577 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
578 defined $stats->{$name}->{numeric} &&
579 $stats->{$name}->{numeric} > $count / 2;
583 my ($self,$name) = @_;
584 warn "_remove_filter $name\n";
586 my $path = $self->session('path');
588 delete $filters->{$name};
589 delete $loaded->{$path}->{filters}->{$name};
590 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
594 keys %{ $loaded->{$path}->{filtered} }
596 delete $loaded->{$path}->{filtered}->{$_};
597 warn "remove filtered cache $_";
604 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
606 if ( my $name = $self->param('remove') ) {
607 $self->_remove_filter( $name );
608 $self->redirect_to( '/data/items' );
612 my $name = $self->param('name') || die "no name";
614 my $all = $self->_perm_scalar('all', 1);
615 my $data = $self->_loaded('data');
617 my $filters = $self->_current_filters;
618 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
619 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
620 if defined $loaded->{$path}->{filtered}->{$all_filters};
622 if ( ! $filtered || $all ) {
623 $filtered = [ 0 .. $#{ $data->{items} } ];
624 warn "filter all values\n";
626 warn "filter using $all_filters\n";
629 foreach my $i ( @$filtered ) {
630 my $item = $data->{items}->[$i];
631 if ( ! exists $item->{$name} ) {
632 $facet->{ _missing }++;
633 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
634 $facet->{$_}++ foreach @{ $item->{$name} };
636 $facet->{ $item->{$name} }++;
640 # warn "# facet $name ",dump $facet;
645 : defined $filters->{$name} ? @{ $filters->{$name} }
648 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
650 my $numeric = $self->_is_numeric($name);
652 my $sort = $self->param('sort');
653 # sort numeric facets with more than 5 values ascending
654 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
656 @facet_names = sort {
658 if ( $sort eq 'a' ) {
659 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
660 } elsif ( $sort eq 'd' ) {
661 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
662 } elsif ( $sort eq 'c' ) {
663 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
665 warn "unknown sort: $sort";
668 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
672 $self->render( name => $name, facet => $facet, checked => $checked,
673 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
679 my $content = $self->param('content');
681 my $i = $self->param('_row_id');
682 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
683 my $path = $self->param('path') || die "no path";
684 my $name = $self->param('name') || die "no name";
685 my $status = 200; # 200 = OK, 201 = Created
687 if ( defined $loaded->{$path}->{data}->{items}->[$i]->{$name} ) {
688 $content =~ s/^\s+//s;
689 $content =~ s/\s+$//s;
691 if ( $content =~ /\xB6/ ) { # para
692 $v = [ split(/\s*\xB6\s*/, $content) ];
697 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
699 if ( $old ne $new ) {
700 warn "# update $path $i $old -> $new\n";
701 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
703 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
704 delete $loaded->{$path}->{sorted}->{$name};
705 warn "# invalidate $path sorted $name\n";
708 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
709 delete $loaded->{$path}->{filtered}->{$_};
710 warn "# invalidate $path filtered $_\n";
713 $status = 201; # created
714 $self->session('save_path' => $path);
717 warn "# unchanged $path $i $old\n";
721 $content = "$path $i $name doesn't exist\n";
735 my $path = $self->param('path');
736 $path ||= $self->session('path');
738 my $dump_path = $self->_save( $path );
739 $self->session('save_path' => 0);
741 $self->redirect_to( '/data/items' );