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, $path ) = @_;
52 my $dir = $self->app->home->rel_dir('data');
55 my $dump_path = '/tmp/mojo_facets.' . $name . '.storable';
57 warn "save loaded to $dump_path";
58 my $info = $loaded->{$path};
59 store $info, $dump_path;
62 my $mtime = $loaded->{$path}->{mtime};
63 utime $mtime, $mtime, $dump_path;
65 warn $dump_path, ' ', -s $dump_path, " bytes\n";
69 my ( $self, $path ) = @_;
71 my $full_path = $self->app->home->rel_file( 'data/' . $path );
72 die "$full_path $!" unless -r $full_path;
74 if ( defined $loaded->{$path}->{data} ) {
75 my $mtime = (stat($full_path))[9];
76 return if $loaded->{$path}->{mtime} == $mtime;
77 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
80 # we could use Mojo::JSON here, but it's too slow
81 # $data = from_json read_file $path;
82 my $data = read_file $full_path;
83 warn "# data snippet: ", substr($data,0,200);
85 if ( $path =~ m/\.js/ ) {
86 Encode::_utf8_on($data);
87 $data = from_json $data;
88 } elsif ( $path =~ m/\.txt/ ) {
89 my @lines = split(/\r?\n/, $data);
90 $data = { items => [] };
92 my $header_line = shift @lines;
93 my $multiline = $header_line =~ s/\^//g;
94 @header = split(/\|/, $header_line );
95 warn "# header ", dump( @header );
96 while ( my $line = shift @lines ) {
99 my @v = split(/\|/, $line);
100 while ( @lines && $#v < $#header ) {
104 my @more_v = split(/\|/, $line);
105 if ( $#v + $#more_v > $#header ) {
106 warn "short line: ",dump( @v );
110 $v[ $#v ] .= shift @more_v if @more_v;
111 push @v, @more_v if @more_v;
113 if ( $#v > $#header ) {
114 die "# splice $#header ", dump( @v );
115 @v = splice @v, 0, $#header;
119 foreach my $i ( 0 .. $#v ) {
120 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
122 push @{ $data->{items} }, $item;
125 warn "file format unknown $path";
130 foreach my $e ( @{ $data->{items} } ) {
131 foreach my $n ( keys %$e ) {
132 $stats->{$n}->{count}++;
134 if ( ref $e->{$n} eq 'ARRAY' ) {
135 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
141 foreach my $x ( @v ) {
142 $stats->{$n}->{numeric}++
143 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
144 $stats->{$n}->{empty}++
145 if length $x == 0; # faster than $x =~ m/^\s*$/;
151 foreach my $n ( keys %$stats ) {
152 next unless defined $stats->{$n}->{array};
153 delete $stats->{$n}->{array}
154 if $stats->{$n}->{array} == $stats->{$n}->{count};
158 if ( defined $data->{header} ) {
159 if ( ref $data->{header} eq 'ARRAY' ) {
160 @header = @{ $data->{header} };
162 warn "header not array ", dump( $data->{header} );
168 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
169 grep { defined $stats->{$_}->{count} } keys %$stats
175 header => [ @header ],
177 full_path => $full_path,
178 size => -s $full_path,
179 mtime => (stat($full_path))[9],
183 $loaded->{ $path } = $info;
184 $self->_save( $path );
192 my @paths = $self->param('paths');
193 warn "# paths ", dump @paths;
194 $self->_load_path( $_ ) foreach @paths;
196 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
197 warn "# path $path\n";
198 $self->session('path' => $path);
199 $self->_load_path( $path );
201 $self->session( 'header' => $loaded->{$path}->{header} );
202 if ( ! defined $loaded->{$path}->{columns} ) {
203 $self->session( 'columns' => $loaded->{$path}->{header} );
204 $self->session( 'order' => $loaded->{$path}->{header}->[0] );
205 $self->redirect_to( '/data/columns' );
207 $self->session( 'columns' => $loaded->{$path}->{columns} );
208 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
209 $self->redirect_to( '/data/items' );
215 my ( $self, $name ) = @_;
216 my $path = $self->session('path');
217 if ( ! defined $loaded->{$path}->{$name} ) {
218 warn "$path $name doesn't exist in loaded ",dump( $loaded );
219 $self->redirect_to('/data/index');
221 return $loaded->{$path}->{$name};
228 $checked->{$_}++ foreach @_;
229 # warn "# _checked ",dump($checked);
237 if ( $self->param('columns') ) {
238 $self->_perm_array('columns');
239 $self->redirect_to('/data/items');
242 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
245 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
247 foreach my $c ( @{ $self->session( 'header' ) } ) {
248 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
252 message => 'Select columns to display',
254 columns => \@columns,
255 checked => $self->_checked( $self->_perm_array('columns') ),
260 my ($self,$name) = @_;
262 my @array = $self->param($name);
263 my $path = $self->session('path');
266 $self->session($name => [ @array ]);
267 $loaded->{$path}->{$name} = [ @array ];
268 } elsif ( my $session = $self->session($name) ) {
269 if ( ref $session eq 'ARRAY' ) {
272 die "$name not array ",dump($session);
275 #warn "# $name ",dump @array;
280 my ($self,$name,$default) = @_;
282 my $scalar = $self->param($name);
284 if ( defined $scalar ) {
285 $self->session($name => $scalar);
287 $scalar = $self->session($name);
290 if ( ! defined $scalar ) {
292 die "no default for $name" unless defined $scalar;
293 $self->session($name => $scalar);
296 warn "# _perm_scalar $name ",dump $scalar;
303 my $name = $self->param('filter_name') || die "name?";
304 my @vals = $self->param('filter_vals');
306 $self->_remove_filter( $name );
307 $self->_filter_on_data( $name, @vals ) if @vals;
309 $self->session( 'offset' => 0 );
311 $self->redirect_to('/data/items');
314 sub _filter_on_data {
315 my ( $self, $name, @vals ) = @_;
317 my $path = $self->session('path');
319 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
320 warn "filter $name not found in data set";
324 $filters->{$name} = [ @vals ];
325 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
328 $filter_hash->{$_}++ foreach @vals;
330 #warn "# filter_hash ",dump( $filter_hash );
332 my $items = $self->_loaded('data')->{items};
334 my $include_missing = defined $filter_hash->{_missing};
337 foreach my $i ( 0 .. $#$items ) {
339 if ( defined $items->[$i]->{$name} ) {
340 foreach my $v ( @{ $items->[$i]->{$name} } ) {
341 if ( defined $filter_hash->{ $v } ) {
342 $filtered_items->{$i}++;
345 } elsif ( $include_missing ) {
346 $filtered_items->{$i}++;
350 #warn "# filter $name ",dump($filtered_items);
352 $loaded->{$path}->{filters}->{$name} = $filtered_items;
353 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
357 my ( $self, $all ) = @_;
358 my $data = $self->_loaded( 'data' );
360 return @{ $data->{items} } if $all == 1;
362 my $filters = $self->_current_filters;
364 foreach my $f ( keys %$filters ) {
365 foreach my $n ( @{ $filters->{$f} } ) {
366 $filter_value->{$f}->{$n} = 1;
369 my @items = @{ $data->{items} };
373 foreach my $n ( keys %$filter_value ) {
374 if ( ! exists $i->{$n} ) {
375 if ( defined $filter_value->{$n}->{_missing} ) {
383 # and match any of values in element
385 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
386 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
388 if ( ! $have_values ) {
394 } @items if $filter_value;
399 sub _current_filters {
402 $current_filters->{ $_ } = $filters->{ $_ }
404 grep { defined $filters->{ $_ } }
405 @{ $self->_loaded('header') }
407 #warn "# current_filters ",dump($current_filters);
408 return $current_filters;
411 sub _data_sorted_by {
412 my ( $self, $order ) = @_;
414 my $path = $self->session('path');
416 warn "_data_sorted_by $order from $path";
418 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
419 return $loaded->{$path}->{sorted}->{$order};
422 my $data = $self->_loaded( 'data' );
423 my $numeric = $self->_is_numeric($order);
424 my $missing = $numeric ? 0 : '';
425 no warnings qw(numeric);
436 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
437 } @{ $data->{items} }
440 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
441 #warn "# sorted ",dump( @sorted );
443 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
450 my $path = $self->session('path');
451 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
453 my @columns = $self->_perm_array('columns');
454 $self->redirect_to('/data/columns') unless @columns;
455 my $order = $self->_perm_scalar('order', $columns[0]);
456 my $sort = $self->_perm_scalar('sort', 'a');
457 my $offset = $self->_perm_scalar('offset', 0);
458 my $limit = $self->_perm_scalar('limit', 20);
459 $self->_perm_scalar('show', 'table');
461 # fix offset when changing limit
462 $offset = int( $offset / $limit ) * $limit;
464 my $sorted = $self->_data_sorted_by( $order );
468 foreach my $name ( keys %$filters ) {
469 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
470 warn "skip filter $name not found in $path\n";
473 push @filter_names, $name;
475 warn "filter_names ",dump( @filter_names );
476 foreach my $name ( @filter_names ) {
477 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
478 $self->_filter_on_data( $name, @{ $filters->{$name} } );
482 my $all_filters = join(' ',sort @filter_names,'order:',$order);
484 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
486 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
488 my $path_filters = $loaded->{$path}->{filters};
490 warn "create combined filter for $all_filters\n";
493 foreach my $i ( 0 .. $#$sorted ) {
494 my $pos = $sorted->[$i];
496 if ( $#filter_names == -1 ) {
497 push @filtered, $pos;
502 foreach ( @filter_names ) {
503 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
507 push @filtered, $pos;
510 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
513 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
514 if defined $loaded->{$path}->{filtered}->{$all_filters};
516 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
519 my $data = $self->_loaded('data');
520 my $from_end = $sort eq 'd' ? $#$filtered : 0;
521 foreach ( 0 .. $limit ) {
522 my $i = $_ + $offset;
523 last unless defined $filtered->[$i];
524 $i = $from_end - $i if $from_end;
525 my $id = $filtered->[$i];
527 my $item = $data->{items}->[ $id ];
528 $item->{_row_id} ||= $id;
531 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
537 sorted => $sorted_items,
538 columns => [ @columns ],
539 rows => $#$filtered + 1,
540 numeric => { map { $_, $self->_is_numeric($_) } @columns },
541 filters => $self->_current_filters,
549 $self->session('order', $self->param('order'));
550 $self->session('sort', $self->param('sort'));
551 $self->redirect_to('/data/items');
555 my ( $self, $name ) = @_;
557 my $stats = $self->_loaded( 'stats' );
559 # sort facet numerically if more >50% elements are numeric
560 my $count = $stats->{$name}->{count};
561 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
562 defined $stats->{$name}->{numeric} &&
563 $stats->{$name}->{numeric} > $count / 2;
567 my ($self,$name) = @_;
568 warn "_remove_filter $name\n";
570 my $path = $self->session('path');
572 delete $filters->{$name};
573 delete $loaded->{$path}->{filters}->{$name};
574 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
578 keys %{ $loaded->{$path}->{filtered} }
580 delete $loaded->{$path}->{filtered}->{$_};
581 warn "remove filtered cache $_";
588 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
590 if ( my $name = $self->param('remove') ) {
591 $self->_remove_filter( $name );
592 $self->redirect_to( '/data/items' );
596 my $name = $self->param('name') || die "no name";
598 my $all = $self->_perm_scalar('all', 1);
599 my $data = $self->_loaded('data');
601 my $filters = $self->_current_filters;
602 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
603 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
604 if defined $loaded->{$path}->{filtered}->{$all_filters};
606 if ( ! $filtered || $all ) {
607 $filtered = [ 0 .. $#{ $data->{items} } ];
608 warn "filter all values\n";
610 warn "filter using $all_filters\n";
613 foreach my $i ( @$filtered ) {
614 my $item = $data->{items}->[$i];
615 if ( ! exists $item->{$name} ) {
616 $facet->{ _missing }++;
617 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
618 $facet->{$_}++ foreach @{ $item->{$name} };
620 $facet->{ $item->{$name} }++;
624 # warn "# facet $name ",dump $facet;
629 : defined $filters->{$name} ? @{ $filters->{$name} }
632 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
634 my $numeric = $self->_is_numeric($name);
636 my $sort = $self->param('sort');
637 # sort numeric facets with more than 5 values ascending
638 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
640 @facet_names = sort {
642 if ( $sort eq 'a' ) {
643 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
644 } elsif ( $sort eq 'd' ) {
645 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
646 } elsif ( $sort eq 'c' ) {
647 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
649 warn "unknown sort: $sort";
652 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
656 $self->render( name => $name, facet => $facet, checked => $checked,
657 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
663 my $content = $self->param('content');
665 my $i = $self->param('_row_id') || die "no _row_id";
666 my $path = $self->param('path') || die "no path";
667 my $name = $self->param('name') || die "no name";
669 if ( defined $loaded->{$path}->{data}->{items}->[$i]->{$name} ) {
670 $content =~ s/^\s+//s;
671 $content =~ s/\s+$//s;
672 if ( $content =~ /\xB6/ ) { # para
673 $content = [ split(/\s*\xB6\s*/, $content) ];
675 $content = [ $content ];
678 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
679 my $new = dump $content;
680 if ( $old ne $new ) {
681 warn "# update $path $i $old -> $new\n";
682 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $content;
684 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
685 delete $loaded->{$path}->{sorted}->{$name};
686 warn "# invalidate $path sorted $name\n";
689 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
690 delete $loaded->{$path}->{filtered}->{$_};
691 warn "# invalidate $path filtered $_\n";
695 warn "# unchanged $path $i $old\n";
698 $content = "$path $i $name doesn't exist\n";