2a35b2db8e37baccd6c3631f7494a6e4e99985ad
[MojoFacets.git] / lib / MojoFacets / Data.pm
1 package MojoFacets::Data;
2
3 use strict;
4 use warnings;
5
6 use base 'Mojolicious::Controller';
7
8 use Data::Dump qw(dump);
9 use File::Slurp;
10 use JSON;
11 use Encode;
12 use locale;
13 use File::Find;
14
15 our $loaded;
16 our $filters;
17
18 sub index {
19         my $self = shift;
20
21         my $path = $self->app->home->rel_dir('data');
22         die "no data dir $path" unless -d $path;
23
24         my @files;
25         find( sub {
26                 my $file = $File::Find::name;
27                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
28                         $file =~ s/$path\/*//;
29                         push @files, $file;
30                 } else {
31                         warn "IGNORE: $file\n";
32                 }
33         }, $path);
34
35         @files = sort { lc $a cmp lc $b } @files;
36         my $size;
37         $size->{$_} = -s "$path/$_" foreach @files;
38
39         $self->render(
40                 files => [ @files ],
41                 size => $size,
42                 loaded => $loaded,
43                 filters => $filters,
44         );
45 }
46
47 sub _load_path {
48         my ( $self, $path ) = @_;
49
50         return if defined $loaded->{$path}->{data};
51
52         my $full_path = $self->app->home->rel_file( 'data/' . $path );
53         die "$full_path $!" unless -r $full_path;
54
55         # we could use Mojo::JSON here, but it's too slow
56 #       $data = from_json read_file $path;
57         my $data = read_file $full_path;
58         warn "# data snippet: ", substr($data,0,200);
59         my @header;
60         if ( $path =~ m/\.js/ ) {
61                 Encode::_utf8_on($data);
62                 $data = from_json $data;
63         } elsif ( $path =~ m/\.txt/ ) {
64                 my @lines = split(/\r?\n/, $data);
65                 $data = { items => [] };
66
67                 my $header_line = shift @lines;
68                 my $multiline = $header_line =~ s/\^//g;
69                 @header = split(/\|/, $header_line );
70                 warn "# header ", dump( @header );
71                 while ( my $line = shift @lines ) {
72                         $line =~ s/\^//g;
73                         chomp $line;
74                         my @v = split(/\|/, $line);
75                         while ( @lines && $#v < $#header ) {
76                                 $line = $lines[0];
77                                 $line =~ s/\^//g;
78                                 chomp $line;
79                                 my @more_v = split(/\|/, $line);
80                                 if ( $#v + $#more_v > $#header ) {
81                                         warn "short line: ",dump( @v );
82                                         last;
83                                 }
84                                 shift @lines;
85                                 $v[ $#v ] .= shift @more_v if @more_v;
86                                 push @v, @more_v if @more_v;
87
88                                 if ( $#v > $#header ) {
89                                         die "# splice $#header ", dump( @v );
90                                         @v = splice @v, 0, $#header;
91                                 }
92                         }
93                         my $item;
94                         $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
95                         push @{ $data->{items} }, $item;
96                 }
97         } else {
98                 warn "file format unknown $path";
99         }
100
101         my $stats;
102
103         foreach my $e ( @{ $data->{items} } ) {
104                 foreach my $n ( keys %$e ) {
105                         $stats->{$n}->{count}++;
106                         my @v;
107                         if ( ref $e->{$n} eq 'ARRAY' ) {
108                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
109                                 @v = @{ $e->{$n} };
110                         } else {
111                                 @v = ( $e->{$n} );
112                         }
113
114                         foreach my $x ( @v ) {
115                                 $stats->{$n}->{numeric}++
116                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
117                                 $stats->{$n}->{empty}++
118                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
119                         }
120
121                 }
122         }
123
124         foreach my $n ( keys %$stats ) {
125                 next unless defined $stats->{$n}->{array};
126                 delete $stats->{$n}->{array}
127                         if $stats->{$n}->{array} == $stats->{$n}->{count};
128         }
129
130         @header =
131                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
132                 grep { defined $stats->{$_}->{count} } keys %$stats
133                 unless @header;
134
135         warn dump($stats);
136
137         $loaded->{ $path } = {
138                 header => [ @header ],
139                 stats  => $stats,
140                 full_path => $full_path,
141                 size => -s $full_path,
142                 data => $data,
143         };
144
145 }
146
147
148 sub load {
149         my $self = shift;
150
151         my @paths = $self->param('paths');
152         warn "# paths ", dump @paths;
153         $self->_load_path( $_ ) foreach @paths;
154
155         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
156         warn "# path $path\n";
157         $self->session('path' => $path);
158         $self->_load_path( $path );
159
160         $self->session( 'header' => $loaded->{$path}->{header} );
161         if ( ! defined $loaded->{$path}->{columns} ) {
162                 $self->session( 'columns' => $loaded->{$path}->{header} );
163                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
164                 $self->redirect_to( '/data/columns' );
165         } else {
166                 $self->session( 'columns' => $loaded->{$path}->{columns} );
167                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
168                 $self->redirect_to( '/data/items' );
169         }
170 }
171
172
173 sub _loaded {
174         my ( $self, $name ) = @_;
175         my $path = $self->session('path');
176         die "$path $name doesn't exist in loaded ",dump( $loaded )
177                 unless defined $loaded->{$path}->{$name};
178         return $loaded->{$path}->{$name};
179 }
180
181
182 sub _checked {
183         my $self = shift;
184         my $checked;
185         $checked->{$_}++ foreach @_;
186 #       warn "# _checked ",dump($checked);
187         return $checked;
188 }
189
190
191 sub columns {
192     my $self = shift;
193
194         if ( $self->param('columns') ) {
195                 $self->_perm_array('columns');
196                 $self->redirect_to('/data/items');
197         }
198
199         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
200
201         my @columns;
202         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
203
204         foreach my $c ( @{ $self->session( 'header' ) } ) {
205                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
206         }
207
208     $self->render(
209                 message => 'Select columns to display',
210                 stats => $stats,
211                 columns => \@columns,
212                 checked => $self->_checked( $self->_perm_array('columns') ),
213         );
214 }
215
216 sub _perm_array {
217     my ($self,$name) = @_;
218
219         my @array = $self->param($name);
220         my $path  = $self->session('path');
221
222         if ( @array ) {
223                 $self->session($name => [ @array ]);
224                 $loaded->{$path}->{$name} = [ @array ];
225         } elsif ( my $session = $self->session($name) ) {
226                 if ( ref $session eq 'ARRAY' ) {
227                         @array = @$session;
228                 } else {
229                         die "$name not array ",dump($session);
230                 }
231         }
232         #warn "# $name ",dump @array;
233         return @array;
234 }
235
236 sub _perm_scalar {
237     my ($self,$name,$default) = @_;
238
239         my $scalar = $self->param($name);
240
241         if ( defined $scalar ) {
242                 $self->session($name => $scalar);
243         } else {
244                 $scalar = $self->session($name);
245         }
246
247         if ( ! defined $scalar ) {
248                 $scalar = $default;
249                 die "no default for $name" unless defined $scalar;
250                 $self->session($name => $scalar);
251         }
252
253         warn "# _perm_scalar $name ",dump $scalar;
254         return $scalar;
255 }
256
257 sub filter {
258         my $self = shift;
259
260         my $name = $self->param('filter_name') || die "name?";
261         my @vals = $self->param('filter_vals');
262
263         $self->_remove_filter( $name );
264         $self->_filter_on_data( $name, @vals ) if @vals;
265
266         $self->session( 'offset' => 0 );
267
268         $self->redirect_to('/data/items');
269 }
270
271 sub _filter_on_data {
272         my ( $self, $name, @vals ) = @_;
273
274         my $path = $self->session('path');
275
276         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
277                 warn "filter $name not found in data set";
278                 return;
279         }
280
281         $filters->{$name} = [ @vals ];
282         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
283
284         my $filter_hash;
285         $filter_hash->{$_}++ foreach @vals;
286
287         #warn "# filter_hash ",dump( $filter_hash );
288
289         my $items = $self->_loaded('data')->{items};
290
291         my $include_missing = defined $filter_hash->{_missing};
292         my $filtered_items;
293
294         foreach my $i ( 0 .. $#$items ) {
295
296                 if ( defined $items->[$i]->{$name} ) {
297                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
298                                 if ( defined $filter_hash->{ $v } ) {
299                                         $filtered_items->{$i}++;
300                                 }
301                         }
302                 } elsif ( $include_missing ) {
303                         $filtered_items->{$i}++;
304                 }
305         }
306
307         #warn "# filter $name ",dump($filtered_items);
308
309         $loaded->{$path}->{filters}->{$name} = $filtered_items;
310         warn "generated ", scalar keys %$filtered_items, " filtered items\n";
311 }
312
313 sub _data_items {
314         my ( $self, $all ) = @_;
315         my $data = $self->_loaded( 'data' );
316
317         return @{ $data->{items} } if $all == 1;
318
319         my $filters = $self->_current_filters;
320         my $filter_value;
321         foreach my $f ( keys %$filters ) {
322                 foreach my $n ( @{ $filters->{$f} } ) {
323                         $filter_value->{$f}->{$n} = 1;
324                 }
325         }
326         my @items = @{ $data->{items} };
327         @items = grep {
328                 my $i = $_;
329                 my $pass = 1;
330                 foreach my $n ( keys %$filter_value ) {
331                         if ( ! exists $i->{$n} ) {
332                                 if ( defined $filter_value->{$n}->{_missing} ) {
333                                         $pass = 1;
334                                         next;
335                                 } else {
336                                         $pass = 0;
337                                         last;
338                                 }
339                         }
340                         # and match any of values in element
341                         my $have_values = 0;
342                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
343                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
344                         }
345                         if ( ! $have_values ) {
346                                 $pass = 0;
347                                 last;
348                         }
349                 }
350                 $pass;
351         } @items if $filter_value;
352         return @items;
353 }
354
355
356 sub _current_filters {
357         my $self = shift;
358         my $current_filters;
359         $current_filters->{ $_ } = $filters->{ $_ }
360                 foreach (
361                         grep { defined $filters->{ $_ } }
362                         @{ $self->_loaded('header') }
363                 );
364         #warn "# current_filters ",dump($current_filters);
365         return $current_filters;
366 }
367
368 sub _data_sorted_by {
369         my ( $self, $order ) = @_;
370
371         my $path = $self->session('path');
372
373         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
374                 return $loaded->{$path}->{sorted}->{$order};
375         }
376
377         my $data = $self->_loaded( 'data' );
378         my $numeric = $self->_is_numeric($order);
379         my $missing = $numeric ? 0 : '';
380         no warnings qw(numeric);
381         my $nr = 0;
382         my @sorted = map {
383                 $_->[0]
384         } sort {
385                 if ( $numeric ) {
386                         $a->[1] <=> $b->[1]
387                 } else {
388                         $a->[1] cmp $b->[1]
389                 }
390         } map {
391                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
392         } @{ $data->{items} }
393         ;
394
395         warn "sorted $order"; # ,dump( @sorted );
396
397         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
398 }
399
400
401 sub items {
402         my $self = shift;
403
404         my $path = $self->session('path');
405         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
406
407         my @columns = $self->_perm_array('columns');
408         $self->redirect_to('/data/columns') unless @columns;
409         my $order   = $self->_perm_scalar('order', $columns[0]);
410         my $sort    = $self->_perm_scalar('sort', 'a');
411         my $offset  = $self->_perm_scalar('offset', 0);
412         my $limit   = $self->_perm_scalar('limit', 20);
413         $self->_perm_scalar('show', 'table');
414
415         # fix offset when changing limit
416         $offset = int( $offset / $limit ) * $limit;
417
418         my $sorted = $self->_data_sorted_by( $order );
419
420         my @filter_names;
421         if ( $filters ) {
422                 foreach my $name ( keys %$filters ) {
423                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
424                                 warn "skip filter $name not found in $path\n";
425                                 next;
426                         }
427                         push @filter_names, $name;
428                 }
429                 warn "filter_names ",dump( @filter_names );
430                 foreach my $name ( @filter_names ) {
431                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
432                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
433                 }
434         }
435
436         my $filtered_names = join(' ',sort @filter_names);
437
438 #       warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
439
440         if ( ! defined $loaded->{$path}->{filtered}->{$filtered_names} ) {
441
442                 my $path_filters = $loaded->{$path}->{filters};
443
444                 warn "create combined filter for $filtered_names\n";
445
446                 my @filtered;
447                 foreach my $i ( 0 .. $#$sorted ) {
448                         my $pos = $sorted->[$i];
449
450                         if ( $#filter_names == -1 ) {
451                                 push @filtered, $pos;
452                                 next;
453                         }
454
455                         my $skip = 0;
456                         foreach ( @filter_names ) {
457                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
458                         }
459                         next if $skip;
460
461                         push @filtered, $pos;
462                 }
463
464                 $loaded->{$path}->{filtered}->{$filtered_names} = [ @filtered ];
465         }
466
467         my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
468                 if defined $loaded->{$path}->{filtered}->{$filtered_names};
469
470         warn "filters $filtered_names" if $filtered;
471
472         my $sorted_items;
473         my $data = $self->_loaded('data');
474         my $sort_start = $sort eq 'd' ? $#$filtered : 0;
475         foreach ( 0 .. $limit ) {
476                 my $i = $sort_start - ( $_ + $offset );
477                 last unless defined $filtered->[$i];
478                 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
479         }
480
481         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit";
482
483         $self->render(
484                 order => $order,
485                 offset => $offset,
486                 limit => $limit,
487                 sorted => $sorted_items,
488                 columns => [ @columns ],
489                 rows => $#$filtered + 1,
490                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
491                 filters => $self->_current_filters,
492         );
493
494 }
495
496
497 sub order {
498         my $self = shift;
499         $self->session('order', $self->param('order'));
500         $self->session('sort', $self->param('sort'));
501         $self->redirect_to('/data/items');
502 }
503
504 sub _is_numeric {
505         my ( $self, $name ) = @_;
506
507         my $stats = $self->_loaded( 'stats' );
508
509         # sort facet numerically if more >50% elements are numeric
510         my $count = $stats->{$name}->{count};
511         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
512         defined $stats->{$name}->{numeric} &&
513                 $stats->{$name}->{numeric} > $count / 2;
514 }
515
516 sub _remove_filter {
517         my ($self,$name) = @_;
518         warn "_remove_filter $name\n";
519
520         my $path = $self->session('path');
521
522         delete $filters->{$name};
523         delete $loaded->{$path}->{filters}->{$name};
524         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
525
526         foreach (
527                         grep { /\b$name\b/ }
528                         keys %{ $loaded->{$path}->{filtered} }
529         ) {
530                 delete $loaded->{$path}->{filtered}->{$_};
531                 warn "remove filtered cache $_";
532         }
533 }
534
535 sub facet {
536         my $self = shift;
537
538         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
539
540         if ( my $name = $self->param('remove') ) {
541                 $self->_remove_filter( $name );
542                 $self->redirect_to( '/data/items' );
543         }
544
545         my $facet;
546         my $name = $self->param('name') || die "no name";
547
548         my $all = $self->_perm_scalar('all', 1);
549         my $data = $self->_loaded('data');
550
551         my $filters = $self->_current_filters;
552         my $filtered_names = join(' ',sort keys %$filters);
553 #       warn "# filtered_names $filtered_names ", dump( $loaded->{$path}->{filtered}->{$filtered_names} );
554         my $filtered = $loaded->{$path}->{filtered}->{$filtered_names}
555                 if defined $loaded->{$path}->{filtered}->{$filtered_names};
556
557         if ( ! $filtered || $all ) {
558                 $filtered = [ 0 .. $#{ $data->{items} } ];
559                 warn "filter all values\n";
560         } else {
561                 warn "filter using $filtered_names\n";
562         }
563
564         foreach my $i ( @$filtered ) {
565                 my $item = $data->{items}->[$i];
566                 if ( ! exists $item->{$name} ) {
567                         $facet->{ _missing }++;
568                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
569                         $facet->{$_}++ foreach @{ $item->{$name} };
570                 } else {
571                         $facet->{ $item->{$name} }++;
572                 }
573         }
574
575 #       warn "# facet $name ",dump $facet;
576
577         my $checked;
578         my @facet_names =
579                   $all                      ? keys %$facet
580                 : defined $filters->{$name} ? @{ $filters->{$name} }
581                 : keys %$facet;
582
583         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
584
585         my $sort = $self->param('sort') || 'c';
586
587         # sort facet numerically if more >50% elements are numeric
588         my $numeric = $self->_is_numeric($name);
589
590         @facet_names = sort {
591                 if ( $sort =~ m/a/i ) {
592                         $numeric ? $a <=> $b : lc $a cmp lc $b;
593                 } elsif ( $sort =~ m/d/i ) {
594                         $numeric ? $b <=> $a : lc $b cmp lc $a;
595                 } elsif ( $sort =~ m/c/i ) {
596                         ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
597                 } else {
598                         warn "unknown sort: $sort";
599                         $a cmp $b;
600                 }
601         } @facet_names;
602
603         $self->render( name => $name, facet => $facet, checked => $checked,
604                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
605         );
606 }
607
608 sub edit {
609         my $self = shift;
610         my $content = $self->param('content');
611
612         $self->render(
613                 content => $content
614         );
615 }
616
617 1;