a5ae43f7b795da05c3e170f3360373067a2cb58a
[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         if ( ! @header ) {
131                 if ( defined $data->{header} ) {
132                         if ( ref $data->{header} eq 'ARRAY' ) {
133                                 @header = @{ $data->{header} };
134                         } else {
135                                 warn "header not array ", dump( $data->{header} );
136                         }
137                 }
138         }
139
140         @header =
141                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
142                 grep { defined $stats->{$_}->{count} } keys %$stats
143                 unless @header;
144
145         warn dump($stats);
146
147         $loaded->{ $path } = {
148                 header => [ @header ],
149                 stats  => $stats,
150                 full_path => $full_path,
151                 size => -s $full_path,
152                 data => $data,
153         };
154
155 }
156
157
158 sub load {
159         my $self = shift;
160
161         my @paths = $self->param('paths');
162         warn "# paths ", dump @paths;
163         $self->_load_path( $_ ) foreach @paths;
164
165         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
166         warn "# path $path\n";
167         $self->session('path' => $path);
168         $self->_load_path( $path );
169
170         $self->session( 'header' => $loaded->{$path}->{header} );
171         if ( ! defined $loaded->{$path}->{columns} ) {
172                 $self->session( 'columns' => $loaded->{$path}->{header} );
173                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
174                 $self->redirect_to( '/data/columns' );
175         } else {
176                 $self->session( 'columns' => $loaded->{$path}->{columns} );
177                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
178                 $self->redirect_to( '/data/items' );
179         }
180 }
181
182
183 sub _loaded {
184         my ( $self, $name ) = @_;
185         my $path = $self->session('path');
186         die "$path $name doesn't exist in loaded ",dump( $loaded )
187                 unless defined $loaded->{$path}->{$name};
188         return $loaded->{$path}->{$name};
189 }
190
191
192 sub _checked {
193         my $self = shift;
194         my $checked;
195         $checked->{$_}++ foreach @_;
196 #       warn "# _checked ",dump($checked);
197         return $checked;
198 }
199
200
201 sub columns {
202     my $self = shift;
203
204         if ( $self->param('columns') ) {
205                 $self->_perm_array('columns');
206                 $self->redirect_to('/data/items');
207         }
208
209         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
210
211         my @columns;
212         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
213
214         foreach my $c ( @{ $self->session( 'header' ) } ) {
215                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
216         }
217
218     $self->render(
219                 message => 'Select columns to display',
220                 stats => $stats,
221                 columns => \@columns,
222                 checked => $self->_checked( $self->_perm_array('columns') ),
223         );
224 }
225
226 sub _perm_array {
227     my ($self,$name) = @_;
228
229         my @array = $self->param($name);
230         my $path  = $self->session('path');
231
232         if ( @array ) {
233                 $self->session($name => [ @array ]);
234                 $loaded->{$path}->{$name} = [ @array ];
235         } elsif ( my $session = $self->session($name) ) {
236                 if ( ref $session eq 'ARRAY' ) {
237                         @array = @$session;
238                 } else {
239                         die "$name not array ",dump($session);
240                 }
241         }
242         #warn "# $name ",dump @array;
243         return @array;
244 }
245
246 sub _perm_scalar {
247     my ($self,$name,$default) = @_;
248
249         my $scalar = $self->param($name);
250
251         if ( defined $scalar ) {
252                 $self->session($name => $scalar);
253         } else {
254                 $scalar = $self->session($name);
255         }
256
257         if ( ! defined $scalar ) {
258                 $scalar = $default;
259                 die "no default for $name" unless defined $scalar;
260                 $self->session($name => $scalar);
261         }
262
263         warn "# _perm_scalar $name ",dump $scalar;
264         return $scalar;
265 }
266
267 sub filter {
268         my $self = shift;
269
270         my $name = $self->param('filter_name') || die "name?";
271         my @vals = $self->param('filter_vals');
272
273         $self->_remove_filter( $name );
274         $self->_filter_on_data( $name, @vals ) if @vals;
275
276         $self->session( 'offset' => 0 );
277
278         $self->redirect_to('/data/items');
279 }
280
281 sub _filter_on_data {
282         my ( $self, $name, @vals ) = @_;
283
284         my $path = $self->session('path');
285
286         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
287                 warn "filter $name not found in data set";
288                 return;
289         }
290
291         $filters->{$name} = [ @vals ];
292         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
293
294         my $filter_hash;
295         $filter_hash->{$_}++ foreach @vals;
296
297         #warn "# filter_hash ",dump( $filter_hash );
298
299         my $items = $self->_loaded('data')->{items};
300
301         my $include_missing = defined $filter_hash->{_missing};
302         my $filtered_items;
303
304         foreach my $i ( 0 .. $#$items ) {
305
306                 if ( defined $items->[$i]->{$name} ) {
307                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
308                                 if ( defined $filter_hash->{ $v } ) {
309                                         $filtered_items->{$i}++;
310                                 }
311                         }
312                 } elsif ( $include_missing ) {
313                         $filtered_items->{$i}++;
314                 }
315         }
316
317         #warn "# filter $name ",dump($filtered_items);
318
319         $loaded->{$path}->{filters}->{$name} = $filtered_items;
320         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
321 }
322
323 sub _data_items {
324         my ( $self, $all ) = @_;
325         my $data = $self->_loaded( 'data' );
326
327         return @{ $data->{items} } if $all == 1;
328
329         my $filters = $self->_current_filters;
330         my $filter_value;
331         foreach my $f ( keys %$filters ) {
332                 foreach my $n ( @{ $filters->{$f} } ) {
333                         $filter_value->{$f}->{$n} = 1;
334                 }
335         }
336         my @items = @{ $data->{items} };
337         @items = grep {
338                 my $i = $_;
339                 my $pass = 1;
340                 foreach my $n ( keys %$filter_value ) {
341                         if ( ! exists $i->{$n} ) {
342                                 if ( defined $filter_value->{$n}->{_missing} ) {
343                                         $pass = 1;
344                                         next;
345                                 } else {
346                                         $pass = 0;
347                                         last;
348                                 }
349                         }
350                         # and match any of values in element
351                         my $have_values = 0;
352                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
353                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
354                         }
355                         if ( ! $have_values ) {
356                                 $pass = 0;
357                                 last;
358                         }
359                 }
360                 $pass;
361         } @items if $filter_value;
362         return @items;
363 }
364
365
366 sub _current_filters {
367         my $self = shift;
368         my $current_filters;
369         $current_filters->{ $_ } = $filters->{ $_ }
370                 foreach (
371                         grep { defined $filters->{ $_ } }
372                         @{ $self->_loaded('header') }
373                 );
374         #warn "# current_filters ",dump($current_filters);
375         return $current_filters;
376 }
377
378 sub _data_sorted_by {
379         my ( $self, $order ) = @_;
380
381         my $path = $self->session('path');
382
383         warn "_data_sorted_by $order from $path";
384
385         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
386                 return $loaded->{$path}->{sorted}->{$order};
387         }
388
389         my $data = $self->_loaded( 'data' );
390         my $numeric = $self->_is_numeric($order);
391         my $missing = $numeric ? 0 : '';
392         no warnings qw(numeric);
393         my $nr = 0;
394         my @sorted = map {
395                 $_->[0]
396         } sort {
397                 if ( $numeric ) {
398                         $a->[1] <=> $b->[1]
399                 } else {
400                         $a->[1] cmp $b->[1]
401                 }
402         } map {
403                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
404         } @{ $data->{items} }
405         ;
406
407         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
408         #warn "# sorted ",dump( @sorted );
409
410         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
411 }
412
413
414 sub items {
415         my $self = shift;
416
417         my $path = $self->session('path');
418         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
419
420         my @columns = $self->_perm_array('columns');
421         $self->redirect_to('/data/columns') unless @columns;
422         my $order   = $self->_perm_scalar('order', $columns[0]);
423         my $sort    = $self->_perm_scalar('sort', 'a');
424         my $offset  = $self->_perm_scalar('offset', 0);
425         my $limit   = $self->_perm_scalar('limit', 20);
426         $self->_perm_scalar('show', 'table');
427
428         # fix offset when changing limit
429         $offset = int( $offset / $limit ) * $limit;
430
431         my $sorted = $self->_data_sorted_by( $order );
432
433         my @filter_names;
434         if ( $filters ) {
435                 foreach my $name ( keys %$filters ) {
436                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
437                                 warn "skip filter $name not found in $path\n";
438                                 next;
439                         }
440                         push @filter_names, $name;
441                 }
442                 warn "filter_names ",dump( @filter_names );
443                 foreach my $name ( @filter_names ) {
444                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
445                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
446                 }
447         }
448
449         my $all_filters = join(' ',sort @filter_names,'order:',$order);
450
451 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
452
453         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
454
455                 my $path_filters = $loaded->{$path}->{filters};
456
457                 warn "create combined filter for $all_filters\n";
458
459                 my @filtered;
460                 foreach my $i ( 0 .. $#$sorted ) {
461                         my $pos = $sorted->[$i];
462
463                         if ( $#filter_names == -1 ) {
464                                 push @filtered, $pos;
465                                 next;
466                         }
467
468                         my $skip = 0;
469                         foreach ( @filter_names ) {
470                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
471                         }
472                         next if $skip;
473
474                         push @filtered, $pos;
475                 }
476
477                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
478         }
479
480         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
481                 if defined $loaded->{$path}->{filtered}->{$all_filters};
482
483         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
484
485         my $sorted_items;
486         my $data = $self->_loaded('data');
487         my $from_end = $sort eq 'd' ? $#$filtered : 0;
488         foreach ( 0 .. $limit ) {
489                 my $i = $_ + $offset;
490                 last unless defined $filtered->[$i];
491                 $i = $from_end - $i if $from_end;
492                 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
493         }
494
495         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
496
497         $self->render(
498                 order => $order,
499                 offset => $offset,
500                 limit => $limit,
501                 sorted => $sorted_items,
502                 columns => [ @columns ],
503                 rows => $#$filtered + 1,
504                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
505                 filters => $self->_current_filters,
506         );
507
508 }
509
510
511 sub order {
512         my $self = shift;
513         $self->session('order', $self->param('order'));
514         $self->session('sort', $self->param('sort'));
515         $self->redirect_to('/data/items');
516 }
517
518 sub _is_numeric {
519         my ( $self, $name ) = @_;
520
521         my $stats = $self->_loaded( 'stats' );
522
523         # sort facet numerically if more >50% elements are numeric
524         my $count = $stats->{$name}->{count};
525         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
526         defined $stats->{$name}->{numeric} &&
527                 $stats->{$name}->{numeric} > $count / 2;
528 }
529
530 sub _remove_filter {
531         my ($self,$name) = @_;
532         warn "_remove_filter $name\n";
533
534         my $path = $self->session('path');
535
536         delete $filters->{$name};
537         delete $loaded->{$path}->{filters}->{$name};
538         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
539
540         foreach (
541                         grep { /\b$name\b/ }
542                         keys %{ $loaded->{$path}->{filtered} }
543         ) {
544                 delete $loaded->{$path}->{filtered}->{$_};
545                 warn "remove filtered cache $_";
546         }
547 }
548
549 sub facet {
550         my $self = shift;
551
552         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
553
554         if ( my $name = $self->param('remove') ) {
555                 $self->_remove_filter( $name );
556                 $self->redirect_to( '/data/items' );
557         }
558
559         my $facet;
560         my $name = $self->param('name') || die "no name";
561
562         my $all = $self->_perm_scalar('all', 1);
563         my $data = $self->_loaded('data');
564
565         my $filters = $self->_current_filters;
566         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
567         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
568                 if defined $loaded->{$path}->{filtered}->{$all_filters};
569
570         if ( ! $filtered || $all ) {
571                 $filtered = [ 0 .. $#{ $data->{items} } ];
572                 warn "filter all values\n";
573         } else {
574                 warn "filter using $all_filters\n";
575         }
576
577         foreach my $i ( @$filtered ) {
578                 my $item = $data->{items}->[$i];
579                 if ( ! exists $item->{$name} ) {
580                         $facet->{ _missing }++;
581                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
582                         $facet->{$_}++ foreach @{ $item->{$name} };
583                 } else {
584                         $facet->{ $item->{$name} }++;
585                 }
586         }
587
588 #       warn "# facet $name ",dump $facet;
589
590         my $checked;
591         my @facet_names =
592                   $all                      ? keys %$facet
593                 : defined $filters->{$name} ? @{ $filters->{$name} }
594                 : keys %$facet;
595
596         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
597
598         my $numeric = $self->_is_numeric($name);
599
600         my $sort = $self->param('sort');
601         $sort ||= $numeric ? 'a' : 'c';
602
603         @facet_names = sort {
604                 if ( $sort =~ m/a/i ) {
605                         $numeric ? $a <=> $b : lc $a cmp lc $b;
606                 } elsif ( $sort =~ m/d/i ) {
607                         $numeric ? $b <=> $a : lc $b cmp lc $a;
608                 } elsif ( $sort =~ m/c/i ) {
609                         ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
610                 } else {
611                         warn "unknown sort: $sort";
612                         $a cmp $b;
613                 }
614         } @facet_names;
615
616         $self->render( name => $name, facet => $facet, checked => $checked,
617                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
618         );
619 }
620
621 sub edit {
622         my $self = shift;
623         my $content = $self->param('content');
624
625         $self->render(
626                 content => $content
627         );
628 }
629
630 1;