redirect to load when leeded
[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         my $full_path = $self->app->home->rel_file( 'data/' . $path );
51         die "$full_path $!" unless -r $full_path;
52
53         if ( defined $loaded->{$path}->{data} ) {
54                 my $mtime = (stat($full_path))[9];
55                 return if $loaded->{$path}->{mtime} == $mtime;
56                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
57         }
58
59         # we could use Mojo::JSON here, but it's too slow
60 #       $data = from_json read_file $path;
61         my $data = read_file $full_path;
62         warn "# data snippet: ", substr($data,0,200);
63         my @header;
64         if ( $path =~ m/\.js/ ) {
65                 Encode::_utf8_on($data);
66                 $data = from_json $data;
67         } elsif ( $path =~ m/\.txt/ ) {
68                 my @lines = split(/\r?\n/, $data);
69                 $data = { items => [] };
70
71                 my $header_line = shift @lines;
72                 my $multiline = $header_line =~ s/\^//g;
73                 @header = split(/\|/, $header_line );
74                 warn "# header ", dump( @header );
75                 while ( my $line = shift @lines ) {
76                         $line =~ s/\^//g;
77                         chomp $line;
78                         my @v = split(/\|/, $line);
79                         while ( @lines && $#v < $#header ) {
80                                 $line = $lines[0];
81                                 $line =~ s/\^//g;
82                                 chomp $line;
83                                 my @more_v = split(/\|/, $line);
84                                 if ( $#v + $#more_v > $#header ) {
85                                         warn "short line: ",dump( @v );
86                                         last;
87                                 }
88                                 shift @lines;
89                                 $v[ $#v ] .= shift @more_v if @more_v;
90                                 push @v, @more_v if @more_v;
91
92                                 if ( $#v > $#header ) {
93                                         die "# splice $#header ", dump( @v );
94                                         @v = splice @v, 0, $#header;
95                                 }
96                         }
97                         my $item;
98                         foreach my $i ( 0 .. $#v ) {
99                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
100                         }
101                         push @{ $data->{items} }, $item;
102                 }
103         } else {
104                 warn "file format unknown $path";
105         }
106
107         my $stats;
108
109         foreach my $e ( @{ $data->{items} } ) {
110                 foreach my $n ( keys %$e ) {
111                         $stats->{$n}->{count}++;
112                         my @v;
113                         if ( ref $e->{$n} eq 'ARRAY' ) {
114                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
115                                 @v = @{ $e->{$n} };
116                         } else {
117                                 @v = ( $e->{$n} );
118                         }
119
120                         foreach my $x ( @v ) {
121                                 $stats->{$n}->{numeric}++
122                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
123                                 $stats->{$n}->{empty}++
124                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
125                         }
126
127                 }
128         }
129
130         foreach my $n ( keys %$stats ) {
131                 next unless defined $stats->{$n}->{array};
132                 delete $stats->{$n}->{array}
133                         if $stats->{$n}->{array} == $stats->{$n}->{count};
134         }
135
136         if ( ! @header ) {
137                 if ( defined $data->{header} ) {
138                         if ( ref $data->{header} eq 'ARRAY' ) {
139                                 @header = @{ $data->{header} };
140                         } else {
141                                 warn "header not array ", dump( $data->{header} );
142                         }
143                 }
144         }
145
146         @header =
147                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
148                 grep { defined $stats->{$_}->{count} } keys %$stats
149                 unless @header;
150
151         warn dump($stats);
152
153         $loaded->{ $path } = {
154                 header => [ @header ],
155                 stats  => $stats,
156                 full_path => $full_path,
157                 size => -s $full_path,
158                 mtime => (stat($full_path))[9],
159                 data => $data,
160         };
161
162 }
163
164
165 sub load {
166         my $self = shift;
167
168         my @paths = $self->param('paths');
169         warn "# paths ", dump @paths;
170         $self->_load_path( $_ ) foreach @paths;
171
172         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
173         warn "# path $path\n";
174         $self->session('path' => $path);
175         $self->_load_path( $path );
176
177         $self->session( 'header' => $loaded->{$path}->{header} );
178         if ( ! defined $loaded->{$path}->{columns} ) {
179                 $self->session( 'columns' => $loaded->{$path}->{header} );
180                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
181                 $self->redirect_to( '/data/columns' );
182         } else {
183                 $self->session( 'columns' => $loaded->{$path}->{columns} );
184                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
185                 $self->redirect_to( '/data/items' );
186         }
187 }
188
189
190 sub _loaded {
191         my ( $self, $name ) = @_;
192         my $path = $self->session('path');
193         if ( ! defined $loaded->{$path}->{$name} ) {
194                 warn "$path $name doesn't exist in loaded ",dump( $loaded );
195                 $self->redirect_to('/data/index');
196         }
197         return $loaded->{$path}->{$name};
198 }
199
200
201 sub _checked {
202         my $self = shift;
203         my $checked;
204         $checked->{$_}++ foreach @_;
205 #       warn "# _checked ",dump($checked);
206         return $checked;
207 }
208
209
210 sub columns {
211     my $self = shift;
212
213         if ( $self->param('columns') ) {
214                 $self->_perm_array('columns');
215                 $self->redirect_to('/data/items');
216         }
217
218         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
219
220         my @columns;
221         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
222
223         foreach my $c ( @{ $self->session( 'header' ) } ) {
224                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
225         }
226
227     $self->render(
228                 message => 'Select columns to display',
229                 stats => $stats,
230                 columns => \@columns,
231                 checked => $self->_checked( $self->_perm_array('columns') ),
232         );
233 }
234
235 sub _perm_array {
236     my ($self,$name) = @_;
237
238         my @array = $self->param($name);
239         my $path  = $self->session('path');
240
241         if ( @array ) {
242                 $self->session($name => [ @array ]);
243                 $loaded->{$path}->{$name} = [ @array ];
244         } elsif ( my $session = $self->session($name) ) {
245                 if ( ref $session eq 'ARRAY' ) {
246                         @array = @$session;
247                 } else {
248                         die "$name not array ",dump($session);
249                 }
250         }
251         #warn "# $name ",dump @array;
252         return @array;
253 }
254
255 sub _perm_scalar {
256     my ($self,$name,$default) = @_;
257
258         my $scalar = $self->param($name);
259
260         if ( defined $scalar ) {
261                 $self->session($name => $scalar);
262         } else {
263                 $scalar = $self->session($name);
264         }
265
266         if ( ! defined $scalar ) {
267                 $scalar = $default;
268                 die "no default for $name" unless defined $scalar;
269                 $self->session($name => $scalar);
270         }
271
272         warn "# _perm_scalar $name ",dump $scalar;
273         return $scalar;
274 }
275
276 sub filter {
277         my $self = shift;
278
279         my $name = $self->param('filter_name') || die "name?";
280         my @vals = $self->param('filter_vals');
281
282         $self->_remove_filter( $name );
283         $self->_filter_on_data( $name, @vals ) if @vals;
284
285         $self->session( 'offset' => 0 );
286
287         $self->redirect_to('/data/items');
288 }
289
290 sub _filter_on_data {
291         my ( $self, $name, @vals ) = @_;
292
293         my $path = $self->session('path');
294
295         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
296                 warn "filter $name not found in data set";
297                 return;
298         }
299
300         $filters->{$name} = [ @vals ];
301         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
302
303         my $filter_hash;
304         $filter_hash->{$_}++ foreach @vals;
305
306         #warn "# filter_hash ",dump( $filter_hash );
307
308         my $items = $self->_loaded('data')->{items};
309
310         my $include_missing = defined $filter_hash->{_missing};
311         my $filtered_items;
312
313         foreach my $i ( 0 .. $#$items ) {
314
315                 if ( defined $items->[$i]->{$name} ) {
316                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
317                                 if ( defined $filter_hash->{ $v } ) {
318                                         $filtered_items->{$i}++;
319                                 }
320                         }
321                 } elsif ( $include_missing ) {
322                         $filtered_items->{$i}++;
323                 }
324         }
325
326         #warn "# filter $name ",dump($filtered_items);
327
328         $loaded->{$path}->{filters}->{$name} = $filtered_items;
329         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
330 }
331
332 sub _data_items {
333         my ( $self, $all ) = @_;
334         my $data = $self->_loaded( 'data' );
335
336         return @{ $data->{items} } if $all == 1;
337
338         my $filters = $self->_current_filters;
339         my $filter_value;
340         foreach my $f ( keys %$filters ) {
341                 foreach my $n ( @{ $filters->{$f} } ) {
342                         $filter_value->{$f}->{$n} = 1;
343                 }
344         }
345         my @items = @{ $data->{items} };
346         @items = grep {
347                 my $i = $_;
348                 my $pass = 1;
349                 foreach my $n ( keys %$filter_value ) {
350                         if ( ! exists $i->{$n} ) {
351                                 if ( defined $filter_value->{$n}->{_missing} ) {
352                                         $pass = 1;
353                                         next;
354                                 } else {
355                                         $pass = 0;
356                                         last;
357                                 }
358                         }
359                         # and match any of values in element
360                         my $have_values = 0;
361                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
362                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
363                         }
364                         if ( ! $have_values ) {
365                                 $pass = 0;
366                                 last;
367                         }
368                 }
369                 $pass;
370         } @items if $filter_value;
371         return @items;
372 }
373
374
375 sub _current_filters {
376         my $self = shift;
377         my $current_filters;
378         $current_filters->{ $_ } = $filters->{ $_ }
379                 foreach (
380                         grep { defined $filters->{ $_ } }
381                         @{ $self->_loaded('header') }
382                 );
383         #warn "# current_filters ",dump($current_filters);
384         return $current_filters;
385 }
386
387 sub _data_sorted_by {
388         my ( $self, $order ) = @_;
389
390         my $path = $self->session('path');
391
392         warn "_data_sorted_by $order from $path";
393
394         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
395                 return $loaded->{$path}->{sorted}->{$order};
396         }
397
398         my $data = $self->_loaded( 'data' );
399         my $numeric = $self->_is_numeric($order);
400         my $missing = $numeric ? 0 : '';
401         no warnings qw(numeric);
402         my $nr = 0;
403         my @sorted = map {
404                 $_->[0]
405         } sort {
406                 if ( $numeric ) {
407                         $a->[1] <=> $b->[1]
408                 } else {
409                         $a->[1] cmp $b->[1]
410                 }
411         } map {
412                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
413         } @{ $data->{items} }
414         ;
415
416         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
417         #warn "# sorted ",dump( @sorted );
418
419         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
420 }
421
422
423 sub items {
424         my $self = shift;
425
426         my $path = $self->session('path');
427         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
428
429         my @columns = $self->_perm_array('columns');
430         $self->redirect_to('/data/columns') unless @columns;
431         my $order   = $self->_perm_scalar('order', $columns[0]);
432         my $sort    = $self->_perm_scalar('sort', 'a');
433         my $offset  = $self->_perm_scalar('offset', 0);
434         my $limit   = $self->_perm_scalar('limit', 20);
435         $self->_perm_scalar('show', 'table');
436
437         # fix offset when changing limit
438         $offset = int( $offset / $limit ) * $limit;
439
440         my $sorted = $self->_data_sorted_by( $order );
441
442         my @filter_names;
443         if ( $filters ) {
444                 foreach my $name ( keys %$filters ) {
445                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
446                                 warn "skip filter $name not found in $path\n";
447                                 next;
448                         }
449                         push @filter_names, $name;
450                 }
451                 warn "filter_names ",dump( @filter_names );
452                 foreach my $name ( @filter_names ) {
453                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
454                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
455                 }
456         }
457
458         my $all_filters = join(' ',sort @filter_names,'order:',$order);
459
460 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
461
462         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
463
464                 my $path_filters = $loaded->{$path}->{filters};
465
466                 warn "create combined filter for $all_filters\n";
467
468                 my @filtered;
469                 foreach my $i ( 0 .. $#$sorted ) {
470                         my $pos = $sorted->[$i];
471
472                         if ( $#filter_names == -1 ) {
473                                 push @filtered, $pos;
474                                 next;
475                         }
476
477                         my $skip = 0;
478                         foreach ( @filter_names ) {
479                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
480                         }
481                         next if $skip;
482
483                         push @filtered, $pos;
484                 }
485
486                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
487         }
488
489         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
490                 if defined $loaded->{$path}->{filtered}->{$all_filters};
491
492         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
493
494         my $sorted_items;
495         my $data = $self->_loaded('data');
496         my $from_end = $sort eq 'd' ? $#$filtered : 0;
497         foreach ( 0 .. $limit ) {
498                 my $i = $_ + $offset;
499                 last unless defined $filtered->[$i];
500                 $i = $from_end - $i if $from_end;
501                 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
502         }
503
504         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
505
506         $self->render(
507                 order => $order,
508                 offset => $offset,
509                 limit => $limit,
510                 sorted => $sorted_items,
511                 columns => [ @columns ],
512                 rows => $#$filtered + 1,
513                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
514                 filters => $self->_current_filters,
515         );
516
517 }
518
519
520 sub order {
521         my $self = shift;
522         $self->session('order', $self->param('order'));
523         $self->session('sort', $self->param('sort'));
524         $self->redirect_to('/data/items');
525 }
526
527 sub _is_numeric {
528         my ( $self, $name ) = @_;
529
530         my $stats = $self->_loaded( 'stats' );
531
532         # sort facet numerically if more >50% elements are numeric
533         my $count = $stats->{$name}->{count};
534         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
535         defined $stats->{$name}->{numeric} &&
536                 $stats->{$name}->{numeric} > $count / 2;
537 }
538
539 sub _remove_filter {
540         my ($self,$name) = @_;
541         warn "_remove_filter $name\n";
542
543         my $path = $self->session('path');
544
545         delete $filters->{$name};
546         delete $loaded->{$path}->{filters}->{$name};
547         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
548
549         foreach (
550                         grep { /\b$name\b/ }
551                         keys %{ $loaded->{$path}->{filtered} }
552         ) {
553                 delete $loaded->{$path}->{filtered}->{$_};
554                 warn "remove filtered cache $_";
555         }
556 }
557
558 sub facet {
559         my $self = shift;
560
561         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
562
563         if ( my $name = $self->param('remove') ) {
564                 $self->_remove_filter( $name );
565                 $self->redirect_to( '/data/items' );
566         }
567
568         my $facet;
569         my $name = $self->param('name') || die "no name";
570
571         my $all = $self->_perm_scalar('all', 1);
572         my $data = $self->_loaded('data');
573
574         my $filters = $self->_current_filters;
575         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
576         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
577                 if defined $loaded->{$path}->{filtered}->{$all_filters};
578
579         if ( ! $filtered || $all ) {
580                 $filtered = [ 0 .. $#{ $data->{items} } ];
581                 warn "filter all values\n";
582         } else {
583                 warn "filter using $all_filters\n";
584         }
585
586         foreach my $i ( @$filtered ) {
587                 my $item = $data->{items}->[$i];
588                 if ( ! exists $item->{$name} ) {
589                         $facet->{ _missing }++;
590                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
591                         $facet->{$_}++ foreach @{ $item->{$name} };
592                 } else {
593                         $facet->{ $item->{$name} }++;
594                 }
595         }
596
597 #       warn "# facet $name ",dump $facet;
598
599         my $checked;
600         my @facet_names =
601                   $all                      ? keys %$facet
602                 : defined $filters->{$name} ? @{ $filters->{$name} }
603                 : keys %$facet;
604
605         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
606
607         my $numeric = $self->_is_numeric($name);
608
609         my $sort = $self->param('sort');
610         # sort numeric facets with more than 5 values ascending
611         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
612
613         @facet_names = sort {
614                 my $result;
615                 if ( $sort eq 'a' ) {
616                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
617                 } elsif ( $sort eq 'd' ) {
618                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
619                 } elsif ( $sort eq 'c' ) {
620                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
621                 } else {
622                         warn "unknown sort: $sort";
623                         $result = $a cmp $b;
624                 }
625                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
626                 $result;
627         } @facet_names;
628
629         $self->render( name => $name, facet => $facet, checked => $checked,
630                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
631         );
632 }
633
634 sub edit {
635         my $self = shift;
636         my $content = $self->param('content');
637
638         $self->render(
639                 content => $content
640         );
641 }
642
643 1;