a515b82fe0dbd5ea06e04a866e577d8f689bc09c
[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 use Storable;
15
16 our $loaded;
17 our $filters;
18
19 sub index {
20         my $self = shift;
21
22         my $path = $self->app->home->rel_dir('data');
23         die "no data dir $path" unless -d $path;
24
25         my @files;
26         find( sub {
27                 my $file = $File::Find::name;
28                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
29                         $file =~ s/$path\/*//;
30                         push @files, $file;
31                 } else {
32                         warn "IGNORE: $file\n";
33                 }
34         }, $path);
35
36         @files = sort { lc $a cmp lc $b } @files;
37         my $size;
38         $size->{$_} = -s "$path/$_" foreach @files;
39
40         if ( my $save_path = $self->session('save_path') ) {
41                 $self->session( 'save_path' => 0 )
42                 if ! defined $loaded->{$save_path};
43         }
44
45         $self->render(
46                 files => [ @files ],
47                 size => $size,
48                 loaded => $loaded,
49                 filters => $filters,
50                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
51         );
52 }
53
54 sub _dump_path {
55         my ( $self, $name ) = @_;
56         my $dir = $self->app->home->rel_dir('data');
57         $name =~ s/^$dir//;
58         $name =~ s/\/+/_/g;
59         return '/tmp/mojo_facets.' . $name . '.storable';
60 }
61
62 sub _save {
63         my ( $self, $path ) = @_;
64
65         my $dump_path = $self->_dump_path( $path );
66         warn "save loaded to $dump_path";
67         my $info = $loaded->{$path};
68         store $info, $dump_path;
69
70         # sync timestamp
71         #my $mtime = $loaded->{$path}->{mtime};
72         #utime $mtime, $mtime, $dump_path;
73
74         warn $dump_path, ' ', -s $dump_path, " bytes\n";
75         return $dump_path;
76 }
77
78 sub _load_path {
79         my ( $self, $path ) = @_;
80
81         my $full_path = $self->app->home->rel_file( 'data/' . $path );
82         die "$full_path $!" unless -r $full_path;
83
84         my $dump_path = $self->_dump_path( $path );
85
86         if ( defined $loaded->{$path}->{data} ) {
87                 my $mtime = (stat($full_path))[9];
88                 return if $loaded->{$path}->{mtime} == $mtime;
89                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
90         } elsif ( -e $dump_path ) {
91                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
92                 my $info = retrieve $dump_path;
93                 $loaded->{ $path } = $info;
94                 return;
95         }
96
97         # we could use Mojo::JSON here, but it's too slow
98 #       $data = from_json read_file $path;
99         my $data = read_file $full_path;
100         warn "# data snippet: ", substr($data,0,200);
101         my @header;
102         if ( $path =~ m/\.js/ ) {
103                 Encode::_utf8_on($data);
104                 $data = from_json $data;
105         } elsif ( $path =~ m/\.txt/ ) {
106                 my @lines = split(/\r?\n/, $data);
107                 $data = { items => [] };
108
109                 my $header_line = shift @lines;
110                 my $multiline = $header_line =~ s/\^//g;
111                 @header = split(/\|/, $header_line );
112                 warn "# header ", dump( @header );
113                 while ( my $line = shift @lines ) {
114                         $line =~ s/\^//g;
115                         chomp $line;
116                         my @v = split(/\|/, $line);
117                         while ( @lines && $#v < $#header ) {
118                                 $line = $lines[0];
119                                 $line =~ s/\^//g;
120                                 chomp $line;
121                                 my @more_v = split(/\|/, $line);
122                                 if ( $#v + $#more_v > $#header ) {
123                                         warn "short line: ",dump( @v );
124                                         last;
125                                 }
126                                 shift @lines;
127                                 $v[ $#v ] .= shift @more_v if @more_v;
128                                 push @v, @more_v if @more_v;
129
130                                 if ( $#v > $#header ) {
131                                         die "# splice $#header ", dump( @v );
132                                         @v = splice @v, 0, $#header;
133                                 }
134                         }
135                         my $item;
136                         foreach my $i ( 0 .. $#v ) {
137                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
138                         }
139                         push @{ $data->{items} }, $item;
140                 }
141         } else {
142                 warn "file format unknown $path";
143         }
144
145         my $stats;
146
147         foreach my $e ( @{ $data->{items} } ) {
148                 foreach my $n ( keys %$e ) {
149                         $stats->{$n}->{count}++;
150                         my @v;
151                         if ( ref $e->{$n} eq 'ARRAY' ) {
152                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
153                                 @v = @{ $e->{$n} };
154                         } else {
155                                 @v = ( $e->{$n} );
156                         }
157
158                         foreach my $x ( @v ) {
159                                 $stats->{$n}->{numeric}++
160                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
161                                 $stats->{$n}->{empty}++
162                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
163                         }
164
165                 }
166         }
167
168         foreach my $n ( keys %$stats ) {
169                 next unless defined $stats->{$n}->{array};
170                 delete $stats->{$n}->{array}
171                         if $stats->{$n}->{array} == $stats->{$n}->{count};
172         }
173
174         if ( ! @header ) {
175                 if ( defined $data->{header} ) {
176                         if ( ref $data->{header} eq 'ARRAY' ) {
177                                 @header = @{ $data->{header} };
178                         } else {
179                                 warn "header not array ", dump( $data->{header} );
180                         }
181                 }
182         }
183
184         @header =
185                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
186                 grep { defined $stats->{$_}->{count} } keys %$stats
187                 unless @header;
188
189         warn dump($stats);
190
191         my $info = {
192                 header => [ @header ],
193                 stats  => $stats,
194                 full_path => $full_path,
195                 size => -s $full_path,
196                 mtime => (stat($full_path))[9],
197                 data => $data,
198         };
199
200         $loaded->{ $path } = $info;
201         $self->_save( $path );
202
203 }
204
205
206 sub load {
207         my $self = shift;
208
209         my @paths = $self->param('paths');
210         warn "# paths ", dump @paths;
211         $self->_load_path( $_ ) foreach @paths;
212
213         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
214         warn "# path $path\n";
215         $self->session('path' => $path);
216         $self->_load_path( $path );
217
218         $self->session( 'header' => $loaded->{$path}->{header} );
219         if ( ! defined $loaded->{$path}->{columns} ) {
220                 $self->session( 'columns' => $loaded->{$path}->{header} );
221                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
222                 $self->redirect_to( '/data/columns' );
223         } else {
224                 $self->session( 'columns' => $loaded->{$path}->{columns} );
225                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
226                 $self->redirect_to( '/data/items' );
227         }
228 }
229
230
231 sub _loaded {
232         my ( $self, $name ) = @_;
233         my $path = $self->session('path');
234         $self->redirect_to('/data/index') unless $path;
235         if ( ! defined $loaded->{$path}->{$name} ) {
236                 warn "$path $name isn't loaded\n";
237                 $self->_load_path( $path );
238                 $self->redirect_to('/data/index')
239                         unless defined $loaded->{$path}->{$name};
240         }
241         return $loaded->{$path}->{$name};
242 }
243
244
245 sub _checked {
246         my $self = shift;
247         my $checked;
248         $checked->{$_}++ foreach @_;
249 #       warn "# _checked ",dump($checked);
250         return $checked;
251 }
252
253
254 sub columns {
255     my $self = shift;
256
257         if ( $self->param('columns') ) {
258                 $self->_perm_array('columns');
259                 $self->redirect_to('/data/items');
260         }
261
262         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
263
264         my @columns;
265         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
266
267         foreach my $c ( @{ $self->session( 'header' ) } ) {
268                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
269         }
270
271     $self->render(
272                 message => 'Select columns to display',
273                 stats => $stats,
274                 columns => \@columns,
275                 checked => $self->_checked( $self->_perm_array('columns') ),
276         );
277 }
278
279 sub _perm_array {
280     my ($self,$name) = @_;
281
282         my @array = $self->param($name);
283         my $path  = $self->session('path');
284
285         if ( @array ) {
286                 $self->session($name => [ @array ]);
287                 $loaded->{$path}->{$name} = [ @array ];
288         } elsif ( my $session = $self->session($name) ) {
289                 if ( ref $session eq 'ARRAY' ) {
290                         @array = @$session;
291                 } else {
292                         die "$name not array ",dump($session);
293                 }
294         }
295         #warn "# $name ",dump @array;
296         return @array;
297 }
298
299 sub _perm_scalar {
300     my ($self,$name,$default) = @_;
301
302         my $scalar = $self->param($name);
303
304         if ( defined $scalar ) {
305                 $self->session($name => $scalar);
306         } else {
307                 $scalar = $self->session($name);
308         }
309
310         if ( ! defined $scalar ) {
311                 $scalar = $default;
312                 die "no default for $name" unless defined $scalar;
313                 $self->session($name => $scalar);
314         }
315
316         warn "# _perm_scalar $name ",dump $scalar;
317         return $scalar;
318 }
319
320 sub filter {
321         my $self = shift;
322
323         my $name = $self->param('filter_name') || die "name?";
324         my @vals = $self->param('filter_vals');
325
326         $self->_remove_filter( $name );
327         $self->_filter_on_data( $name, @vals ) if @vals;
328
329         $self->session( 'offset' => 0 );
330
331         $self->redirect_to('/data/items');
332 }
333
334 sub _filter_on_data {
335         my ( $self, $name, @vals ) = @_;
336
337         my $path = $self->session('path');
338
339         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
340                 warn "filter $name not found in data set";
341                 return;
342         }
343
344         $filters->{$name} = [ @vals ];
345         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
346
347         my $filter_hash;
348         $filter_hash->{$_}++ foreach @vals;
349
350         #warn "# filter_hash ",dump( $filter_hash );
351
352         my $items = $self->_loaded('data')->{items};
353
354         my $include_missing = defined $filter_hash->{_missing};
355         my $filtered_items;
356
357         foreach my $i ( 0 .. $#$items ) {
358
359                 if ( defined $items->[$i]->{$name} ) {
360                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
361                                 if ( defined $filter_hash->{ $v } ) {
362                                         $filtered_items->{$i}++;
363                                 }
364                         }
365                 } elsif ( $include_missing ) {
366                         $filtered_items->{$i}++;
367                 }
368         }
369
370         #warn "# filter $name ",dump($filtered_items);
371
372         $loaded->{$path}->{filters}->{$name} = $filtered_items;
373         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
374 }
375
376 sub _data_items {
377         my ( $self, $all ) = @_;
378         my $data = $self->_loaded( 'data' );
379
380         return @{ $data->{items} } if $all == 1;
381
382         my $filters = $self->_current_filters;
383         my $filter_value;
384         foreach my $f ( keys %$filters ) {
385                 foreach my $n ( @{ $filters->{$f} } ) {
386                         $filter_value->{$f}->{$n} = 1;
387                 }
388         }
389         my @items = @{ $data->{items} };
390         @items = grep {
391                 my $i = $_;
392                 my $pass = 1;
393                 foreach my $n ( keys %$filter_value ) {
394                         if ( ! exists $i->{$n} ) {
395                                 if ( defined $filter_value->{$n}->{_missing} ) {
396                                         $pass = 1;
397                                         next;
398                                 } else {
399                                         $pass = 0;
400                                         last;
401                                 }
402                         }
403                         # and match any of values in element
404                         my $have_values = 0;
405                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
406                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
407                         }
408                         if ( ! $have_values ) {
409                                 $pass = 0;
410                                 last;
411                         }
412                 }
413                 $pass;
414         } @items if $filter_value;
415         return @items;
416 }
417
418
419 sub _current_filters {
420         my $self = shift;
421         my $current_filters;
422         $current_filters->{ $_ } = $filters->{ $_ }
423                 foreach (
424                         grep { defined $filters->{ $_ } }
425                         @{ $self->_loaded('header') }
426                 );
427         #warn "# current_filters ",dump($current_filters);
428         return $current_filters;
429 }
430
431 sub _data_sorted_by {
432         my ( $self, $order ) = @_;
433
434         my $path = $self->session('path');
435
436         warn "_data_sorted_by $order from $path";
437
438         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
439                 return $loaded->{$path}->{sorted}->{$order};
440         }
441
442         my $data = $self->_loaded( 'data' );
443         my $numeric = $self->_is_numeric($order);
444         my $missing = $numeric ? 0 : '';
445         no warnings qw(numeric);
446         my $nr = 0;
447         my @sorted = map {
448                 $_->[0]
449         } sort {
450                 if ( $numeric ) {
451                         $a->[1] <=> $b->[1]
452                 } else {
453                         $a->[1] cmp $b->[1]
454                 }
455         } map {
456                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
457         } @{ $data->{items} }
458         ;
459
460         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
461         #warn "# sorted ",dump( @sorted );
462
463         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
464 }
465
466
467 sub items {
468         my $self = shift;
469
470         my $path = $self->session('path');
471         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
472
473         my @columns = $self->_perm_array('columns');
474         $self->redirect_to('/data/columns') unless @columns;
475         my $order   = $self->_perm_scalar('order', $columns[0]);
476         my $sort    = $self->_perm_scalar('sort', 'a');
477         my $offset  = $self->_perm_scalar('offset', 0);
478         my $limit   = $self->_perm_scalar('limit', 20);
479         $self->_perm_scalar('show', 'table');
480
481         # fix offset when changing limit
482         $offset = int( $offset / $limit ) * $limit;
483
484         my $sorted = $self->_data_sorted_by( $order );
485
486         my @filter_names;
487         if ( $filters ) {
488                 foreach my $name ( keys %$filters ) {
489                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
490                                 warn "skip filter $name not found in $path\n";
491                                 next;
492                         }
493                         push @filter_names, $name;
494                 }
495                 warn "filter_names ",dump( @filter_names );
496                 foreach my $name ( @filter_names ) {
497                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
498                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
499                 }
500         }
501
502         my $all_filters = join(' ',sort @filter_names,'order:',$order);
503
504 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
505
506         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
507
508                 my $path_filters = $loaded->{$path}->{filters};
509
510                 warn "create combined filter for $all_filters\n";
511
512                 my @filtered;
513                 foreach my $i ( 0 .. $#$sorted ) {
514                         my $pos = $sorted->[$i];
515
516                         if ( $#filter_names == -1 ) {
517                                 push @filtered, $pos;
518                                 next;
519                         }
520
521                         my $skip = 0;
522                         foreach ( @filter_names ) {
523                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
524                         }
525                         next if $skip;
526
527                         push @filtered, $pos;
528                 }
529
530                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
531         }
532
533         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
534                 if defined $loaded->{$path}->{filtered}->{$all_filters};
535
536         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
537
538         my $sorted_items;
539         my $data = $self->_loaded('data');
540         my $from_end = $sort eq 'd' ? $#$filtered : 0;
541         foreach ( 0 .. $limit ) {
542                 my $i = $_ + $offset;
543                 last unless defined $filtered->[$i];
544                 $i = $from_end - $i if $from_end;
545                 my $id = $filtered->[$i];
546                 push @$sorted_items,
547                 my $item = $data->{items}->[ $id ];
548                 $item->{_row_id} ||= $id;
549         }
550
551         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
552
553         $self->render(
554                 order => $order,
555                 offset => $offset,
556                 limit => $limit,
557                 sorted => $sorted_items,
558                 columns => [ @columns ],
559                 rows => $#$filtered + 1,
560                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
561                 filters => $self->_current_filters,
562         );
563
564 }
565
566
567 sub order {
568         my $self = shift;
569         $self->session('order', $self->param('order'));
570         $self->session('sort', $self->param('sort'));
571         $self->redirect_to('/data/items');
572 }
573
574 sub _is_numeric {
575         my ( $self, $name ) = @_;
576
577         my $stats = $self->_loaded( 'stats' );
578
579         # sort facet numerically if more >50% elements are numeric
580         my $count = $stats->{$name}->{count};
581         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
582         defined $stats->{$name}->{numeric} &&
583                 $stats->{$name}->{numeric} > $count / 2;
584 }
585
586 sub _remove_filter {
587         my ($self,$name) = @_;
588         warn "_remove_filter $name\n";
589
590         my $path = $self->session('path');
591
592         delete $filters->{$name};
593         delete $loaded->{$path}->{filters}->{$name};
594         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
595
596         foreach (
597                         grep { /\b$name\b/ }
598                         keys %{ $loaded->{$path}->{filtered} }
599         ) {
600                 delete $loaded->{$path}->{filtered}->{$_};
601                 warn "remove filtered cache $_";
602         }
603 }
604
605 sub facet {
606         my $self = shift;
607
608         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
609
610         if ( my $name = $self->param('remove') ) {
611                 $self->_remove_filter( $name );
612                 $self->redirect_to( '/data/items' );
613         }
614
615         my $facet;
616         my $name = $self->param('name') || die "no name";
617
618         my $all = $self->_perm_scalar('all', 1);
619         my $data = $self->_loaded('data');
620
621         my $filters = $self->_current_filters;
622         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
623         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
624                 if defined $loaded->{$path}->{filtered}->{$all_filters};
625
626         if ( ! $filtered || $all ) {
627                 $filtered = [ 0 .. $#{ $data->{items} } ];
628                 warn "filter all values\n";
629         } else {
630                 warn "filter using $all_filters\n";
631         }
632
633         foreach my $i ( @$filtered ) {
634                 my $item = $data->{items}->[$i];
635                 if ( ! exists $item->{$name} ) {
636                         $facet->{ _missing }++;
637                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
638                         $facet->{$_}++ foreach @{ $item->{$name} };
639                 } else {
640                         $facet->{ $item->{$name} }++;
641                 }
642         }
643
644 #       warn "# facet $name ",dump $facet;
645
646         my $checked;
647         my @facet_names =
648                   $all                      ? keys %$facet
649                 : defined $filters->{$name} ? @{ $filters->{$name} }
650                 : keys %$facet;
651
652         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
653
654         my $numeric = $self->_is_numeric($name);
655
656         my $sort = $self->param('sort');
657         # sort numeric facets with more than 5 values ascending
658         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
659
660         @facet_names = sort {
661                 my $result;
662                 if ( $sort eq 'a' ) {
663                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
664                 } elsif ( $sort eq 'd' ) {
665                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
666                 } elsif ( $sort eq 'c' ) {
667                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
668                 } else {
669                         warn "unknown sort: $sort";
670                         $result = $a cmp $b;
671                 }
672                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
673                 $result;
674         } @facet_names;
675
676         $self->render( name => $name, facet => $facet, checked => $checked,
677                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
678         );
679 }
680
681 sub edit {
682         my $self = shift;
683         my $content = $self->param('content');
684
685         my $i = $self->param('_row_id');
686         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
687         my $path = $self->param('path') || die "no path";
688         my $name = $self->param('name') || die "no name";
689         my $status = 200; # 200 = OK, 201 = Created
690
691         my $data = $self->_loaded('data');
692
693         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
694                 $content =~ s/^\s+//s;
695                 $content =~ s/\s+$//s;
696                 my $v;
697                 if ( $content =~ /\xB6/ ) {     # para
698                         $v = [ split(/\s*\xB6\s*/, $content) ];
699                 } else {
700                         $v = [ $content ];
701                 }
702
703                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
704                 my $new = dump $v;
705                 if ( $old ne $new
706                         && ! ( $old eq 'undef' && length($content) == 0 ) # new value empty, previous undef
707                 ) {
708                         warn "# update $path $i $old -> $new\n";
709                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
710
711                         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
712                             delete $loaded->{$path}->{sorted}->{$name};
713                                 warn "# invalidate $path sorted $name\n";
714                         }
715
716                         foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
717                             delete $loaded->{$path}->{filtered}->{$_};
718                                 warn "# invalidate $path filtered $_\n";
719                         }
720
721                         $status = 201; # created
722                         $self->session('save_path' => $path);
723
724                 } else {
725                         warn "# unchanged $path $i $old\n";
726                         $status = 304;
727                 }
728         } else {
729                 $content = "$path $i $name doesn't exist\n";
730                 $status = 404;
731         }
732
733         warn "# edit $status $content";
734
735         $self->render(
736                 status => $status,
737                 content => $content,
738         );
739 }
740
741
742 sub save {
743         my $self = shift;
744         my $path = $self->param('path');
745         $path  ||= $self->session('path');
746
747         my $dump_path = $self->_save( $path );
748         $self->session('save_path' => 0);
749
750         $self->redirect_to( '/data/items' );
751 }
752
753 1;