move automatic loader into _loaded
[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         my $dump_path;
41         $dump_path->{$_} = $self->_dump_path( $_ ) foreach @files;
42
43         if ( my $save_path = $self->session('save_path') ) {
44                 $self->session( 'save_path' => 0 )
45                 if ! defined $loaded->{$save_path};
46         }
47
48         $self->render(
49                 files => [ @files ],
50                 size => $size,
51                 loaded => $loaded,
52                 filters => $filters,
53                 dump_path => $dump_path,
54         );
55 }
56
57 sub _dump_path {
58         my ( $self, $name ) = @_;
59         my $dir = $self->app->home->rel_dir('data');
60         $name =~ s/^$dir//;
61         $name =~ s/\/+/_/g;
62         return '/tmp/mojo_facets.' . $name . '.storable';
63 }
64
65 sub _save {
66         my ( $self, $path ) = @_;
67
68         my $dump_path = $self->_dump_path( $path );
69         warn "save loaded to $dump_path";
70         my $info = $loaded->{$path};
71         store $info, $dump_path;
72
73         # sync timestamp
74         #my $mtime = $loaded->{$path}->{mtime};
75         #utime $mtime, $mtime, $dump_path;
76
77         warn $dump_path, ' ', -s $dump_path, " bytes\n";
78         return $dump_path;
79 }
80
81 sub _load_path {
82         my ( $self, $path ) = @_;
83
84         my $full_path = $self->app->home->rel_file( 'data/' . $path );
85         die "$full_path $!" unless -r $full_path;
86
87         my $dump_path = $self->_dump_path( $path );
88
89         if ( defined $loaded->{$path}->{data} ) {
90                 my $mtime = (stat($full_path))[9];
91                 return if $loaded->{$path}->{mtime} == $mtime;
92                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
93         } elsif ( -e $dump_path ) {
94                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
95                 my $info = retrieve $dump_path;
96                 $loaded->{ $path } = $info;
97                 return;
98         }
99
100         # we could use Mojo::JSON here, but it's too slow
101 #       $data = from_json read_file $path;
102         my $data = read_file $full_path;
103         warn "# data snippet: ", substr($data,0,200);
104         my @header;
105         if ( $path =~ m/\.js/ ) {
106                 Encode::_utf8_on($data);
107                 $data = from_json $data;
108         } elsif ( $path =~ m/\.txt/ ) {
109                 my @lines = split(/\r?\n/, $data);
110                 $data = { items => [] };
111
112                 my $header_line = shift @lines;
113                 my $multiline = $header_line =~ s/\^//g;
114                 @header = split(/\|/, $header_line );
115                 warn "# header ", dump( @header );
116                 while ( my $line = shift @lines ) {
117                         $line =~ s/\^//g;
118                         chomp $line;
119                         my @v = split(/\|/, $line);
120                         while ( @lines && $#v < $#header ) {
121                                 $line = $lines[0];
122                                 $line =~ s/\^//g;
123                                 chomp $line;
124                                 my @more_v = split(/\|/, $line);
125                                 if ( $#v + $#more_v > $#header ) {
126                                         warn "short line: ",dump( @v );
127                                         last;
128                                 }
129                                 shift @lines;
130                                 $v[ $#v ] .= shift @more_v if @more_v;
131                                 push @v, @more_v if @more_v;
132
133                                 if ( $#v > $#header ) {
134                                         die "# splice $#header ", dump( @v );
135                                         @v = splice @v, 0, $#header;
136                                 }
137                         }
138                         my $item;
139                         foreach my $i ( 0 .. $#v ) {
140                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
141                         }
142                         push @{ $data->{items} }, $item;
143                 }
144         } else {
145                 warn "file format unknown $path";
146         }
147
148         my $stats;
149
150         foreach my $e ( @{ $data->{items} } ) {
151                 foreach my $n ( keys %$e ) {
152                         $stats->{$n}->{count}++;
153                         my @v;
154                         if ( ref $e->{$n} eq 'ARRAY' ) {
155                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
156                                 @v = @{ $e->{$n} };
157                         } else {
158                                 @v = ( $e->{$n} );
159                         }
160
161                         foreach my $x ( @v ) {
162                                 $stats->{$n}->{numeric}++
163                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
164                                 $stats->{$n}->{empty}++
165                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
166                         }
167
168                 }
169         }
170
171         foreach my $n ( keys %$stats ) {
172                 next unless defined $stats->{$n}->{array};
173                 delete $stats->{$n}->{array}
174                         if $stats->{$n}->{array} == $stats->{$n}->{count};
175         }
176
177         if ( ! @header ) {
178                 if ( defined $data->{header} ) {
179                         if ( ref $data->{header} eq 'ARRAY' ) {
180                                 @header = @{ $data->{header} };
181                         } else {
182                                 warn "header not array ", dump( $data->{header} );
183                         }
184                 }
185         }
186
187         @header =
188                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
189                 grep { defined $stats->{$_}->{count} } keys %$stats
190                 unless @header;
191
192         warn dump($stats);
193
194         my $info = {
195                 header => [ @header ],
196                 stats  => $stats,
197                 full_path => $full_path,
198                 size => -s $full_path,
199                 mtime => (stat($full_path))[9],
200                 data => $data,
201         };
202
203         $loaded->{ $path } = $info;
204         $self->_save( $path );
205
206 }
207
208
209 sub load {
210         my $self = shift;
211
212         my @paths = $self->param('paths');
213         warn "# paths ", dump @paths;
214         $self->_load_path( $_ ) foreach @paths;
215
216         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
217         warn "# path $path\n";
218         $self->session('path' => $path);
219         $self->_load_path( $path );
220
221         $self->session( 'header' => $loaded->{$path}->{header} );
222         if ( ! defined $loaded->{$path}->{columns} ) {
223                 $self->session( 'columns' => $loaded->{$path}->{header} );
224                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
225                 $self->redirect_to( '/data/columns' );
226         } else {
227                 $self->session( 'columns' => $loaded->{$path}->{columns} );
228                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
229                 $self->redirect_to( '/data/items' );
230         }
231 }
232
233
234 sub _loaded {
235         my ( $self, $name ) = @_;
236         my $path = $self->session('path');
237         $self->redirect_to('/data/index') unless $path;
238         if ( ! defined $loaded->{$path}->{$name} ) {
239                 warn "$path $name isn't loaded\n";
240                 $self->_load_path( $path );
241                 $self->redirect_to('/data/index')
242                         unless defined $loaded->{$path}->{$name};
243         }
244         return $loaded->{$path}->{$name};
245 }
246
247
248 sub _checked {
249         my $self = shift;
250         my $checked;
251         $checked->{$_}++ foreach @_;
252 #       warn "# _checked ",dump($checked);
253         return $checked;
254 }
255
256
257 sub columns {
258     my $self = shift;
259
260         if ( $self->param('columns') ) {
261                 $self->_perm_array('columns');
262                 $self->redirect_to('/data/items');
263         }
264
265         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
266
267         my @columns;
268         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
269
270         foreach my $c ( @{ $self->session( 'header' ) } ) {
271                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
272         }
273
274     $self->render(
275                 message => 'Select columns to display',
276                 stats => $stats,
277                 columns => \@columns,
278                 checked => $self->_checked( $self->_perm_array('columns') ),
279         );
280 }
281
282 sub _perm_array {
283     my ($self,$name) = @_;
284
285         my @array = $self->param($name);
286         my $path  = $self->session('path');
287
288         if ( @array ) {
289                 $self->session($name => [ @array ]);
290                 $loaded->{$path}->{$name} = [ @array ];
291         } elsif ( my $session = $self->session($name) ) {
292                 if ( ref $session eq 'ARRAY' ) {
293                         @array = @$session;
294                 } else {
295                         die "$name not array ",dump($session);
296                 }
297         }
298         #warn "# $name ",dump @array;
299         return @array;
300 }
301
302 sub _perm_scalar {
303     my ($self,$name,$default) = @_;
304
305         my $scalar = $self->param($name);
306
307         if ( defined $scalar ) {
308                 $self->session($name => $scalar);
309         } else {
310                 $scalar = $self->session($name);
311         }
312
313         if ( ! defined $scalar ) {
314                 $scalar = $default;
315                 die "no default for $name" unless defined $scalar;
316                 $self->session($name => $scalar);
317         }
318
319         warn "# _perm_scalar $name ",dump $scalar;
320         return $scalar;
321 }
322
323 sub filter {
324         my $self = shift;
325
326         my $name = $self->param('filter_name') || die "name?";
327         my @vals = $self->param('filter_vals');
328
329         $self->_remove_filter( $name );
330         $self->_filter_on_data( $name, @vals ) if @vals;
331
332         $self->session( 'offset' => 0 );
333
334         $self->redirect_to('/data/items');
335 }
336
337 sub _filter_on_data {
338         my ( $self, $name, @vals ) = @_;
339
340         my $path = $self->session('path');
341
342         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
343                 warn "filter $name not found in data set";
344                 return;
345         }
346
347         $filters->{$name} = [ @vals ];
348         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
349
350         my $filter_hash;
351         $filter_hash->{$_}++ foreach @vals;
352
353         #warn "# filter_hash ",dump( $filter_hash );
354
355         my $items = $self->_loaded('data')->{items};
356
357         my $include_missing = defined $filter_hash->{_missing};
358         my $filtered_items;
359
360         foreach my $i ( 0 .. $#$items ) {
361
362                 if ( defined $items->[$i]->{$name} ) {
363                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
364                                 if ( defined $filter_hash->{ $v } ) {
365                                         $filtered_items->{$i}++;
366                                 }
367                         }
368                 } elsif ( $include_missing ) {
369                         $filtered_items->{$i}++;
370                 }
371         }
372
373         #warn "# filter $name ",dump($filtered_items);
374
375         $loaded->{$path}->{filters}->{$name} = $filtered_items;
376         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
377 }
378
379 sub _data_items {
380         my ( $self, $all ) = @_;
381         my $data = $self->_loaded( 'data' );
382
383         return @{ $data->{items} } if $all == 1;
384
385         my $filters = $self->_current_filters;
386         my $filter_value;
387         foreach my $f ( keys %$filters ) {
388                 foreach my $n ( @{ $filters->{$f} } ) {
389                         $filter_value->{$f}->{$n} = 1;
390                 }
391         }
392         my @items = @{ $data->{items} };
393         @items = grep {
394                 my $i = $_;
395                 my $pass = 1;
396                 foreach my $n ( keys %$filter_value ) {
397                         if ( ! exists $i->{$n} ) {
398                                 if ( defined $filter_value->{$n}->{_missing} ) {
399                                         $pass = 1;
400                                         next;
401                                 } else {
402                                         $pass = 0;
403                                         last;
404                                 }
405                         }
406                         # and match any of values in element
407                         my $have_values = 0;
408                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
409                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
410                         }
411                         if ( ! $have_values ) {
412                                 $pass = 0;
413                                 last;
414                         }
415                 }
416                 $pass;
417         } @items if $filter_value;
418         return @items;
419 }
420
421
422 sub _current_filters {
423         my $self = shift;
424         my $current_filters;
425         $current_filters->{ $_ } = $filters->{ $_ }
426                 foreach (
427                         grep { defined $filters->{ $_ } }
428                         @{ $self->_loaded('header') }
429                 );
430         #warn "# current_filters ",dump($current_filters);
431         return $current_filters;
432 }
433
434 sub _data_sorted_by {
435         my ( $self, $order ) = @_;
436
437         my $path = $self->session('path');
438
439         warn "_data_sorted_by $order from $path";
440
441         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
442                 return $loaded->{$path}->{sorted}->{$order};
443         }
444
445         my $data = $self->_loaded( 'data' );
446         my $numeric = $self->_is_numeric($order);
447         my $missing = $numeric ? 0 : '';
448         no warnings qw(numeric);
449         my $nr = 0;
450         my @sorted = map {
451                 $_->[0]
452         } sort {
453                 if ( $numeric ) {
454                         $a->[1] <=> $b->[1]
455                 } else {
456                         $a->[1] cmp $b->[1]
457                 }
458         } map {
459                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
460         } @{ $data->{items} }
461         ;
462
463         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
464         #warn "# sorted ",dump( @sorted );
465
466         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
467 }
468
469
470 sub items {
471         my $self = shift;
472
473         my $path = $self->session('path');
474         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
475
476         my @columns = $self->_perm_array('columns');
477         $self->redirect_to('/data/columns') unless @columns;
478         my $order   = $self->_perm_scalar('order', $columns[0]);
479         my $sort    = $self->_perm_scalar('sort', 'a');
480         my $offset  = $self->_perm_scalar('offset', 0);
481         my $limit   = $self->_perm_scalar('limit', 20);
482         $self->_perm_scalar('show', 'table');
483
484         # fix offset when changing limit
485         $offset = int( $offset / $limit ) * $limit;
486
487         my $sorted = $self->_data_sorted_by( $order );
488
489         my @filter_names;
490         if ( $filters ) {
491                 foreach my $name ( keys %$filters ) {
492                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
493                                 warn "skip filter $name not found in $path\n";
494                                 next;
495                         }
496                         push @filter_names, $name;
497                 }
498                 warn "filter_names ",dump( @filter_names );
499                 foreach my $name ( @filter_names ) {
500                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
501                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
502                 }
503         }
504
505         my $all_filters = join(' ',sort @filter_names,'order:',$order);
506
507 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
508
509         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
510
511                 my $path_filters = $loaded->{$path}->{filters};
512
513                 warn "create combined filter for $all_filters\n";
514
515                 my @filtered;
516                 foreach my $i ( 0 .. $#$sorted ) {
517                         my $pos = $sorted->[$i];
518
519                         if ( $#filter_names == -1 ) {
520                                 push @filtered, $pos;
521                                 next;
522                         }
523
524                         my $skip = 0;
525                         foreach ( @filter_names ) {
526                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
527                         }
528                         next if $skip;
529
530                         push @filtered, $pos;
531                 }
532
533                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
534         }
535
536         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
537                 if defined $loaded->{$path}->{filtered}->{$all_filters};
538
539         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
540
541         my $sorted_items;
542         my $data = $self->_loaded('data');
543         my $from_end = $sort eq 'd' ? $#$filtered : 0;
544         foreach ( 0 .. $limit ) {
545                 my $i = $_ + $offset;
546                 last unless defined $filtered->[$i];
547                 $i = $from_end - $i if $from_end;
548                 my $id = $filtered->[$i];
549                 push @$sorted_items,
550                 my $item = $data->{items}->[ $id ];
551                 $item->{_row_id} ||= $id;
552         }
553
554         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
555
556         $self->render(
557                 order => $order,
558                 offset => $offset,
559                 limit => $limit,
560                 sorted => $sorted_items,
561                 columns => [ @columns ],
562                 rows => $#$filtered + 1,
563                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
564                 filters => $self->_current_filters,
565         );
566
567 }
568
569
570 sub order {
571         my $self = shift;
572         $self->session('order', $self->param('order'));
573         $self->session('sort', $self->param('sort'));
574         $self->redirect_to('/data/items');
575 }
576
577 sub _is_numeric {
578         my ( $self, $name ) = @_;
579
580         my $stats = $self->_loaded( 'stats' );
581
582         # sort facet numerically if more >50% elements are numeric
583         my $count = $stats->{$name}->{count};
584         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
585         defined $stats->{$name}->{numeric} &&
586                 $stats->{$name}->{numeric} > $count / 2;
587 }
588
589 sub _remove_filter {
590         my ($self,$name) = @_;
591         warn "_remove_filter $name\n";
592
593         my $path = $self->session('path');
594
595         delete $filters->{$name};
596         delete $loaded->{$path}->{filters}->{$name};
597         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
598
599         foreach (
600                         grep { /\b$name\b/ }
601                         keys %{ $loaded->{$path}->{filtered} }
602         ) {
603                 delete $loaded->{$path}->{filtered}->{$_};
604                 warn "remove filtered cache $_";
605         }
606 }
607
608 sub facet {
609         my $self = shift;
610
611         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
612
613         if ( my $name = $self->param('remove') ) {
614                 $self->_remove_filter( $name );
615                 $self->redirect_to( '/data/items' );
616         }
617
618         my $facet;
619         my $name = $self->param('name') || die "no name";
620
621         my $all = $self->_perm_scalar('all', 1);
622         my $data = $self->_loaded('data');
623
624         my $filters = $self->_current_filters;
625         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
626         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
627                 if defined $loaded->{$path}->{filtered}->{$all_filters};
628
629         if ( ! $filtered || $all ) {
630                 $filtered = [ 0 .. $#{ $data->{items} } ];
631                 warn "filter all values\n";
632         } else {
633                 warn "filter using $all_filters\n";
634         }
635
636         foreach my $i ( @$filtered ) {
637                 my $item = $data->{items}->[$i];
638                 if ( ! exists $item->{$name} ) {
639                         $facet->{ _missing }++;
640                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
641                         $facet->{$_}++ foreach @{ $item->{$name} };
642                 } else {
643                         $facet->{ $item->{$name} }++;
644                 }
645         }
646
647 #       warn "# facet $name ",dump $facet;
648
649         my $checked;
650         my @facet_names =
651                   $all                      ? keys %$facet
652                 : defined $filters->{$name} ? @{ $filters->{$name} }
653                 : keys %$facet;
654
655         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
656
657         my $numeric = $self->_is_numeric($name);
658
659         my $sort = $self->param('sort');
660         # sort numeric facets with more than 5 values ascending
661         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
662
663         @facet_names = sort {
664                 my $result;
665                 if ( $sort eq 'a' ) {
666                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
667                 } elsif ( $sort eq 'd' ) {
668                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
669                 } elsif ( $sort eq 'c' ) {
670                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
671                 } else {
672                         warn "unknown sort: $sort";
673                         $result = $a cmp $b;
674                 }
675                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
676                 $result;
677         } @facet_names;
678
679         $self->render( name => $name, facet => $facet, checked => $checked,
680                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
681         );
682 }
683
684 sub edit {
685         my $self = shift;
686         my $content = $self->param('content');
687
688         my $i = $self->param('_row_id');
689         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
690         my $path = $self->param('path') || die "no path";
691         my $name = $self->param('name') || die "no name";
692         my $status = 200; # 200 = OK, 201 = Created
693
694         my $data = $self->_loaded('data');
695
696         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
697                 $content =~ s/^\s+//s;
698                 $content =~ s/\s+$//s;
699                 my $v;
700                 if ( $content =~ /\xB6/ ) {     # para
701                         $v = [ split(/\s*\xB6\s*/, $content) ];
702                 } else {
703                         $v = [ $content ];
704                 }
705
706                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
707                 my $new = dump $v;
708                 if ( $old ne $new
709                         && ! ( $old eq 'undef' && length($content) == 0 ) # new value empty, previous undef
710                 ) {
711                         warn "# update $path $i $old -> $new\n";
712                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
713
714                         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
715                             delete $loaded->{$path}->{sorted}->{$name};
716                                 warn "# invalidate $path sorted $name\n";
717                         }
718
719                         foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
720                             delete $loaded->{$path}->{filtered}->{$_};
721                                 warn "# invalidate $path filtered $_\n";
722                         }
723
724                         $status = 201; # created
725                         $self->session('save_path' => $path);
726
727                 } else {
728                         warn "# unchanged $path $i $old\n";
729                         $status = 304;
730                 }
731         } else {
732                 $content = "$path $i $name doesn't exist\n";
733                 $status = 404;
734         }
735
736         warn "# edit $status $content";
737
738         $self->render(
739                 status => $status,
740                 content => $content,
741         );
742 }
743
744
745 sub save {
746         my $self = shift;
747         my $path = $self->param('path');
748         $path  ||= $self->session('path');
749
750         my $dump_path = $self->_save( $path );
751         $self->session('save_path' => 0);
752
753         $self->redirect_to( '/data/items' );
754 }
755
756 1;