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