e69c16e5fbac008e77afa35f88b6fbe7a2a862c5
[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                                 my $v = $v[$i];
100                                 # unpack numeric values separated by commas
101                                 my $a = $v =~ m/\d+\s*,\s*\d+/ ? [ split(/\,\s*/,$v) ] : [ $v ];
102                                 $item->{ $header[$i] || "f_$i" } = $a;
103                         }
104                         push @{ $data->{items} }, $item;
105                 }
106         } else {
107                 warn "file format unknown $path";
108         }
109
110         my $stats;
111
112         foreach my $e ( @{ $data->{items} } ) {
113                 foreach my $n ( keys %$e ) {
114                         $stats->{$n}->{count}++;
115                         my @v;
116                         if ( ref $e->{$n} eq 'ARRAY' ) {
117                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
118                                 @v = @{ $e->{$n} };
119                         } else {
120                                 @v = ( $e->{$n} );
121                         }
122
123                         foreach my $x ( @v ) {
124                                 $stats->{$n}->{numeric}++
125                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
126                                 $stats->{$n}->{empty}++
127                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
128                         }
129
130                 }
131         }
132
133         foreach my $n ( keys %$stats ) {
134                 next unless defined $stats->{$n}->{array};
135                 delete $stats->{$n}->{array}
136                         if $stats->{$n}->{array} == $stats->{$n}->{count};
137         }
138
139         if ( ! @header ) {
140                 if ( defined $data->{header} ) {
141                         if ( ref $data->{header} eq 'ARRAY' ) {
142                                 @header = @{ $data->{header} };
143                         } else {
144                                 warn "header not array ", dump( $data->{header} );
145                         }
146                 }
147         }
148
149         @header =
150                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
151                 grep { defined $stats->{$_}->{count} } keys %$stats
152                 unless @header;
153
154         warn dump($stats);
155
156         $loaded->{ $path } = {
157                 header => [ @header ],
158                 stats  => $stats,
159                 full_path => $full_path,
160                 size => -s $full_path,
161                 mtime => (stat($full_path))[9],
162                 data => $data,
163         };
164
165 }
166
167
168 sub load {
169         my $self = shift;
170
171         my @paths = $self->param('paths');
172         warn "# paths ", dump @paths;
173         $self->_load_path( $_ ) foreach @paths;
174
175         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
176         warn "# path $path\n";
177         $self->session('path' => $path);
178         $self->_load_path( $path );
179
180         $self->session( 'header' => $loaded->{$path}->{header} );
181         if ( ! defined $loaded->{$path}->{columns} ) {
182                 $self->session( 'columns' => $loaded->{$path}->{header} );
183                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
184                 $self->redirect_to( '/data/columns' );
185         } else {
186                 $self->session( 'columns' => $loaded->{$path}->{columns} );
187                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
188                 $self->redirect_to( '/data/items' );
189         }
190 }
191
192
193 sub _loaded {
194         my ( $self, $name ) = @_;
195         my $path = $self->session('path');
196         die "$path $name doesn't exist in loaded ",dump( $loaded )
197                 unless defined $loaded->{$path}->{$name};
198         return $loaded->{$path}->{$name};
199 }
200
201
202 sub _checked {
203         my $self = shift;
204         my $checked;
205         $checked->{$_}++ foreach @_;
206 #       warn "# _checked ",dump($checked);
207         return $checked;
208 }
209
210
211 sub columns {
212     my $self = shift;
213
214         if ( $self->param('columns') ) {
215                 $self->_perm_array('columns');
216                 $self->redirect_to('/data/items');
217         }
218
219         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
220
221         my @columns;
222         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
223
224         foreach my $c ( @{ $self->session( 'header' ) } ) {
225                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
226         }
227
228     $self->render(
229                 message => 'Select columns to display',
230                 stats => $stats,
231                 columns => \@columns,
232                 checked => $self->_checked( $self->_perm_array('columns') ),
233         );
234 }
235
236 sub _perm_array {
237     my ($self,$name) = @_;
238
239         my @array = $self->param($name);
240         my $path  = $self->session('path');
241
242         if ( @array ) {
243                 $self->session($name => [ @array ]);
244                 $loaded->{$path}->{$name} = [ @array ];
245         } elsif ( my $session = $self->session($name) ) {
246                 if ( ref $session eq 'ARRAY' ) {
247                         @array = @$session;
248                 } else {
249                         die "$name not array ",dump($session);
250                 }
251         }
252         #warn "# $name ",dump @array;
253         return @array;
254 }
255
256 sub _perm_scalar {
257     my ($self,$name,$default) = @_;
258
259         my $scalar = $self->param($name);
260
261         if ( defined $scalar ) {
262                 $self->session($name => $scalar);
263         } else {
264                 $scalar = $self->session($name);
265         }
266
267         if ( ! defined $scalar ) {
268                 $scalar = $default;
269                 die "no default for $name" unless defined $scalar;
270                 $self->session($name => $scalar);
271         }
272
273         warn "# _perm_scalar $name ",dump $scalar;
274         return $scalar;
275 }
276
277 sub filter {
278         my $self = shift;
279
280         my $name = $self->param('filter_name') || die "name?";
281         my @vals = $self->param('filter_vals');
282
283         $self->_remove_filter( $name );
284         $self->_filter_on_data( $name, @vals ) if @vals;
285
286         $self->session( 'offset' => 0 );
287
288         $self->redirect_to('/data/items');
289 }
290
291 sub _filter_on_data {
292         my ( $self, $name, @vals ) = @_;
293
294         my $path = $self->session('path');
295
296         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
297                 warn "filter $name not found in data set";
298                 return;
299         }
300
301         $filters->{$name} = [ @vals ];
302         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
303
304         my $filter_hash;
305         $filter_hash->{$_}++ foreach @vals;
306
307         #warn "# filter_hash ",dump( $filter_hash );
308
309         my $items = $self->_loaded('data')->{items};
310
311         my $include_missing = defined $filter_hash->{_missing};
312         my $filtered_items;
313
314         foreach my $i ( 0 .. $#$items ) {
315
316                 if ( defined $items->[$i]->{$name} ) {
317                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
318                                 if ( defined $filter_hash->{ $v } ) {
319                                         $filtered_items->{$i}++;
320                                 }
321                         }
322                 } elsif ( $include_missing ) {
323                         $filtered_items->{$i}++;
324                 }
325         }
326
327         #warn "# filter $name ",dump($filtered_items);
328
329         $loaded->{$path}->{filters}->{$name} = $filtered_items;
330         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
331 }
332
333 sub _data_items {
334         my ( $self, $all ) = @_;
335         my $data = $self->_loaded( 'data' );
336
337         return @{ $data->{items} } if $all == 1;
338
339         my $filters = $self->_current_filters;
340         my $filter_value;
341         foreach my $f ( keys %$filters ) {
342                 foreach my $n ( @{ $filters->{$f} } ) {
343                         $filter_value->{$f}->{$n} = 1;
344                 }
345         }
346         my @items = @{ $data->{items} };
347         @items = grep {
348                 my $i = $_;
349                 my $pass = 1;
350                 foreach my $n ( keys %$filter_value ) {
351                         if ( ! exists $i->{$n} ) {
352                                 if ( defined $filter_value->{$n}->{_missing} ) {
353                                         $pass = 1;
354                                         next;
355                                 } else {
356                                         $pass = 0;
357                                         last;
358                                 }
359                         }
360                         # and match any of values in element
361                         my $have_values = 0;
362                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
363                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
364                         }
365                         if ( ! $have_values ) {
366                                 $pass = 0;
367                                 last;
368                         }
369                 }
370                 $pass;
371         } @items if $filter_value;
372         return @items;
373 }
374
375
376 sub _current_filters {
377         my $self = shift;
378         my $current_filters;
379         $current_filters->{ $_ } = $filters->{ $_ }
380                 foreach (
381                         grep { defined $filters->{ $_ } }
382                         @{ $self->_loaded('header') }
383                 );
384         #warn "# current_filters ",dump($current_filters);
385         return $current_filters;
386 }
387
388 sub _data_sorted_by {
389         my ( $self, $order ) = @_;
390
391         my $path = $self->session('path');
392
393         warn "_data_sorted_by $order from $path";
394
395         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
396                 return $loaded->{$path}->{sorted}->{$order};
397         }
398
399         my $data = $self->_loaded( 'data' );
400         my $numeric = $self->_is_numeric($order);
401         my $missing = $numeric ? 0 : '';
402         no warnings qw(numeric);
403         my $nr = 0;
404         my @sorted = map {
405                 $_->[0]
406         } sort {
407                 if ( $numeric ) {
408                         $a->[1] <=> $b->[1]
409                 } else {
410                         $a->[1] cmp $b->[1]
411                 }
412         } map {
413                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
414         } @{ $data->{items} }
415         ;
416
417         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
418         #warn "# sorted ",dump( @sorted );
419
420         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
421 }
422
423
424 sub items {
425         my $self = shift;
426
427         my $path = $self->session('path');
428         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
429
430         my @columns = $self->_perm_array('columns');
431         $self->redirect_to('/data/columns') unless @columns;
432         my $order   = $self->_perm_scalar('order', $columns[0]);
433         my $sort    = $self->_perm_scalar('sort', 'a');
434         my $offset  = $self->_perm_scalar('offset', 0);
435         my $limit   = $self->_perm_scalar('limit', 20);
436         $self->_perm_scalar('show', 'table');
437
438         # fix offset when changing limit
439         $offset = int( $offset / $limit ) * $limit;
440
441         my $sorted = $self->_data_sorted_by( $order );
442
443         my @filter_names;
444         if ( $filters ) {
445                 foreach my $name ( keys %$filters ) {
446                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
447                                 warn "skip filter $name not found in $path\n";
448                                 next;
449                         }
450                         push @filter_names, $name;
451                 }
452                 warn "filter_names ",dump( @filter_names );
453                 foreach my $name ( @filter_names ) {
454                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
455                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
456                 }
457         }
458
459         my $all_filters = join(' ',sort @filter_names,'order:',$order);
460
461 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
462
463         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
464
465                 my $path_filters = $loaded->{$path}->{filters};
466
467                 warn "create combined filter for $all_filters\n";
468
469                 my @filtered;
470                 foreach my $i ( 0 .. $#$sorted ) {
471                         my $pos = $sorted->[$i];
472
473                         if ( $#filter_names == -1 ) {
474                                 push @filtered, $pos;
475                                 next;
476                         }
477
478                         my $skip = 0;
479                         foreach ( @filter_names ) {
480                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
481                         }
482                         next if $skip;
483
484                         push @filtered, $pos;
485                 }
486
487                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
488         }
489
490         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
491                 if defined $loaded->{$path}->{filtered}->{$all_filters};
492
493         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
494
495         my $sorted_items;
496         my $data = $self->_loaded('data');
497         my $from_end = $sort eq 'd' ? $#$filtered : 0;
498         foreach ( 0 .. $limit ) {
499                 my $i = $_ + $offset;
500                 last unless defined $filtered->[$i];
501                 $i = $from_end - $i if $from_end;
502                 push @$sorted_items, $data->{items}->[ $filtered->[$i] ];
503         }
504
505         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
506
507         $self->render(
508                 order => $order,
509                 offset => $offset,
510                 limit => $limit,
511                 sorted => $sorted_items,
512                 columns => [ @columns ],
513                 rows => $#$filtered + 1,
514                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
515                 filters => $self->_current_filters,
516         );
517
518 }
519
520
521 sub order {
522         my $self = shift;
523         $self->session('order', $self->param('order'));
524         $self->session('sort', $self->param('sort'));
525         $self->redirect_to('/data/items');
526 }
527
528 sub _is_numeric {
529         my ( $self, $name ) = @_;
530
531         my $stats = $self->_loaded( 'stats' );
532
533         # sort facet numerically if more >50% elements are numeric
534         my $count = $stats->{$name}->{count};
535         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
536         defined $stats->{$name}->{numeric} &&
537                 $stats->{$name}->{numeric} > $count / 2;
538 }
539
540 sub _remove_filter {
541         my ($self,$name) = @_;
542         warn "_remove_filter $name\n";
543
544         my $path = $self->session('path');
545
546         delete $filters->{$name};
547         delete $loaded->{$path}->{filters}->{$name};
548         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
549
550         foreach (
551                         grep { /\b$name\b/ }
552                         keys %{ $loaded->{$path}->{filtered} }
553         ) {
554                 delete $loaded->{$path}->{filtered}->{$_};
555                 warn "remove filtered cache $_";
556         }
557 }
558
559 sub facet {
560         my $self = shift;
561
562         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
563
564         if ( my $name = $self->param('remove') ) {
565                 $self->_remove_filter( $name );
566                 $self->redirect_to( '/data/items' );
567         }
568
569         my $facet;
570         my $name = $self->param('name') || die "no name";
571
572         my $all = $self->_perm_scalar('all', 1);
573         my $data = $self->_loaded('data');
574
575         my $filters = $self->_current_filters;
576         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
577         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
578                 if defined $loaded->{$path}->{filtered}->{$all_filters};
579
580         if ( ! $filtered || $all ) {
581                 $filtered = [ 0 .. $#{ $data->{items} } ];
582                 warn "filter all values\n";
583         } else {
584                 warn "filter using $all_filters\n";
585         }
586
587         foreach my $i ( @$filtered ) {
588                 my $item = $data->{items}->[$i];
589                 if ( ! exists $item->{$name} ) {
590                         $facet->{ _missing }++;
591                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
592                         $facet->{$_}++ foreach @{ $item->{$name} };
593                 } else {
594                         $facet->{ $item->{$name} }++;
595                 }
596         }
597
598 #       warn "# facet $name ",dump $facet;
599
600         my $checked;
601         my @facet_names =
602                   $all                      ? keys %$facet
603                 : defined $filters->{$name} ? @{ $filters->{$name} }
604                 : keys %$facet;
605
606         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
607
608         my $numeric = $self->_is_numeric($name);
609
610         my $sort = $self->param('sort');
611         # sort numeric facets with more than 5 values ascending
612         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
613
614         @facet_names = sort {
615                 my $result;
616                 if ( $sort eq 'a' ) {
617                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
618                 } elsif ( $sort eq 'd' ) {
619                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
620                 } elsif ( $sort eq 'c' ) {
621                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
622                 } else {
623                         warn "unknown sort: $sort";
624                         $result = $a cmp $b;
625                 }
626                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
627                 $result;
628         } @facet_names;
629
630         $self->render( name => $name, facet => $facet, checked => $checked,
631                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
632         );
633 }
634
635 sub edit {
636         my $self = shift;
637         my $content = $self->param('content');
638
639         $self->render(
640                 content => $content
641         );
642 }
643
644 1;