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