return correct status codes and content
[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.' . $name . '.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         $self->redirect_to('/data/index') unless $path;
218         if ( ! defined $loaded->{$path}->{$name} ) {
219                 warn "$path $name doesn't exist in loaded ",dump( $loaded );
220                 $self->redirect_to('/data/index');
221         }
222         return $loaded->{$path}->{$name};
223 }
224
225
226 sub _checked {
227         my $self = shift;
228         my $checked;
229         $checked->{$_}++ foreach @_;
230 #       warn "# _checked ",dump($checked);
231         return $checked;
232 }
233
234
235 sub columns {
236     my $self = shift;
237
238         if ( $self->param('columns') ) {
239                 $self->_perm_array('columns');
240                 $self->redirect_to('/data/items');
241         }
242
243         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
244
245         my @columns;
246         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
247
248         foreach my $c ( @{ $self->session( 'header' ) } ) {
249                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
250         }
251
252     $self->render(
253                 message => 'Select columns to display',
254                 stats => $stats,
255                 columns => \@columns,
256                 checked => $self->_checked( $self->_perm_array('columns') ),
257         );
258 }
259
260 sub _perm_array {
261     my ($self,$name) = @_;
262
263         my @array = $self->param($name);
264         my $path  = $self->session('path');
265
266         if ( @array ) {
267                 $self->session($name => [ @array ]);
268                 $loaded->{$path}->{$name} = [ @array ];
269         } elsif ( my $session = $self->session($name) ) {
270                 if ( ref $session eq 'ARRAY' ) {
271                         @array = @$session;
272                 } else {
273                         die "$name not array ",dump($session);
274                 }
275         }
276         #warn "# $name ",dump @array;
277         return @array;
278 }
279
280 sub _perm_scalar {
281     my ($self,$name,$default) = @_;
282
283         my $scalar = $self->param($name);
284
285         if ( defined $scalar ) {
286                 $self->session($name => $scalar);
287         } else {
288                 $scalar = $self->session($name);
289         }
290
291         if ( ! defined $scalar ) {
292                 $scalar = $default;
293                 die "no default for $name" unless defined $scalar;
294                 $self->session($name => $scalar);
295         }
296
297         warn "# _perm_scalar $name ",dump $scalar;
298         return $scalar;
299 }
300
301 sub filter {
302         my $self = shift;
303
304         my $name = $self->param('filter_name') || die "name?";
305         my @vals = $self->param('filter_vals');
306
307         $self->_remove_filter( $name );
308         $self->_filter_on_data( $name, @vals ) if @vals;
309
310         $self->session( 'offset' => 0 );
311
312         $self->redirect_to('/data/items');
313 }
314
315 sub _filter_on_data {
316         my ( $self, $name, @vals ) = @_;
317
318         my $path = $self->session('path');
319
320         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
321                 warn "filter $name not found in data set";
322                 return;
323         }
324
325         $filters->{$name} = [ @vals ];
326         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
327
328         my $filter_hash;
329         $filter_hash->{$_}++ foreach @vals;
330
331         #warn "# filter_hash ",dump( $filter_hash );
332
333         my $items = $self->_loaded('data')->{items};
334
335         my $include_missing = defined $filter_hash->{_missing};
336         my $filtered_items;
337
338         foreach my $i ( 0 .. $#$items ) {
339
340                 if ( defined $items->[$i]->{$name} ) {
341                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
342                                 if ( defined $filter_hash->{ $v } ) {
343                                         $filtered_items->{$i}++;
344                                 }
345                         }
346                 } elsif ( $include_missing ) {
347                         $filtered_items->{$i}++;
348                 }
349         }
350
351         #warn "# filter $name ",dump($filtered_items);
352
353         $loaded->{$path}->{filters}->{$name} = $filtered_items;
354         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
355 }
356
357 sub _data_items {
358         my ( $self, $all ) = @_;
359         my $data = $self->_loaded( 'data' );
360
361         return @{ $data->{items} } if $all == 1;
362
363         my $filters = $self->_current_filters;
364         my $filter_value;
365         foreach my $f ( keys %$filters ) {
366                 foreach my $n ( @{ $filters->{$f} } ) {
367                         $filter_value->{$f}->{$n} = 1;
368                 }
369         }
370         my @items = @{ $data->{items} };
371         @items = grep {
372                 my $i = $_;
373                 my $pass = 1;
374                 foreach my $n ( keys %$filter_value ) {
375                         if ( ! exists $i->{$n} ) {
376                                 if ( defined $filter_value->{$n}->{_missing} ) {
377                                         $pass = 1;
378                                         next;
379                                 } else {
380                                         $pass = 0;
381                                         last;
382                                 }
383                         }
384                         # and match any of values in element
385                         my $have_values = 0;
386                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
387                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
388                         }
389                         if ( ! $have_values ) {
390                                 $pass = 0;
391                                 last;
392                         }
393                 }
394                 $pass;
395         } @items if $filter_value;
396         return @items;
397 }
398
399
400 sub _current_filters {
401         my $self = shift;
402         my $current_filters;
403         $current_filters->{ $_ } = $filters->{ $_ }
404                 foreach (
405                         grep { defined $filters->{ $_ } }
406                         @{ $self->_loaded('header') }
407                 );
408         #warn "# current_filters ",dump($current_filters);
409         return $current_filters;
410 }
411
412 sub _data_sorted_by {
413         my ( $self, $order ) = @_;
414
415         my $path = $self->session('path');
416
417         warn "_data_sorted_by $order from $path";
418
419         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
420                 return $loaded->{$path}->{sorted}->{$order};
421         }
422
423         my $data = $self->_loaded( 'data' );
424         my $numeric = $self->_is_numeric($order);
425         my $missing = $numeric ? 0 : '';
426         no warnings qw(numeric);
427         my $nr = 0;
428         my @sorted = map {
429                 $_->[0]
430         } sort {
431                 if ( $numeric ) {
432                         $a->[1] <=> $b->[1]
433                 } else {
434                         $a->[1] cmp $b->[1]
435                 }
436         } map {
437                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
438         } @{ $data->{items} }
439         ;
440
441         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
442         #warn "# sorted ",dump( @sorted );
443
444         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
445 }
446
447
448 sub items {
449         my $self = shift;
450
451         my $path = $self->session('path');
452         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
453
454         my @columns = $self->_perm_array('columns');
455         $self->redirect_to('/data/columns') unless @columns;
456         my $order   = $self->_perm_scalar('order', $columns[0]);
457         my $sort    = $self->_perm_scalar('sort', 'a');
458         my $offset  = $self->_perm_scalar('offset', 0);
459         my $limit   = $self->_perm_scalar('limit', 20);
460         $self->_perm_scalar('show', 'table');
461
462         # fix offset when changing limit
463         $offset = int( $offset / $limit ) * $limit;
464
465         my $sorted = $self->_data_sorted_by( $order );
466
467         my @filter_names;
468         if ( $filters ) {
469                 foreach my $name ( keys %$filters ) {
470                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
471                                 warn "skip filter $name not found in $path\n";
472                                 next;
473                         }
474                         push @filter_names, $name;
475                 }
476                 warn "filter_names ",dump( @filter_names );
477                 foreach my $name ( @filter_names ) {
478                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
479                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
480                 }
481         }
482
483         my $all_filters = join(' ',sort @filter_names,'order:',$order);
484
485 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
486
487         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
488
489                 my $path_filters = $loaded->{$path}->{filters};
490
491                 warn "create combined filter for $all_filters\n";
492
493                 my @filtered;
494                 foreach my $i ( 0 .. $#$sorted ) {
495                         my $pos = $sorted->[$i];
496
497                         if ( $#filter_names == -1 ) {
498                                 push @filtered, $pos;
499                                 next;
500                         }
501
502                         my $skip = 0;
503                         foreach ( @filter_names ) {
504                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
505                         }
506                         next if $skip;
507
508                         push @filtered, $pos;
509                 }
510
511                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
512         }
513
514         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
515                 if defined $loaded->{$path}->{filtered}->{$all_filters};
516
517         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
518
519         my $sorted_items;
520         my $data = $self->_loaded('data');
521         my $from_end = $sort eq 'd' ? $#$filtered : 0;
522         foreach ( 0 .. $limit ) {
523                 my $i = $_ + $offset;
524                 last unless defined $filtered->[$i];
525                 $i = $from_end - $i if $from_end;
526                 my $id = $filtered->[$i];
527                 push @$sorted_items,
528                 my $item = $data->{items}->[ $id ];
529                 $item->{_row_id} ||= $id;
530         }
531
532         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
533
534         $self->render(
535                 order => $order,
536                 offset => $offset,
537                 limit => $limit,
538                 sorted => $sorted_items,
539                 columns => [ @columns ],
540                 rows => $#$filtered + 1,
541                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
542                 filters => $self->_current_filters,
543         );
544
545 }
546
547
548 sub order {
549         my $self = shift;
550         $self->session('order', $self->param('order'));
551         $self->session('sort', $self->param('sort'));
552         $self->redirect_to('/data/items');
553 }
554
555 sub _is_numeric {
556         my ( $self, $name ) = @_;
557
558         my $stats = $self->_loaded( 'stats' );
559
560         # sort facet numerically if more >50% elements are numeric
561         my $count = $stats->{$name}->{count};
562         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
563         defined $stats->{$name}->{numeric} &&
564                 $stats->{$name}->{numeric} > $count / 2;
565 }
566
567 sub _remove_filter {
568         my ($self,$name) = @_;
569         warn "_remove_filter $name\n";
570
571         my $path = $self->session('path');
572
573         delete $filters->{$name};
574         delete $loaded->{$path}->{filters}->{$name};
575         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
576
577         foreach (
578                         grep { /\b$name\b/ }
579                         keys %{ $loaded->{$path}->{filtered} }
580         ) {
581                 delete $loaded->{$path}->{filtered}->{$_};
582                 warn "remove filtered cache $_";
583         }
584 }
585
586 sub facet {
587         my $self = shift;
588
589         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
590
591         if ( my $name = $self->param('remove') ) {
592                 $self->_remove_filter( $name );
593                 $self->redirect_to( '/data/items' );
594         }
595
596         my $facet;
597         my $name = $self->param('name') || die "no name";
598
599         my $all = $self->_perm_scalar('all', 1);
600         my $data = $self->_loaded('data');
601
602         my $filters = $self->_current_filters;
603         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
604         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
605                 if defined $loaded->{$path}->{filtered}->{$all_filters};
606
607         if ( ! $filtered || $all ) {
608                 $filtered = [ 0 .. $#{ $data->{items} } ];
609                 warn "filter all values\n";
610         } else {
611                 warn "filter using $all_filters\n";
612         }
613
614         foreach my $i ( @$filtered ) {
615                 my $item = $data->{items}->[$i];
616                 if ( ! exists $item->{$name} ) {
617                         $facet->{ _missing }++;
618                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
619                         $facet->{$_}++ foreach @{ $item->{$name} };
620                 } else {
621                         $facet->{ $item->{$name} }++;
622                 }
623         }
624
625 #       warn "# facet $name ",dump $facet;
626
627         my $checked;
628         my @facet_names =
629                   $all                      ? keys %$facet
630                 : defined $filters->{$name} ? @{ $filters->{$name} }
631                 : keys %$facet;
632
633         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
634
635         my $numeric = $self->_is_numeric($name);
636
637         my $sort = $self->param('sort');
638         # sort numeric facets with more than 5 values ascending
639         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
640
641         @facet_names = sort {
642                 my $result;
643                 if ( $sort eq 'a' ) {
644                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
645                 } elsif ( $sort eq 'd' ) {
646                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
647                 } elsif ( $sort eq 'c' ) {
648                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
649                 } else {
650                         warn "unknown sort: $sort";
651                         $result = $a cmp $b;
652                 }
653                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
654                 $result;
655         } @facet_names;
656
657         $self->render( name => $name, facet => $facet, checked => $checked,
658                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
659         );
660 }
661
662 sub edit {
663         my $self = shift;
664         my $content = $self->param('content');
665
666         my $i = $self->param('_row_id');
667         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
668         my $path = $self->param('path') || die "no path";
669         my $name = $self->param('name') || die "no name";
670         my $status = 200; # 200 = OK, 201 = Created
671
672         if ( defined $loaded->{$path}->{data}->{items}->[$i]->{$name} ) {
673                 $content =~ s/^\s+//s;
674                 $content =~ s/\s+$//s;
675                 my $v;
676                 if ( $content =~ /\xB6/ ) {     # para
677                         $v = [ split(/\s*\xB6\s*/, $content) ];
678                 } else {
679                         $v = [ $content ];
680                 }
681
682                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
683                 my $new = dump $v;
684                 if ( $old ne $new ) {
685                         warn "# update $path $i $old -> $new\n";
686                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
687
688                         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
689                             delete $loaded->{$path}->{sorted}->{$name};
690                                 warn "# invalidate $path sorted $name\n";
691                         }
692
693                         foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
694                             delete $loaded->{$path}->{filtered}->{$_};
695                                 warn "# invalidate $path filtered $_\n";
696                         }
697
698                         $status = 201; # created
699
700                 } else {
701                         warn "# unchanged $path $i $old\n";
702                         $status = 304;
703                 }
704         } else {
705                 $content = "$path $i $name doesn't exist\n";
706                 $status = 404;
707                 warn "# $content\n";
708         }
709
710         $self->render(
711                 status => $status,
712                 content => $content,
713         );
714 }
715
716 1;