commit $out changes into new dataset
[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 Encode;
11 use locale;
12 use File::Find;
13 use Storable;
14 use Time::HiRes qw(time);
15 use File::Path qw(mkpath);
16
17 use MojoFacets::Import::File;
18 use MojoFacets::Import::HTMLTable;
19
20 our $loaded;
21 our $filters;
22
23 sub index {
24         my $self = shift;
25
26         my $data_dir = $self->app->home->rel_dir('data');
27         die "no data dir $data_dir" unless -d $data_dir;
28
29         my @files;
30         my $changes;
31         find( sub {
32                 my $file = $File::Find::name;
33                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
34                         $file =~ s/$data_dir\/*//;
35                         push @files, $file;
36                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
37                         push @{ $changes->{$1} }, $2
38                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
39                         $file =~ s/$data_dir\/*//;
40                         push @files, $file;
41                 } else {
42                         #warn "IGNORE: $file\n";
43                 }
44         }, $data_dir);
45
46         @files = sort { lc $a cmp lc $b } @files;
47         unshift @files, grep { defined $loaded->{$_}->{generated} } keys %$loaded;
48         my $size;
49         $size->{$_} = -s "$data_dir/$_" foreach @files;
50
51         $self->render(
52                 files => [ @files ],
53                 size => $size,
54                 loaded => $loaded,
55                 filters => $filters,
56                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
57                 changes => $changes,
58         );
59 }
60
61 sub _dump_path {
62         my ( $self, $name ) = @_;
63         my $dir = $self->app->home->rel_dir('data');
64         $name =~ s/^$dir//;
65         $name =~ s/\/+/_/g;
66         return '/tmp/mojo_facets.' . $name . '.storable';
67 }
68
69 sub _save {
70         my ( $self, $path ) = @_;
71
72         my $dump_path = $self->_dump_path( $path );
73         my $first_load = ! -e $dump_path;
74         warn "save loaded to $dump_path";
75         my $info = $loaded->{$path};
76         store $info, $dump_path;
77
78         if ( $first_load ) {
79                 my $mtime = $loaded->{$path}->{mtime};
80                 utime $mtime, $mtime, $dump_path;
81                 warn "sync time to $path at $mtime\n";
82         }
83
84         warn $dump_path, ' ', -s $dump_path, " bytes\n";
85         return $dump_path;
86 }
87
88
89 sub __stats {
90
91         my $stats;
92
93         my $nr_items = $#{ $_[0] } + 1;
94
95         warn "__stats $nr_items\n";
96
97         foreach my $e ( @{ $_[0] } ) {
98                 foreach my $n ( keys %$e ) {
99                         $stats->{$n}->{count}++;
100                         my @v;
101                         if ( ref $e->{$n} eq 'ARRAY' ) {
102                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
103                                 @v = @{ $e->{$n} };
104                         } else {
105                                 @v = ( $e->{$n} );
106                         }
107
108                         foreach my $x ( @v ) {
109                                 $stats->{$n}->{numeric}++
110                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
111                                 $stats->{$n}->{empty}++
112                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
113                         }
114
115                 }
116         }
117
118         foreach my $n ( keys %$stats ) {
119                 my $s = $stats->{$n};
120                 next unless defined $s->{array};
121                 if ( $s->{array} == $s->{count} ) {
122                         delete $s->{array};
123                         if ( $s->{count} == $nr_items ) {
124                                 warn "check $n for uniqeness\n";
125                                 my $unique;
126                                 foreach my $e ( @{ $_[0] } ) {
127                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
128                                                 $unique = 0;
129                                                 last;
130                                         }
131                                 }
132                                 if ( $unique ) {
133                                         $stats->{$n}->{unique} = 1;
134                                         #warn "# $n unique ",dump( $unique );
135                                 }
136                         }
137                 }
138         }
139
140         warn "# __stats ",dump($stats);
141
142         return $stats;
143 }
144
145 sub _param_or_session {
146         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
147 }
148
149 sub stats {
150         my $self = shift;
151         my $path = $self->_param_or_session('path');
152         warn "stats $path\n";
153         delete $loaded->{$path}->{stats};
154         $self->redirect_to( '/data/columns' );
155 }
156
157
158 sub _load_path {
159         my ( $self, $path ) = @_;
160
161         return if defined $loaded->{$path}->{'generated'};
162
163         my $full_path = $self->app->home->rel_file( 'data/' . $path );
164         die "$full_path $!" unless -r $full_path;
165
166         my $dump_path = $self->_dump_path( $path );
167
168         if ( defined $loaded->{$path}->{data} ) {
169                 my $mtime = (stat($full_path))[9];
170                 return if $loaded->{$path}->{mtime} == $mtime;
171                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
172         } elsif ( -e $dump_path ) {
173                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
174                 my $info = retrieve $dump_path;
175                 $loaded->{ $path } = $info;
176                 return;
177         }
178
179         my $data;
180         if ( -f $full_path ) {
181                 $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
182         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
183                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
184         } else {
185                 die "can't load $full_path";
186         }
187
188         my @header;
189
190         if ( defined $data->{header} ) {
191                 if ( ref $data->{header} eq 'ARRAY' ) {
192                         @header = @{ $data->{header} };
193                 } else {
194                         warn "header not array ", dump( $data->{header} );
195                 }
196         }
197
198         my $stats = __stats( $data->{items} );
199
200         @header =
201                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
202                 grep { defined $stats->{$_}->{count} } keys %$stats
203                 unless @header;
204
205         my $info = {
206                 header => [ @header ],
207                 stats  => $stats,
208                 full_path => $full_path,
209                 size => -s $full_path,
210                 mtime => (stat($full_path))[9],
211                 data => $data,
212         };
213
214         $loaded->{ $path } = $info;
215         $self->_save( $path );
216
217 }
218
219
220 sub load {
221         my $self = shift;
222
223         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
224
225         my @paths = $self->param('paths');
226         warn "# paths ", dump @paths;
227
228         foreach my $p ( keys %$loaded ) {
229                 next if grep { /^\Q$p\E$/ } @paths;
230                 warn "remove $p from memory\n";
231                 delete $loaded->{$p};
232         }
233
234         $self->_load_path( $_ ) foreach @paths;
235
236         warn "# path $path\n";
237         $self->_load_path( $path );
238
239         $self->session( 'path' => $path );
240
241         my $redirect_to = '/data/items';
242
243         $self->session( 'header' => $loaded->{$path}->{header} );
244         if ( ! defined $loaded->{$path}->{columns} ) {
245                 my $columns_path = $self->_permanent_path( 'columns' );
246                 if ( -e $columns_path ) {
247                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path;
248                         $loaded->{$path}->{columns} = [ @columns ];
249                         warn "# columns_path $columns_path ",dump(@columns);
250                 } else {
251                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
252                 }
253
254                 $redirect_to = '/data/columns';
255         }
256         $self->session( 'columns' => $loaded->{$path}->{columns} );
257         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
258         $self->redirect_to( $redirect_to );
259 }
260
261
262 sub _loaded {
263         my ( $self, $name ) = @_;
264         my $path = $self->session('path') || $self->param('path');
265         $self->redirect_to('/data/index') unless $path;
266
267         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
268                 my $caller = (caller(1))[3];
269                 if ( $caller =~ m/::edit/ ) {
270                         warn "rebuild stats for $path ignored caller $caller\n";
271                 } else {
272                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
273                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
274                         $loaded->{$path}->{modified} = 1;
275                 }
276         }
277
278         if ( ! defined $loaded->{$path}->{$name} ) {
279                 warn "$path $name isn't loaded\n";
280                 $self->_load_path( $path );
281                 if ( ! defined $loaded->{$path}->{stats} ) {
282                         warn "rebuild stats for $path\n";
283                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
284                 }
285                 if ( ! defined $loaded->{$path}->{$name} ) {
286                         warn "MISSING $name for $path\n";
287                         $self->redirect_to('/data/index')
288                 }
289         }
290
291         $self->session( 'modified' => $loaded->{$path}->{modified} );
292
293         return $loaded->{$path}->{$name};
294 }
295
296
297 sub _checked {
298         my $self = shift;
299         my $checked;
300         $checked->{$_}++ foreach @_;
301 #       warn "# _checked ",dump($checked);
302         return $checked;
303 }
304
305 sub _permanent_path {
306         my $self = shift;
307         my $path = $self->_param_or_session('path');
308         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
309 }
310
311 sub _export_path {
312         my $self = shift;
313         my $path = $self->_param_or_session('path');
314         if ( ! $path ) {
315                 warn "no path in param or session";
316                 return;
317         }
318         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
319         mkpath $dir unless -e $dir;
320         $dir . '/' . join('.', @_);
321 }
322
323 sub columns {
324     my $self = shift;
325
326         if ( $self->param('columns') ) {
327                 my @columns = $self->_param_array('columns');
328                 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
329                 $self->redirect_to('/data/items');
330         }
331
332         my $stats = $self->_loaded( 'stats' );
333
334         my @columns;
335         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
336
337         foreach my $c ( @{ $self->session( 'header' ) } ) {
338                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
339         }
340
341     $self->render(
342                 message => 'Select columns to display',
343                 stats => $stats,
344                 columns => \@columns,
345                 checked => $self->_checked( $self->_param_array('columns') ),
346         );
347 }
348
349 sub _param_array {
350     my ($self,$name) = @_;
351
352         my @array = $self->param($name);
353         my $path  = $self->session('path');
354
355         if ( @array ) {
356                 $self->session($name => [ @array ]);
357                 $loaded->{$path}->{$name} = [ @array ];
358         } elsif ( my $session = $self->session($name) ) {
359                 if ( ref $session eq 'ARRAY' ) {
360                         @array = @$session;
361                 } else {
362                         die "$name not array ",dump($session);
363                 }
364         }
365         #warn "# $name ",dump @array;
366         return @array;
367 }
368
369 sub _param_scalar {
370     my ($self,$name,$default) = @_;
371
372         my $scalar = $self->param($name);
373
374         if ( defined $scalar ) {
375                 $self->session($name => $scalar);
376         } else {
377                 $scalar = $self->session($name);
378         }
379
380         if ( ! defined $scalar ) {
381                 $scalar = $default;
382                 die "no default for $name" unless defined $scalar;
383                 $self->session($name => $scalar);
384         }
385
386         warn "# _param_scalar $name ",dump $scalar;
387         return $scalar;
388 }
389
390 sub filter {
391         my $self = shift;
392
393         my $name = $self->param('filter_name') || die "name?";
394         my @vals = $self->param('filter_vals');
395
396         $self->_remove_filter( $name );
397         if ( @vals ) {
398                 $self->_filter_on_data( $name, @vals );
399                 if ( my $permanent = $self->param('_permanent') ) {
400                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
401                         write_file $permanent_path, map { "$_\n" } @vals;
402                         warn "permanent filter $permanent_path ", -s $permanent_path;
403                 }
404         }
405
406         $self->session( 'offset' => 0 );
407
408         $self->redirect_to('/data/items');
409 }
410
411 sub _filter_on_data {
412         my ( $self, $name, @vals ) = @_;
413
414         my $path = $self->session('path');
415
416         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
417                 warn "filter $name not found in data set";
418                 return;
419         }
420
421         $filters->{$name} = [ @vals ];
422         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
423
424         my $filter_hash;
425         $filter_hash->{$_}++ foreach @vals;
426
427         #warn "# filter_hash ",dump( $filter_hash );
428
429         my $items = $self->_loaded('data')->{items};
430
431         my $include_missing = defined $filter_hash->{_missing};
432         my $filtered_items;
433
434         foreach my $i ( 0 .. $#$items ) {
435
436                 if ( defined $items->[$i]->{$name} ) {
437                         my $row = $items->[$i]->{$name};
438                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
439                         foreach my $v ( @$row ) {
440                                 if ( defined $filter_hash->{ $v } ) {
441                                         $filtered_items->{$i}++;
442                                 }
443                         }
444                 } elsif ( $include_missing ) {
445                         $filtered_items->{$i}++;
446                 }
447         }
448
449         #warn "# filter $name ",dump($filtered_items);
450
451         $loaded->{$path}->{filters}->{$name} = $filtered_items;
452         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
453 }
454
455
456 sub _current_filters {
457         my $self = shift;
458         my $current_filters;
459         my $columns = $self->_loaded('header');
460         if ( my $sc = $self->session('columns') ) {
461                 $columns = $sc;
462         }
463
464         $current_filters->{ $_ } = $filters->{ $_ }
465                 foreach ( grep { defined $filters->{ $_ } } @$columns )
466         ;
467         warn "# _current_filters ",dump($columns);
468         return $current_filters;
469 }
470
471 sub _data_sorted_by {
472         my ( $self, $order ) = @_;
473
474         my $path = $self->session('path');
475
476         warn "_data_sorted_by $order from $path";
477
478         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
479                 return $loaded->{$path}->{sorted}->{$order};
480         }
481
482         my $data = $self->_loaded( 'data' );
483         my $numeric = $self->_is_numeric($order);
484         my $missing = $numeric ? 0 : '';
485         no warnings qw(numeric);
486         my $nr = 0;
487         my @sorted = map {
488                 $_->[0]
489         } sort {
490                 if ( $numeric ) {
491                         $a->[1] <=> $b->[1]
492                 } else {
493                         $a->[1] cmp $b->[1]
494                 }
495         } map {
496                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
497         } grep { ref $_->{$order} eq 'ARRAY' } @{ $data->{items} }
498         ;
499
500         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
501         #warn "# sorted ",dump( @sorted );
502
503         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
504 }
505
506
507 sub items {
508         my $self = shift;
509
510         if ( my $show = $self->param('id') ) {
511                 $self->param('show', $show);
512                 warn "show $show\n";
513         }
514
515         my $path = $self->session('path');
516
517         my @columns = $self->_param_array('columns');
518         $self->redirect_to('/data/columns') unless @columns;
519         my $order   = $self->_param_scalar('order', $columns[0]);
520         my $sort    = $self->_param_scalar('sort', 'a');
521         my $offset  = $self->_param_scalar('offset', 0);
522         my $limit   = $self->_param_scalar('limit', 20);
523         $self->_param_scalar('show', 'table');
524
525         # fix offset when changing limit
526         $offset = int( $offset / $limit ) * $limit;
527
528         if ( ! grep { /^\Q$order\E$/ } @columns ) {
529                 $order = $columns[0];
530                 $self->session( order => $order );
531         }
532         my $sorted = $self->_data_sorted_by( $order );
533
534         my @filter_names;
535         if ( $filters ) {
536                 foreach my $name ( keys %$filters ) {
537                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
538                                 warn "skip filter $name not found in $path\n";
539                                 next;
540                         }
541                         push @filter_names, $name;
542                 }
543                 warn "filter_names ",dump( @filter_names );
544                 foreach my $name ( @filter_names ) {
545                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
546                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
547                 }
548         }
549
550         my $all_filters = join(' ',sort @filter_names,'order:',$order);
551
552 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
553
554         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
555
556                 my $path_filters = $loaded->{$path}->{filters};
557
558                 warn "create combined filter for $all_filters\n";
559
560                 my @filtered;
561                 foreach my $i ( 0 .. $#$sorted ) {
562                         my $pos = $sorted->[$i];
563
564                         if ( $#filter_names == -1 ) {
565                                 push @filtered, $pos;
566                                 next;
567                         }
568
569                         my $skip = 0;
570                         foreach ( @filter_names ) {
571                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
572                         }
573                         next if $skip;
574
575                         push @filtered, $pos;
576                 }
577
578                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
579         }
580
581         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
582                 if defined $loaded->{$path}->{filtered}->{$all_filters};
583
584         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
585
586         my $data = $self->_loaded('data');
587
588         my $code = $self->_param_scalar('code','');
589         $code =~ s{[\r\n]+$}{}s;
590
591         my $commit = $self->param('commit');
592         my $test = $self->param('test');
593
594         my $cols_changed;
595
596         if ( $code && ( $test || $commit ) ) {
597                 # XXX find columns used in code snippet and show them to user
598                 foreach my $column ( $code =~ m/\$row->{(.+?)}/g ) {
599                         if ( $column =~ s/^(['"])// ) {
600                                 $column =~ s/$1$//;
601                         }
602                         $cols_changed->{$column}++;
603                         next if grep { /$column/ } @columns;
604                         $cols_changed->{$column}++;
605                         unshift @columns, $column;
606                         if ( $commit ) {
607                                 $self->session('columns', [ @columns ]);
608                                 $loaded->{$path}->{columns} = [ @columns ];
609                                 __path_modified( $path, 2 );
610                         }
611                 }
612         }
613
614         my $code_path = $self->app->home->rel_dir('public') . "/code";
615         if ( $commit ) {
616                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
617                 my $out;
618                 foreach ( 0 .. $#$filtered ) {
619                         my $i = $filtered->[$_];
620                         my $row = $data->{items}->[$i];
621                         eval $code;
622                 }
623                 if ( my $description = $self->param('code_description') ) {
624                         my $depends = $self->param('code_depends') || die "no code_depends?";
625                         my $path = "$code_path/$depends.$description.pl";
626                         if ( -e $path && ! $self->param('overwrite') ) {
627                                 warn "# code $path not saved\n";
628                         } else {
629                                 write_file $path, $code;
630                                 warn "code $path ", -s $path, " bytes saved\n";
631                         }
632                 }
633                 $code = '';
634                 if ( $out ) {
635                         my $commit_dataset = join('.'
636                                 , $self->param('code_depends')
637                                 , $self->param('code_description')
638                         );
639                         my $key = $self->param('code_depends');
640                         $key =~ s/,.+$//;
641                         $key ||= 'key';
642                         my $items;
643                         foreach my $n ( keys %$out ) {
644                                 my $i = { $key => [ $n ] };
645                                 $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
646                                 push @$items, $i;
647                         };
648                         undef $out;
649                         my $stats = __stats( $items );
650                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
651                         unshift @columns, $key;
652
653                         $loaded->{$commit_dataset} = {
654                                 header => [ @columns ],
655                                 columns => [ @columns ],
656                                 mtime => time(),
657                                 data => { items => $items },
658                                 stats => $stats,
659                                 generated => 1,
660                         };
661                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
662                         $self->session('path', $commit_dataset);
663                         $self->session('columns', [ @columns ]);
664                         $self->session('order', $key);
665                         $self->redirect_to('/data/items');
666                 }
667         }
668
669         my $sorted_items;
670         my $from_end = $sort eq 'd' ? $#$filtered : 0;
671         my $out;
672         foreach ( 0 .. $limit ) {
673                 my $i = $_ + $offset;
674                 last unless defined $filtered->[$i];
675                 $i = $from_end - $i if $from_end;
676                 my $id = $filtered->[$i];
677                 my $row = $data->{items}->[ $id ];
678                 if ( $code && $test ) {
679                         $row = Storable::dclone $row;
680                         eval $code;
681                         if ( $@ ) {
682                                 warn "ERROR evaling\n$code\n$@";
683                                 $self->stash('eval_error', $@) if $@;
684                         } else {
685                                 warn "EVAL ",dump($row);
686                         }
687                 }
688                 $row->{_row_id} ||= $id;
689                 push @$sorted_items, $row;
690         }
691
692         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
693
694         my $code_depends = $self->param('code_depends')||
695         join(',', grep { defined $cols_changed->{$_} && $cols_changed->{$_} == 1 } @columns );
696         my $code_description = $self->param('code_description') ||
697         join(',', grep { defined $cols_changed->{$_} && $cols_changed->{$_} == 2 } @columns );
698
699         $code_depends ||= $code_description; # self-modifing
700
701         warn "# cols_changed ",dump( $cols_changed, $code_depends, $code_description );
702
703         $self->render(
704                 order => $order,
705                 offset => $offset,
706                 limit => $limit,
707                 sorted => $sorted_items,
708                 columns => [ @columns ],
709                 rows => $#$filtered + 1,
710                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
711                 filters => $self->_current_filters,
712                 code => $code,
713                 cols_changed => $cols_changed,
714                 code_depends => $code_depends,
715                 code_description => $code_description,
716                 code_path => $code_path,
717                 out => $out,
718         );
719
720 }
721
722
723 sub order {
724         my $self = shift;
725         $self->session('order', $self->param('order'));
726         $self->session('sort', $self->param('sort'));
727         $self->redirect_to('/data/items');
728 }
729
730 sub _is_numeric {
731         my ( $self, $name ) = @_;
732
733         my $stats = $self->_loaded( 'stats' );
734
735         # sort facet numerically if more >50% elements are numeric
736         my $count = $stats->{$name}->{count};
737         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
738         defined $stats->{$name}->{numeric} &&
739                 $stats->{$name}->{numeric} > $count / 2;
740 }
741
742 sub _remove_filter {
743         my ($self,$name) = @_;
744         warn "_remove_filter $name\n";
745
746         my $path = $self->session('path');
747
748         delete $filters->{$name};
749         delete $loaded->{$path}->{filters}->{$name};
750         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
751
752         foreach (
753                         grep { /\Q$name\E/ }
754                         keys %{ $loaded->{$path}->{filtered} }
755         ) {
756                 delete $loaded->{$path}->{filtered}->{$_};
757                 warn "remove filtered cache $_";
758         }
759 }
760
761 sub facet {
762         my $self = shift;
763
764         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
765
766         if ( my $name = $self->param('remove') ) {
767                 $self->_remove_filter( $name );
768                 $self->redirect_to( '/data/items' );
769         }
770
771         my $facet;
772         my $name = $self->param('name') || die "no name";
773
774         my $all = $self->_param_scalar('all', 1);
775         my $data = $self->_loaded('data');
776
777         my $filters = $self->_current_filters;
778         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
779         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
780                 if defined $loaded->{$path}->{filtered}->{$all_filters};
781
782         if ( ! $filtered || $all ) {
783                 $filtered = [ 0 .. $#{ $data->{items} } ];
784                 warn "filter all values\n";
785         } else {
786                 warn "filter using $all_filters\n";
787         }
788
789         foreach my $i ( @$filtered ) {
790                 my $item = $data->{items}->[$i];
791                 if ( ! exists $item->{$name} ) {
792                         $facet->{ _missing }++;
793                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
794                         $facet->{$_}++ foreach @{ $item->{$name} };
795                 } else {
796                         $facet->{ $item->{$name} }++;
797                 }
798         }
799
800 #       warn "# facet $name ",dump $facet;
801
802         my $checked;
803         my @facet_names =
804                   $all                      ? keys %$facet
805                 : defined $filters->{$name} ? @{ $filters->{$name} }
806                 : keys %$facet;
807
808         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
809
810         my $numeric = $self->_is_numeric($name);
811
812         my $sort = $self->param('sort');
813         # sort numeric facets with more than 5 values ascending
814         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
815
816         @facet_names = sort {
817                 my $result;
818                 if ( $sort eq 'a' ) {
819                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
820                 } elsif ( $sort eq 'd' ) {
821                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
822                 } elsif ( $sort eq 'c' ) {
823                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
824                 } else {
825                         warn "unknown sort: $sort";
826                         $result = $a cmp $b;
827                 }
828                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
829                 $result;
830         } @facet_names;
831
832         $self->render( name => $name, facet => $facet, checked => $checked,
833                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
834         );
835 }
836
837
838 sub __invalidate_path_column {
839         my ( $path, $name ) = @_;
840
841         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
842                 delete $loaded->{$path}->{sorted}->{$name};
843                 warn "# invalidate $path sorted $name\n";
844         }
845
846         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
847                 delete $loaded->{$path}->{filtered}->{$_};
848                 warn "# invalidate $path filtered $_\n";
849         }
850 }
851
852 sub __path_modified {
853         my ( $path, $value ) = @_;
854         $value = 1 unless defined $value;
855         
856         $loaded->{$path}->{modified}  = $value;
857
858         warn "# __path_modified $path $value\n";
859 }
860
861 sub edit {
862         my $self = shift;
863         my $new_content = $self->param('new_content');
864         $new_content  ||= $self->param('content'); # backward compatibility with old actions
865
866         my $i = $self->param('_row_id');
867         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
868         my $path = $self->param('path') || die "no path";
869         my $name = $self->param('name') || die "no name";
870         my $status = 200; # 200 = OK, 201 = Created
871
872         my $data = $self->_loaded('data');
873
874         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
875                 $new_content =~ s/^\s+//s;
876                 $new_content =~ s/\s+$//s;
877                 my $v;
878                 if ( $new_content =~ /\xB6/ ) { # para
879                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
880                 } else {
881                         $v = [ $new_content ];
882                 }
883
884                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
885                 my $new = dump $v;
886                 if ( $old ne $new
887                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
888                 ) {
889                         my $change = {
890                                 path => $path,
891                                 column => $name,
892                                 pos => $i,
893                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
894                                 new => $v,
895                                 time => $self->param('time') || time(),
896                                 user => $self->param('user') || $ENV{'LOGNAME'},
897                                 unique => {
898                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
899                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
900                                         keys %{ $loaded->{$path}->{stats} }
901                                 },
902                         };
903                         my $change_path = $self->_permanent_path( 'changes' );
904                         mkdir $change_path unless -d $change_path;
905                         $change_path .= '/' . $change->{time};
906                         store $change, $change_path;
907                         utime $change->{time}, $change->{time}, $change_path;
908                         warn "# $change_path ", dump($change);
909
910                         warn "# change $path $i $old -> $new\n";
911                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
912
913                         __invalidate_path_column( $path, $name );
914
915                         $status = 201; # created
916                         # modified = 2 -- force rebuild of stats
917                         __path_modified( $path, 2 );
918         
919                         $new_content = join("\xB6",@$v);
920
921                 } else {
922                         warn "# unchanged $path $i $old\n";
923                         $status = 304;
924                 }
925         } else {
926                 $new_content = "$path $i $name doesn't exist\n";
927                 $status = 404;
928         }
929
930         warn "# edit $status ", dump $new_content;
931
932         $self->render(
933                 status => $status,
934                 new_content => scalar $new_content,
935         );
936 }
937
938
939 sub save {
940         my $self = shift;
941         my $path = $self->_param_or_session('path');
942         my $dump_path = $self->_save( $path );
943         __path_modified( $path, 0 );
944
945         $self->redirect_to( '/data/items' );
946 }
947
948 sub export {
949         my $self = shift;
950
951         if ( my $import = $self->param('import') ) {
952
953                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
954                         my $name = $1;
955                         my @vals = map { chomp; $_ }
956                                 read_file $self->app->home->rel_dir('public') . "/export/$import";
957                         $self->_remove_filter( $name );
958                         $self->_filter_on_data( $name, @vals );
959                         $self->session( 'offset' => 0 );
960                         $self->redirect_to('/data/items');
961                 } else {
962                         warn "UNKNOWN IMPORT $import";
963                 }
964         }
965
966         $self->render( export => [
967                 glob( $self->_export_path . '*' )
968         ] );
969 }
970
971 sub __loaded_paths {
972         return
973                 grep { defined $loaded->{$_}->{data} }
974                 keys %$loaded;
975 }
976
977 1;