86d71bcefce900ebc4c65769fc979d480aa98927
[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 use Text::Unaccent::PurePerl;
17
18 use MojoFacets::Import::File;
19 use MojoFacets::Import::HTMLTable;
20 use MojoFacets::Import::CSV;
21
22 our $loaded;
23 our $filters;
24
25 sub index {
26         my $self = shift;
27
28         my $data_dir = $self->app->home->rel_dir('data');
29         die "no data dir $data_dir" unless -d $data_dir;
30
31         my @files;
32         my $changes;
33         find( sub {
34                 my $file = $File::Find::name;
35                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
36                         $file =~ s/$data_dir\/*//;
37                         push @files, $file;
38                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
39                         push @{ $changes->{$1} }, $2
40                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
41                         $file =~ s/$data_dir\/*//;
42                         push @files, $file;
43                 } elsif ( -f $file && $file =~ m/\.csv$/i ) {
44                         $file =~ s/$data_dir\/*//;
45                         push @files, $file;
46                 } elsif ( -f $file && $file =~ m/\.storable/i ) {
47                         $file =~ s/$data_dir\/*//;
48                         push @files, $file;
49                 } else {
50                         #warn "IGNORE: $file\n";
51                 }
52         }, $data_dir);
53
54         no warnings qw(uninitialized); # mtime
55         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
56                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
57         my $size;
58         $size->{$_} = -s "$data_dir/$_" foreach @files;
59
60         $self->render(
61                 files => [ @files ],
62                 size => $size,
63                 loaded => $loaded,
64                 filters => $filters,
65                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
66                 changes => $changes,
67         );
68 }
69
70 sub _dump_path {
71         my ( $self, $name ) = @_;
72         my $dir = $self->app->home->rel_dir('data');
73         $name =~ s/^$dir//;
74         $name =~ s/\/+/_/g;
75         return '/tmp/mojo_facets.' . $name . '.storable';
76 }
77
78 sub _save {
79         my ( $self, $path ) = @_;
80
81         my $dump_path = $self->_dump_path( $path );
82         my $first_load = ! -e $dump_path;
83         warn "save loaded to $dump_path";
84         my $info = $loaded->{$path};
85         store $info, $dump_path;
86
87         if ( $first_load ) {
88                 my $mtime = $loaded->{$path}->{mtime};
89                 utime $mtime, $mtime, $dump_path;
90                 warn "sync time to $path at $mtime\n";
91         }
92
93         warn $dump_path, ' ', -s $dump_path, " bytes\n";
94         return $dump_path;
95 }
96
97
98 sub __stats {
99
100         my $stats;
101
102         my $nr_items = $#{ $_[0] } + 1;
103
104         warn "__stats $nr_items\n";
105
106         foreach my $e ( @{ $_[0] } ) {
107                 foreach my $n ( keys %$e ) {
108                         $stats->{$n}->{count}++;
109                         my @v;
110                         if ( ref $e->{$n} eq 'ARRAY' ) {
111                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
112                                 @v = @{ $e->{$n} };
113                         } else {
114                                 @v = ( $e->{$n} );
115                         }
116
117                         foreach my $x ( @v ) {
118                                 $stats->{$n}->{numeric}++
119                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
120                                 $stats->{$n}->{empty}++
121                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
122                         }
123
124                 }
125         }
126
127         foreach my $n ( keys %$stats ) {
128                 my $s = $stats->{$n};
129                 next unless defined $s->{array};
130                 if ( $s->{array} == $s->{count} ) {
131                         delete $s->{array};
132                         if ( $s->{count} == $nr_items ) {
133                                 warn "check $n for uniqeness\n";
134                                 my $unique;
135                                 foreach my $e ( @{ $_[0] } ) {
136                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
137                                                 $unique = 0;
138                                                 last;
139                                         }
140                                 }
141                                 if ( $unique ) {
142                                         $stats->{$n}->{unique} = 1;
143                                         #warn "# $n unique ",dump( $unique );
144                                 }
145                         }
146                 }
147         }
148
149         warn "# __stats ",dump($stats);
150
151         return $stats;
152 }
153
154 sub _param_or_session {
155         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
156 }
157
158 sub stats {
159         my $self = shift;
160         my $path = $self->_param_or_session('path');
161         warn "stats $path\n";
162         delete $loaded->{$path}->{stats};
163         $self->redirect_to( '/data/columns' );
164 }
165
166
167 sub _load_path {
168         my ( $self, $path ) = @_;
169
170         return if defined $loaded->{$path}->{'generated'};
171
172         my $full_path = $self->app->home->rel_file( 'data/' . $path );
173         $self->redirect_to('/data/index') unless -r $full_path;
174
175         my $dump_path = $self->_dump_path( $path );
176
177         if ( defined $loaded->{$path}->{data} ) {
178                 my $mtime = (stat($full_path))[9];
179                 return if $loaded->{$path}->{mtime} == $mtime;
180                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
181         } elsif ( -e $dump_path ) {
182                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
183                 my $info = retrieve $dump_path;
184                 $loaded->{ $path } = $info;
185                 return;
186         }
187
188         my $data;
189         if ( -f $full_path ) {
190                 if ( $full_path =~ m/.csv/i ) {
191                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
192                 } elsif ( $full_path =~ m/.storable/ ) {
193                         warn "open $full_path ", -s $full_path, " bytes";
194                         open(my $pipe, "<", $full_path) || die $!;
195                         while ( my $o = eval { Storable::fd_retrieve $pipe } ) {
196                                 if ( defined $o->{item} ) {
197                                         push @{ $data->{items} }, $o->{item};
198                                 } else {
199                                         warn "SKIP ",dump($o);
200                                 }
201                         }
202                         close($pipe);
203                         warn "loaded ", $#{ $data->{items} } + 1, " items from $full_path\n";
204                 } else {
205                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
206                 }
207         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
208                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
209         } else {
210                 die "can't load $full_path";
211         }
212
213         my @header;
214
215         if ( defined $data->{header} ) {
216                 if ( ref $data->{header} eq 'ARRAY' ) {
217                         @header = @{ $data->{header} };
218                 } else {
219                         warn "header not array ", dump( $data->{header} );
220                 }
221         }
222
223         my $stats = __stats( $data->{items} );
224
225         @header =
226                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
227                 grep { defined $stats->{$_}->{count} } keys %$stats
228                 unless @header;
229
230         my $info = {
231                 header => [ @header ],
232                 stats  => $stats,
233                 full_path => $full_path,
234                 size => -s $full_path,
235                 mtime => (stat($full_path))[9],
236                 data => $data,
237         };
238
239         $loaded->{ $path } = $info;
240         $self->_save( $path );
241
242 }
243
244
245 sub load {
246         my $self = shift;
247
248         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
249
250         my @paths = $self->param('paths');
251         warn "# paths ", dump @paths;
252
253         foreach my $p ( keys %$loaded ) {
254                 next if grep { /^\Q$p\E$/ } @paths;
255                 warn "remove $p from memory\n";
256                 delete $loaded->{$p};
257         }
258
259         $self->_load_path( $_ ) foreach @paths;
260
261         warn "# path $path\n";
262         $self->_load_path( $path );
263
264         $self->session( 'path' => $path );
265
266         my $redirect_to = '/data/items';
267
268         $self->session( 'header' => $loaded->{$path}->{header} );
269         if ( ! defined $loaded->{$path}->{columns} ) {
270                 my $columns_path = $self->_permanent_path( 'columns' );
271                 if ( -e $columns_path ) {
272                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
273                         $loaded->{$path}->{columns} = [ @columns ];
274                         warn "# columns_path $columns_path ",dump(@columns);
275                 } else {
276                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
277                 }
278
279                 $redirect_to = '/data/columns';
280         }
281         $self->session( 'columns' => $loaded->{$path}->{columns} );
282         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
283         $self->redirect_to( $redirect_to );
284 }
285
286
287 sub _loaded {
288         my ( $self, $name ) = @_;
289         my $path = $self->session('path') || $self->param('path');
290         $self->redirect_to('/data/index') unless $path;
291
292         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
293                 my $caller = (caller(1))[3];
294                 if ( $caller =~ m/::edit/ ) {
295                         warn "rebuild stats for $path ignored caller $caller\n";
296                 } else {
297                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
298 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
299                         $loaded->{$path}->{rebuild_stats} = 1;
300                         $loaded->{$path}->{modified} = 1;
301                 }
302         }
303
304         if ( defined $loaded->{$path}->{rebuild_stats} ) {
305                 warn "rebuild_stats $path";
306                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
307                 delete $loaded->{$path}->{rebuild_stats};
308         }
309
310         if ( ! defined $loaded->{$path}->{$name} ) {
311                 warn "$path $name isn't loaded\n";
312                 $self->_load_path( $path );
313                 if ( ! defined $loaded->{$path}->{stats} ) {
314                         warn "rebuild stats for $path\n";
315                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
316                 }
317                 if ( ! defined $loaded->{$path}->{$name} ) {
318                         warn "MISSING $name for $path\n";
319                         $self->redirect_to('/data/index')
320                 }
321         }
322
323         $self->session( 'modified' => $loaded->{$path}->{modified} );
324
325         return $loaded->{$path}->{$name};
326 }
327
328
329 sub _checked {
330         my $self = shift;
331         my $checked;
332         $checked->{$_}++ foreach @_;
333 #       warn "# _checked ",dump($checked);
334         return $checked;
335 }
336
337 sub _permanent_path {
338         my $self = shift;
339         my $path = $self->_param_or_session('path');
340         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
341 }
342
343 sub __unac {
344         my $n = shift;
345         $n = unac_string($n);
346         $n =~ s/\W+/_/g;
347         return $n;
348 }
349
350 sub _column_from_unac {
351         my ($self,$name) = @_;
352
353         my $stats = $self->_loaded('stats');
354         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
355
356         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
357 }
358
359 sub _export_path {
360         my $self = shift;
361         my $path = $self->_param_or_session('path');
362         if ( ! $path ) {
363                 warn "no path in param or session";
364                 return;
365         }
366         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
367         mkpath $dir unless -e $dir;
368         my $name = join('.', map { __unac($_) } @_ );
369         my $full = $dir . '/' . $name;
370         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
371         return $full;
372 }
373
374 sub columns {
375     my $self = shift;
376
377         if ( $self->param('columns') ) {
378                 my @columns = $self->_param_array('columns');
379                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
380                 $self->redirect_to('/data/items');
381         }
382
383         my $stats = $self->_loaded( 'stats' );
384
385         my @columns;
386         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
387
388         foreach my $c ( @{ $self->session( 'header' ) } ) {
389                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
390         }
391
392     $self->render(
393                 message => 'Select columns to display',
394                 stats => $stats,
395                 columns => \@columns,
396                 checked => $self->_checked( $self->_param_array('columns') ),
397         );
398 }
399
400 sub _param_array {
401     my ($self,$name) = @_;
402
403         my @array = $self->param($name);
404         my $path  = $self->session('path');
405
406         if ( @array ) {
407                 $self->session($name => [ @array ]);
408                 $loaded->{$path}->{$name} = [ @array ];
409         } elsif ( my $session = $self->session($name) ) {
410                 if ( ref $session eq 'ARRAY' ) {
411                         @array = @$session;
412                 } else {
413                         die "$name not array ",dump($session);
414                 }
415         }
416         #warn "# $name ",dump @array;
417         return @array;
418 }
419
420 sub _param_scalar {
421     my ($self,$name,$default) = @_;
422
423         my $scalar = $self->param($name);
424
425         if ( defined $scalar ) {
426                 $self->session($name => $scalar);
427         } else {
428                 $scalar = $self->session($name);
429         }
430
431         if ( ! defined $scalar ) {
432                 $scalar = $default;
433                 die "no default for $name" unless defined $scalar;
434                 $self->session($name => $scalar);
435         }
436
437         warn "# _param_scalar $name ",dump $scalar;
438         return $scalar;
439 }
440
441 sub filter {
442         my $self = shift;
443
444         my $name = $self->param('filter_name') || die "name?";
445         my @vals = $self->param('filter_vals');
446
447         $self->_remove_filter( $name );
448         if ( @vals ) {
449                 $self->_filter_on_data( $name, @vals );
450                 if ( my $permanent = $self->param('_permanent') ) {
451                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
452                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
453                         warn "permanent filter $permanent_path ", -s $permanent_path;
454                 }
455         }
456
457         $self->session( 'offset' => 0 );
458
459         $self->redirect_to('/data/items');
460 }
461
462 sub _filter_on_data {
463         my ( $self, $name, @vals ) = @_;
464
465         my $path = $self->session('path');
466
467         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
468                 warn "filter $name not found in data set";
469                 return;
470         }
471
472         $filters->{$name} = [ @vals ];
473         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
474
475         my $filter_hash;
476         $filter_hash->{$_}++ foreach @vals;
477
478         #warn "# filter_hash ",dump( $filter_hash );
479
480         my $items = $self->_loaded('data')->{items};
481
482         my $include_missing = defined $filter_hash->{_missing};
483         my $filtered_items;
484
485         foreach my $i ( 0 .. $#$items ) {
486
487                 if ( defined $items->[$i]->{$name} ) {
488                         my $row = $items->[$i]->{$name};
489                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
490                         foreach my $v ( @$row ) {
491                                 if ( defined $filter_hash->{ $v } ) {
492                                         $filtered_items->{$i}++;
493                                 }
494                         }
495                 } elsif ( $include_missing ) {
496                         $filtered_items->{$i}++;
497                 }
498         }
499
500         #warn "# filter $name ",dump($filtered_items);
501
502         $loaded->{$path}->{filters}->{$name} = $filtered_items;
503         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
504 }
505
506
507 sub _current_filters {
508         my $self = shift;
509         my $current_filters;
510         my $stats = $self->_loaded('stats');
511
512         $current_filters->{ $_ } = $filters->{ $_ }
513                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
514         ;
515         warn "# _current_filters ",dump( keys %$current_filters );
516         return $current_filters;
517 }
518
519 sub _data_sorted_by {
520         my ( $self, $order ) = @_;
521
522         my $path = $self->session('path');
523
524         warn "_data_sorted_by $order from $path";
525
526         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
527                 return $loaded->{$path}->{sorted}->{$order};
528         }
529
530         my $data = $self->_loaded( 'data' );
531         my $numeric = $self->_is_numeric($order);
532         my $missing = $numeric ? 0 : '';
533         no warnings qw(numeric);
534         my $nr = 0;
535         my @sorted = map {
536                 $_->[0]
537         } sort {
538                 if ( $numeric ) {
539                         $a->[1] <=> $b->[1]
540                 } else {
541                         $a->[1] cmp $b->[1]
542                 }
543         } map {
544                 my $v;
545                 if ( ! exists $_->{$order} ) {
546                         $v = $missing;
547                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
548                         $v = join('', @{$_->{$order}});
549                 } else {
550                         $v = $_->{$order};
551                 }
552                 [ $nr++, $v ]
553         } @{ $data->{items} }
554         ;
555
556         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
557         #warn "# sorted ",dump( @sorted );
558
559         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
560 }
561
562
563 sub __all_filters {
564         my $order = pop @_;
565         join(',', sort(@_), 'order', $order);
566 }
567
568 our ($out, $key,$value);
569
570 sub __commit_path_code {
571         my ( $path, $i, $code, $commit_changed ) = @_;
572
573         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
574         my $row = $items->[$i];
575         my $update;
576         eval $code;
577         foreach ( keys %$update ) {
578                 $$commit_changed->{$_}++;
579                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
580         }
581         #warn "__commit_path_code $path $i ",dump( $update );
582 }
583
584 sub items {
585         my $self = shift;
586
587         if ( my $show = $self->param('id') ) {
588                 $self->param('show', $show);
589                 warn "show $show\n";
590         }
591
592         my $path = $self->session('path');
593
594         my @columns = $self->_param_array('columns');
595         $self->redirect_to('/data/columns') unless @columns;
596         my $order   = $self->_param_scalar('order', $columns[0]);
597         my $sort    = $self->_param_scalar('sort', 'a');
598         my $offset  = $self->_param_scalar('offset', 0);
599         my $limit   = $self->_param_scalar('limit', 20);
600         $self->_param_scalar('show', 'table');
601
602         # fix offset when changing limit
603         $offset = int( $offset / $limit ) * $limit;
604
605         if ( ! grep { /^\Q$order\E$/ } @columns ) {
606                 $order = $columns[0];
607                 $self->session( order => $order );
608         }
609         my $sorted = $self->_data_sorted_by( $order );
610
611         my @filter_names;
612         if ( $filters ) {
613                 foreach my $name ( keys %$filters ) {
614                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
615                                 warn "skip filter $name not found in $path\n";
616                                 next;
617                         }
618                         push @filter_names, $name;
619                 }
620                 warn "filter_names ",dump( @filter_names );
621                 foreach my $name ( @filter_names ) {
622                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
623                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
624                 }
625         }
626
627         my $all_filters = __all_filters( @filter_names,$order );
628
629 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
630
631         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
632
633                 my $path_filters = $loaded->{$path}->{filters};
634
635                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
636
637                 my @filtered;
638                 foreach my $i ( 0 .. $#$sorted ) {
639                         my $pos = $sorted->[$i];
640
641                         if ( $#filter_names == -1 ) {
642                                 push @filtered, $pos;
643                                 next;
644                         }
645
646                         my $skip = 0;
647                         foreach ( @filter_names ) {
648                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
649                         }
650                         next if $skip;
651
652                         push @filtered, $pos;
653                 }
654
655                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
656         }
657
658         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
659                 if defined $loaded->{$path}->{filtered}->{$all_filters};
660
661         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
662
663         my $data = $self->_loaded('data');
664
665         my $code = $self->_param_scalar('code','');
666         $code =~ s{\r}{}gs;
667         $code =~ s{\n+$}{\n}s;
668
669         my $commit = $self->param('commit');
670         my $test = $self->param('test');
671
672         my $commit_changed;
673
674         if ( $code && ( $test || $commit ) ) {
675                 # XXX find columns used in code snippet and show them to user
676                 my $order = 0;
677                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
678                         if ( $column =~ s/^(['"])// ) {
679                                 $column =~ s/$1$//;
680                         }
681                         next if $column =~ m/\$/; # hide columns with vars in them
682                         $commit_changed->{$column} = 0;
683                 }
684         }
685
686         my $code_path = $self->app->home->rel_dir('public') . "/code";
687         if ( $commit ) {
688
689                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
690                 ( $key, $value, $out ) = ( 'key', 'value' );
691                 foreach ( 0 .. $#$filtered ) {
692                         my $i = $filtered->[$_];
693                         __commit_path_code( $path, $i, $code, \$commit_changed );
694                 }
695
696                 $self->_save_change({
697                         path => $path,
698                         time => $self->param('time') || time(),
699                         user => $self->param('user') || $ENV{'LOGNAME'},
700                         code => $code,
701                         commit_changed => $commit_changed,
702                 });
703
704                 if ( my $description = $self->param('code_description') ) {
705                         my $depends = $self->param('code_depends') || die "no code_depends?";
706                         my $path = "$code_path/$depends.$description.pl";
707                         if ( -e $path && ! $self->param('overwrite') ) {
708                                 warn "# code $path not saved\n";
709                         } else {
710                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
711                                 warn "code $path ", -s $path, " bytes saved\n";
712                         }
713                 }
714
715                 # remove console
716                 $code = '';
717                 if ( $out ) {
718                         my $commit_dataset = join('.'
719                                 , $self->param('code_depends')
720                                 , $self->param('code_description')
721                                 , time()
722                         );
723                         $key ||= 'key';
724                         $value ||= 'value';
725                         warn "key $key value $value";
726                         my $items;
727                         foreach my $n ( keys %$out ) {
728                                 my $i = { $key => [ $n ] };
729                                 my $ref = ref $out->{$n};
730                                 if ( $ref eq 'HASH' ) {
731                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
732                                 } elsif ( $ref eq 'ARRAY' ) {
733                                         $i->{$_} = $out->{$n};
734                                 } elsif ( ! $ref ) {
735                                         $i->{$value} = [ $out->{$n} ];
736                                 } else {
737                                         $i->{_error} = [ dump($out->{$n}) ];
738                                 }
739                                 push @$items, $i;
740                         };
741                         undef $out;
742                         my $stats = __stats( $items );
743                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
744                         unshift @columns, $key;
745
746                         $loaded->{$commit_dataset} = {
747                                 header => [ @columns ],
748                                 columns => [ @columns ],
749                                 mtime => time(),
750                                 data => { items => $items },
751                                 stats => $stats,
752                                 generated => 1,
753                         };
754                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
755                         $self->session('path', $commit_dataset);
756                         $self->session('columns', [ @columns ]);
757                         $self->session('order', $key);
758                         $self->redirect_to('/data/items');
759                         return; # FIXME needed to correctly show columns
760                 }
761
762                 # this might move before $out to recalculate stats on source dataset?
763                 __path_rebuild_stats( $path );
764                 my $c = { map { $_ => 1 } @columns };
765                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
766                 warn "# added_columns ",dump( @added_columns );
767                 unshift @columns, @added_columns;
768
769                 $self->session('columns', [ @columns ]);
770                 $loaded->{$path}->{columns} = [ @columns ];
771                 warn "# new columns ",dump( @columns );
772
773                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
774         }
775
776         my $sorted_items;
777         my $from_end = $sort eq 'd' ? $#$filtered : 0;
778         my $test_changed;
779         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
780         foreach ( 0 .. $limit ) {
781                 my $i = $_ + $offset;
782                 last unless defined $filtered->[$i];
783                 $i = $from_end - $i if $from_end;
784                 my $id = $filtered->[$i];
785                 my $row = Storable::dclone $data->{items}->[ $id ];
786                 if ( $code && $test ) {
787                         my $update;
788                         eval $code;
789                         if ( $@ ) {
790                                 warn "ERROR evaling $@", dump($code);
791                                 $self->stash('eval_error', $@) if $@;
792                         } else {
793                                 warn "EVAL ",dump($update);
794                                 foreach ( keys %$update ) {
795                                         $test_changed->{$_}++;
796                                         $row->{$_} = $update->{$_};
797                                 }
798                         }
799                 }
800                 $row->{_row_id} ||= $id;
801                 push @$sorted_items, $row;
802         }
803
804         if ( $self->param('export') ) {
805                 my $export_path = $self->_export_path( 'items', @columns);
806                 open(my $fh, '>', $export_path) || warn "ERROR: can't open $export_path: $!";
807                 foreach my $f ( 0 .. $#$filtered ) {
808                         print $fh join("\t", map {
809                                 my $i = $data->{items}->[ $filtered->[$f] ];
810                                 my $v = '\N';
811                                 if ( ! defined $i->{$_} ) {
812                                         # nop
813                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
814                                         $v =join(',', @{ $i->{$_} });
815                                         $v = '\N' if length($v) == 0;
816                                 } else {
817                                         $v = dump $i->{$_};
818                                 }
819                                 $v;
820                         } @columns),"\n";
821                 }
822                 close($fh);
823                 warn "export $export_path ", -s $export_path, " bytes\n";
824         }
825
826         warn "# test_changed ",dump( $test_changed );
827         my $c = { map { $_ => 1 } @columns };
828         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
829         unshift @columns, @added_columns;
830
831         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
832
833         my $code_depends = $self->param('code_depends')||
834         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
835         my $code_description = $self->param('code_description') ||
836         join(',', @added_columns);
837
838         $code_depends ||= $code_description; # self-modifing
839         if ( ! $code_depends && $out ) {
840                 $code_depends = $key;
841                 $code_description = $value;
842         }
843
844         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
845
846         $self->render(
847                 order => $order,
848                 offset => $offset,
849                 limit => $limit,
850                 sorted => $sorted_items,
851                 columns => [ @columns ],
852                 rows => $#$filtered + 1,
853                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
854                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
855                 filters => $self->_current_filters,
856                 code => $code,
857                 cols_changed => $commit ? $commit_changed : $test_changed,
858                 code_depends => $code_depends,
859                 code_description => $code_description,
860                 code_path => $code_path,
861                 out => $out,
862         );
863
864 }
865
866
867 sub order {
868         my $self = shift;
869         $self->session('order', $self->param('order'));
870         $self->session('sort', $self->param('sort'));
871         $self->redirect_to('/data/items');
872 }
873
874 sub _is_numeric {
875         my ( $self, $name ) = @_;
876
877         my $stats = $self->_loaded( 'stats' );
878
879         # sort facet numerically if more >50% elements are numeric
880         my $count = $stats->{$name}->{count};
881         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
882         defined $stats->{$name}->{numeric} &&
883                 $stats->{$name}->{numeric} > $count / 2;
884 }
885
886 sub _is_unique {
887         my ( $self, $name ) = @_;
888         my $stats = $self->_loaded( 'stats' );
889         defined $stats->{$name}->{unique};
890 }
891
892 sub _remove_filter {
893         my ($self,$name) = @_;
894         warn "_remove_filter $name\n";
895
896         my $path = $self->session('path');
897
898         delete $filters->{$name};
899         delete $loaded->{$path}->{filters}->{$name};
900         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
901
902         foreach (
903                         grep { /\Q$name\E/ }
904                         keys %{ $loaded->{$path}->{filtered} }
905         ) {
906                 delete $loaded->{$path}->{filtered}->{$_};
907                 warn "remove filtered cache $_";
908         }
909 }
910
911 sub facet {
912         my $self = shift;
913
914         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
915
916         if ( my $name = $self->param('remove') ) {
917                 $self->_remove_filter( $name );
918                 $self->redirect_to( '/data/items' );
919         }
920
921         my $facet;
922         my $name = $self->param('name') || die "no name";
923
924         my $all = $self->_param_scalar('all', 1);
925         my $data = $self->_loaded('data');
926
927         my $filters = $self->_current_filters;
928         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
929         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
930                 if defined $loaded->{$path}->{filtered}->{$all_filters};
931
932         if ( ! $filtered || $all ) {
933                 $filtered = [ 0 .. $#{ $data->{items} } ];
934                 warn "filter all values\n";
935         } else {
936                 warn "filter using $all_filters\n";
937         }
938
939         foreach my $i ( @$filtered ) {
940                 my $item = $data->{items}->[$i];
941                 if ( ! exists $item->{$name} ) {
942                         $facet->{ _missing }++;
943                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
944                         $facet->{$_}++ foreach @{ $item->{$name} };
945                 } else {
946                         $facet->{ $item->{$name} }++;
947                 }
948         }
949
950         if ( my $code = $self->param('code') ) {
951                 my $out;
952                 foreach my $v ( keys %$facet ) {
953                         my $c = $facet->{$v};
954                         eval $code;
955                         if ( $@ ) {
956                                 $out = $@;
957                                 warn "ERROR: $@\n$code\n";
958                                 last;
959                         }
960                 }
961                 warn "out ",dump( $out );
962                 $self->stash( out => $out );
963         }
964
965 #       warn "# facet $name ",dump $facet;
966
967         my $checked;
968         my @facet_names =
969                   $all                      ? keys %$facet
970                 : defined $filters->{$name} ? @{ $filters->{$name} }
971                 : keys %$facet;
972
973         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
974
975         my $numeric = $self->_is_numeric($name);
976
977         my $sort = $self->param('sort');
978         # sort numeric facets with more than 5 values ascending
979         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
980
981         @facet_names = sort {
982                 my $result;
983                 if ( $sort eq 'a' ) {
984                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
985                 } elsif ( $sort eq 'd' ) {
986                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
987                 } elsif ( $sort eq 'c' ) {
988                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
989                 } else {
990                         warn "unknown sort: $sort";
991                         $result = $a cmp $b;
992                 }
993                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
994                 $result;
995         } @facet_names;
996
997         $self->render( name => $name, facet => $facet, checked => $checked,
998                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
999         );
1000 }
1001
1002
1003 sub __invalidate_path_column {
1004         my ( $path, $name ) = @_;
1005
1006         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1007                 delete $loaded->{$path}->{sorted}->{$name};
1008                 warn "# invalidate $path sorted $name\n";
1009         }
1010
1011         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1012                 delete $loaded->{$path}->{filtered}->{$_};
1013                 warn "# invalidate $path filtered $_\n";
1014         }
1015 }
1016
1017 sub __path_modified {
1018         my ( $path, $value ) = @_;
1019         $value = 1 unless defined $value;
1020         
1021         $loaded->{$path}->{modified}  = $value;
1022
1023         warn "# __path_modified $path $value\n";
1024 }
1025
1026 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1027
1028 sub _save_change {
1029         my ($self,$change) = @_;
1030
1031         my $change_path = $self->_permanent_path( 'changes' );
1032         mkdir $change_path unless -d $change_path;
1033         $change_path .= '/' . $change->{time};
1034         store $change, $change_path;
1035         utime $change->{time}, $change->{time}, $change_path;
1036         warn "_save_change $change_path ", dump($change);
1037 }
1038
1039 sub edit {
1040         my $self = shift;
1041         my $new_content = $self->param('new_content');
1042         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1043
1044         my $i = $self->param('_row_id');
1045         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1046         my $path = $self->param('path') || die "no path";
1047         my $name = $self->param('name') || die "no name";
1048         my $status = 200; # 200 = OK, 201 = Created
1049
1050         my $data = $self->_loaded('data');
1051
1052         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1053                 $new_content =~ s/^\s+//s;
1054                 $new_content =~ s/\s+$//s;
1055                 my $v;
1056                 if ( $new_content =~ /\xB6/ ) { # para
1057                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1058                 } else {
1059                         $v = [ $new_content ];
1060                 }
1061
1062                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1063                 my $new = dump $v;
1064                 if ( $old ne $new
1065                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1066                 ) {
1067                         $self->_save_change({
1068                                 path => $path,
1069                                 column => $name,
1070                                 pos => $i,
1071                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1072                                 new => $v,
1073                                 time => $self->param('time') || time(),
1074                                 user => $self->param('user') || $ENV{'LOGNAME'},
1075                                 unique => {
1076                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1077                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1078                                         keys %{ $loaded->{$path}->{stats} }
1079                                 },
1080                         });
1081
1082                         warn "# change $path $i $old -> $new\n";
1083                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1084
1085                         __invalidate_path_column( $path, $name );
1086
1087                         $status = 201; # created
1088                         __path_rebuild_stats( $path );
1089         
1090                         $new_content = join("\xB6",@$v);
1091
1092                 } else {
1093                         warn "# unchanged $path $i $old\n";
1094                         $status = 304;
1095                 }
1096         } else {
1097                 $new_content = "$path $i $name doesn't exist\n";
1098                 $status = 404;
1099         }
1100
1101         warn "# edit $status ", dump $new_content;
1102
1103         $self->render(
1104                 status => $status,
1105                 new_content => scalar $new_content,
1106         );
1107 }
1108
1109
1110 sub save {
1111         my $self = shift;
1112         my $path = $self->_param_or_session('path');
1113         my $dump_path = $self->_save( $path );
1114         __path_modified( $path, 0 );
1115
1116         $self->redirect_to( '/data/items' );
1117 }
1118
1119 sub export {
1120         my $self = shift;
1121
1122         my $dir = $self->app->home->rel_dir('public');
1123
1124         if ( my $import = $self->param('import') ) {
1125
1126                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1127                         my $name = $self->_column_from_unac( $1 );
1128
1129                         my @vals = map { chomp; $_ }
1130                                 read_file "$dir/export/$import", binmode => ':utf8';
1131
1132                         $self->_remove_filter( $name );
1133                         $self->_filter_on_data( $name, @vals );
1134                         $self->session( 'offset' => 0 );
1135                         $self->redirect_to('/data/items');
1136                 } else {
1137                         warn "UNKNOWN IMPORT $import";
1138                 }
1139         }
1140
1141         if ( my $remove = $self->param('remove') ) {
1142                 my $path = "$dir/export/$remove";
1143                 unlink $path if -e $path;
1144                 $path .= '.png';
1145                 unlink $path if -e $path;
1146         }
1147
1148         my $path = $self->_export_path || $self->redirect_to('/data/index');
1149
1150         my @files = grep { ! /\.png$/ } glob "$path/*";
1151         my $mtime = { map { $_ => (stat($_))[9] } @files };
1152         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1153         $self->render( export => [ @files ] );
1154 }
1155
1156 sub __loaded_paths {
1157         return
1158                 grep { defined $loaded->{$_}->{data} }
1159                 keys %$loaded;
1160 }
1161
1162 1;