ededa5881819ca7b7dc20fbcb5bc8b8052d2bdc6
[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                 } else {
47                         #warn "IGNORE: $file\n";
48                 }
49         }, $data_dir);
50
51         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
52                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
53         my $size;
54         $size->{$_} = -s "$data_dir/$_" foreach @files;
55
56         $self->render(
57                 files => [ @files ],
58                 size => $size,
59                 loaded => $loaded,
60                 filters => $filters,
61                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
62                 changes => $changes,
63         );
64 }
65
66 sub _dump_path {
67         my ( $self, $name ) = @_;
68         my $dir = $self->app->home->rel_dir('data');
69         $name =~ s/^$dir//;
70         $name =~ s/\/+/_/g;
71         return '/tmp/mojo_facets.' . $name . '.storable';
72 }
73
74 sub _save {
75         my ( $self, $path ) = @_;
76
77         my $dump_path = $self->_dump_path( $path );
78         my $first_load = ! -e $dump_path;
79         warn "save loaded to $dump_path";
80         my $info = $loaded->{$path};
81         store $info, $dump_path;
82
83         if ( $first_load ) {
84                 my $mtime = $loaded->{$path}->{mtime};
85                 utime $mtime, $mtime, $dump_path;
86                 warn "sync time to $path at $mtime\n";
87         }
88
89         warn $dump_path, ' ', -s $dump_path, " bytes\n";
90         return $dump_path;
91 }
92
93
94 sub __stats {
95
96         my $stats;
97
98         my $nr_items = $#{ $_[0] } + 1;
99
100         warn "__stats $nr_items\n";
101
102         foreach my $e ( @{ $_[0] } ) {
103                 foreach my $n ( keys %$e ) {
104                         $stats->{$n}->{count}++;
105                         my @v;
106                         if ( ref $e->{$n} eq 'ARRAY' ) {
107                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
108                                 @v = @{ $e->{$n} };
109                         } else {
110                                 @v = ( $e->{$n} );
111                         }
112
113                         foreach my $x ( @v ) {
114                                 $stats->{$n}->{numeric}++
115                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
116                                 $stats->{$n}->{empty}++
117                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
118                         }
119
120                 }
121         }
122
123         foreach my $n ( keys %$stats ) {
124                 my $s = $stats->{$n};
125                 next unless defined $s->{array};
126                 if ( $s->{array} == $s->{count} ) {
127                         delete $s->{array};
128                         if ( $s->{count} == $nr_items ) {
129                                 warn "check $n for uniqeness\n";
130                                 my $unique;
131                                 foreach my $e ( @{ $_[0] } ) {
132                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
133                                                 $unique = 0;
134                                                 last;
135                                         }
136                                 }
137                                 if ( $unique ) {
138                                         $stats->{$n}->{unique} = 1;
139                                         #warn "# $n unique ",dump( $unique );
140                                 }
141                         }
142                 }
143         }
144
145         warn "# __stats ",dump($stats);
146
147         return $stats;
148 }
149
150 sub _param_or_session {
151         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
152 }
153
154 sub stats {
155         my $self = shift;
156         my $path = $self->_param_or_session('path');
157         warn "stats $path\n";
158         delete $loaded->{$path}->{stats};
159         $self->redirect_to( '/data/columns' );
160 }
161
162
163 sub _load_path {
164         my ( $self, $path ) = @_;
165
166         return if defined $loaded->{$path}->{'generated'};
167
168         my $full_path = $self->app->home->rel_file( 'data/' . $path );
169         $self->redirect_to('/data/index') unless -r $full_path;
170
171         my $dump_path = $self->_dump_path( $path );
172
173         if ( defined $loaded->{$path}->{data} ) {
174                 my $mtime = (stat($full_path))[9];
175                 return if $loaded->{$path}->{mtime} == $mtime;
176                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
177         } elsif ( -e $dump_path ) {
178                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
179                 my $info = retrieve $dump_path;
180                 $loaded->{ $path } = $info;
181                 return;
182         }
183
184         my $data;
185         if ( -f $full_path ) {
186                 if ( $full_path =~ m/.csv/i ) {
187                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
188                 } else {
189                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
190                 }
191         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
192                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
193         } else {
194                 die "can't load $full_path";
195         }
196
197         my @header;
198
199         if ( defined $data->{header} ) {
200                 if ( ref $data->{header} eq 'ARRAY' ) {
201                         @header = @{ $data->{header} };
202                 } else {
203                         warn "header not array ", dump( $data->{header} );
204                 }
205         }
206
207         my $stats = __stats( $data->{items} );
208
209         @header =
210                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
211                 grep { defined $stats->{$_}->{count} } keys %$stats
212                 unless @header;
213
214         my $info = {
215                 header => [ @header ],
216                 stats  => $stats,
217                 full_path => $full_path,
218                 size => -s $full_path,
219                 mtime => (stat($full_path))[9],
220                 data => $data,
221         };
222
223         $loaded->{ $path } = $info;
224         $self->_save( $path );
225
226 }
227
228
229 sub load {
230         my $self = shift;
231
232         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
233
234         my @paths = $self->param('paths');
235         warn "# paths ", dump @paths;
236
237         foreach my $p ( keys %$loaded ) {
238                 next if grep { /^\Q$p\E$/ } @paths;
239                 warn "remove $p from memory\n";
240                 delete $loaded->{$p};
241         }
242
243         $self->_load_path( $_ ) foreach @paths;
244
245         warn "# path $path\n";
246         $self->_load_path( $path );
247
248         $self->session( 'path' => $path );
249
250         my $redirect_to = '/data/items';
251
252         $self->session( 'header' => $loaded->{$path}->{header} );
253         if ( ! defined $loaded->{$path}->{columns} ) {
254                 my $columns_path = $self->_permanent_path( 'columns' );
255                 if ( -e $columns_path ) {
256                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
257                         $loaded->{$path}->{columns} = [ @columns ];
258                         warn "# columns_path $columns_path ",dump(@columns);
259                 } else {
260                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
261                 }
262
263                 $redirect_to = '/data/columns';
264         }
265         $self->session( 'columns' => $loaded->{$path}->{columns} );
266         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
267         $self->redirect_to( $redirect_to );
268 }
269
270
271 sub _loaded {
272         my ( $self, $name ) = @_;
273         my $path = $self->session('path') || $self->param('path');
274         $self->redirect_to('/data/index') unless $path;
275
276         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
277                 my $caller = (caller(1))[3];
278                 if ( $caller =~ m/::edit/ ) {
279                         warn "rebuild stats for $path ignored caller $caller\n";
280                 } else {
281                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
282 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
283                         $loaded->{$path}->{rebuild_stats} = 1;
284                         $loaded->{$path}->{modified} = 1;
285                 }
286         }
287
288         if ( defined $loaded->{$path}->{rebuild_stats} ) {
289                 warn "rebuild_stats $path";
290                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
291                 delete $loaded->{$path}->{rebuild_stats};
292         }
293
294         if ( ! defined $loaded->{$path}->{$name} ) {
295                 warn "$path $name isn't loaded\n";
296                 $self->_load_path( $path );
297                 if ( ! defined $loaded->{$path}->{stats} ) {
298                         warn "rebuild stats for $path\n";
299                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
300                 }
301                 if ( ! defined $loaded->{$path}->{$name} ) {
302                         warn "MISSING $name for $path\n";
303                         $self->redirect_to('/data/index')
304                 }
305         }
306
307         $self->session( 'modified' => $loaded->{$path}->{modified} );
308
309         return $loaded->{$path}->{$name};
310 }
311
312
313 sub _checked {
314         my $self = shift;
315         my $checked;
316         $checked->{$_}++ foreach @_;
317 #       warn "# _checked ",dump($checked);
318         return $checked;
319 }
320
321 sub _permanent_path {
322         my $self = shift;
323         my $path = $self->_param_or_session('path');
324         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
325 }
326
327 sub __unac {
328         my $n = shift;
329         $n = unac_string($n);
330         $n =~ s/\W+/_/g;
331         return $n;
332 }
333
334 sub _column_from_unac {
335         my ($self,$name) = @_;
336
337         my $stats = $self->_loaded('stats');
338         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
339
340         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
341 }
342
343 sub _export_path {
344         my $self = shift;
345         my $path = $self->_param_or_session('path');
346         if ( ! $path ) {
347                 warn "no path in param or session";
348                 return;
349         }
350         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
351         mkpath $dir unless -e $dir;
352         my $name = join('.', map { __unac($_) } @_ );
353         my $full = $dir . '/' . $name;
354         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
355         return $full;
356 }
357
358 sub columns {
359     my $self = shift;
360
361         if ( $self->param('columns') ) {
362                 my @columns = $self->_param_array('columns');
363                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
364                 $self->redirect_to('/data/items');
365         }
366
367         my $stats = $self->_loaded( 'stats' );
368
369         my @columns;
370         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
371
372         foreach my $c ( @{ $self->session( 'header' ) } ) {
373                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
374         }
375
376     $self->render(
377                 message => 'Select columns to display',
378                 stats => $stats,
379                 columns => \@columns,
380                 checked => $self->_checked( $self->_param_array('columns') ),
381         );
382 }
383
384 sub _param_array {
385     my ($self,$name) = @_;
386
387         my @array = $self->param($name);
388         my $path  = $self->session('path');
389
390         if ( @array ) {
391                 $self->session($name => [ @array ]);
392                 $loaded->{$path}->{$name} = [ @array ];
393         } elsif ( my $session = $self->session($name) ) {
394                 if ( ref $session eq 'ARRAY' ) {
395                         @array = @$session;
396                 } else {
397                         die "$name not array ",dump($session);
398                 }
399         }
400         #warn "# $name ",dump @array;
401         return @array;
402 }
403
404 sub _param_scalar {
405     my ($self,$name,$default) = @_;
406
407         my $scalar = $self->param($name);
408
409         if ( defined $scalar ) {
410                 $self->session($name => $scalar);
411         } else {
412                 $scalar = $self->session($name);
413         }
414
415         if ( ! defined $scalar ) {
416                 $scalar = $default;
417                 die "no default for $name" unless defined $scalar;
418                 $self->session($name => $scalar);
419         }
420
421         warn "# _param_scalar $name ",dump $scalar;
422         return $scalar;
423 }
424
425 sub filter {
426         my $self = shift;
427
428         my $name = $self->param('filter_name') || die "name?";
429         my @vals = $self->param('filter_vals');
430
431         $self->_remove_filter( $name );
432         if ( @vals ) {
433                 $self->_filter_on_data( $name, @vals );
434                 if ( my $permanent = $self->param('_permanent') ) {
435                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
436                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
437                         warn "permanent filter $permanent_path ", -s $permanent_path;
438                 }
439         }
440
441         $self->session( 'offset' => 0 );
442
443         $self->redirect_to('/data/items');
444 }
445
446 sub _filter_on_data {
447         my ( $self, $name, @vals ) = @_;
448
449         my $path = $self->session('path');
450
451         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
452                 warn "filter $name not found in data set";
453                 return;
454         }
455
456         $filters->{$name} = [ @vals ];
457         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
458
459         my $filter_hash;
460         $filter_hash->{$_}++ foreach @vals;
461
462         #warn "# filter_hash ",dump( $filter_hash );
463
464         my $items = $self->_loaded('data')->{items};
465
466         my $include_missing = defined $filter_hash->{_missing};
467         my $filtered_items;
468
469         foreach my $i ( 0 .. $#$items ) {
470
471                 if ( defined $items->[$i]->{$name} ) {
472                         my $row = $items->[$i]->{$name};
473                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
474                         foreach my $v ( @$row ) {
475                                 if ( defined $filter_hash->{ $v } ) {
476                                         $filtered_items->{$i}++;
477                                 }
478                         }
479                 } elsif ( $include_missing ) {
480                         $filtered_items->{$i}++;
481                 }
482         }
483
484         #warn "# filter $name ",dump($filtered_items);
485
486         $loaded->{$path}->{filters}->{$name} = $filtered_items;
487         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
488 }
489
490
491 sub _current_filters {
492         my $self = shift;
493         my $current_filters;
494         my $stats = $self->_loaded('stats');
495
496         $current_filters->{ $_ } = $filters->{ $_ }
497                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
498         ;
499         warn "# _current_filters ",dump( keys %$current_filters );
500         return $current_filters;
501 }
502
503 sub _data_sorted_by {
504         my ( $self, $order ) = @_;
505
506         my $path = $self->session('path');
507
508         warn "_data_sorted_by $order from $path";
509
510         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
511                 return $loaded->{$path}->{sorted}->{$order};
512         }
513
514         my $data = $self->_loaded( 'data' );
515         my $numeric = $self->_is_numeric($order);
516         my $missing = $numeric ? 0 : '';
517         no warnings qw(numeric);
518         my $nr = 0;
519         my @sorted = map {
520                 $_->[0]
521         } sort {
522                 if ( $numeric ) {
523                         $a->[1] <=> $b->[1]
524                 } else {
525                         $a->[1] cmp $b->[1]
526                 }
527         } map {
528                 my $v;
529                 if ( ! exists $_->{$order} ) {
530                         $v = $missing;
531                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
532                         $v = join('', @{$_->{$order}});
533                 } else {
534                         $v = $_->{$order};
535                 }
536                 [ $nr++, $v ]
537         } @{ $data->{items} }
538         ;
539
540         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
541         #warn "# sorted ",dump( @sorted );
542
543         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
544 }
545
546
547 sub __all_filters {
548         my $order = pop @_;
549         join(',', sort(@_), 'order', $order);
550 }
551
552 our $out;
553
554 sub __commit_path_code {
555         my ( $path, $i, $code, $commit_changed, $out_ref ) = @_;
556
557         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
558         my $row = $items->[$i];
559         my $update;
560         eval $code;
561         foreach ( keys %$update ) {
562                 $$commit_changed->{$_}++;
563                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
564         }
565         #warn "__commit_path_code $path $i ",dump( $update );
566 }
567
568 sub items {
569         my $self = shift;
570
571         if ( my $show = $self->param('id') ) {
572                 $self->param('show', $show);
573                 warn "show $show\n";
574         }
575
576         my $path = $self->session('path');
577
578         my @columns = $self->_param_array('columns');
579         $self->redirect_to('/data/columns') unless @columns;
580         my $order   = $self->_param_scalar('order', $columns[0]);
581         my $sort    = $self->_param_scalar('sort', 'a');
582         my $offset  = $self->_param_scalar('offset', 0);
583         my $limit   = $self->_param_scalar('limit', 20);
584         $self->_param_scalar('show', 'table');
585
586         # fix offset when changing limit
587         $offset = int( $offset / $limit ) * $limit;
588
589         if ( ! grep { /^\Q$order\E$/ } @columns ) {
590                 $order = $columns[0];
591                 $self->session( order => $order );
592         }
593         my $sorted = $self->_data_sorted_by( $order );
594
595         my @filter_names;
596         if ( $filters ) {
597                 foreach my $name ( keys %$filters ) {
598                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
599                                 warn "skip filter $name not found in $path\n";
600                                 next;
601                         }
602                         push @filter_names, $name;
603                 }
604                 warn "filter_names ",dump( @filter_names );
605                 foreach my $name ( @filter_names ) {
606                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
607                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
608                 }
609         }
610
611         my $all_filters = __all_filters( @filter_names,$order );
612
613 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
614
615         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
616
617                 my $path_filters = $loaded->{$path}->{filters};
618
619                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
620
621                 my @filtered;
622                 foreach my $i ( 0 .. $#$sorted ) {
623                         my $pos = $sorted->[$i];
624
625                         if ( $#filter_names == -1 ) {
626                                 push @filtered, $pos;
627                                 next;
628                         }
629
630                         my $skip = 0;
631                         foreach ( @filter_names ) {
632                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
633                         }
634                         next if $skip;
635
636                         push @filtered, $pos;
637                 }
638
639                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
640         }
641
642         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
643                 if defined $loaded->{$path}->{filtered}->{$all_filters};
644
645         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
646
647         my $data = $self->_loaded('data');
648
649         my $code = $self->_param_scalar('code','');
650         $code =~ s{[\r\n]+$}{\n}s;
651
652         my $commit = $self->param('commit');
653         my $test = $self->param('test');
654
655         my $commit_changed;
656
657         if ( $code && ( $test || $commit ) ) {
658                 # XXX find columns used in code snippet and show them to user
659                 my $order = 0;
660                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
661                         if ( $column =~ s/^(['"])// ) {
662                                 $column =~ s/$1$//;
663                         }
664                         next if $column =~ m/\$/; # hide columns with vars in them
665                         $commit_changed->{$column} = 0;
666                 }
667         }
668
669         my $code_path = $self->app->home->rel_dir('public') . "/code";
670         if ( $commit ) {
671
672                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
673                 $out = undef;
674                 foreach ( 0 .. $#$filtered ) {
675                         my $i = $filtered->[$_];
676                         __commit_path_code( $path, $i, $code, \$commit_changed );
677                 }
678
679                 warn "out ",dump($out);
680
681                 $self->_save_change({
682                         path => $path,
683                         time => $self->param('time') || time(),
684                         user => $self->param('user') || $ENV{'LOGNAME'},
685                         code => $code,
686                         commit_changed => $commit_changed,
687                 });
688
689                 if ( my $description = $self->param('code_description') ) {
690                         my $depends = $self->param('code_depends') || die "no code_depends?";
691                         my $path = "$code_path/$depends.$description.pl";
692                         if ( -e $path && ! $self->param('overwrite') ) {
693                                 warn "# code $path not saved\n";
694                         } else {
695                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
696                                 warn "code $path ", -s $path, " bytes saved\n";
697                         }
698                 }
699
700                 # remove console
701                 $code = '';
702                 if ( $out ) {
703                         my $commit_dataset = join('.'
704                                 , $self->param('code_depends')
705                                 , $self->param('code_description')
706                                 , time()
707                         );
708                         my $key = $self->param('code_depends');
709                         $key =~ s/,.+$//;
710                         $key ||= 'key';
711                         my $items;
712                         foreach my $n ( keys %$out ) {
713                                 my $i = { $key => [ $n ] };
714                                 my $ref = ref $out->{$n};
715                                 if ( $ref eq 'HASH' ) {
716                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
717                                 } elsif ( $ref eq 'ARRAY' ) {
718                                         $i->{$_} = $out->{$n};
719                                 } elsif ( ! $ref ) {
720                                         $i->{value} = [ $out->{$n} ];
721                                 } else {
722                                         $i->{_error} = [ dump($out->{$n}) ];
723                                 }
724                                 push @$items, $i;
725                         };
726                         undef $out;
727                         my $stats = __stats( $items );
728                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
729                         unshift @columns, $key;
730
731                         $loaded->{$commit_dataset} = {
732                                 header => [ @columns ],
733                                 columns => [ @columns ],
734                                 mtime => time(),
735                                 data => { items => $items },
736                                 stats => $stats,
737                                 generated => 1,
738                         };
739                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
740                         $self->session('path', $commit_dataset);
741                         $self->session('columns', [ @columns ]);
742                         $self->session('order', $key);
743                         $self->redirect_to('/data/items');
744                         return; # FIXME needed to correctly show columns
745                 }
746
747                 # this might move before $out to recalculate stats on source dataset?
748                 __path_rebuild_stats( $path );
749                 my $c = { map { $_ => 1 } @columns };
750                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
751                 warn "# added_columns ",dump( @added_columns );
752                 unshift @columns, @added_columns;
753
754                 $self->session('columns', [ @columns ]);
755                 $loaded->{$path}->{columns} = [ @columns ];
756                 warn "# new columns ",dump( @columns );
757
758                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
759         }
760
761         my $sorted_items;
762         my $from_end = $sort eq 'd' ? $#$filtered : 0;
763         my $test_changed;
764         my $out;
765         foreach ( 0 .. $limit ) {
766                 my $i = $_ + $offset;
767                 last unless defined $filtered->[$i];
768                 $i = $from_end - $i if $from_end;
769                 my $id = $filtered->[$i];
770                 my $row = Storable::dclone $data->{items}->[ $id ];
771                 if ( $code && $test ) {
772                         my $update;
773                         eval $code;
774                         if ( $@ ) {
775                                 warn "ERROR evaling\n$code\n$@";
776                                 $self->stash('eval_error', $@) if $@;
777                         } else {
778                                 warn "EVAL ",dump($update);
779                                 foreach ( keys %$update ) {
780                                         $test_changed->{$_}++;
781                                         $row->{$_} = $update->{$_};
782                                 }
783                         }
784                 }
785                 $row->{_row_id} ||= $id;
786                 push @$sorted_items, $row;
787         }
788
789         if ( $self->param('export') ) {
790                 my $export_path = $self->_export_path( 'items', @columns);
791                 open(my $fh, '>', $export_path) || warn "ERROR: can't open $export_path: $!";
792                 foreach my $f ( 0 .. $#$filtered ) {
793                         print $fh join("\t", map {
794                                 my $i = $data->{items}->[ $filtered->[$f] ];
795                                 my $v = '\N';
796                                 if ( ! defined $i->{$_} ) {
797                                         # nop
798                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
799                                         $v =join(',', @{ $i->{$_} });
800                                         $v = '\N' if length($v) == 0;
801                                 } else {
802                                         $v = dump $i->{$_};
803                                 }
804                                 $v;
805                         } @columns),"\n";
806                 }
807                 close($fh);
808                 warn "export $export_path ", -s $export_path, " bytes\n";
809         }
810
811         warn "# test_changed ",dump( $test_changed );
812         my $c = { map { $_ => 1 } @columns };
813         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
814         unshift @columns, @added_columns;
815
816         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
817
818         my $code_depends = $self->param('code_depends')||
819         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
820         my $code_description = $self->param('code_description') ||
821         join(',', @added_columns);
822
823         $code_depends ||= $code_description; # self-modifing
824
825         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
826
827         $self->render(
828                 order => $order,
829                 offset => $offset,
830                 limit => $limit,
831                 sorted => $sorted_items,
832                 columns => [ @columns ],
833                 rows => $#$filtered + 1,
834                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
835                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
836                 filters => $self->_current_filters,
837                 code => $code,
838                 cols_changed => $commit ? $commit_changed : $test_changed,
839                 code_depends => $code_depends,
840                 code_description => $code_description,
841                 code_path => $code_path,
842                 out => $out,
843         );
844
845 }
846
847
848 sub order {
849         my $self = shift;
850         $self->session('order', $self->param('order'));
851         $self->session('sort', $self->param('sort'));
852         $self->redirect_to('/data/items');
853 }
854
855 sub _is_numeric {
856         my ( $self, $name ) = @_;
857
858         my $stats = $self->_loaded( 'stats' );
859
860         # sort facet numerically if more >50% elements are numeric
861         my $count = $stats->{$name}->{count};
862         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
863         defined $stats->{$name}->{numeric} &&
864                 $stats->{$name}->{numeric} > $count / 2;
865 }
866
867 sub _is_unique {
868         my ( $self, $name ) = @_;
869         my $stats = $self->_loaded( 'stats' );
870         defined $stats->{$name}->{unique};
871 }
872
873 sub _remove_filter {
874         my ($self,$name) = @_;
875         warn "_remove_filter $name\n";
876
877         my $path = $self->session('path');
878
879         delete $filters->{$name};
880         delete $loaded->{$path}->{filters}->{$name};
881         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
882
883         foreach (
884                         grep { /\Q$name\E/ }
885                         keys %{ $loaded->{$path}->{filtered} }
886         ) {
887                 delete $loaded->{$path}->{filtered}->{$_};
888                 warn "remove filtered cache $_";
889         }
890 }
891
892 sub facet {
893         my $self = shift;
894
895         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
896
897         if ( my $name = $self->param('remove') ) {
898                 $self->_remove_filter( $name );
899                 $self->redirect_to( '/data/items' );
900         }
901
902         my $facet;
903         my $name = $self->param('name') || die "no name";
904
905         my $all = $self->_param_scalar('all', 1);
906         my $data = $self->_loaded('data');
907
908         my $filters = $self->_current_filters;
909         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
910         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
911                 if defined $loaded->{$path}->{filtered}->{$all_filters};
912
913         if ( ! $filtered || $all ) {
914                 $filtered = [ 0 .. $#{ $data->{items} } ];
915                 warn "filter all values\n";
916         } else {
917                 warn "filter using $all_filters\n";
918         }
919
920         foreach my $i ( @$filtered ) {
921                 my $item = $data->{items}->[$i];
922                 if ( ! exists $item->{$name} ) {
923                         $facet->{ _missing }++;
924                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
925                         $facet->{$_}++ foreach @{ $item->{$name} };
926                 } else {
927                         $facet->{ $item->{$name} }++;
928                 }
929         }
930
931 #       warn "# facet $name ",dump $facet;
932
933         my $checked;
934         my @facet_names =
935                   $all                      ? keys %$facet
936                 : defined $filters->{$name} ? @{ $filters->{$name} }
937                 : keys %$facet;
938
939         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
940
941         my $numeric = $self->_is_numeric($name);
942
943         my $sort = $self->param('sort');
944         # sort numeric facets with more than 5 values ascending
945         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
946
947         @facet_names = sort {
948                 my $result;
949                 if ( $sort eq 'a' ) {
950                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
951                 } elsif ( $sort eq 'd' ) {
952                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
953                 } elsif ( $sort eq 'c' ) {
954                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
955                 } else {
956                         warn "unknown sort: $sort";
957                         $result = $a cmp $b;
958                 }
959                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
960                 $result;
961         } @facet_names;
962
963         $self->render( name => $name, facet => $facet, checked => $checked,
964                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
965         );
966 }
967
968
969 sub __invalidate_path_column {
970         my ( $path, $name ) = @_;
971
972         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
973                 delete $loaded->{$path}->{sorted}->{$name};
974                 warn "# invalidate $path sorted $name\n";
975         }
976
977         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
978                 delete $loaded->{$path}->{filtered}->{$_};
979                 warn "# invalidate $path filtered $_\n";
980         }
981 }
982
983 sub __path_modified {
984         my ( $path, $value ) = @_;
985         $value = 1 unless defined $value;
986         
987         $loaded->{$path}->{modified}  = $value;
988
989         warn "# __path_modified $path $value\n";
990 }
991
992 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
993
994 sub _save_change {
995         my ($self,$change) = @_;
996
997         my $change_path = $self->_permanent_path( 'changes' );
998         mkdir $change_path unless -d $change_path;
999         $change_path .= '/' . $change->{time};
1000         store $change, $change_path;
1001         utime $change->{time}, $change->{time}, $change_path;
1002         warn "_save_change $change_path ", dump($change);
1003 }
1004
1005 sub edit {
1006         my $self = shift;
1007         my $new_content = $self->param('new_content');
1008         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1009
1010         my $i = $self->param('_row_id');
1011         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1012         my $path = $self->param('path') || die "no path";
1013         my $name = $self->param('name') || die "no name";
1014         my $status = 200; # 200 = OK, 201 = Created
1015
1016         my $data = $self->_loaded('data');
1017
1018         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1019                 $new_content =~ s/^\s+//s;
1020                 $new_content =~ s/\s+$//s;
1021                 my $v;
1022                 if ( $new_content =~ /\xB6/ ) { # para
1023                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1024                 } else {
1025                         $v = [ $new_content ];
1026                 }
1027
1028                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1029                 my $new = dump $v;
1030                 if ( $old ne $new
1031                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1032                 ) {
1033                         $self->_save_change({
1034                                 path => $path,
1035                                 column => $name,
1036                                 pos => $i,
1037                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1038                                 new => $v,
1039                                 time => $self->param('time') || time(),
1040                                 user => $self->param('user') || $ENV{'LOGNAME'},
1041                                 unique => {
1042                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1043                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1044                                         keys %{ $loaded->{$path}->{stats} }
1045                                 },
1046                         });
1047
1048                         warn "# change $path $i $old -> $new\n";
1049                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1050
1051                         __invalidate_path_column( $path, $name );
1052
1053                         $status = 201; # created
1054                         __path_rebuild_stats( $path );
1055         
1056                         $new_content = join("\xB6",@$v);
1057
1058                 } else {
1059                         warn "# unchanged $path $i $old\n";
1060                         $status = 304;
1061                 }
1062         } else {
1063                 $new_content = "$path $i $name doesn't exist\n";
1064                 $status = 404;
1065         }
1066
1067         warn "# edit $status ", dump $new_content;
1068
1069         $self->render(
1070                 status => $status,
1071                 new_content => scalar $new_content,
1072         );
1073 }
1074
1075
1076 sub save {
1077         my $self = shift;
1078         my $path = $self->_param_or_session('path');
1079         my $dump_path = $self->_save( $path );
1080         __path_modified( $path, 0 );
1081
1082         $self->redirect_to( '/data/items' );
1083 }
1084
1085 sub export {
1086         my $self = shift;
1087
1088         my $dir = $self->app->home->rel_dir('public');
1089
1090         if ( my $import = $self->param('import') ) {
1091
1092                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1093                         my $name = $self->_column_from_unac( $1 );
1094
1095                         my @vals = map { chomp; $_ }
1096                                 read_file "$dir/export/$import", binmode => ':utf8';
1097
1098                         $self->_remove_filter( $name );
1099                         $self->_filter_on_data( $name, @vals );
1100                         $self->session( 'offset' => 0 );
1101                         $self->redirect_to('/data/items');
1102                 } else {
1103                         warn "UNKNOWN IMPORT $import";
1104                 }
1105         }
1106
1107         if ( my $remove = $self->param('remove') ) {
1108                 my $path = "$dir/export/$remove";
1109                 unlink $path if -e $path;
1110                 $path .= '.png';
1111                 unlink $path if -e $path;
1112         }
1113
1114         my $path = $self->_export_path || $self->redirect_to('/data/index');
1115
1116         my @files = grep { ! /\.png$/ } glob "$path/*";
1117         my $mtime = { map { $_ => (stat($_))[9] } @files };
1118         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1119         $self->render( export => [ @files ] );
1120 }
1121
1122 sub __loaded_paths {
1123         return
1124                 grep { defined $loaded->{$_}->{data} }
1125                 keys %$loaded;
1126 }
1127
1128 1;