3681e0f761d0b1e6319707afffdb7f9fbe640f8d
[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}->{modified} = 1;
284                 }
285         }
286
287         if ( ! defined $loaded->{$path}->{$name} ) {
288                 warn "$path $name isn't loaded\n";
289                 $self->_load_path( $path );
290                 if ( ! defined $loaded->{$path}->{stats} ) {
291                         warn "rebuild stats for $path\n";
292                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
293                 }
294                 if ( ! defined $loaded->{$path}->{$name} ) {
295                         warn "MISSING $name for $path\n";
296                         $self->redirect_to('/data/index')
297                 }
298         }
299
300         $self->session( 'modified' => $loaded->{$path}->{modified} );
301
302         return $loaded->{$path}->{$name};
303 }
304
305
306 sub _checked {
307         my $self = shift;
308         my $checked;
309         $checked->{$_}++ foreach @_;
310 #       warn "# _checked ",dump($checked);
311         return $checked;
312 }
313
314 sub _permanent_path {
315         my $self = shift;
316         my $path = $self->_param_or_session('path');
317         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
318 }
319
320 sub __unac {
321         my $n = shift;
322         $n = unac_string($n);
323         $n =~ s/\W+/_/g;
324         return $n;
325 }
326
327 sub _column_from_unac {
328         my ($self,$name) = @_;
329
330         my $stats = $self->_loaded('stats');
331         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
332
333         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
334 }
335
336 sub _export_path {
337         my $self = shift;
338         my $path = $self->_param_or_session('path');
339         if ( ! $path ) {
340                 warn "no path in param or session";
341                 return;
342         }
343         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
344         mkpath $dir unless -e $dir;
345         my $name = join('.', map { __unac($_) } @_ );
346         $dir . '/' . $name;
347 }
348
349 sub columns {
350     my $self = shift;
351
352         if ( $self->param('columns') ) {
353                 my @columns = $self->_param_array('columns');
354                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
355                 $self->redirect_to('/data/items');
356         }
357
358         my $stats = $self->_loaded( 'stats' );
359
360         my @columns;
361         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
362
363         foreach my $c ( @{ $self->session( 'header' ) } ) {
364                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
365         }
366
367     $self->render(
368                 message => 'Select columns to display',
369                 stats => $stats,
370                 columns => \@columns,
371                 checked => $self->_checked( $self->_param_array('columns') ),
372         );
373 }
374
375 sub _param_array {
376     my ($self,$name) = @_;
377
378         my @array = $self->param($name);
379         my $path  = $self->session('path');
380
381         if ( @array ) {
382                 $self->session($name => [ @array ]);
383                 $loaded->{$path}->{$name} = [ @array ];
384         } elsif ( my $session = $self->session($name) ) {
385                 if ( ref $session eq 'ARRAY' ) {
386                         @array = @$session;
387                 } else {
388                         die "$name not array ",dump($session);
389                 }
390         }
391         #warn "# $name ",dump @array;
392         return @array;
393 }
394
395 sub _param_scalar {
396     my ($self,$name,$default) = @_;
397
398         my $scalar = $self->param($name);
399
400         if ( defined $scalar ) {
401                 $self->session($name => $scalar);
402         } else {
403                 $scalar = $self->session($name);
404         }
405
406         if ( ! defined $scalar ) {
407                 $scalar = $default;
408                 die "no default for $name" unless defined $scalar;
409                 $self->session($name => $scalar);
410         }
411
412         warn "# _param_scalar $name ",dump $scalar;
413         return $scalar;
414 }
415
416 sub filter {
417         my $self = shift;
418
419         my $name = $self->param('filter_name') || die "name?";
420         my @vals = $self->param('filter_vals');
421
422         $self->_remove_filter( $name );
423         if ( @vals ) {
424                 $self->_filter_on_data( $name, @vals );
425                 if ( my $permanent = $self->param('_permanent') ) {
426                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
427                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
428                         warn "permanent filter $permanent_path ", -s $permanent_path;
429                 }
430         }
431
432         $self->session( 'offset' => 0 );
433
434         $self->redirect_to('/data/items');
435 }
436
437 sub _filter_on_data {
438         my ( $self, $name, @vals ) = @_;
439
440         my $path = $self->session('path');
441
442         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
443                 warn "filter $name not found in data set";
444                 return;
445         }
446
447         $filters->{$name} = [ @vals ];
448         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
449
450         my $filter_hash;
451         $filter_hash->{$_}++ foreach @vals;
452
453         #warn "# filter_hash ",dump( $filter_hash );
454
455         my $items = $self->_loaded('data')->{items};
456
457         my $include_missing = defined $filter_hash->{_missing};
458         my $filtered_items;
459
460         foreach my $i ( 0 .. $#$items ) {
461
462                 if ( defined $items->[$i]->{$name} ) {
463                         my $row = $items->[$i]->{$name};
464                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
465                         foreach my $v ( @$row ) {
466                                 if ( defined $filter_hash->{ $v } ) {
467                                         $filtered_items->{$i}++;
468                                 }
469                         }
470                 } elsif ( $include_missing ) {
471                         $filtered_items->{$i}++;
472                 }
473         }
474
475         #warn "# filter $name ",dump($filtered_items);
476
477         $loaded->{$path}->{filters}->{$name} = $filtered_items;
478         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
479 }
480
481
482 sub _current_filters {
483         my $self = shift;
484         my $current_filters;
485         my $stats = $self->_loaded('stats');
486
487         $current_filters->{ $_ } = $filters->{ $_ }
488                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
489         ;
490         warn "# _current_filters ",dump( keys %$current_filters );
491         return $current_filters;
492 }
493
494 sub _data_sorted_by {
495         my ( $self, $order ) = @_;
496
497         my $path = $self->session('path');
498
499         warn "_data_sorted_by $order from $path";
500
501         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
502                 return $loaded->{$path}->{sorted}->{$order};
503         }
504
505         my $data = $self->_loaded( 'data' );
506         my $numeric = $self->_is_numeric($order);
507         my $missing = $numeric ? 0 : '';
508         no warnings qw(numeric);
509         my $nr = 0;
510         my @sorted = map {
511                 $_->[0]
512         } sort {
513                 if ( $numeric ) {
514                         $a->[1] <=> $b->[1]
515                 } else {
516                         $a->[1] cmp $b->[1]
517                 }
518         } map {
519                 my $v;
520                 if ( ! exists $_->{$order} ) {
521                         $v = $missing;
522                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
523                         $v = join('', @{$_->{$order}});
524                 } else {
525                         $v = $_->{$order};
526                 }
527                 [ $nr++, $v ]
528         } @{ $data->{items} }
529         ;
530
531         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
532         #warn "# sorted ",dump( @sorted );
533
534         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
535 }
536
537
538 sub __all_filters {
539         my $order = pop @_;
540         join(',', sort(@_), 'order', $order);
541 }
542
543 sub items {
544         my $self = shift;
545
546         if ( my $show = $self->param('id') ) {
547                 $self->param('show', $show);
548                 warn "show $show\n";
549         }
550
551         my $path = $self->session('path');
552
553         my @columns = $self->_param_array('columns');
554         $self->redirect_to('/data/columns') unless @columns;
555         my $order   = $self->_param_scalar('order', $columns[0]);
556         my $sort    = $self->_param_scalar('sort', 'a');
557         my $offset  = $self->_param_scalar('offset', 0);
558         my $limit   = $self->_param_scalar('limit', 20);
559         $self->_param_scalar('show', 'table');
560
561         # fix offset when changing limit
562         $offset = int( $offset / $limit ) * $limit;
563
564         if ( ! grep { /^\Q$order\E$/ } @columns ) {
565                 $order = $columns[0];
566                 $self->session( order => $order );
567         }
568         my $sorted = $self->_data_sorted_by( $order );
569
570         my @filter_names;
571         if ( $filters ) {
572                 foreach my $name ( keys %$filters ) {
573                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
574                                 warn "skip filter $name not found in $path\n";
575                                 next;
576                         }
577                         push @filter_names, $name;
578                 }
579                 warn "filter_names ",dump( @filter_names );
580                 foreach my $name ( @filter_names ) {
581                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
582                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
583                 }
584         }
585
586         my $all_filters = __all_filters( @filter_names,$order );
587
588 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
589
590         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
591
592                 my $path_filters = $loaded->{$path}->{filters};
593
594                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
595
596                 my @filtered;
597                 foreach my $i ( 0 .. $#$sorted ) {
598                         my $pos = $sorted->[$i];
599
600                         if ( $#filter_names == -1 ) {
601                                 push @filtered, $pos;
602                                 next;
603                         }
604
605                         my $skip = 0;
606                         foreach ( @filter_names ) {
607                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
608                         }
609                         next if $skip;
610
611                         push @filtered, $pos;
612                 }
613
614                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
615         }
616
617         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
618                 if defined $loaded->{$path}->{filtered}->{$all_filters};
619
620         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
621
622         my $data = $self->_loaded('data');
623
624         my $code = $self->_param_scalar('code','');
625         $code =~ s{[\r\n]+$}{}s;
626
627         my $commit = $self->param('commit');
628         my $test = $self->param('test');
629
630         my $commit_changed;
631
632         if ( $code && ( $test || $commit ) ) {
633                 # XXX find columns used in code snippet and show them to user
634                 my $order = 0;
635                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
636                         if ( $column =~ s/^(['"])// ) {
637                                 $column =~ s/$1$//;
638                         }
639                         next if $column =~ m/\$/; # hide columns with vars in them
640                         $commit_changed->{$column} = 0;
641                 }
642         }
643
644         my $code_path = $self->app->home->rel_dir('public') . "/code";
645         if ( $commit ) {
646
647                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
648                 my $out;
649                 foreach ( 0 .. $#$filtered ) {
650                         my $i = $filtered->[$_];
651                         my $row = $data->{items}->[$i];
652                         my $update;
653                         eval $code;
654                         foreach ( keys %$update ) {
655                                 $commit_changed->{$_}++;
656                                 $row->{$_} = $update->{$_};
657                         }
658                 }
659                 if ( my $description = $self->param('code_description') ) {
660                         my $depends = $self->param('code_depends') || die "no code_depends?";
661                         my $path = "$code_path/$depends.$description.pl";
662                         if ( -e $path && ! $self->param('overwrite') ) {
663                                 warn "# code $path not saved\n";
664                         } else {
665                                 write_file(  $path, { binmode => ':utf8' }, $code );
666                                 warn "code $path ", -s $path, " bytes saved\n";
667                         }
668                 }
669                 $code = '';
670                 if ( $out ) {
671                         my $commit_dataset = join('.'
672                                 , $self->param('code_depends')
673                                 , $self->param('code_description')
674                                 , time()
675                         );
676                         my $key = $self->param('code_depends');
677                         $key =~ s/,.+$//;
678                         $key ||= 'key';
679                         my $items;
680                         foreach my $n ( keys %$out ) {
681                                 my $i = { $key => [ $n ] };
682                                 my $ref = ref $out->{$n};
683                                 if ( $ref eq 'HASH' ) {
684                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
685                                 } elsif ( $ref eq 'ARRAY' ) {
686                                         $i->{$_} = $out->{$n};
687                                 } elsif ( ! $ref ) {
688                                         $i->{value} = [ $out->{$n} ];
689                                 } else {
690                                         $i->{_error} = [ dump($out->{$n}) ];
691                                 }
692                                 push @$items, $i;
693                         };
694                         undef $out;
695                         my $stats = __stats( $items );
696                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
697                         unshift @columns, $key;
698
699                         $loaded->{$commit_dataset} = {
700                                 header => [ @columns ],
701                                 columns => [ @columns ],
702                                 mtime => time(),
703                                 data => { items => $items },
704                                 stats => $stats,
705                                 generated => 1,
706                         };
707                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
708                         $self->session('path', $commit_dataset);
709                         $self->session('columns', [ @columns ]);
710                         $self->session('order', $key);
711                         $self->redirect_to('/data/items');
712                         return; # FIXME needed to correctly show columns
713                 }
714
715                 # this might move before $out to recalculate stats on source dataset?
716                 __path_modified( $path, 2 );
717                 my $c = { map { $_ => 1 } @columns };
718                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
719                 warn "# added_columns ",dump( @added_columns );
720                 unshift @columns, @added_columns;
721
722                 $self->session('columns', [ @columns ]);
723                 $loaded->{$path}->{columns} = [ @columns ];
724                 warn "# new columns ",dump( @columns );
725
726                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
727         }
728
729         my $sorted_items;
730         my $from_end = $sort eq 'd' ? $#$filtered : 0;
731         my $test_changed;
732         my $out;
733         foreach ( 0 .. $limit ) {
734                 my $i = $_ + $offset;
735                 last unless defined $filtered->[$i];
736                 $i = $from_end - $i if $from_end;
737                 my $id = $filtered->[$i];
738                 my $row = Storable::dclone $data->{items}->[ $id ];
739                 if ( $code && $test ) {
740                         my $update;
741                         eval $code;
742                         if ( $@ ) {
743                                 warn "ERROR evaling\n$code\n$@";
744                                 $self->stash('eval_error', $@) if $@;
745                         } else {
746                                 warn "EVAL ",dump($update);
747                                 foreach ( keys %$update ) {
748                                         $test_changed->{$_}++;
749                                         $row->{$_} = $update->{$_};
750                                 }
751                         }
752                 }
753                 $row->{_row_id} ||= $id;
754                 push @$sorted_items, $row;
755         }
756
757         if ( $self->param('export') ) {
758                 my $export_path = $self->_export_path( 'items', @columns);
759                 open(my $fh, '>', $export_path) || warn "ERROR: can't open $export_path: $!";
760                 foreach my $f ( 0 .. $#$filtered ) {
761                         print $fh join("\t", map {
762                                 my $i = $data->{items}->[ $filtered->[$f] ];
763                                 my $v = '\N';
764                                 if ( ! defined $i->{$_} ) {
765                                         # nop
766                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
767                                         $v =join(',', @{ $i->{$_} });
768                                         $v = '\N' if length($v) == 0;
769                                 } else {
770                                         $v = dump $i->{$_};
771                                 }
772                                 $v;
773                         } @columns),"\n";
774                 }
775                 close($fh);
776                 warn "export $export_path ", -s $export_path, " bytes\n";
777         }
778
779         warn "# test_changed ",dump( $test_changed );
780         my $c = { map { $_ => 1 } @columns };
781         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
782         unshift @columns, @added_columns;
783
784         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
785
786         my $code_depends = $self->param('code_depends')||
787         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
788         my $code_description = $self->param('code_description') ||
789         join(',', @added_columns);
790
791         $code_depends ||= $code_description; # self-modifing
792
793         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
794
795         $self->render(
796                 order => $order,
797                 offset => $offset,
798                 limit => $limit,
799                 sorted => $sorted_items,
800                 columns => [ @columns ],
801                 rows => $#$filtered + 1,
802                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
803                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
804                 filters => $self->_current_filters,
805                 code => $code,
806                 cols_changed => $commit ? $commit_changed : $test_changed,
807                 code_depends => $code_depends,
808                 code_description => $code_description,
809                 code_path => $code_path,
810                 out => $out,
811         );
812
813 }
814
815
816 sub order {
817         my $self = shift;
818         $self->session('order', $self->param('order'));
819         $self->session('sort', $self->param('sort'));
820         $self->redirect_to('/data/items');
821 }
822
823 sub _is_numeric {
824         my ( $self, $name ) = @_;
825
826         my $stats = $self->_loaded( 'stats' );
827
828         # sort facet numerically if more >50% elements are numeric
829         my $count = $stats->{$name}->{count};
830         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
831         defined $stats->{$name}->{numeric} &&
832                 $stats->{$name}->{numeric} > $count / 2;
833 }
834
835 sub _is_unique {
836         my ( $self, $name ) = @_;
837         my $stats = $self->_loaded( 'stats' );
838         defined $stats->{$name}->{unique};
839 }
840
841 sub _remove_filter {
842         my ($self,$name) = @_;
843         warn "_remove_filter $name\n";
844
845         my $path = $self->session('path');
846
847         delete $filters->{$name};
848         delete $loaded->{$path}->{filters}->{$name};
849         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
850
851         foreach (
852                         grep { /\Q$name\E/ }
853                         keys %{ $loaded->{$path}->{filtered} }
854         ) {
855                 delete $loaded->{$path}->{filtered}->{$_};
856                 warn "remove filtered cache $_";
857         }
858 }
859
860 sub facet {
861         my $self = shift;
862
863         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
864
865         if ( my $name = $self->param('remove') ) {
866                 $self->_remove_filter( $name );
867                 $self->redirect_to( '/data/items' );
868         }
869
870         my $facet;
871         my $name = $self->param('name') || die "no name";
872
873         my $all = $self->_param_scalar('all', 1);
874         my $data = $self->_loaded('data');
875
876         my $filters = $self->_current_filters;
877         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
878         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
879                 if defined $loaded->{$path}->{filtered}->{$all_filters};
880
881         if ( ! $filtered || $all ) {
882                 $filtered = [ 0 .. $#{ $data->{items} } ];
883                 warn "filter all values\n";
884         } else {
885                 warn "filter using $all_filters\n";
886         }
887
888         foreach my $i ( @$filtered ) {
889                 my $item = $data->{items}->[$i];
890                 if ( ! exists $item->{$name} ) {
891                         $facet->{ _missing }++;
892                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
893                         $facet->{$_}++ foreach @{ $item->{$name} };
894                 } else {
895                         $facet->{ $item->{$name} }++;
896                 }
897         }
898
899 #       warn "# facet $name ",dump $facet;
900
901         my $checked;
902         my @facet_names =
903                   $all                      ? keys %$facet
904                 : defined $filters->{$name} ? @{ $filters->{$name} }
905                 : keys %$facet;
906
907         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
908
909         my $numeric = $self->_is_numeric($name);
910
911         my $sort = $self->param('sort');
912         # sort numeric facets with more than 5 values ascending
913         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
914
915         @facet_names = sort {
916                 my $result;
917                 if ( $sort eq 'a' ) {
918                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
919                 } elsif ( $sort eq 'd' ) {
920                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
921                 } elsif ( $sort eq 'c' ) {
922                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
923                 } else {
924                         warn "unknown sort: $sort";
925                         $result = $a cmp $b;
926                 }
927                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
928                 $result;
929         } @facet_names;
930
931         $self->render( name => $name, facet => $facet, checked => $checked,
932                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
933         );
934 }
935
936
937 sub __invalidate_path_column {
938         my ( $path, $name ) = @_;
939
940         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
941                 delete $loaded->{$path}->{sorted}->{$name};
942                 warn "# invalidate $path sorted $name\n";
943         }
944
945         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
946                 delete $loaded->{$path}->{filtered}->{$_};
947                 warn "# invalidate $path filtered $_\n";
948         }
949 }
950
951 sub __path_modified {
952         my ( $path, $value ) = @_;
953         $value = 1 unless defined $value;
954         
955         $loaded->{$path}->{modified}  = $value;
956
957         warn "# __path_modified $path $value\n";
958 }
959
960 sub edit {
961         my $self = shift;
962         my $new_content = $self->param('new_content');
963         $new_content  ||= $self->param('content'); # backward compatibility with old actions
964
965         my $i = $self->param('_row_id');
966         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
967         my $path = $self->param('path') || die "no path";
968         my $name = $self->param('name') || die "no name";
969         my $status = 200; # 200 = OK, 201 = Created
970
971         my $data = $self->_loaded('data');
972
973         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
974                 $new_content =~ s/^\s+//s;
975                 $new_content =~ s/\s+$//s;
976                 my $v;
977                 if ( $new_content =~ /\xB6/ ) { # para
978                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
979                 } else {
980                         $v = [ $new_content ];
981                 }
982
983                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
984                 my $new = dump $v;
985                 if ( $old ne $new
986                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
987                 ) {
988                         my $change = {
989                                 path => $path,
990                                 column => $name,
991                                 pos => $i,
992                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
993                                 new => $v,
994                                 time => $self->param('time') || time(),
995                                 user => $self->param('user') || $ENV{'LOGNAME'},
996                                 unique => {
997                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
998                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
999                                         keys %{ $loaded->{$path}->{stats} }
1000                                 },
1001                         };
1002                         my $change_path = $self->_permanent_path( 'changes' );
1003                         mkdir $change_path unless -d $change_path;
1004                         $change_path .= '/' . $change->{time};
1005                         store $change, $change_path;
1006                         utime $change->{time}, $change->{time}, $change_path;
1007                         warn "# $change_path ", dump($change);
1008
1009                         warn "# change $path $i $old -> $new\n";
1010                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1011
1012                         __invalidate_path_column( $path, $name );
1013
1014                         $status = 201; # created
1015                         # modified = 2 -- force rebuild of stats
1016                         __path_modified( $path, 2 );
1017         
1018                         $new_content = join("\xB6",@$v);
1019
1020                 } else {
1021                         warn "# unchanged $path $i $old\n";
1022                         $status = 304;
1023                 }
1024         } else {
1025                 $new_content = "$path $i $name doesn't exist\n";
1026                 $status = 404;
1027         }
1028
1029         warn "# edit $status ", dump $new_content;
1030
1031         $self->render(
1032                 status => $status,
1033                 new_content => scalar $new_content,
1034         );
1035 }
1036
1037
1038 sub save {
1039         my $self = shift;
1040         my $path = $self->_param_or_session('path');
1041         my $dump_path = $self->_save( $path );
1042         __path_modified( $path, 0 );
1043
1044         $self->redirect_to( '/data/items' );
1045 }
1046
1047 sub export {
1048         my $self = shift;
1049
1050         my $dir = $self->app->home->rel_dir('public');
1051
1052         if ( my $import = $self->param('import') ) {
1053
1054                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1055                         my $name = $self->_column_from_unac( $1 );
1056
1057                         my @vals = map { chomp; $_ }
1058                                 read_file "$dir/export/$import", binmode => ':utf8';
1059
1060                         $self->_remove_filter( $name );
1061                         $self->_filter_on_data( $name, @vals );
1062                         $self->session( 'offset' => 0 );
1063                         $self->redirect_to('/data/items');
1064                 } else {
1065                         warn "UNKNOWN IMPORT $import";
1066                 }
1067         }
1068
1069         if ( my $remove = $self->param('remove') ) {
1070                 my $path = "$dir/export/$remove";
1071                 unlink $path if -e $path;
1072                 $path .= '.png';
1073                 unlink $path if -e $path;
1074         }
1075
1076         my $path = $self->_export_path || $self->redirect_to('/data/index');
1077
1078         my @files = grep { ! /\.png$/ } glob "$path/*";
1079         my $mtime = { map { $_ => (stat($_))[9] } @files };
1080         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1081         $self->render( export => [ @files ] );
1082 }
1083
1084 sub __loaded_paths {
1085         return
1086                 grep { defined $loaded->{$_}->{data} }
1087                 keys %$loaded;
1088 }
1089
1090 1;