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