take path from params in /data/items
[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 use Digest::MD5;
18
19 use MojoFacets::Import::File;
20 use MojoFacets::Import::HTMLTable;
21 use MojoFacets::Import::CSV;
22
23 our $loaded;
24 our $filters;
25
26 sub index {
27         my $self = shift;
28
29         my $data_dir = $self->app->home->rel_dir('data');
30         die "no data dir $data_dir" unless -d $data_dir;
31
32         my @files;
33         my $changes;
34         find( sub {
35                 my $file = $File::Find::name;
36                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
37                         $file =~ s/$data_dir\/*//;
38                         push @files, $file;
39                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
40                         push @{ $changes->{$1} }, $2
41                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
42                         $file =~ s/$data_dir\/*//;
43                         push @files, $file;
44                 } elsif ( -f $file && $file =~ m/\.csv$/i ) {
45                         $file =~ s/$data_dir\/*//;
46                         push @files, $file;
47                 } elsif ( -f $file && $file =~ m/\.storable/i ) {
48                         $file =~ s/$data_dir\/*//;
49                         push @files, $file;
50                 } else {
51                         #warn "IGNORE: $file\n";
52                 }
53         }, $data_dir);
54
55         no warnings qw(uninitialized); # mtime
56         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
57                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
58         my $size;
59         $size->{$_} = -s "$data_dir/$_" foreach @files;
60
61         $self->render(
62                 files => [ @files ],
63                 size => $size,
64                 loaded => $loaded,
65                 filters => $filters,
66                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
67                 changes => $changes,
68         );
69 }
70
71 sub _dump_path {
72         my ( $self, $name ) = @_;
73         my $dir = $self->app->home->rel_dir('data');
74         $name =~ s/^$dir//;
75         $name =~ s/\/+/_/g;
76         return '/tmp/mojo_facets.' . $name . '.storable';
77 }
78
79 sub _save {
80         my ( $self, $path ) = @_;
81
82         my $dump_path = $self->_dump_path( $path );
83         my $first_load = ! -e $dump_path;
84         warn "save loaded to $dump_path";
85         my $info = $loaded->{$path};
86         store $info, $dump_path;
87
88         if ( $first_load ) {
89                 my $mtime = $loaded->{$path}->{mtime};
90                 utime $mtime, $mtime, $dump_path;
91                 warn "sync time to $path at $mtime\n";
92         }
93
94         warn $dump_path, ' ', -s $dump_path, " bytes\n";
95         return $dump_path;
96 }
97
98
99 sub __stats {
100
101         my $stats;
102
103         my $nr_items = $#{ $_[0] } + 1;
104
105         warn "__stats $nr_items\n";
106
107         foreach my $e ( @{ $_[0] } ) {
108                 foreach my $n ( keys %$e ) {
109                         $stats->{$n}->{count}++;
110                         my @v;
111                         if ( ref $e->{$n} eq 'ARRAY' ) {
112                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
113                                 @v = @{ $e->{$n} };
114                         } else {
115                                 @v = ( $e->{$n} );
116                         }
117
118                         foreach my $x ( @v ) {
119                                 $stats->{$n}->{numeric}++
120                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
121                                 $stats->{$n}->{empty}++
122                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
123                         }
124
125                 }
126         }
127
128         foreach my $n ( keys %$stats ) {
129                 my $s = $stats->{$n};
130                 next unless defined $s->{array};
131                 if ( $s->{array} == $s->{count} ) {
132                         delete $s->{array};
133                         if ( $s->{count} == $nr_items ) {
134                                 warn "check $n for uniqeness\n";
135                                 my $unique;
136                                 foreach my $e ( @{ $_[0] } ) {
137                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
138                                                 $unique = 0;
139                                                 last;
140                                         }
141                                 }
142                                 if ( $unique ) {
143                                         $stats->{$n}->{unique} = 1;
144                                         #warn "# $n unique ",dump( $unique );
145                                 }
146                         }
147                 }
148         }
149
150         warn "# __stats ",dump($stats);
151
152         return $stats;
153 }
154
155 sub _param_or_session {
156         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
157 }
158
159 sub stats {
160         my $self = shift;
161         my $path = $self->_param_or_session('path');
162         warn "stats $path\n";
163         delete $loaded->{$path}->{stats};
164         $self->redirect_to( '/data/columns' );
165 }
166
167
168 sub _load_path {
169         my ( $self, $path ) = @_;
170
171         return if defined $loaded->{$path}->{generated};
172
173         my $full_path = $self->app->home->rel_file( 'data/' . $path );
174         $self->redirect_to('/data/index') unless -r $full_path;
175
176         my $dump_path = $self->_dump_path( $path );
177
178         if ( defined $loaded->{$path}->{data} ) {
179                 my $mtime = (stat($full_path))[9];
180                 return if $loaded->{$path}->{mtime} == $mtime;
181                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
182         } elsif ( -e $dump_path ) {
183                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
184                 my $info = retrieve $dump_path;
185                 $loaded->{ $path } = $info;
186                 return;
187         }
188
189         my $data;
190         if ( -f $full_path ) {
191                 if ( $full_path =~ m/.csv/i ) {
192                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
193                 } elsif ( $full_path =~ m/.storable/ ) {
194                         warn "open $full_path ", -s $full_path, " bytes";
195                         open(my $pipe, "<", $full_path) || die $!;
196                         while ( my $o = eval { Storable::fd_retrieve $pipe } ) {
197                                 if ( defined $o->{item} ) {
198                                         push @{ $data->{items} }, $o->{item};
199                                 } else {
200                                         warn "SKIP ",dump($o);
201                                 }
202                         }
203                         close($pipe);
204                         warn "loaded ", $#{ $data->{items} } + 1, " items from $full_path\n";
205                         $data->{generated}++;
206                 } else {
207                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
208                 }
209         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
210                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
211         } else {
212                 die "can't load $full_path";
213         }
214
215         my @header;
216
217         if ( defined $data->{header} ) {
218                 if ( ref $data->{header} eq 'ARRAY' ) {
219                         @header = @{ $data->{header} };
220                 } else {
221                         warn "header not array ", dump( $data->{header} );
222                 }
223         }
224
225         my $stats = __stats( $data->{items} );
226
227         @header =
228                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
229                 grep { defined $stats->{$_}->{count} } keys %$stats
230                 unless @header;
231
232         my $info = {
233                 header => [ @header ],
234                 stats  => $stats,
235                 full_path => $full_path,
236                 size => -s $full_path,
237                 mtime => (stat($full_path))[9],
238                 data => $data,
239                 defined $data->{generated} ? ( generated => 1 ) : (),
240         };
241
242         $loaded->{ $path } = $info;
243         $self->_save( $path ) unless $info->{generated};
244
245 }
246
247
248 sub load {
249         my $self = shift;
250
251         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
252
253         my @paths = $self->param('paths');
254         warn "# paths ", dump @paths;
255
256         foreach my $p ( keys %$loaded ) {
257                 next if grep { /^\Q$p\E$/ } @paths;
258                 warn "remove $p from memory\n";
259                 delete $loaded->{$p};
260         }
261
262         $self->_load_path( $_ ) foreach @paths;
263
264         warn "# path $path\n";
265         $self->_load_path( $path );
266
267         $self->session( 'path' => $path );
268
269         my $redirect_to = '/data/items';
270
271         $self->session( 'header' => $loaded->{$path}->{header} );
272         if ( ! defined $loaded->{$path}->{columns} ) {
273                 my $columns_path = $self->_permanent_path( 'columns' );
274                 if ( -e $columns_path ) {
275                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
276                         $loaded->{$path}->{columns} = [ @columns ];
277                         warn "# columns_path $columns_path ",dump(@columns);
278                 } else {
279                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
280                 }
281
282                 $redirect_to = '/data/columns';
283         }
284         $self->session( 'columns' => $loaded->{$path}->{columns} );
285         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
286         $self->redirect_to( $redirect_to );
287 }
288
289
290 sub _loaded {
291         my ( $self, $name ) = @_;
292         my $path = $self->session('path') || $self->param('path');
293         $self->redirect_to('/data/index') unless $path;
294
295         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
296                 my $caller = (caller(1))[3];
297                 if ( $caller =~ m/::edit/ ) {
298                         warn "rebuild stats for $path ignored caller $caller\n";
299                 } else {
300                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
301 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
302                         $loaded->{$path}->{rebuild_stats} = 1;
303                         $loaded->{$path}->{modified} = 1;
304                 }
305         }
306
307         if ( defined $loaded->{$path}->{rebuild_stats} ) {
308                 warn "rebuild_stats $path";
309                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
310                 delete $loaded->{$path}->{rebuild_stats};
311         }
312
313         if ( ! defined $loaded->{$path}->{$name} ) {
314                 warn "$path $name isn't loaded\n";
315                 $self->_load_path( $path );
316                 if ( ! defined $loaded->{$path}->{stats} ) {
317                         warn "rebuild stats for $path\n";
318                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
319                 }
320                 if ( ! defined $loaded->{$path}->{$name} ) {
321                         warn "MISSING $name for $path\n";
322                         $self->redirect_to('/data/index')
323                 }
324         }
325
326         $self->session( 'modified' => $loaded->{$path}->{modified} );
327
328         return $loaded->{$path}->{$name};
329 }
330
331
332 sub _checked {
333         my $self = shift;
334         my $checked;
335         $checked->{$_}++ foreach @_;
336 #       warn "# _checked ",dump($checked);
337         return $checked;
338 }
339
340 sub _permanent_path {
341         my $self = shift;
342         my $path = $self->_param_or_session('path');
343         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
344 }
345
346 sub __unac {
347         my $n = shift;
348         $n = unac_string($n);
349         $n =~ s/\W+/_/g;
350         return $n;
351 }
352
353 sub _column_from_unac {
354         my ($self,$name) = @_;
355
356         my $stats = $self->_loaded('stats');
357         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
358
359         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
360 }
361
362 sub _export_path {
363         my $max_length = 80;
364
365         my $self = shift;
366         my $path = $self->_param_or_session('path');
367         if ( ! $path ) {
368                 warn "no path in param or session";
369                 return;
370         }
371         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
372         mkpath $dir unless -e $dir;
373         my $name = join('.', map { __unac($_) } @_ );
374         if ( length($name) > $max_length ) {
375                 $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
376         }
377         my $full = $dir . '/' . $name;
378         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
379         return $full;
380 }
381
382 sub columns {
383     my $self = shift;
384
385         my $view_path = $self->_permanent_path( 'views' );
386
387         if ( $self->param('columns') ) {
388                 my @columns = $self->_param_array('columns');
389                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
390                 if ( my $view = $self->param('view') ) {
391                         mkdir $view_path unless -e $view_path;
392                         write_file( "$view_path/$view", { binmode => ':utf8' }, map { "$_\n" } @columns );
393                         warn "view $view_path/$view ", -s "$view_path/$view", " bytes\n";
394                 }
395
396                 $self->redirect_to('/data/items');
397         }
398
399         if ( my $id = $self->param('id') ) {
400                 my $view_full = "$view_path/$id";
401                 if ( -f $view_full ) {
402                         my @columns = map { chomp; $_ } read_file $view_full, binmode => ':utf8';
403                         warn "view $view_full loaded ", dump @columns;
404                         $self->session( 'columns' => [ @columns ] );
405                         $self->redirect_to('/data/items');
406                 }
407         }
408
409         my $stats = $self->_loaded( 'stats' );
410
411         my @columns;
412         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
413
414         foreach my $c ( @{ $self->session( 'header' ) } ) {
415                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
416         }
417
418         my @views;
419         if ( -d $view_path ) {
420                 @views = map { s{^\Q$view_path\E/*}{}; $_ } glob "$view_path/*";
421                 warn "# views ",dump @views;
422         }
423
424     $self->render(
425                 message => 'Select columns to display',
426                 stats => $stats,
427                 columns => \@columns,
428                 checked => $self->_checked( $self->_param_array('columns') ),
429                 views => \@views,
430         );
431 }
432
433 sub _param_array {
434     my ($self,$name) = @_;
435
436         my @array = $self->param($name);
437         my $path  = $self->session('path');
438
439         if ( @array ) {
440                 $self->session($name => [ @array ]);
441                 $loaded->{$path}->{$name} = [ @array ];
442         } elsif ( my $session = $self->session($name) ) {
443                 if ( ref $session eq 'ARRAY' ) {
444                         @array = @$session;
445                 } else {
446                         die "$name not array ",dump($session);
447                 }
448         }
449         #warn "# $name ",dump @array;
450         return @array;
451 }
452
453 sub _param_scalar {
454     my ($self,$name,$default) = @_;
455
456         my $scalar = $self->param($name);
457
458         if ( defined $scalar ) {
459                 $self->session($name => $scalar);
460         } else {
461                 $scalar = $self->session($name);
462         }
463
464         if ( ! defined $scalar ) {
465                 $scalar = $default;
466                 die "no default for $name" unless defined $scalar;
467                 $self->session($name => $scalar);
468         }
469
470         warn "# _param_scalar $name ",dump $scalar;
471         return $scalar;
472 }
473
474 sub filter {
475         my $self = shift;
476
477         my $name = $self->param('filter_name') || die "name?";
478         my @vals = $self->param('filter_vals');
479
480         $self->_remove_filter( $name );
481         if ( @vals ) {
482                 $self->_filter_on_data( $name, @vals );
483                 if ( my $permanent = $self->param('_permanent') ) {
484                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
485                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
486                         warn "permanent filter $permanent_path ", -s $permanent_path;
487                 }
488         }
489
490         $self->session( 'offset' => 0 );
491
492         $self->redirect_to('/data/items');
493 }
494
495 sub _filter_on_data {
496         my ( $self, $name, @vals ) = @_;
497
498         my $path = $self->session('path');
499
500         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
501                 warn "filter $name not found in data set";
502                 return;
503         }
504
505         $filters->{$name} = [ @vals ];
506         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
507
508         my $filter_hash;
509         $filter_hash->{$_}++ foreach @vals;
510
511         #warn "# filter_hash ",dump( $filter_hash );
512
513         my $items = $self->_loaded('data')->{items};
514
515         my $include_missing = defined $filter_hash->{_missing};
516         my $filtered_items;
517
518         foreach my $i ( 0 .. $#$items ) {
519
520                 if ( defined $items->[$i]->{$name} ) {
521                         my $row = $items->[$i]->{$name};
522                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
523                         foreach my $v ( @$row ) {
524                                 if ( defined $filter_hash->{ $v } ) {
525                                         $filtered_items->{$i}++;
526                                 }
527                         }
528                 } elsif ( $include_missing ) {
529                         $filtered_items->{$i}++;
530                 }
531         }
532
533         #warn "# filter $name ",dump($filtered_items);
534
535         $loaded->{$path}->{filters}->{$name} = $filtered_items;
536         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
537 }
538
539
540 sub _current_filters {
541         my $self = shift;
542         my $current_filters;
543         my $stats = $self->_loaded('stats');
544
545         $current_filters->{ $_ } = $filters->{ $_ }
546                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
547         ;
548         warn "# _current_filters ",dump( keys %$current_filters );
549         return $current_filters;
550 }
551
552 sub _data_sorted_by {
553         my ( $self, $order ) = @_;
554
555         my $path = $self->session('path');
556
557         warn "_data_sorted_by $order from $path";
558
559         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
560                 return $loaded->{$path}->{sorted}->{$order};
561         }
562
563         my $data = $self->_loaded( 'data' );
564         my $numeric = $self->_is_numeric($order);
565         my $missing = $numeric ? 0 : '';
566         no warnings qw(numeric);
567         my $nr = 0;
568         my @sorted = map {
569                 $_->[0]
570         } sort {
571                 if ( $numeric ) {
572                         $a->[1] <=> $b->[1]
573                 } else {
574                         $a->[1] cmp $b->[1]
575                 }
576         } map {
577                 my $v;
578                 if ( ! exists $_->{$order} ) {
579                         $v = $missing;
580                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
581                         $v = join('', @{$_->{$order}});
582                 } else {
583                         $v = $_->{$order};
584                 }
585                 [ $nr++, $v ]
586         } @{ $data->{items} }
587         ;
588
589         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
590         #warn "# sorted ",dump( @sorted );
591
592         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
593 }
594
595
596 sub __all_filters {
597         my $order = pop @_;
598         join(',', sort(@_), 'order', $order);
599 }
600
601 our ($out, $key,$value);
602
603 sub __commit_path_code {
604         my ( $path, $i, $code, $commit_changed ) = @_;
605
606         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
607         my $row = $items->[$i];
608         my $update;
609         eval $code;
610         foreach ( keys %$update ) {
611                 $$commit_changed->{$_}++;
612                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
613         }
614         #warn "__commit_path_code $path $i ",dump( $update );
615 }
616
617 sub items {
618         my $self = shift;
619
620         if ( my $show = $self->param('id') ) {
621                 $self->param('show', $show);
622                 warn "show $show\n";
623         }
624
625         my $path = $self->_param_scalar('path');
626
627         my @columns = $self->_param_array('columns');
628         $self->redirect_to('/data/columns') unless @columns;
629         my $order   = $self->_param_scalar('order', $columns[0]);
630         my $sort    = $self->_param_scalar('sort', 'a');
631         my $offset  = $self->_param_scalar('offset', 0);
632         my $limit   = $self->_param_scalar('limit', 20);
633         $self->_param_scalar('show', 'table');
634
635         # fix offset when changing limit
636         $offset = int( $offset / $limit ) * $limit;
637
638         if ( ! grep { /^\Q$order\E$/ } @columns ) {
639                 $order = $columns[0];
640                 $self->session( order => $order );
641         }
642         my $sorted = $self->_data_sorted_by( $order );
643
644         my @filter_names;
645         if ( $filters ) {
646                 foreach my $name ( keys %$filters ) {
647                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
648                                 warn "skip filter $name not found in $path\n";
649                                 next;
650                         }
651                         push @filter_names, $name;
652                 }
653                 warn "filter_names ",dump( @filter_names );
654                 foreach my $name ( @filter_names ) {
655                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
656                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
657                 }
658         }
659
660         my $all_filters = __all_filters( @filter_names,$order );
661
662 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
663
664         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
665
666                 my $path_filters = $loaded->{$path}->{filters};
667
668                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
669
670                 my @filtered;
671                 foreach my $i ( 0 .. $#$sorted ) {
672                         my $pos = $sorted->[$i];
673
674                         if ( $#filter_names == -1 ) {
675                                 push @filtered, $pos;
676                                 next;
677                         }
678
679                         my $skip = 0;
680                         foreach ( @filter_names ) {
681                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
682                         }
683                         next if $skip;
684
685                         push @filtered, $pos;
686                 }
687
688                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
689         }
690
691         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
692                 if defined $loaded->{$path}->{filtered}->{$all_filters};
693
694         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
695
696         my $data = $self->_loaded('data');
697
698         my $code = $self->_param_scalar('code','');
699         $code =~ s{\r}{}gs;
700         $code =~ s{\n+$}{\n}s;
701
702         my $commit = $self->param('commit');
703         my $test = $self->param('test');
704
705         my $commit_changed;
706
707         if ( $code && ( $test || $commit ) ) {
708                 # XXX find columns used in code snippet and show them to user
709                 my $order = 0;
710                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
711                         if ( $column =~ s/^(['"])// ) {
712                                 $column =~ s/$1$//;
713                         }
714                         next if $column =~ m/\$/; # hide columns with vars in them
715                         $commit_changed->{$column} = 0;
716                 }
717         }
718
719         my $code_path = $self->app->home->rel_dir('public') . "/code";
720         if ( $commit ) {
721
722                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
723                 ( $key, $value, $out ) = ( 'key', 'value' );
724                 foreach ( 0 .. $#$filtered ) {
725                         my $i = $filtered->[$_];
726                         __commit_path_code( $path, $i, $code, \$commit_changed );
727                 }
728
729                 $self->_save_change({
730                         path => $path,
731                         time => $self->param('time') || time(),
732                         user => $self->param('user') || $ENV{'LOGNAME'},
733                         code => $code,
734                         commit_changed => $commit_changed,
735                 });
736
737                 if ( my $description = $self->param('code_description') ) {
738                         my $depends = $self->param('code_depends') || die "no code_depends?";
739                         my $path = "$code_path/$depends.$description.pl";
740                         if ( -e $path && ! $self->param('overwrite') ) {
741                                 warn "# code $path not saved\n";
742                         } else {
743                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
744                                 warn "code $path ", -s $path, " bytes saved\n";
745                         }
746                 }
747
748                 # remove console
749                 $code = '';
750                 if ( $out ) {
751                         my $commit_dataset = join('.'
752                                 , $self->param('code_depends')
753                                 , $self->param('code_description')
754                                 , time()
755                         );
756                         $key ||= 'key';
757                         $value ||= 'value';
758                         warn "key $key value $value";
759                         my $items;
760                         foreach my $n ( keys %$out ) {
761                                 my $i = { $key => [ $n ] };
762                                 my $ref = ref $out->{$n};
763                                 if ( $ref eq 'HASH' ) {
764                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
765                                 } elsif ( $ref eq 'ARRAY' ) {
766                                         $i->{$_} = $out->{$n};
767                                 } elsif ( ! $ref ) {
768                                         $i->{$value} = [ $out->{$n} ];
769                                 } else {
770                                         $i->{_error} = [ dump($out->{$n}) ];
771                                 }
772                                 push @$items, $i;
773                         };
774                         undef $out;
775                         my $stats = __stats( $items );
776                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
777                         unshift @columns, $key;
778
779                         $loaded->{$commit_dataset} = {
780                                 header => [ @columns ],
781                                 columns => [ @columns ],
782                                 mtime => time(),
783                                 data => { items => $items },
784                                 stats => $stats,
785                                 generated => 1,
786                         };
787                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
788                         $self->session('path', $commit_dataset);
789                         $self->session('columns', [ @columns ]);
790                         $self->session('order', $key);
791                         $self->redirect_to('/data/items');
792                         return; # FIXME needed to correctly show columns
793                 }
794
795                 # this might move before $out to recalculate stats on source dataset?
796                 __path_rebuild_stats( $path );
797                 my $c = { map { $_ => 1 } @columns };
798                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
799                 warn "# added_columns ",dump( @added_columns );
800                 unshift @columns, @added_columns;
801
802                 $self->session('columns', [ @columns ]);
803                 $loaded->{$path}->{columns} = [ @columns ];
804                 warn "# new columns ",dump( @columns );
805
806                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
807         }
808
809         my $sorted_items;
810         my $from_end = $sort eq 'd' ? $#$filtered : 0;
811         my $test_changed;
812         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
813         foreach ( 0 .. $limit ) {
814                 my $i = $_ + $offset;
815                 last unless defined $filtered->[$i];
816                 $i = $from_end - $i if $from_end;
817                 my $id = $filtered->[$i];
818                 my $row = Storable::dclone $data->{items}->[ $id ];
819                 if ( $code && $test ) {
820                         my $update;
821                         eval $code;
822                         if ( $@ ) {
823                                 warn "ERROR evaling $@", dump($code);
824                                 $self->stash('eval_error', $@) if $@;
825                         } else {
826                                 warn "EVAL ",dump($update);
827                                 foreach ( keys %$update ) {
828                                         $test_changed->{$_}++;
829                                         $row->{$_} = $update->{$_};
830                                 }
831                         }
832                 }
833                 $row->{_row_id} ||= $id;
834                 push @$sorted_items, $row;
835         }
836
837         if ( $self->param('export') ) {
838                 my $export_path = $self->_export_path( 'items', @columns);
839                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
840                 print $fh "#",join("\t",@columns),"\n";
841                 foreach my $f ( 0 .. $#$filtered ) {
842                         print $fh join("\t", map {
843                                 my $i = $data->{items}->[ $filtered->[$f] ];
844                                 my $v = '\N';
845                                 if ( ! defined $i->{$_} ) {
846                                         # nop
847                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
848                                         $v =join(',', @{ $i->{$_} });
849                                         $v = '\N' if length($v) == 0;
850                                 } elsif ( ! ref $i->{$_} ) {
851                                         $v = $i->{$_};
852                                 } else {
853                                         $v = dump $i->{$_};
854                                 }
855                                 $v;
856                         } @columns),"\n";
857                 }
858                 close($fh);
859                 warn "export $export_path ", -s $export_path, " bytes\n";
860         }
861
862         warn "# test_changed ",dump( $test_changed );
863         my $c = { map { $_ => 1 } @columns };
864         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
865         unshift @columns, @added_columns;
866
867         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
868
869         my $code_depends = $self->param('code_depends')||
870         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
871         my $code_description = $self->param('code_description') ||
872         join(',', @added_columns);
873
874         $code_depends ||= $code_description; # self-modifing
875         if ( ! $code_depends && $out ) {
876                 $code_depends = $key;
877                 $code_description = $value;
878         }
879
880         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
881
882         $self->render(
883                 order => $order,
884                 offset => $offset,
885                 limit => $limit,
886                 sorted => $sorted_items,
887                 columns => [ @columns ],
888                 rows => $#$filtered + 1,
889                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
890                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
891                 filters => $self->_current_filters,
892                 code => $code,
893                 cols_changed => $commit ? $commit_changed : $test_changed,
894                 code_depends => $code_depends,
895                 code_description => $code_description,
896                 code_path => $code_path,
897                 out => $out,
898         );
899
900 }
901
902
903 sub order {
904         my $self = shift;
905         $self->session('order', $self->param('order'));
906         $self->session('sort', $self->param('sort'));
907         $self->redirect_to('/data/items');
908 }
909
910 sub _is_numeric {
911         my ( $self, $name ) = @_;
912
913         my $stats = $self->_loaded( 'stats' );
914
915         # sort facet numerically if more >50% elements are numeric
916         my $count = $stats->{$name}->{count};
917         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
918         defined $stats->{$name}->{numeric} &&
919                 $stats->{$name}->{numeric} > $count / 2;
920 }
921
922 sub _is_unique {
923         my ( $self, $name ) = @_;
924         my $stats = $self->_loaded( 'stats' );
925         defined $stats->{$name}->{unique};
926 }
927
928 sub _remove_filter {
929         my ($self,$name) = @_;
930         warn "_remove_filter $name\n";
931
932         my $path = $self->session('path');
933
934         delete $filters->{$name};
935         delete $loaded->{$path}->{filters}->{$name};
936         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
937
938         foreach (
939                         grep { /\Q$name\E/ }
940                         keys %{ $loaded->{$path}->{filtered} }
941         ) {
942                 delete $loaded->{$path}->{filtered}->{$_};
943                 warn "remove filtered cache $_";
944         }
945 }
946
947 sub facet {
948         my $self = shift;
949
950         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
951
952         if ( my $name = $self->param('remove') ) {
953                 $self->_remove_filter( $name );
954                 $self->redirect_to( '/data/items' );
955         }
956
957         my $facet;
958         my $name = $self->param('name') || die "no name";
959
960         my $all = $self->_param_scalar('all', 1);
961         my $data = $self->_loaded('data');
962
963         my $filters = $self->_current_filters;
964         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
965         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
966                 if defined $loaded->{$path}->{filtered}->{$all_filters};
967
968         if ( ! $filtered || $all ) {
969                 $filtered = [ 0 .. $#{ $data->{items} } ];
970                 warn "filter all values\n";
971         } else {
972                 warn "filter using $all_filters\n";
973         }
974
975         foreach my $i ( @$filtered ) {
976                 my $item = $data->{items}->[$i];
977                 if ( ! exists $item->{$name} ) {
978                         $facet->{ _missing }++;
979                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
980                         $facet->{$_}++ foreach @{ $item->{$name} };
981                 } else {
982                         $facet->{ $item->{$name} }++;
983                 }
984         }
985
986         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
987
988         if ( my $code = $self->param('code') ) {
989                 my $out;
990                 foreach my $value ( keys %$facet ) {
991                         my $count = $facet->{$value};
992                         my $checked = $checked_values->{$value};
993                         eval $code;
994                         if ( $@ ) {
995                                 $out = $@;
996                                 warn "ERROR: $@\n$code\n";
997                                 last;
998                         } elsif ( $checked != $checked_values->{$value} ) {
999                                 warn "checked $value $count -> $checked\n";
1000                                 $checked_values->{$value} = $checked;
1001                         }
1002                 }
1003                 warn "out ",dump( $out );
1004                 $self->stash( out => $out );
1005         }
1006
1007 #       warn "# facet $name ",dump $facet;
1008
1009         my @facet_names =
1010                   $all                      ? keys %$facet
1011                 : defined $filters->{$name} ? @{ $filters->{$name} }
1012                 : keys %$facet;
1013
1014         my $numeric = $self->_is_numeric($name);
1015
1016         my $sort = $self->param('sort');
1017         # sort numeric facets with more than 5 values ascending
1018         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
1019
1020         @facet_names = sort {
1021                 my $result;
1022                 if ( $sort eq 'a' ) {
1023                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
1024                 } elsif ( $sort eq 'd' ) {
1025                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1026                 } elsif ( $sort eq 'c' ) {
1027                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1028                 } else {
1029                         warn "unknown sort: $sort";
1030                         $result = $a cmp $b;
1031                 }
1032                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1033                 $result;
1034         } @facet_names;
1035
1036         $self->render( name => $name, facet => $facet, checked => $checked_values,
1037                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1038         );
1039 }
1040
1041
1042 sub __invalidate_path_column {
1043         my ( $path, $name ) = @_;
1044
1045         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1046                 delete $loaded->{$path}->{sorted}->{$name};
1047                 warn "# invalidate $path sorted $name\n";
1048         }
1049
1050         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1051                 delete $loaded->{$path}->{filtered}->{$_};
1052                 warn "# invalidate $path filtered $_\n";
1053         }
1054 }
1055
1056 sub __path_modified {
1057         my ( $path, $value ) = @_;
1058         $value = 1 unless defined $value;
1059         
1060         $loaded->{$path}->{modified}  = $value;
1061
1062         warn "# __path_modified $path $value\n";
1063 }
1064
1065 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1066
1067 sub _save_change {
1068         my ($self,$change) = @_;
1069
1070         my $change_path = $self->_permanent_path( 'changes' );
1071         mkdir $change_path unless -d $change_path;
1072         $change_path .= '/' . $change->{time};
1073         store $change, $change_path;
1074         utime $change->{time}, $change->{time}, $change_path;
1075         warn "_save_change $change_path ", dump($change);
1076 }
1077
1078 sub edit {
1079         my $self = shift;
1080         my $new_content = $self->param('new_content');
1081         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1082
1083         my $i = $self->param('_row_id');
1084         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1085         my $path = $self->param('path') || die "no path";
1086         my $name = $self->param('name') || die "no name";
1087         my $status = 200; # 200 = OK, 201 = Created
1088
1089         my $data = $self->_loaded('data');
1090
1091         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1092                 $new_content =~ s/^\s+//s;
1093                 $new_content =~ s/\s+$//s;
1094                 my $v;
1095                 if ( $new_content =~ /\xB6/ ) { # para
1096                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1097                 } else {
1098                         $v = [ $new_content ];
1099                 }
1100
1101                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1102                 my $new = dump $v;
1103                 if ( $old ne $new
1104                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1105                 ) {
1106                         $self->_save_change({
1107                                 path => $path,
1108                                 column => $name,
1109                                 pos => $i,
1110                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1111                                 new => $v,
1112                                 time => $self->param('time') || time(),
1113                                 user => $self->param('user') || $ENV{'LOGNAME'},
1114                                 unique => {
1115                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1116                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1117                                         keys %{ $loaded->{$path}->{stats} }
1118                                 },
1119                         });
1120
1121                         warn "# change $path $i $old -> $new\n";
1122                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1123
1124                         __invalidate_path_column( $path, $name );
1125
1126                         $status = 201; # created
1127                         __path_rebuild_stats( $path );
1128         
1129                         $new_content = join("\xB6",@$v);
1130
1131                 } else {
1132                         warn "# unchanged $path $i $old\n";
1133                         $status = 304;
1134                 }
1135         } else {
1136                 $new_content = "$path $i $name doesn't exist\n";
1137                 $status = 404;
1138         }
1139
1140         warn "# edit $status ", dump $new_content;
1141
1142         $self->render(
1143                 status => $status,
1144                 new_content => scalar $new_content,
1145         );
1146 }
1147
1148
1149 sub save {
1150         my $self = shift;
1151         my $path = $self->_param_or_session('path');
1152         my $dump_path = $self->_save( $path );
1153         __path_modified( $path, 0 );
1154
1155         $self->redirect_to( '/data/items' );
1156 }
1157
1158 sub export {
1159         my $self = shift;
1160
1161         my $dir = $self->app->home->rel_dir('public');
1162
1163         if ( my $import = $self->param('import') ) {
1164
1165                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1166                         my $name = $self->_column_from_unac( $1 );
1167
1168                         my @vals = map { chomp; $_ }
1169                                 read_file "$dir/export/$import", binmode => ':utf8';
1170
1171                         $self->_remove_filter( $name );
1172                         $self->_filter_on_data( $name, @vals );
1173                         $self->session( 'offset' => 0 );
1174                         $self->redirect_to('/data/items');
1175                 } else {
1176                         warn "UNKNOWN IMPORT $import";
1177                 }
1178         }
1179
1180         if ( my $remove = $self->param('remove') ) {
1181                 my $path = "$dir/export/$remove";
1182                 unlink $path if -e $path;
1183                 $path .= '.png';
1184                 unlink $path if -e $path;
1185         }
1186
1187         my $path = $self->_export_path || $self->redirect_to('/data/index');
1188
1189         my @files = grep { ! /\.png$/ } glob "$path/*";
1190         my $mtime = { map { $_ => (stat($_))[9] } @files };
1191         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1192         $self->render( export => [ @files ] );
1193 }
1194
1195 sub __loaded_paths {
1196         return
1197                 grep { defined $loaded->{$_}->{data} }
1198                 keys %$loaded;
1199 }
1200
1201 1;