761ded7e8513fbf9025b41ca590207e74684881d
[MojoFacets.git] / lib / MojoFacets / Data.pm
1 package MojoFacets::Data;
2
3 use strict;
4 use warnings;
5
6 use base 'Mojolicious::Controller';
7
8 use Data::Dump qw(dump);
9 use File::Slurp;
10 use Encode;
11 use locale;
12 use File::Find;
13 use Storable;
14 use Time::HiRes qw(time);
15 use File::Path qw(mkpath);
16 use Text::Unaccent::PurePerl;
17
18 use MojoFacets::Import::File;
19 use MojoFacets::Import::HTMLTable;
20 use MojoFacets::Import::CSV;
21
22 our $loaded;
23 our $filters;
24
25 sub index {
26         my $self = shift;
27
28         my $data_dir = $self->app->home->rel_dir('data');
29         die "no data dir $data_dir" unless -d $data_dir;
30
31         my @files;
32         my $changes;
33         find( sub {
34                 my $file = $File::Find::name;
35                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
36                         $file =~ s/$data_dir\/*//;
37                         push @files, $file;
38                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
39                         push @{ $changes->{$1} }, $2
40                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
41                         $file =~ s/$data_dir\/*//;
42                         push @files, $file;
43                 } elsif ( -f $file && $file =~ m/\.csv$/i ) {
44                         $file =~ s/$data_dir\/*//;
45                         push @files, $file;
46                 } else {
47                         #warn "IGNORE: $file\n";
48                 }
49         }, $data_dir);
50
51         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
52                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
53         my $size;
54         $size->{$_} = -s "$data_dir/$_" foreach @files;
55
56         $self->render(
57                 files => [ @files ],
58                 size => $size,
59                 loaded => $loaded,
60                 filters => $filters,
61                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
62                 changes => $changes,
63         );
64 }
65
66 sub _dump_path {
67         my ( $self, $name ) = @_;
68         my $dir = $self->app->home->rel_dir('data');
69         $name =~ s/^$dir//;
70         $name =~ s/\/+/_/g;
71         return '/tmp/mojo_facets.' . $name . '.storable';
72 }
73
74 sub _save {
75         my ( $self, $path ) = @_;
76
77         my $dump_path = $self->_dump_path( $path );
78         my $first_load = ! -e $dump_path;
79         warn "save loaded to $dump_path";
80         my $info = $loaded->{$path};
81         store $info, $dump_path;
82
83         if ( $first_load ) {
84                 my $mtime = $loaded->{$path}->{mtime};
85                 utime $mtime, $mtime, $dump_path;
86                 warn "sync time to $path at $mtime\n";
87         }
88
89         warn $dump_path, ' ', -s $dump_path, " bytes\n";
90         return $dump_path;
91 }
92
93
94 sub __stats {
95
96         my $stats;
97
98         my $nr_items = $#{ $_[0] } + 1;
99
100         warn "__stats $nr_items\n";
101
102         foreach my $e ( @{ $_[0] } ) {
103                 foreach my $n ( keys %$e ) {
104                         $stats->{$n}->{count}++;
105                         my @v;
106                         if ( ref $e->{$n} eq 'ARRAY' ) {
107                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
108                                 @v = @{ $e->{$n} };
109                         } else {
110                                 @v = ( $e->{$n} );
111                         }
112
113                         foreach my $x ( @v ) {
114                                 $stats->{$n}->{numeric}++
115                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
116                                 $stats->{$n}->{empty}++
117                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
118                         }
119
120                 }
121         }
122
123         foreach my $n ( keys %$stats ) {
124                 my $s = $stats->{$n};
125                 next unless defined $s->{array};
126                 if ( $s->{array} == $s->{count} ) {
127                         delete $s->{array};
128                         if ( $s->{count} == $nr_items ) {
129                                 warn "check $n for uniqeness\n";
130                                 my $unique;
131                                 foreach my $e ( @{ $_[0] } ) {
132                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
133                                                 $unique = 0;
134                                                 last;
135                                         }
136                                 }
137                                 if ( $unique ) {
138                                         $stats->{$n}->{unique} = 1;
139                                         #warn "# $n unique ",dump( $unique );
140                                 }
141                         }
142                 }
143         }
144
145         warn "# __stats ",dump($stats);
146
147         return $stats;
148 }
149
150 sub _param_or_session {
151         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
152 }
153
154 sub stats {
155         my $self = shift;
156         my $path = $self->_param_or_session('path');
157         warn "stats $path\n";
158         delete $loaded->{$path}->{stats};
159         $self->redirect_to( '/data/columns' );
160 }
161
162
163 sub _load_path {
164         my ( $self, $path ) = @_;
165
166         return if defined $loaded->{$path}->{'generated'};
167
168         my $full_path = $self->app->home->rel_file( 'data/' . $path );
169         $self->redirect_to('/data/index') unless -r $full_path;
170
171         my $dump_path = $self->_dump_path( $path );
172
173         if ( defined $loaded->{$path}->{data} ) {
174                 my $mtime = (stat($full_path))[9];
175                 return if $loaded->{$path}->{mtime} == $mtime;
176                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
177         } elsif ( -e $dump_path ) {
178                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
179                 my $info = retrieve $dump_path;
180                 $loaded->{ $path } = $info;
181                 return;
182         }
183
184         my $data;
185         if ( -f $full_path ) {
186                 if ( $full_path =~ m/.csv/i ) {
187                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
188                 } else {
189                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
190                 }
191         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
192                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
193         } else {
194                 die "can't load $full_path";
195         }
196
197         my @header;
198
199         if ( defined $data->{header} ) {
200                 if ( ref $data->{header} eq 'ARRAY' ) {
201                         @header = @{ $data->{header} };
202                 } else {
203                         warn "header not array ", dump( $data->{header} );
204                 }
205         }
206
207         my $stats = __stats( $data->{items} );
208
209         @header =
210                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
211                 grep { defined $stats->{$_}->{count} } keys %$stats
212                 unless @header;
213
214         my $info = {
215                 header => [ @header ],
216                 stats  => $stats,
217                 full_path => $full_path,
218                 size => -s $full_path,
219                 mtime => (stat($full_path))[9],
220                 data => $data,
221         };
222
223         $loaded->{ $path } = $info;
224         $self->_save( $path );
225
226 }
227
228
229 sub load {
230         my $self = shift;
231
232         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
233
234         my @paths = $self->param('paths');
235         warn "# paths ", dump @paths;
236
237         foreach my $p ( keys %$loaded ) {
238                 next if grep { /^\Q$p\E$/ } @paths;
239                 warn "remove $p from memory\n";
240                 delete $loaded->{$p};
241         }
242
243         $self->_load_path( $_ ) foreach @paths;
244
245         warn "# path $path\n";
246         $self->_load_path( $path );
247
248         $self->session( 'path' => $path );
249
250         my $redirect_to = '/data/items';
251
252         $self->session( 'header' => $loaded->{$path}->{header} );
253         if ( ! defined $loaded->{$path}->{columns} ) {
254                 my $columns_path = $self->_permanent_path( 'columns' );
255                 if ( -e $columns_path ) {
256                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
257                         $loaded->{$path}->{columns} = [ @columns ];
258                         warn "# columns_path $columns_path ",dump(@columns);
259                 } else {
260                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
261                 }
262
263                 $redirect_to = '/data/columns';
264         }
265         $self->session( 'columns' => $loaded->{$path}->{columns} );
266         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
267         $self->redirect_to( $redirect_to );
268 }
269
270
271 sub _loaded {
272         my ( $self, $name ) = @_;
273         my $path = $self->session('path') || $self->param('path');
274         $self->redirect_to('/data/index') unless $path;
275
276         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
277                 my $caller = (caller(1))[3];
278                 if ( $caller =~ m/::edit/ ) {
279                         warn "rebuild stats for $path ignored caller $caller\n";
280                 } else {
281                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
282 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
283                         $loaded->{$path}->{rebuild_stats} = 1;
284                         $loaded->{$path}->{modified} = 1;
285                 }
286         }
287
288         if ( defined $loaded->{$path}->{rebuild_stats} ) {
289                 warn "rebuild_stats $path";
290                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
291                 delete $loaded->{$path}->{rebuild_stats};
292         }
293
294         if ( ! defined $loaded->{$path}->{$name} ) {
295                 warn "$path $name isn't loaded\n";
296                 $self->_load_path( $path );
297                 if ( ! defined $loaded->{$path}->{stats} ) {
298                         warn "rebuild stats for $path\n";
299                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
300                 }
301                 if ( ! defined $loaded->{$path}->{$name} ) {
302                         warn "MISSING $name for $path\n";
303                         $self->redirect_to('/data/index')
304                 }
305         }
306
307         $self->session( 'modified' => $loaded->{$path}->{modified} );
308
309         return $loaded->{$path}->{$name};
310 }
311
312
313 sub _checked {
314         my $self = shift;
315         my $checked;
316         $checked->{$_}++ foreach @_;
317 #       warn "# _checked ",dump($checked);
318         return $checked;
319 }
320
321 sub _permanent_path {
322         my $self = shift;
323         my $path = $self->_param_or_session('path');
324         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
325 }
326
327 sub __unac {
328         my $n = shift;
329         $n = unac_string($n);
330         $n =~ s/\W+/_/g;
331         return $n;
332 }
333
334 sub _column_from_unac {
335         my ($self,$name) = @_;
336
337         my $stats = $self->_loaded('stats');
338         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
339
340         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
341 }
342
343 sub _export_path {
344         my $self = shift;
345         my $path = $self->_param_or_session('path');
346         if ( ! $path ) {
347                 warn "no path in param or session";
348                 return;
349         }
350         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
351         mkpath $dir unless -e $dir;
352         my $name = join('.', map { __unac($_) } @_ );
353         my $full = $dir . '/' . $name;
354         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
355         return $full;
356 }
357
358 sub columns {
359     my $self = shift;
360
361         if ( $self->param('columns') ) {
362                 my @columns = $self->_param_array('columns');
363                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
364                 $self->redirect_to('/data/items');
365         }
366
367         my $stats = $self->_loaded( 'stats' );
368
369         my @columns;
370         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
371
372         foreach my $c ( @{ $self->session( 'header' ) } ) {
373                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
374         }
375
376     $self->render(
377                 message => 'Select columns to display',
378                 stats => $stats,
379                 columns => \@columns,
380                 checked => $self->_checked( $self->_param_array('columns') ),
381         );
382 }
383
384 sub _param_array {
385     my ($self,$name) = @_;
386
387         my @array = $self->param($name);
388         my $path  = $self->session('path');
389
390         if ( @array ) {
391                 $self->session($name => [ @array ]);
392                 $loaded->{$path}->{$name} = [ @array ];
393         } elsif ( my $session = $self->session($name) ) {
394                 if ( ref $session eq 'ARRAY' ) {
395                         @array = @$session;
396                 } else {
397                         die "$name not array ",dump($session);
398                 }
399         }
400         #warn "# $name ",dump @array;
401         return @array;
402 }
403
404 sub _param_scalar {
405     my ($self,$name,$default) = @_;
406
407         my $scalar = $self->param($name);
408
409         if ( defined $scalar ) {
410                 $self->session($name => $scalar);
411         } else {
412                 $scalar = $self->session($name);
413         }
414
415         if ( ! defined $scalar ) {
416                 $scalar = $default;
417                 die "no default for $name" unless defined $scalar;
418                 $self->session($name => $scalar);
419         }
420
421         warn "# _param_scalar $name ",dump $scalar;
422         return $scalar;
423 }
424
425 sub filter {
426         my $self = shift;
427
428         my $name = $self->param('filter_name') || die "name?";
429         my @vals = $self->param('filter_vals');
430
431         $self->_remove_filter( $name );
432         if ( @vals ) {
433                 $self->_filter_on_data( $name, @vals );
434                 if ( my $permanent = $self->param('_permanent') ) {
435                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
436                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
437                         warn "permanent filter $permanent_path ", -s $permanent_path;
438                 }
439         }
440
441         $self->session( 'offset' => 0 );
442
443         $self->redirect_to('/data/items');
444 }
445
446 sub _filter_on_data {
447         my ( $self, $name, @vals ) = @_;
448
449         my $path = $self->session('path');
450
451         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
452                 warn "filter $name not found in data set";
453                 return;
454         }
455
456         $filters->{$name} = [ @vals ];
457         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
458
459         my $filter_hash;
460         $filter_hash->{$_}++ foreach @vals;
461
462         #warn "# filter_hash ",dump( $filter_hash );
463
464         my $items = $self->_loaded('data')->{items};
465
466         my $include_missing = defined $filter_hash->{_missing};
467         my $filtered_items;
468
469         foreach my $i ( 0 .. $#$items ) {
470
471                 if ( defined $items->[$i]->{$name} ) {
472                         my $row = $items->[$i]->{$name};
473                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
474                         foreach my $v ( @$row ) {
475                                 if ( defined $filter_hash->{ $v } ) {
476                                         $filtered_items->{$i}++;
477                                 }
478                         }
479                 } elsif ( $include_missing ) {
480                         $filtered_items->{$i}++;
481                 }
482         }
483
484         #warn "# filter $name ",dump($filtered_items);
485
486         $loaded->{$path}->{filters}->{$name} = $filtered_items;
487         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
488 }
489
490
491 sub _current_filters {
492         my $self = shift;
493         my $current_filters;
494         my $stats = $self->_loaded('stats');
495
496         $current_filters->{ $_ } = $filters->{ $_ }
497                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
498         ;
499         warn "# _current_filters ",dump( keys %$current_filters );
500         return $current_filters;
501 }
502
503 sub _data_sorted_by {
504         my ( $self, $order ) = @_;
505
506         my $path = $self->session('path');
507
508         warn "_data_sorted_by $order from $path";
509
510         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
511                 return $loaded->{$path}->{sorted}->{$order};
512         }
513
514         my $data = $self->_loaded( 'data' );
515         my $numeric = $self->_is_numeric($order);
516         my $missing = $numeric ? 0 : '';
517         no warnings qw(numeric);
518         my $nr = 0;
519         my @sorted = map {
520                 $_->[0]
521         } sort {
522                 if ( $numeric ) {
523                         $a->[1] <=> $b->[1]
524                 } else {
525                         $a->[1] cmp $b->[1]
526                 }
527         } map {
528                 my $v;
529                 if ( ! exists $_->{$order} ) {
530                         $v = $missing;
531                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
532                         $v = join('', @{$_->{$order}});
533                 } else {
534                         $v = $_->{$order};
535                 }
536                 [ $nr++, $v ]
537         } @{ $data->{items} }
538         ;
539
540         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
541         #warn "# sorted ",dump( @sorted );
542
543         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
544 }
545
546
547 sub __all_filters {
548         my $order = pop @_;
549         join(',', sort(@_), 'order', $order);
550 }
551
552 sub __commit_path_code {
553         my ( $path, $i, $code, $commit_changed ) = @_;
554
555         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
556         my $row = $items->[$i];
557         my $update;
558         eval $code;
559         foreach ( keys %$update ) {
560                 $$commit_changed->{$_}++;
561                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
562         }
563         warn "XX ",dump( $loaded->{$path}->{data}->{items}->[$i] );
564         warn "__commit_path_code $path ",dump( $update );
565 }
566
567 sub items {
568         my $self = shift;
569
570         if ( my $show = $self->param('id') ) {
571                 $self->param('show', $show);
572                 warn "show $show\n";
573         }
574
575         my $path = $self->session('path');
576
577         my @columns = $self->_param_array('columns');
578         $self->redirect_to('/data/columns') unless @columns;
579         my $order   = $self->_param_scalar('order', $columns[0]);
580         my $sort    = $self->_param_scalar('sort', 'a');
581         my $offset  = $self->_param_scalar('offset', 0);
582         my $limit   = $self->_param_scalar('limit', 20);
583         $self->_param_scalar('show', 'table');
584
585         # fix offset when changing limit
586         $offset = int( $offset / $limit ) * $limit;
587
588         if ( ! grep { /^\Q$order\E$/ } @columns ) {
589                 $order = $columns[0];
590                 $self->session( order => $order );
591         }
592         my $sorted = $self->_data_sorted_by( $order );
593
594         my @filter_names;
595         if ( $filters ) {
596                 foreach my $name ( keys %$filters ) {
597                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
598                                 warn "skip filter $name not found in $path\n";
599                                 next;
600                         }
601                         push @filter_names, $name;
602                 }
603                 warn "filter_names ",dump( @filter_names );
604                 foreach my $name ( @filter_names ) {
605                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
606                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
607                 }
608         }
609
610         my $all_filters = __all_filters( @filter_names,$order );
611
612 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
613
614         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
615
616                 my $path_filters = $loaded->{$path}->{filters};
617
618                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
619
620                 my @filtered;
621                 foreach my $i ( 0 .. $#$sorted ) {
622                         my $pos = $sorted->[$i];
623
624                         if ( $#filter_names == -1 ) {
625                                 push @filtered, $pos;
626                                 next;
627                         }
628
629                         my $skip = 0;
630                         foreach ( @filter_names ) {
631                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
632                         }
633                         next if $skip;
634
635                         push @filtered, $pos;
636                 }
637
638                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
639         }
640
641         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
642                 if defined $loaded->{$path}->{filtered}->{$all_filters};
643
644         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
645
646         my $data = $self->_loaded('data');
647
648         my $code = $self->_param_scalar('code','');
649         $code =~ s{[\r\n]+$}{\n}s;
650
651         my $commit = $self->param('commit');
652         my $test = $self->param('test');
653
654         my $commit_changed;
655
656         if ( $code && ( $test || $commit ) ) {
657                 # XXX find columns used in code snippet and show them to user
658                 my $order = 0;
659                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
660                         if ( $column =~ s/^(['"])// ) {
661                                 $column =~ s/$1$//;
662                         }
663                         next if $column =~ m/\$/; # hide columns with vars in them
664                         $commit_changed->{$column} = 0;
665                 }
666         }
667
668         my $code_path = $self->app->home->rel_dir('public') . "/code";
669         if ( $commit ) {
670
671                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
672                 my $out;
673                 foreach ( 0 .. $#$filtered ) {
674                         my $i = $filtered->[$_];
675                         __commit_path_code( $path, $i, $code, \$commit_changed );
676                 }
677
678                 $self->_save_change({
679                         path => $path,
680                         time => $self->param('time') || time(),
681                         user => $self->param('user') || $ENV{'LOGNAME'},
682                         code => $code,
683                         commit_changed => $commit_changed,
684                 });
685
686                 if ( my $description = $self->param('code_description') ) {
687                         my $depends = $self->param('code_depends') || die "no code_depends?";
688                         my $path = "$code_path/$depends.$description.pl";
689                         if ( -e $path && ! $self->param('overwrite') ) {
690                                 warn "# code $path not saved\n";
691                         } else {
692                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
693                                 warn "code $path ", -s $path, " bytes saved\n";
694                         }
695                 }
696
697                 # remove console
698                 $code = '';
699                 if ( $out ) {
700                         my $commit_dataset = join('.'
701                                 , $self->param('code_depends')
702                                 , $self->param('code_description')
703                                 , time()
704                         );
705                         my $key = $self->param('code_depends');
706                         $key =~ s/,.+$//;
707                         $key ||= 'key';
708                         my $items;
709                         foreach my $n ( keys %$out ) {
710                                 my $i = { $key => [ $n ] };
711                                 my $ref = ref $out->{$n};
712                                 if ( $ref eq 'HASH' ) {
713                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
714                                 } elsif ( $ref eq 'ARRAY' ) {
715                                         $i->{$_} = $out->{$n};
716                                 } elsif ( ! $ref ) {
717                                         $i->{value} = [ $out->{$n} ];
718                                 } else {
719                                         $i->{_error} = [ dump($out->{$n}) ];
720                                 }
721                                 push @$items, $i;
722                         };
723                         undef $out;
724                         my $stats = __stats( $items );
725                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
726                         unshift @columns, $key;
727
728                         $loaded->{$commit_dataset} = {
729                                 header => [ @columns ],
730                                 columns => [ @columns ],
731                                 mtime => time(),
732                                 data => { items => $items },
733                                 stats => $stats,
734                                 generated => 1,
735                         };
736                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
737                         $self->session('path', $commit_dataset);
738                         $self->session('columns', [ @columns ]);
739                         $self->session('order', $key);
740                         $self->redirect_to('/data/items');
741                         return; # FIXME needed to correctly show columns
742                 }
743
744                 # this might move before $out to recalculate stats on source dataset?
745                 __path_rebuild_stats( $path );
746                 my $c = { map { $_ => 1 } @columns };
747                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
748                 warn "# added_columns ",dump( @added_columns );
749                 unshift @columns, @added_columns;
750
751                 $self->session('columns', [ @columns ]);
752                 $loaded->{$path}->{columns} = [ @columns ];
753                 warn "# new columns ",dump( @columns );
754
755                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
756         }
757
758         my $sorted_items;
759         my $from_end = $sort eq 'd' ? $#$filtered : 0;
760         my $test_changed;
761         my $out;
762         foreach ( 0 .. $limit ) {
763                 my $i = $_ + $offset;
764                 last unless defined $filtered->[$i];
765                 $i = $from_end - $i if $from_end;
766                 my $id = $filtered->[$i];
767                 my $row = Storable::dclone $data->{items}->[ $id ];
768                 if ( $code && $test ) {
769                         my $update;
770                         eval $code;
771                         if ( $@ ) {
772                                 warn "ERROR evaling\n$code\n$@";
773                                 $self->stash('eval_error', $@) if $@;
774                         } else {
775                                 warn "EVAL ",dump($update);
776                                 foreach ( keys %$update ) {
777                                         $test_changed->{$_}++;
778                                         $row->{$_} = $update->{$_};
779                                 }
780                         }
781                 }
782                 $row->{_row_id} ||= $id;
783                 push @$sorted_items, $row;
784         }
785
786         if ( $self->param('export') ) {
787                 my $export_path = $self->_export_path( 'items', @columns);
788                 open(my $fh, '>', $export_path) || warn "ERROR: can't open $export_path: $!";
789                 foreach my $f ( 0 .. $#$filtered ) {
790                         print $fh join("\t", map {
791                                 my $i = $data->{items}->[ $filtered->[$f] ];
792                                 my $v = '\N';
793                                 if ( ! defined $i->{$_} ) {
794                                         # nop
795                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
796                                         $v =join(',', @{ $i->{$_} });
797                                         $v = '\N' if length($v) == 0;
798                                 } else {
799                                         $v = dump $i->{$_};
800                                 }
801                                 $v;
802                         } @columns),"\n";
803                 }
804                 close($fh);
805                 warn "export $export_path ", -s $export_path, " bytes\n";
806         }
807
808         warn "# test_changed ",dump( $test_changed );
809         my $c = { map { $_ => 1 } @columns };
810         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
811         unshift @columns, @added_columns;
812
813         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
814
815         my $code_depends = $self->param('code_depends')||
816         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
817         my $code_description = $self->param('code_description') ||
818         join(',', @added_columns);
819
820         $code_depends ||= $code_description; # self-modifing
821
822         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
823
824         $self->render(
825                 order => $order,
826                 offset => $offset,
827                 limit => $limit,
828                 sorted => $sorted_items,
829                 columns => [ @columns ],
830                 rows => $#$filtered + 1,
831                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
832                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
833                 filters => $self->_current_filters,
834                 code => $code,
835                 cols_changed => $commit ? $commit_changed : $test_changed,
836                 code_depends => $code_depends,
837                 code_description => $code_description,
838                 code_path => $code_path,
839                 out => $out,
840         );
841
842 }
843
844
845 sub order {
846         my $self = shift;
847         $self->session('order', $self->param('order'));
848         $self->session('sort', $self->param('sort'));
849         $self->redirect_to('/data/items');
850 }
851
852 sub _is_numeric {
853         my ( $self, $name ) = @_;
854
855         my $stats = $self->_loaded( 'stats' );
856
857         # sort facet numerically if more >50% elements are numeric
858         my $count = $stats->{$name}->{count};
859         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
860         defined $stats->{$name}->{numeric} &&
861                 $stats->{$name}->{numeric} > $count / 2;
862 }
863
864 sub _is_unique {
865         my ( $self, $name ) = @_;
866         my $stats = $self->_loaded( 'stats' );
867         defined $stats->{$name}->{unique};
868 }
869
870 sub _remove_filter {
871         my ($self,$name) = @_;
872         warn "_remove_filter $name\n";
873
874         my $path = $self->session('path');
875
876         delete $filters->{$name};
877         delete $loaded->{$path}->{filters}->{$name};
878         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
879
880         foreach (
881                         grep { /\Q$name\E/ }
882                         keys %{ $loaded->{$path}->{filtered} }
883         ) {
884                 delete $loaded->{$path}->{filtered}->{$_};
885                 warn "remove filtered cache $_";
886         }
887 }
888
889 sub facet {
890         my $self = shift;
891
892         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
893
894         if ( my $name = $self->param('remove') ) {
895                 $self->_remove_filter( $name );
896                 $self->redirect_to( '/data/items' );
897         }
898
899         my $facet;
900         my $name = $self->param('name') || die "no name";
901
902         my $all = $self->_param_scalar('all', 1);
903         my $data = $self->_loaded('data');
904
905         my $filters = $self->_current_filters;
906         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
907         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
908                 if defined $loaded->{$path}->{filtered}->{$all_filters};
909
910         if ( ! $filtered || $all ) {
911                 $filtered = [ 0 .. $#{ $data->{items} } ];
912                 warn "filter all values\n";
913         } else {
914                 warn "filter using $all_filters\n";
915         }
916
917         foreach my $i ( @$filtered ) {
918                 my $item = $data->{items}->[$i];
919                 if ( ! exists $item->{$name} ) {
920                         $facet->{ _missing }++;
921                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
922                         $facet->{$_}++ foreach @{ $item->{$name} };
923                 } else {
924                         $facet->{ $item->{$name} }++;
925                 }
926         }
927
928 #       warn "# facet $name ",dump $facet;
929
930         my $checked;
931         my @facet_names =
932                   $all                      ? keys %$facet
933                 : defined $filters->{$name} ? @{ $filters->{$name} }
934                 : keys %$facet;
935
936         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
937
938         my $numeric = $self->_is_numeric($name);
939
940         my $sort = $self->param('sort');
941         # sort numeric facets with more than 5 values ascending
942         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
943
944         @facet_names = sort {
945                 my $result;
946                 if ( $sort eq 'a' ) {
947                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
948                 } elsif ( $sort eq 'd' ) {
949                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
950                 } elsif ( $sort eq 'c' ) {
951                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
952                 } else {
953                         warn "unknown sort: $sort";
954                         $result = $a cmp $b;
955                 }
956                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
957                 $result;
958         } @facet_names;
959
960         $self->render( name => $name, facet => $facet, checked => $checked,
961                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
962         );
963 }
964
965
966 sub __invalidate_path_column {
967         my ( $path, $name ) = @_;
968
969         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
970                 delete $loaded->{$path}->{sorted}->{$name};
971                 warn "# invalidate $path sorted $name\n";
972         }
973
974         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
975                 delete $loaded->{$path}->{filtered}->{$_};
976                 warn "# invalidate $path filtered $_\n";
977         }
978 }
979
980 sub __path_modified {
981         my ( $path, $value ) = @_;
982         $value = 1 unless defined $value;
983         
984         $loaded->{$path}->{modified}  = $value;
985
986         warn "# __path_modified $path $value\n";
987 }
988
989 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
990
991 sub _save_change {
992         my ($self,$change) = @_;
993
994         my $change_path = $self->_permanent_path( 'changes' );
995         mkdir $change_path unless -d $change_path;
996         $change_path .= '/' . $change->{time};
997         store $change, $change_path;
998         utime $change->{time}, $change->{time}, $change_path;
999         warn "_save_change $change_path ", dump($change);
1000 }
1001
1002 sub edit {
1003         my $self = shift;
1004         my $new_content = $self->param('new_content');
1005         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1006
1007         my $i = $self->param('_row_id');
1008         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1009         my $path = $self->param('path') || die "no path";
1010         my $name = $self->param('name') || die "no name";
1011         my $status = 200; # 200 = OK, 201 = Created
1012
1013         my $data = $self->_loaded('data');
1014
1015         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1016                 $new_content =~ s/^\s+//s;
1017                 $new_content =~ s/\s+$//s;
1018                 my $v;
1019                 if ( $new_content =~ /\xB6/ ) { # para
1020                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1021                 } else {
1022                         $v = [ $new_content ];
1023                 }
1024
1025                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1026                 my $new = dump $v;
1027                 if ( $old ne $new
1028                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1029                 ) {
1030                         $self->_save_change({
1031                                 path => $path,
1032                                 column => $name,
1033                                 pos => $i,
1034                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1035                                 new => $v,
1036                                 time => $self->param('time') || time(),
1037                                 user => $self->param('user') || $ENV{'LOGNAME'},
1038                                 unique => {
1039                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1040                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1041                                         keys %{ $loaded->{$path}->{stats} }
1042                                 },
1043                         });
1044
1045                         warn "# change $path $i $old -> $new\n";
1046                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1047
1048                         __invalidate_path_column( $path, $name );
1049
1050                         $status = 201; # created
1051                         __path_rebuild_stats( $path );
1052         
1053                         $new_content = join("\xB6",@$v);
1054
1055                 } else {
1056                         warn "# unchanged $path $i $old\n";
1057                         $status = 304;
1058                 }
1059         } else {
1060                 $new_content = "$path $i $name doesn't exist\n";
1061                 $status = 404;
1062         }
1063
1064         warn "# edit $status ", dump $new_content;
1065
1066         $self->render(
1067                 status => $status,
1068                 new_content => scalar $new_content,
1069         );
1070 }
1071
1072
1073 sub save {
1074         my $self = shift;
1075         my $path = $self->_param_or_session('path');
1076         my $dump_path = $self->_save( $path );
1077         __path_modified( $path, 0 );
1078
1079         $self->redirect_to( '/data/items' );
1080 }
1081
1082 sub export {
1083         my $self = shift;
1084
1085         my $dir = $self->app->home->rel_dir('public');
1086
1087         if ( my $import = $self->param('import') ) {
1088
1089                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1090                         my $name = $self->_column_from_unac( $1 );
1091
1092                         my @vals = map { chomp; $_ }
1093                                 read_file "$dir/export/$import", binmode => ':utf8';
1094
1095                         $self->_remove_filter( $name );
1096                         $self->_filter_on_data( $name, @vals );
1097                         $self->session( 'offset' => 0 );
1098                         $self->redirect_to('/data/items');
1099                 } else {
1100                         warn "UNKNOWN IMPORT $import";
1101                 }
1102         }
1103
1104         if ( my $remove = $self->param('remove') ) {
1105                 my $path = "$dir/export/$remove";
1106                 unlink $path if -e $path;
1107                 $path .= '.png';
1108                 unlink $path if -e $path;
1109         }
1110
1111         my $path = $self->_export_path || $self->redirect_to('/data/index');
1112
1113         my @files = grep { ! /\.png$/ } glob "$path/*";
1114         my $mtime = { map { $_ => (stat($_))[9] } @files };
1115         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1116         $self->render( export => [ @files ] );
1117 }
1118
1119 sub __loaded_paths {
1120         return
1121                 grep { defined $loaded->{$_}->{data} }
1122                 keys %$loaded;
1123 }
1124
1125 1;