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