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