all code just loads snippets without dependencies
[MojoFacets.git] / lib / MojoFacets / Data.pm
1 package MojoFacets::Data;
2
3 use strict;
4 use warnings;
5
6 use base 'Mojolicious::Controller';
7
8 use Data::Dump qw(dump);
9 use File::Slurp;
10 use Encode;
11 use locale;
12 use File::Find;
13 use Storable;
14 use Time::HiRes qw(time);
15 use File::Path qw(mkpath);
16 use Text::Unaccent::PurePerl;
17 use Digest::MD5;
18
19 use MojoFacets::Import::File;
20 use MojoFacets::Import::HTMLTable;
21 use MojoFacets::Import::CSV;
22 use MojoFacets::Import::CouchDB;
23 use MojoFacets::Import::SQL;
24
25 our $loaded;
26 our $filters;
27
28 sub index {
29         my $self = shift;
30
31         my $data_dir = $self->app->home->rel_dir('data');
32         die "no data dir $data_dir" unless -d $data_dir;
33
34         my @files;
35         my $changes;
36         find( sub {
37                 my $file = $File::Find::name;
38                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
39                         $file =~ s/$data_dir\/*//;
40                         push @files, $file;
41                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
42                         push @{ $changes->{$1} }, $2
43                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
44                         $file =~ s/$data_dir\/*//;
45                         push @files, $file;
46                 } elsif ( -f $file && $file =~ m/\.(csv|storabe|couchdb|sql)$/i ) {
47                         $file =~ s/$data_dir\/*//;
48                         push @files, $file;
49                 } else {
50                         #warn "IGNORE: $file\n";
51                 }
52         }, $data_dir);
53
54         no warnings qw(uninitialized); # mtime
55         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
56                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
57         my $size;
58         $size->{$_} = -s "$data_dir/$_" foreach @files;
59
60         $self->render(
61                 files => [ @files ],
62                 size => $size,
63                 loaded => $loaded,
64                 filters => $filters,
65                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
66                 changes => $changes,
67         );
68 }
69
70 sub _dump_path {
71         my ( $self, $name ) = @_;
72         my $dir = $self->app->home->rel_dir('data');
73         $name =~ s/^$dir//;
74         $name =~ s/\/+/_/g;
75         return '/tmp/mojo_facets.' . $name . '.storable';
76 }
77
78 sub _save {
79         my ( $self, $path ) = @_;
80
81         my $dump_path = $self->_dump_path( $path );
82         my $first_load = ! -e $dump_path;
83         warn "save loaded to $dump_path";
84         my $info = $loaded->{$path};
85         store $info, $dump_path;
86
87         if ( $first_load ) {
88                 my $mtime = $loaded->{$path}->{mtime};
89                 utime $mtime, $mtime, $dump_path;
90                 warn "sync time to $path at $mtime\n";
91         }
92
93         warn $dump_path, ' ', -s $dump_path, " bytes\n";
94         return $dump_path;
95 }
96
97
98 sub __stats {
99
100         my $stats;
101
102         my $nr_items = $#{ $_[0] } + 1;
103
104         warn "__stats $nr_items\n";
105
106         foreach my $e ( @{ $_[0] } ) {
107                 foreach my $n ( keys %$e ) {
108                         $stats->{$n}->{count}++;
109                         my @v;
110                         if ( ref $e->{$n} eq 'ARRAY' ) {
111                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
112                                 @v = @{ $e->{$n} };
113                         } else {
114                                 @v = ( $e->{$n} );
115                         }
116
117                         foreach my $x ( @v ) {
118                                 $stats->{$n}->{numeric}++
119                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
120                                 $stats->{$n}->{empty}++
121                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
122                         }
123
124                 }
125         }
126
127         foreach my $n ( keys %$stats ) {
128                 my $s = $stats->{$n};
129                 next unless defined $s->{array};
130                 if ( $s->{array} == $s->{count} ) {
131                         delete $s->{array};
132                         if ( $s->{count} == $nr_items ) {
133                                 warn "check $n for uniqeness\n";
134                                 my $unique;
135                                 foreach my $e ( @{ $_[0] } ) {
136                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
137                                                 $unique = 0;
138                                                 last;
139                                         }
140                                 }
141                                 if ( $unique ) {
142                                         $stats->{$n}->{unique} = 1;
143                                         #warn "# $n unique ",dump( $unique );
144                                 }
145                         }
146                 }
147         }
148
149         warn "# __stats ",dump($stats);
150
151         return $stats;
152 }
153
154 sub _param_or_session {
155         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
156 }
157
158 sub stats {
159         my $self = shift;
160         my $path = $self->_param_or_session('path');
161         warn "stats $path\n";
162         delete $loaded->{$path}->{stats};
163         return $self->redirect_to( '/data/columns' );
164 }
165
166
167 sub _load_path {
168         my ( $self, $path ) = @_;
169
170         return if defined $loaded->{$path}->{generated};
171
172         my $full_path = $self->app->home->rel_file( 'data/' . $path );
173         return $self->redirect_to('/data/index') unless -r $full_path;
174
175         my $dump_path = $self->_dump_path( $path );
176
177         if ( defined $loaded->{$path}->{data} ) {
178                 my $mtime = (stat($full_path))[9];
179                 return if $loaded->{$path}->{mtime} == $mtime;
180                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
181         } elsif ( -e $dump_path ) {
182                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
183                 my $info = retrieve $dump_path;
184                 $loaded->{ $path } = $info;
185                 return;
186         }
187
188         my $data;
189         if ( -f $full_path ) {
190                 if ( $full_path =~ m/.csv/i ) {
191                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
192                 } elsif ( $full_path =~ m/.sql/i ) {
193                         $data = MojoFacets::Import::SQL->new( full_path => $full_path )->data;
194                 } elsif ( $full_path =~ m/.couchdb/i ) {
195                         $data = MojoFacets::Import::CouchDB->new( full_path => $full_path )->data;
196                 } elsif ( $full_path =~ m/.storable/ ) {
197                         warn "open $full_path ", -s $full_path, " bytes";
198                         open(my $pipe, "<", $full_path) || die $!;
199                         while ( my $o = eval { Storable::fd_retrieve $pipe } ) {
200                                 if ( defined $o->{item} ) {
201                                         push @{ $data->{items} }, $o->{item};
202                                 } else {
203                                         warn "SKIP ",dump($o);
204                                 }
205                         }
206                         close($pipe);
207                         warn "loaded ", $#{ $data->{items} } + 1, " items from $full_path\n";
208                         $data->{generated}++;
209                 } else {
210                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
211                 }
212         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
213                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
214         } else {
215                 die "can't load $full_path";
216         }
217
218         my @header;
219
220         if ( defined $data->{header} ) {
221                 if ( ref $data->{header} eq 'ARRAY' ) {
222                         @header = @{ $data->{header} };
223                 } else {
224                         warn "header not array ", dump( $data->{header} );
225                 }
226         }
227
228         my $stats = __stats( $data->{items} );
229
230         @header =
231                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
232                 grep { defined $stats->{$_}->{count} } keys %$stats
233                 unless @header;
234
235         my $info = {
236                 header => [ @header ],
237                 stats  => $stats,
238                 full_path => $full_path,
239                 size => -s $full_path,
240                 mtime => (stat($full_path))[9],
241                 data => $data,
242                 defined $data->{generated} ? ( generated => 1 ) : (),
243         };
244
245         $loaded->{ $path } = $info;
246         $self->_save( $path ) unless $info->{generated};
247
248 }
249
250
251 sub load {
252         my $self = shift;
253
254         my $path = $self->param('path') || return $self->redirect_to( '/data/index' );
255
256         my @paths = $self->param('paths');
257         warn "# paths ", dump @paths;
258
259         foreach my $p ( keys %$loaded ) {
260                 next if grep { /^\Q$p\E$/ } @paths;
261                 warn "remove $p from memory\n";
262                 delete $loaded->{$p};
263         }
264
265         $self->_load_path( $_ ) foreach @paths;
266
267         warn "# path $path\n";
268         $self->_load_path( $path );
269
270         $self->session( 'path' => $path );
271
272         my $redirect_to = '/data/items';
273
274         $self->session( 'header' => $loaded->{$path}->{header} );
275         if ( ! defined $loaded->{$path}->{columns} ) {
276                 my $columns_path = $self->_permanent_path( 'columns' );
277                 if ( -e $columns_path ) {
278                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
279                         $loaded->{$path}->{columns} = [ @columns ];
280                         warn "# columns_path $columns_path ",dump(@columns);
281                 } else {
282                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
283                 }
284
285                 $redirect_to = '/data/columns';
286         }
287         $self->session( 'columns' => $loaded->{$path}->{columns} );
288         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
289         return $self->redirect_to( $redirect_to );
290 }
291
292
293 sub _loaded {
294         my ( $self, $name ) = @_;
295         my $path = $self->session('path') || $self->param('path');
296         return $self->redirect_to('/data/index') unless $path;
297
298         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
299                 my $caller = (caller(1))[3];
300                 if ( $caller =~ m/::edit/ ) {
301                         warn "rebuild stats for $path ignored caller $caller\n";
302                 } else {
303                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
304 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
305                         $loaded->{$path}->{rebuild_stats} = 1;
306                         $loaded->{$path}->{modified} = 1;
307                 }
308         }
309
310         if ( defined $loaded->{$path}->{rebuild_stats} ) {
311                 warn "rebuild_stats $path";
312                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
313                 delete $loaded->{$path}->{rebuild_stats};
314         }
315
316         if ( ! defined $loaded->{$path}->{$name} ) {
317                 warn "$path $name isn't loaded\n";
318                 $self->_load_path( $path );
319                 if ( ! defined $loaded->{$path}->{stats} ) {
320                         warn "rebuild stats for $path\n";
321                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
322                 }
323                 if ( ! defined $loaded->{$path}->{$name} ) {
324                         warn "MISSING $name for $path\n";
325                         return $self->redirect_to('/data/index')
326                 }
327         }
328
329         $self->session( 'modified' => $loaded->{$path}->{modified} );
330
331         return $loaded->{$path}->{$name};
332 }
333
334
335 sub _checked {
336         my $self = shift;
337         my $checked;
338         $checked->{$_}++ foreach @_;
339 #       warn "# _checked ",dump($checked);
340         return $checked;
341 }
342
343 sub _permanent_path {
344         my $self = shift;
345         my $path = $self->_param_or_session('path');
346         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
347 }
348
349 sub __unac {
350         my $n = shift;
351         $n = unac_string($n);
352         $n =~ s/\W+/_/g;
353         return $n;
354 }
355
356 sub _column_from_unac {
357         my ($self,$name) = @_;
358
359         my $stats = $self->_loaded('stats');
360         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
361
362         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
363 }
364
365 sub _export_path {
366         my $self = shift;
367         my $path = $self->_param_or_session('path');
368         if ( ! $path ) {
369                 warn "no path in param or session";
370                 return;
371         }
372         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
373         mkpath $dir unless -e $dir;
374         my $name = __export_path_name( $path, @_ );
375         my $full = $dir . '/' . $name;
376         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
377         return $full;
378 }
379
380 sub __export_path_name {
381         my $max_length = 80;
382
383         my $path = shift;
384         my $name = join('.', map { __unac($_) } @_ );
385         if ( length($name) > $max_length ) {
386                 $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
387         }
388         return $name;
389 }
390
391 sub columns {
392     my $self = shift;
393
394         my $view_path = $self->_permanent_path( 'views' );
395
396         if ( $self->param('columns') ) {
397                 my @columns = $self->_param_array('columns');
398                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
399                 if ( my $view = $self->param('view') ) {
400                         mkdir $view_path unless -e $view_path;
401                         write_file( "$view_path/$view", { binmode => ':utf8' }, map { "$_\n" } @columns );
402                         warn "view $view_path/$view ", -s "$view_path/$view", " bytes\n";
403                 }
404
405                 return $self->redirect_to('/data/items');
406
407         } elsif ( ! $self->session('header') ) {
408                 return $self->redirect_to('/data/load');
409         }
410
411         if ( my $id = $self->param('id') ) {
412                 my $view_full = "$view_path/$id";
413                 if ( -f $view_full ) {
414                         my @columns = map { chomp; $_ } read_file $view_full, binmode => ':utf8';
415                         warn "view $view_full loaded ", dump @columns;
416                         $self->session( 'columns' => [ @columns ] );
417                         return $self->redirect_to('/data/items');
418                 }
419         }
420
421         my $stats = $self->_loaded( 'stats' );
422
423         my @columns;
424         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
425
426         foreach my $c ( @{ $self->session( 'header' ) } ) {
427                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
428         }
429
430         my @views;
431         if ( -d $view_path ) {
432                 @views = map { s{^\Q$view_path\E/*}{}; $_ } glob "$view_path/*";
433                 warn "# views ",dump @views;
434         }
435
436     $self->render(
437                 message => 'Select columns to display',
438                 stats => $stats,
439                 columns => \@columns,
440                 checked => $self->_checked( $self->_param_array('columns') ),
441                 views => \@views,
442         );
443 }
444
445 sub _param_array {
446     my ($self,$name) = @_;
447
448         my @array = $self->param($name);
449         my $path  = $self->session('path');
450
451         if ( @array ) {
452                 $self->session($name => [ @array ]);
453                 $loaded->{$path}->{$name} = [ @array ];
454         } elsif ( my $session = $self->session($name) ) {
455                 if ( ref $session eq 'ARRAY' ) {
456                         @array = @$session;
457                 } else {
458                         die "$name not array ",dump($session);
459                 }
460         }
461         #warn "# $name ",dump @array;
462         return @array;
463 }
464
465 sub _param_scalar {
466     my ($self,$name,$default) = @_;
467
468         my $scalar = $self->param($name);
469
470         if ( defined $scalar ) {
471                 $self->session($name => $scalar);
472         } else {
473                 $scalar = $self->session($name);
474         }
475
476         if ( ! defined $scalar ) {
477                 $scalar = $default;
478                 die "no default for $name" unless defined $scalar;
479                 $self->session($name => $scalar);
480         }
481
482         warn "# _param_scalar $name ",dump $scalar;
483         return $scalar;
484 }
485
486 sub filter {
487         my $self = shift;
488
489         my $name = $self->param('filter_name') || die "name?";
490         my @vals = $self->param('filter_vals');
491
492         $self->_remove_filter( $name );
493         if ( @vals ) {
494                 $self->_filter_on_data( $name, @vals );
495                 if ( my $permanent = $self->param('_permanent') ) {
496                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
497                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
498                         warn "permanent filter $permanent_path ", -s $permanent_path;
499                 }
500         }
501
502         $self->session( 'offset' => 0 );
503
504         return $self->redirect_to('/data/items');
505 }
506
507 sub _filter_on_data {
508         my ( $self, $name, @vals ) = @_;
509
510         my $path = $self->session('path');
511
512         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
513                 warn "filter $name not found in data set";
514                 return;
515         }
516
517         $filters->{$name} = [ @vals ];
518         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
519
520         my $filter_hash;
521         $filter_hash->{$_}++ foreach @vals;
522
523         #warn "# filter_hash ",dump( $filter_hash );
524
525         my $items = $self->_loaded('data')->{items};
526
527         my $include_missing = defined $filter_hash->{_missing};
528         my $filtered_items;
529
530         foreach my $i ( 0 .. $#$items ) {
531
532                 if ( defined $items->[$i]->{$name} ) {
533                         my $row = $items->[$i]->{$name};
534                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
535                         foreach my $v ( @$row ) {
536                                 if ( defined $filter_hash->{ $v } ) {
537                                         $filtered_items->{$i}++;
538                                 }
539                         }
540                 } elsif ( $include_missing ) {
541                         $filtered_items->{$i}++;
542                 }
543         }
544
545         #warn "# filter $name ",dump($filtered_items);
546
547         # invalidate filters on other datasets
548         foreach my $dataset ( grep { exists $loaded->{$_}->{filters}->{$name} } keys %$loaded ) {
549                 delete $loaded->{$dataset}->{filters}->{$name};
550                 delete $loaded->{$dataset}->{filtered};
551         }
552
553         $loaded->{$path}->{filters}->{$name} = $filtered_items;
554         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
555 }
556
557
558 sub _current_filters {
559         my $self = shift;
560         my $current_filters;
561         my $stats = $self->_loaded('stats');
562
563         $current_filters->{ $_ } = $filters->{ $_ }
564                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
565         ;
566         warn "# _current_filters ",dump( keys %$current_filters );
567         return $current_filters;
568 }
569
570 sub _data_sorted_by {
571         my ( $self, $order ) = @_;
572
573         my $path = $self->session('path');
574
575         warn "_data_sorted_by $order from $path";
576
577         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
578                 return $loaded->{$path}->{sorted}->{$order};
579         }
580
581         my $data = $self->_loaded( 'data' );
582         my $numeric = $self->_is_numeric($order);
583         my $missing = $numeric ? 0 : '';
584         no warnings qw(numeric);
585         my $nr = 0;
586         my @sorted = map {
587                 $_->[0]
588         } sort {
589                 if ( $numeric ) {
590                         $a->[1] <=> $b->[1]
591                 } else {
592                         $a->[1] cmp $b->[1]
593                 }
594         } map {
595                 my $v;
596                 if ( ! exists $_->{$order} ) {
597                         $v = $missing;
598                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
599                         $v = join('', @{$_->{$order}});
600                 } else {
601                         $v = $_->{$order};
602                 }
603                 [ $nr++, $v ]
604         } @{ $data->{items} }
605         ;
606
607         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
608         #warn "# sorted ",dump( @sorted );
609
610         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
611 }
612
613
614 sub __all_filters {
615         my $order = pop @_;
616         join(',', sort(@_), 'order', $order);
617 }
618
619 our ($out, $key,$value);
620
621 sub __commit_path_code {
622         my ( $path, $i, $code, $commit_changed ) = @_;
623
624         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
625         my $row = $items->[$i];
626         my $update;
627         eval $code;
628         foreach ( keys %$update ) {
629                 $$commit_changed->{$_}++;
630                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
631         }
632         #warn "__commit_path_code $path $i ",dump( $update );
633 }
634
635 # uses templates/admin.html.ep
636 sub _switch_dataset {
637         my $self = shift;
638
639         my $datasets;
640
641         foreach my $path ( keys %$loaded ) {
642                 next unless exists $loaded->{$path}->{data};
643                 push @$datasets, $path;
644         }
645
646         warn "# datasets ",dump($datasets);
647
648         $self->stash( 'datasets' => $datasets );
649 }
650
651 sub items {
652         my $self = shift;
653
654         $self->_switch_dataset;
655
656         if ( my $show = $self->param('id') ) {
657                 $self->param('show', $show);
658                 warn "show $show\n";
659         }
660
661         my $path = $self->_param_scalar('path');
662
663         my @columns = $self->_param_array('columns');
664         return $self->redirect_to('/data/columns') unless @columns;
665         my $order   = $self->_param_scalar('order', $columns[0]);
666         my $sort    = $self->_param_scalar('sort', 'a');
667         my $offset  = $self->_param_scalar('offset', 0);
668         my $limit   = $self->_param_scalar('limit', 20);
669         $self->_param_scalar('show', 'table');
670
671         # fix offset when changing limit
672         $offset = int( $offset / $limit ) * $limit;
673
674         if ( ! grep { /^\Q$order\E$/ } @columns ) {
675                 $order = $columns[0];
676                 $self->session( order => $order );
677         }
678         my $sorted = $self->_data_sorted_by( $order );
679
680         my @filter_names;
681         if ( $filters ) {
682                 foreach my $name ( keys %$filters ) {
683                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
684                                 warn "skip filter $name not found in $path\n";
685                                 next;
686                         }
687                         push @filter_names, $name;
688                 }
689                 warn "filter_names ",dump( @filter_names );
690                 foreach my $name ( @filter_names ) {
691                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
692                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
693                 }
694         }
695
696         my $all_filters = __all_filters( @filter_names,$order );
697
698 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
699
700         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
701
702                 my $path_filters = $loaded->{$path}->{filters};
703
704                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
705
706                 my @filtered;
707                 foreach my $i ( 0 .. $#$sorted ) {
708                         my $pos = $sorted->[$i];
709
710                         if ( $#filter_names == -1 ) {
711                                 push @filtered, $pos;
712                                 next;
713                         }
714
715                         my $skip = 0;
716                         foreach ( @filter_names ) {
717                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
718                         }
719                         next if $skip;
720
721                         push @filtered, $pos;
722                 }
723
724                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
725         }
726
727         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
728                 if defined $loaded->{$path}->{filtered}->{$all_filters};
729
730         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
731
732         my $data = $self->_loaded('data');
733
734         my $code = $self->_param_scalar('code','');
735         $code =~ s{\r}{}gs;
736         $code =~ s{\n+$}{\n}s;
737
738         my $commit = $self->param('commit');
739         my $test = $self->param('test');
740
741         my $commit_changed;
742
743         if ( $code && ( $test || $commit ) ) {
744                 # XXX find columns used in code snippet and show them to user
745                 my $order = 0;
746                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
747                         if ( $column =~ s/^(['"])// ) {
748                                 $column =~ s/$1$//;
749                         }
750                         next if $column =~ m/\$/; # hide columns with vars in them
751                         $commit_changed->{$column} = 0;
752                 }
753         }
754
755         my $code_path = $self->app->home->rel_dir('public') . "/code";
756         if ( $commit ) {
757
758                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
759                 ( $key, $value, $out ) = ( 'key', 'value' );
760                 foreach ( 0 .. $#$filtered ) {
761                         my $i = $filtered->[$_];
762                         __commit_path_code( $path, $i, $code, \$commit_changed );
763                 }
764
765                 # this might move before $out to recalculate stats on source dataset?
766                 __path_rebuild_stats( $path );
767                 my $c = { map { $_ => 1 } @columns };
768                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
769                 warn "# added_columns ",dump( @added_columns );
770                 unshift @columns, @added_columns;
771
772                 $loaded->{$path}->{columns} = [ @columns ];
773                 warn "# new columns ",dump( @columns );
774
775                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
776
777                 $self->_save_change({
778                         path => $path,
779                         time => $self->param('time') || time(),
780                         user => $self->param('user') || $ENV{'LOGNAME'},
781                         code => $code,
782                         commit_changed => $commit_changed,
783                 });
784
785                 if ( my $description = $self->param('code_description') ) {
786                         my $depends = $self->param('code_depends') || die "no code_depends?";
787                         my $path = "$code_path/$depends.$description.pl";
788                         if ( -e $path && ! $self->param('overwrite') ) {
789                                 warn "# code $path not saved\n";
790                         } else {
791                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
792                                 warn "code $path ", -s $path, " bytes saved\n";
793                         }
794                 }
795
796                 # remove console
797                 $code = '';
798                 if ( $out ) {
799                         my $commit_dataset = join('.'
800                                 , $self->param('code_depends')
801                                 , $self->param('code_description')
802                                 , time()
803                         );
804                         $key ||= 'key';
805                         $value ||= 'value';
806                         warn "key $key value $value";
807                         my $items;
808                         foreach my $n ( keys %$out ) {
809                                 my $i = { $key => [ $n ] };
810                                 my $ref = ref $out->{$n};
811                                 if ( $ref eq 'HASH' ) {
812                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
813                                 } elsif ( $ref eq 'ARRAY' ) {
814                                         $i->{$_} = $out->{$n};
815                                 } elsif ( ! $ref ) {
816                                         $i->{$value} = [ $out->{$n} ];
817                                 } else {
818                                         $i->{_error} = [ dump($out->{$n}) ];
819                                 }
820                                 push @$items, $i;
821                         };
822                         undef $out;
823                         my $stats = __stats( $items );
824                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
825                         unshift @columns, $key;
826
827                         $loaded->{$commit_dataset} = {
828                                 header => [ @columns ],
829                                 columns => [ @columns ],
830                                 mtime => time(),
831                                 data => { items => $items },
832                                 stats => $stats,
833                                 generated => 1,
834                         };
835                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
836                         $self->session('path', $commit_dataset);
837                         $self->session('columns', [ @columns ]);
838                         $self->session('order', $key);
839                         return $self->redirect_to('/data/items');
840                 }
841
842                 $self->session('columns', [ @columns ]);
843         }
844
845         my $sorted_items;
846         my $from_end = $sort eq 'd' ? $#$filtered : 0;
847         my $test_changed;
848         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
849         foreach ( 0 .. $limit ) {
850                 my $i = $_ + $offset;
851                 last unless defined $filtered->[$i];
852                 $i = $from_end - $i if $from_end;
853                 my $id = $filtered->[$i];
854                 my $row = Storable::dclone $data->{items}->[ $id ];
855                 if ( $code && $test ) {
856                         my $update;
857                         eval $code;
858                         if ( $@ ) {
859                                 warn "ERROR evaling $@", dump($code);
860                                 $self->stash('eval_error', $@) if $@;
861                         } else {
862                                 warn "EVAL ",dump($update);
863                                 foreach ( keys %$update ) {
864                                         $test_changed->{$_}++;
865                                         $row->{$_} = $update->{$_};
866                                 }
867                         }
868                 }
869                 $row->{_row_id} ||= $id;
870                 push @$sorted_items, $row;
871         }
872
873         if ( $self->param('export') ) {
874                 my $export_path = $self->_export_path( 'items', @columns);
875                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
876                 print $fh "#",join("\t",@columns),"\n";
877                 foreach my $f ( 0 .. $#$filtered ) {
878                         print $fh join("\t", map {
879                                 my $i = $data->{items}->[ $filtered->[$f] ];
880                                 my $v = '\N';
881                                 if ( ! defined $i->{$_} ) {
882                                         # nop
883                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
884                                         $v =join(',', @{ $i->{$_} });
885                                         $v = '\N' if length($v) == 0;
886                                 } elsif ( ! ref $i->{$_} ) {
887                                         $v = $i->{$_};
888                                 } else {
889                                         $v = dump $i->{$_};
890                                 }
891                                 $v;
892                         } @columns),"\n";
893                 }
894                 close($fh);
895                 warn "export $export_path ", -s $export_path, " bytes\n";
896         }
897
898         warn "# test_changed ",dump( $test_changed );
899         my $c = { map { $_ => 1 } @columns };
900         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
901         unshift @columns, @added_columns;
902
903         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
904
905         my $depends_on;
906         my $tmp = $code; $tmp =~ s/\$row->{(['"]?)(\w+)\1/$depends_on->{$2}++/gse;
907         warn "# depends_on ",dump $depends_on;
908
909         my $test_added = Storable::dclone $test_changed;
910         delete $test_added->{$_} foreach keys %$depends_on;
911
912         my $code_depends = $self->param('code_depends')
913         || join(',', keys %$depends_on);
914
915         my $code_description = $self->param('code_description') ||
916         join(',', keys %$test_added);
917
918         $code_depends ||= $code_description; # self-modifing
919         if ( ! $code_depends && $out ) {
920                 $code_depends = $key;
921                 $code_description = $value;
922         }
923
924         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
925
926         $self->render(
927                 order => $order,
928                 offset => $offset,
929                 limit => $limit,
930                 sorted => $sorted_items,
931                 columns => [ @columns ],
932                 rows => $#$filtered + 1,
933                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
934                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
935                 filters => $self->_current_filters,
936                 code => $code,
937                 cols_changed => $commit ? $commit_changed : $test_changed,
938                 code_depends => $code_depends,
939                 code_description => $code_description,
940                 code_path => $code_path,
941                 out => $out,
942         );
943
944 }
945
946
947 sub order {
948         my $self = shift;
949         $self->session('order', $self->param('order'));
950         $self->session('sort', $self->param('sort'));
951         return $self->redirect_to('/data/items');
952 }
953
954 sub _is_numeric {
955         my ( $self, $name ) = @_;
956
957         my $stats = $self->_loaded( 'stats' );
958
959         # sort facet numerically if more >50% elements are numeric
960         my $count = $stats->{$name}->{count};
961         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
962         defined $stats->{$name}->{numeric} &&
963                 $stats->{$name}->{numeric} > $count / 2;
964 }
965
966 sub _is_unique {
967         my ( $self, $name ) = @_;
968         my $stats = $self->_loaded( 'stats' );
969         defined $stats->{$name}->{unique};
970 }
971
972 sub _remove_filter {
973         my ($self,$name) = @_;
974         warn "_remove_filter $name\n";
975
976         my $path = $self->session('path');
977
978         delete $filters->{$name};
979         delete $loaded->{$path}->{filters}->{$name};
980         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
981
982         foreach (
983                         grep { /\Q$name\E/ }
984                         keys %{ $loaded->{$path}->{filtered} }
985         ) {
986                 delete $loaded->{$path}->{filtered}->{$_};
987                 warn "remove filtered cache $_";
988         }
989 }
990
991 sub facet {
992         my $self = shift;
993
994         my $path = $self->session('path') || return $self->redirect_to( '/data/index' );
995
996         if ( my $name = $self->param('remove') ) {
997                 $self->_remove_filter( $name );
998                 return $self->redirect_to( '/data/items' );
999         }
1000
1001         my $facet;
1002         my $name = $self->param('name') || die "no name";
1003
1004         my $all = $self->_param_scalar('all', 1);
1005         my $data = $self->_loaded('data');
1006
1007         my $filters = $self->_current_filters;
1008         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
1009         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
1010                 if defined $loaded->{$path}->{filtered}->{$all_filters};
1011
1012         if ( ! $filtered || $all ) {
1013                 $filtered = [ 0 .. $#{ $data->{items} } ];
1014                 warn "filter all values\n";
1015         } else {
1016                 warn "filter using $all_filters\n";
1017         }
1018
1019         foreach my $i ( @$filtered ) {
1020                 my $item = $data->{items}->[$i];
1021                 if ( ! exists $item->{$name} ) {
1022                         $facet->{ _missing }++;
1023                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
1024                         $facet->{$_}++ foreach @{ $item->{$name} };
1025                 } else {
1026                         $facet->{ $item->{$name} }++;
1027                 }
1028         }
1029
1030         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
1031
1032         if ( my $code = $self->param('code') ) {
1033                 my $out;
1034                 foreach my $value ( keys %$facet ) {
1035                         my $count = $facet->{$value};
1036                         my $checked = $checked_values->{$value};
1037                         eval $code;
1038                         if ( $@ ) {
1039                                 $out = $@;
1040                                 warn "ERROR: $@\n$code\n";
1041                                 last;
1042                         } elsif ( $checked != $checked_values->{$value} ) {
1043                                 warn "checked $value $count -> $checked\n";
1044                                 $checked_values->{$value} = $checked;
1045                         }
1046                 }
1047                 warn "out ",dump( $out );
1048                 $self->stash( out => $out );
1049         }
1050
1051 #       warn "# facet $name ",dump $facet;
1052
1053         my @facet_names =
1054                   $all                      ? keys %$facet
1055                 : defined $filters->{$name} ? @{ $filters->{$name} }
1056                 : keys %$facet;
1057
1058         my $numeric = $self->_is_numeric($name);
1059
1060         my $sort = $self->param('sort');
1061         # sort numeric facets with more than 5 values ascending
1062         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
1063
1064         @facet_names = sort {
1065                 my $result;
1066                 if ( $sort eq 'a' ) {
1067                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
1068                 } elsif ( $sort eq 'd' ) {
1069                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1070                 } elsif ( $sort eq 'c' ) {
1071                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1072                 } else {
1073                         warn "unknown sort: $sort";
1074                         $result = $a cmp $b;
1075                 }
1076                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1077                 $result;
1078         } @facet_names;
1079
1080         $self->render( name => $name, facet => $facet, checked => $checked_values,
1081                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1082         );
1083 }
1084
1085
1086 sub __invalidate_path_column {
1087         my ( $path, $name ) = @_;
1088
1089         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1090                 delete $loaded->{$path}->{sorted}->{$name};
1091                 warn "# invalidate $path sorted $name\n";
1092         }
1093
1094         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1095                 delete $loaded->{$path}->{filtered}->{$_};
1096                 warn "# invalidate $path filtered $_\n";
1097         }
1098 }
1099
1100 sub __path_modified {
1101         my ( $path, $value ) = @_;
1102         $value = 1 unless defined $value;
1103         
1104         $loaded->{$path}->{modified}  = $value;
1105
1106         warn "# __path_modified $path $value\n";
1107 }
1108
1109 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1110
1111 sub _save_change {
1112         my ($self,$change) = @_;
1113
1114         my $change_path = $self->_permanent_path( 'changes' );
1115         mkdir $change_path unless -d $change_path;
1116         $change_path .= '/' . $change->{time};
1117         store $change, $change_path;
1118         utime $change->{time}, $change->{time}, $change_path;
1119         warn "_save_change $change_path ", dump($change);
1120 }
1121
1122 sub edit {
1123         my $self = shift;
1124         my $new_content = $self->param('new_content');
1125         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1126
1127         my $i = $self->param('_row_id');
1128         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1129         my $path = $self->param('path') || die "no path";
1130         my $name = $self->param('name') || die "no name";
1131         my $status = 200; # 200 = OK, 201 = Created
1132
1133         my $data = $self->_loaded('data');
1134
1135         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1136                 $new_content =~ s/^\s+//s;
1137                 $new_content =~ s/\s+$//s;
1138                 my $v;
1139                 if ( $new_content =~ /\xB6/ ) { # para
1140                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1141                 } else {
1142                         $v = [ $new_content ];
1143                 }
1144
1145                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1146                 my $new = dump $v;
1147                 if ( $old ne $new
1148                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1149                 ) {
1150                         $self->_save_change({
1151                                 path => $path,
1152                                 column => $name,
1153                                 pos => $i,
1154                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1155                                 new => $v,
1156                                 time => $self->param('time') || time(),
1157                                 user => $self->param('user') || $ENV{'LOGNAME'},
1158                                 unique => {
1159                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1160                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1161                                         keys %{ $loaded->{$path}->{stats} }
1162                                 },
1163                         });
1164
1165                         warn "# change $path $i $old -> $new\n";
1166                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1167
1168                         __invalidate_path_column( $path, $name );
1169
1170                         $status = 201; # created
1171                         __path_rebuild_stats( $path );
1172         
1173                         $new_content = join("\xB6",@$v);
1174
1175                 } else {
1176                         warn "# unchanged $path $i $old\n";
1177                         $status = 304;
1178                 }
1179         } else {
1180                 $new_content = "$path $i $name doesn't exist\n";
1181                 $status = 404;
1182         }
1183
1184         warn "# edit $status ", dump $new_content;
1185
1186         $self->render(
1187                 status => $status,
1188                 new_content => scalar $new_content,
1189         );
1190 }
1191
1192
1193 sub save {
1194         my $self = shift;
1195         my $path = $self->_param_or_session('path');
1196         my $dump_path = $self->_save( $path );
1197         __path_modified( $path, 0 );
1198
1199         return $self->redirect_to( '/data/items' );
1200 }
1201
1202 sub export {
1203         my $self = shift;
1204
1205         my $dir = $self->app->home->rel_dir('public');
1206
1207         if ( my $import = $self->param('import') ) {
1208
1209                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1210                         my $name = $self->_column_from_unac( $1 );
1211
1212                         my @vals = map { chomp; $_ }
1213                                 read_file "$dir/export/$import", binmode => ':utf8';
1214
1215                         $self->_remove_filter( $name );
1216                         $self->_filter_on_data( $name, @vals );
1217                         $self->session( 'offset' => 0 );
1218                         return $self->redirect_to('/data/items');
1219                 } else {
1220                         warn "UNKNOWN IMPORT $import";
1221                 }
1222         }
1223
1224         if ( my $remove = $self->param('remove') ) {
1225                 my $path = "$dir/export/$remove";
1226                 unlink $path if -e $path;
1227                 $path .= '.png';
1228                 unlink $path if -e $path;
1229         }
1230
1231         my $path = $self->_export_path || return $self->redirect_to('/data/index');
1232
1233         my @files = grep { ! /\.png$/ } glob "$path/*";
1234         my $mtime = { map { $_ => (stat($_))[9] } @files };
1235         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1236         $self->render( export => [ @files ] );
1237 }
1238
1239 sub __loaded_paths {
1240         return
1241                 grep { defined $loaded->{$_}->{data} }
1242                 keys %$loaded;
1243 }
1244
1245 sub remove {
1246         my $self = shift;
1247         my $path = $self->param('path');
1248         if ( $path =~ m{^/tmp/mojo_facets\.} ) {
1249                 unlink $path;
1250                 warn "# unlink $path";
1251                 if ( my $name = $self->param('name') ) {
1252                         delete $loaded->{$name};
1253                         warn "# remove $name from memory";
1254                 }
1255         } else {
1256                 warn "WARNING: $path unlink ignored";
1257         }
1258         return $self->redirect_to( '/data/load' );
1259 }
1260
1261 1;