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