d9c1dfd469e9670813fa71c516cbf2e7663c2138
[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
23 our $loaded;
24 our $filters;
25
26 sub index {
27         my $self = shift;
28
29         my $data_dir = $self->app->home->rel_dir('data');
30         die "no data dir $data_dir" unless -d $data_dir;
31
32         my @files;
33         my $changes;
34         find( sub {
35                 my $file = $File::Find::name;
36                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
37                         $file =~ s/$data_dir\/*//;
38                         push @files, $file;
39                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
40                         push @{ $changes->{$1} }, $2
41                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
42                         $file =~ s/$data_dir\/*//;
43                         push @files, $file;
44                 } elsif ( -f $file && $file =~ m/\.csv$/i ) {
45                         $file =~ s/$data_dir\/*//;
46                         push @files, $file;
47                 } elsif ( -f $file && $file =~ m/\.storable/i ) {
48                         $file =~ s/$data_dir\/*//;
49                         push @files, $file;
50                 } else {
51                         #warn "IGNORE: $file\n";
52                 }
53         }, $data_dir);
54
55         no warnings qw(uninitialized); # mtime
56         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
57                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
58         my $size;
59         $size->{$_} = -s "$data_dir/$_" foreach @files;
60
61         $self->render(
62                 files => [ @files ],
63                 size => $size,
64                 loaded => $loaded,
65                 filters => $filters,
66                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
67                 changes => $changes,
68         );
69 }
70
71 sub _dump_path {
72         my ( $self, $name ) = @_;
73         my $dir = $self->app->home->rel_dir('data');
74         $name =~ s/^$dir//;
75         $name =~ s/\/+/_/g;
76         return '/tmp/mojo_facets.' . $name . '.storable';
77 }
78
79 sub _save {
80         my ( $self, $path ) = @_;
81
82         my $dump_path = $self->_dump_path( $path );
83         my $first_load = ! -e $dump_path;
84         warn "save loaded to $dump_path";
85         my $info = $loaded->{$path};
86         store $info, $dump_path;
87
88         if ( $first_load ) {
89                 my $mtime = $loaded->{$path}->{mtime};
90                 utime $mtime, $mtime, $dump_path;
91                 warn "sync time to $path at $mtime\n";
92         }
93
94         warn $dump_path, ' ', -s $dump_path, " bytes\n";
95         return $dump_path;
96 }
97
98
99 sub __stats {
100
101         my $stats;
102
103         my $nr_items = $#{ $_[0] } + 1;
104
105         warn "__stats $nr_items\n";
106
107         foreach my $e ( @{ $_[0] } ) {
108                 foreach my $n ( keys %$e ) {
109                         $stats->{$n}->{count}++;
110                         my @v;
111                         if ( ref $e->{$n} eq 'ARRAY' ) {
112                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
113                                 @v = @{ $e->{$n} };
114                         } else {
115                                 @v = ( $e->{$n} );
116                         }
117
118                         foreach my $x ( @v ) {
119                                 $stats->{$n}->{numeric}++
120                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
121                                 $stats->{$n}->{empty}++
122                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
123                         }
124
125                 }
126         }
127
128         foreach my $n ( keys %$stats ) {
129                 my $s = $stats->{$n};
130                 next unless defined $s->{array};
131                 if ( $s->{array} == $s->{count} ) {
132                         delete $s->{array};
133                         if ( $s->{count} == $nr_items ) {
134                                 warn "check $n for uniqeness\n";
135                                 my $unique;
136                                 foreach my $e ( @{ $_[0] } ) {
137                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
138                                                 $unique = 0;
139                                                 last;
140                                         }
141                                 }
142                                 if ( $unique ) {
143                                         $stats->{$n}->{unique} = 1;
144                                         #warn "# $n unique ",dump( $unique );
145                                 }
146                         }
147                 }
148         }
149
150         warn "# __stats ",dump($stats);
151
152         return $stats;
153 }
154
155 sub _param_or_session {
156         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
157 }
158
159 sub stats {
160         my $self = shift;
161         my $path = $self->_param_or_session('path');
162         warn "stats $path\n";
163         delete $loaded->{$path}->{stats};
164         $self->redirect_to( '/data/columns' );
165 }
166
167
168 sub _load_path {
169         my ( $self, $path ) = @_;
170
171         return if defined $loaded->{$path}->{generated};
172
173         my $full_path = $self->app->home->rel_file( 'data/' . $path );
174         $self->redirect_to('/data/index') unless -r $full_path;
175
176         my $dump_path = $self->_dump_path( $path );
177
178         if ( defined $loaded->{$path}->{data} ) {
179                 my $mtime = (stat($full_path))[9];
180                 return if $loaded->{$path}->{mtime} == $mtime;
181                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
182         } elsif ( -e $dump_path ) {
183                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
184                 my $info = retrieve $dump_path;
185                 $loaded->{ $path } = $info;
186                 return;
187         }
188
189         my $data;
190         if ( -f $full_path ) {
191                 if ( $full_path =~ m/.csv/i ) {
192                         $data = MojoFacets::Import::CSV->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') || $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         $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         $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                         $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 $max_length = 80;
364
365         my $self = shift;
366         my $path = $self->_param_or_session('path');
367         if ( ! $path ) {
368                 warn "no path in param or session";
369                 return;
370         }
371         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
372         mkpath $dir unless -e $dir;
373         my $name = join('.', map { __unac($_) } @_ );
374         if ( length($name) > $max_length ) {
375                 $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
376         }
377         my $full = $dir . '/' . $name;
378         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
379         return $full;
380 }
381
382 sub columns {
383     my $self = shift;
384
385         if ( $self->param('columns') ) {
386                 my @columns = $self->_param_array('columns');
387                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
388                 $self->redirect_to('/data/items');
389         }
390
391         my $stats = $self->_loaded( 'stats' );
392
393         my @columns;
394         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
395
396         foreach my $c ( @{ $self->session( 'header' ) } ) {
397                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
398         }
399
400     $self->render(
401                 message => 'Select columns to display',
402                 stats => $stats,
403                 columns => \@columns,
404                 checked => $self->_checked( $self->_param_array('columns') ),
405         );
406 }
407
408 sub _param_array {
409     my ($self,$name) = @_;
410
411         my @array = $self->param($name);
412         my $path  = $self->session('path');
413
414         if ( @array ) {
415                 $self->session($name => [ @array ]);
416                 $loaded->{$path}->{$name} = [ @array ];
417         } elsif ( my $session = $self->session($name) ) {
418                 if ( ref $session eq 'ARRAY' ) {
419                         @array = @$session;
420                 } else {
421                         die "$name not array ",dump($session);
422                 }
423         }
424         #warn "# $name ",dump @array;
425         return @array;
426 }
427
428 sub _param_scalar {
429     my ($self,$name,$default) = @_;
430
431         my $scalar = $self->param($name);
432
433         if ( defined $scalar ) {
434                 $self->session($name => $scalar);
435         } else {
436                 $scalar = $self->session($name);
437         }
438
439         if ( ! defined $scalar ) {
440                 $scalar = $default;
441                 die "no default for $name" unless defined $scalar;
442                 $self->session($name => $scalar);
443         }
444
445         warn "# _param_scalar $name ",dump $scalar;
446         return $scalar;
447 }
448
449 sub filter {
450         my $self = shift;
451
452         my $name = $self->param('filter_name') || die "name?";
453         my @vals = $self->param('filter_vals');
454
455         $self->_remove_filter( $name );
456         if ( @vals ) {
457                 $self->_filter_on_data( $name, @vals );
458                 if ( my $permanent = $self->param('_permanent') ) {
459                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
460                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
461                         warn "permanent filter $permanent_path ", -s $permanent_path;
462                 }
463         }
464
465         $self->session( 'offset' => 0 );
466
467         $self->redirect_to('/data/items');
468 }
469
470 sub _filter_on_data {
471         my ( $self, $name, @vals ) = @_;
472
473         my $path = $self->session('path');
474
475         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
476                 warn "filter $name not found in data set";
477                 return;
478         }
479
480         $filters->{$name} = [ @vals ];
481         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
482
483         my $filter_hash;
484         $filter_hash->{$_}++ foreach @vals;
485
486         #warn "# filter_hash ",dump( $filter_hash );
487
488         my $items = $self->_loaded('data')->{items};
489
490         my $include_missing = defined $filter_hash->{_missing};
491         my $filtered_items;
492
493         foreach my $i ( 0 .. $#$items ) {
494
495                 if ( defined $items->[$i]->{$name} ) {
496                         my $row = $items->[$i]->{$name};
497                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
498                         foreach my $v ( @$row ) {
499                                 if ( defined $filter_hash->{ $v } ) {
500                                         $filtered_items->{$i}++;
501                                 }
502                         }
503                 } elsif ( $include_missing ) {
504                         $filtered_items->{$i}++;
505                 }
506         }
507
508         #warn "# filter $name ",dump($filtered_items);
509
510         $loaded->{$path}->{filters}->{$name} = $filtered_items;
511         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
512 }
513
514
515 sub _current_filters {
516         my $self = shift;
517         my $current_filters;
518         my $stats = $self->_loaded('stats');
519
520         $current_filters->{ $_ } = $filters->{ $_ }
521                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
522         ;
523         warn "# _current_filters ",dump( keys %$current_filters );
524         return $current_filters;
525 }
526
527 sub _data_sorted_by {
528         my ( $self, $order ) = @_;
529
530         my $path = $self->session('path');
531
532         warn "_data_sorted_by $order from $path";
533
534         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
535                 return $loaded->{$path}->{sorted}->{$order};
536         }
537
538         my $data = $self->_loaded( 'data' );
539         my $numeric = $self->_is_numeric($order);
540         my $missing = $numeric ? 0 : '';
541         no warnings qw(numeric);
542         my $nr = 0;
543         my @sorted = map {
544                 $_->[0]
545         } sort {
546                 if ( $numeric ) {
547                         $a->[1] <=> $b->[1]
548                 } else {
549                         $a->[1] cmp $b->[1]
550                 }
551         } map {
552                 my $v;
553                 if ( ! exists $_->{$order} ) {
554                         $v = $missing;
555                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
556                         $v = join('', @{$_->{$order}});
557                 } else {
558                         $v = $_->{$order};
559                 }
560                 [ $nr++, $v ]
561         } @{ $data->{items} }
562         ;
563
564         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
565         #warn "# sorted ",dump( @sorted );
566
567         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
568 }
569
570
571 sub __all_filters {
572         my $order = pop @_;
573         join(',', sort(@_), 'order', $order);
574 }
575
576 our ($out, $key,$value);
577
578 sub __commit_path_code {
579         my ( $path, $i, $code, $commit_changed ) = @_;
580
581         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
582         my $row = $items->[$i];
583         my $update;
584         eval $code;
585         foreach ( keys %$update ) {
586                 $$commit_changed->{$_}++;
587                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
588         }
589         #warn "__commit_path_code $path $i ",dump( $update );
590 }
591
592 sub items {
593         my $self = shift;
594
595         if ( my $show = $self->param('id') ) {
596                 $self->param('show', $show);
597                 warn "show $show\n";
598         }
599
600         my $path = $self->session('path');
601
602         my @columns = $self->_param_array('columns');
603         $self->redirect_to('/data/columns') unless @columns;
604         my $order   = $self->_param_scalar('order', $columns[0]);
605         my $sort    = $self->_param_scalar('sort', 'a');
606         my $offset  = $self->_param_scalar('offset', 0);
607         my $limit   = $self->_param_scalar('limit', 20);
608         $self->_param_scalar('show', 'table');
609
610         # fix offset when changing limit
611         $offset = int( $offset / $limit ) * $limit;
612
613         if ( ! grep { /^\Q$order\E$/ } @columns ) {
614                 $order = $columns[0];
615                 $self->session( order => $order );
616         }
617         my $sorted = $self->_data_sorted_by( $order );
618
619         my @filter_names;
620         if ( $filters ) {
621                 foreach my $name ( keys %$filters ) {
622                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
623                                 warn "skip filter $name not found in $path\n";
624                                 next;
625                         }
626                         push @filter_names, $name;
627                 }
628                 warn "filter_names ",dump( @filter_names );
629                 foreach my $name ( @filter_names ) {
630                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
631                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
632                 }
633         }
634
635         my $all_filters = __all_filters( @filter_names,$order );
636
637 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
638
639         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
640
641                 my $path_filters = $loaded->{$path}->{filters};
642
643                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
644
645                 my @filtered;
646                 foreach my $i ( 0 .. $#$sorted ) {
647                         my $pos = $sorted->[$i];
648
649                         if ( $#filter_names == -1 ) {
650                                 push @filtered, $pos;
651                                 next;
652                         }
653
654                         my $skip = 0;
655                         foreach ( @filter_names ) {
656                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
657                         }
658                         next if $skip;
659
660                         push @filtered, $pos;
661                 }
662
663                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
664         }
665
666         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
667                 if defined $loaded->{$path}->{filtered}->{$all_filters};
668
669         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
670
671         my $data = $self->_loaded('data');
672
673         my $code = $self->_param_scalar('code','');
674         $code =~ s{\r}{}gs;
675         $code =~ s{\n+$}{\n}s;
676
677         my $commit = $self->param('commit');
678         my $test = $self->param('test');
679
680         my $commit_changed;
681
682         if ( $code && ( $test || $commit ) ) {
683                 # XXX find columns used in code snippet and show them to user
684                 my $order = 0;
685                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
686                         if ( $column =~ s/^(['"])// ) {
687                                 $column =~ s/$1$//;
688                         }
689                         next if $column =~ m/\$/; # hide columns with vars in them
690                         $commit_changed->{$column} = 0;
691                 }
692         }
693
694         my $code_path = $self->app->home->rel_dir('public') . "/code";
695         if ( $commit ) {
696
697                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
698                 ( $key, $value, $out ) = ( 'key', 'value' );
699                 foreach ( 0 .. $#$filtered ) {
700                         my $i = $filtered->[$_];
701                         __commit_path_code( $path, $i, $code, \$commit_changed );
702                 }
703
704                 $self->_save_change({
705                         path => $path,
706                         time => $self->param('time') || time(),
707                         user => $self->param('user') || $ENV{'LOGNAME'},
708                         code => $code,
709                         commit_changed => $commit_changed,
710                 });
711
712                 if ( my $description = $self->param('code_description') ) {
713                         my $depends = $self->param('code_depends') || die "no code_depends?";
714                         my $path = "$code_path/$depends.$description.pl";
715                         if ( -e $path && ! $self->param('overwrite') ) {
716                                 warn "# code $path not saved\n";
717                         } else {
718                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
719                                 warn "code $path ", -s $path, " bytes saved\n";
720                         }
721                 }
722
723                 # remove console
724                 $code = '';
725                 if ( $out ) {
726                         my $commit_dataset = join('.'
727                                 , $self->param('code_depends')
728                                 , $self->param('code_description')
729                                 , time()
730                         );
731                         $key ||= 'key';
732                         $value ||= 'value';
733                         warn "key $key value $value";
734                         my $items;
735                         foreach my $n ( keys %$out ) {
736                                 my $i = { $key => [ $n ] };
737                                 my $ref = ref $out->{$n};
738                                 if ( $ref eq 'HASH' ) {
739                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
740                                 } elsif ( $ref eq 'ARRAY' ) {
741                                         $i->{$_} = $out->{$n};
742                                 } elsif ( ! $ref ) {
743                                         $i->{$value} = [ $out->{$n} ];
744                                 } else {
745                                         $i->{_error} = [ dump($out->{$n}) ];
746                                 }
747                                 push @$items, $i;
748                         };
749                         undef $out;
750                         my $stats = __stats( $items );
751                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
752                         unshift @columns, $key;
753
754                         $loaded->{$commit_dataset} = {
755                                 header => [ @columns ],
756                                 columns => [ @columns ],
757                                 mtime => time(),
758                                 data => { items => $items },
759                                 stats => $stats,
760                                 generated => 1,
761                         };
762                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
763                         $self->session('path', $commit_dataset);
764                         $self->session('columns', [ @columns ]);
765                         $self->session('order', $key);
766                         $self->redirect_to('/data/items');
767                         return; # FIXME needed to correctly show columns
768                 }
769
770                 # this might move before $out to recalculate stats on source dataset?
771                 __path_rebuild_stats( $path );
772                 my $c = { map { $_ => 1 } @columns };
773                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
774                 warn "# added_columns ",dump( @added_columns );
775                 unshift @columns, @added_columns;
776
777                 $self->session('columns', [ @columns ]);
778                 $loaded->{$path}->{columns} = [ @columns ];
779                 warn "# new columns ",dump( @columns );
780
781                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
782         }
783
784         my $sorted_items;
785         my $from_end = $sort eq 'd' ? $#$filtered : 0;
786         my $test_changed;
787         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
788         foreach ( 0 .. $limit ) {
789                 my $i = $_ + $offset;
790                 last unless defined $filtered->[$i];
791                 $i = $from_end - $i if $from_end;
792                 my $id = $filtered->[$i];
793                 my $row = Storable::dclone $data->{items}->[ $id ];
794                 if ( $code && $test ) {
795                         my $update;
796                         eval $code;
797                         if ( $@ ) {
798                                 warn "ERROR evaling $@", dump($code);
799                                 $self->stash('eval_error', $@) if $@;
800                         } else {
801                                 warn "EVAL ",dump($update);
802                                 foreach ( keys %$update ) {
803                                         $test_changed->{$_}++;
804                                         $row->{$_} = $update->{$_};
805                                 }
806                         }
807                 }
808                 $row->{_row_id} ||= $id;
809                 push @$sorted_items, $row;
810         }
811
812         if ( $self->param('export') ) {
813                 my $export_path = $self->_export_path( 'items', @columns);
814                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
815                 foreach my $f ( 0 .. $#$filtered ) {
816                         print $fh join("\t", map {
817                                 my $i = $data->{items}->[ $filtered->[$f] ];
818                                 my $v = '\N';
819                                 if ( ! defined $i->{$_} ) {
820                                         # nop
821                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
822                                         $v =join(',', @{ $i->{$_} });
823                                         $v = '\N' if length($v) == 0;
824                                 } elsif ( ! ref $i->{$_} ) {
825                                         $v = $i->{$_};
826                                 } else {
827                                         $v = dump $i->{$_};
828                                 }
829                                 $v;
830                         } @columns),"\n";
831                 }
832                 close($fh);
833                 warn "export $export_path ", -s $export_path, " bytes\n";
834         }
835
836         warn "# test_changed ",dump( $test_changed );
837         my $c = { map { $_ => 1 } @columns };
838         my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
839         unshift @columns, @added_columns;
840
841         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
842
843         my $code_depends = $self->param('code_depends')||
844         join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
845         my $code_description = $self->param('code_description') ||
846         join(',', @added_columns);
847
848         $code_depends ||= $code_description; # self-modifing
849         if ( ! $code_depends && $out ) {
850                 $code_depends = $key;
851                 $code_description = $value;
852         }
853
854         warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
855
856         $self->render(
857                 order => $order,
858                 offset => $offset,
859                 limit => $limit,
860                 sorted => $sorted_items,
861                 columns => [ @columns ],
862                 rows => $#$filtered + 1,
863                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
864                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
865                 filters => $self->_current_filters,
866                 code => $code,
867                 cols_changed => $commit ? $commit_changed : $test_changed,
868                 code_depends => $code_depends,
869                 code_description => $code_description,
870                 code_path => $code_path,
871                 out => $out,
872         );
873
874 }
875
876
877 sub order {
878         my $self = shift;
879         $self->session('order', $self->param('order'));
880         $self->session('sort', $self->param('sort'));
881         $self->redirect_to('/data/items');
882 }
883
884 sub _is_numeric {
885         my ( $self, $name ) = @_;
886
887         my $stats = $self->_loaded( 'stats' );
888
889         # sort facet numerically if more >50% elements are numeric
890         my $count = $stats->{$name}->{count};
891         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
892         defined $stats->{$name}->{numeric} &&
893                 $stats->{$name}->{numeric} > $count / 2;
894 }
895
896 sub _is_unique {
897         my ( $self, $name ) = @_;
898         my $stats = $self->_loaded( 'stats' );
899         defined $stats->{$name}->{unique};
900 }
901
902 sub _remove_filter {
903         my ($self,$name) = @_;
904         warn "_remove_filter $name\n";
905
906         my $path = $self->session('path');
907
908         delete $filters->{$name};
909         delete $loaded->{$path}->{filters}->{$name};
910         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
911
912         foreach (
913                         grep { /\Q$name\E/ }
914                         keys %{ $loaded->{$path}->{filtered} }
915         ) {
916                 delete $loaded->{$path}->{filtered}->{$_};
917                 warn "remove filtered cache $_";
918         }
919 }
920
921 sub facet {
922         my $self = shift;
923
924         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
925
926         if ( my $name = $self->param('remove') ) {
927                 $self->_remove_filter( $name );
928                 $self->redirect_to( '/data/items' );
929         }
930
931         my $facet;
932         my $name = $self->param('name') || die "no name";
933
934         my $all = $self->_param_scalar('all', 1);
935         my $data = $self->_loaded('data');
936
937         my $filters = $self->_current_filters;
938         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
939         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
940                 if defined $loaded->{$path}->{filtered}->{$all_filters};
941
942         if ( ! $filtered || $all ) {
943                 $filtered = [ 0 .. $#{ $data->{items} } ];
944                 warn "filter all values\n";
945         } else {
946                 warn "filter using $all_filters\n";
947         }
948
949         foreach my $i ( @$filtered ) {
950                 my $item = $data->{items}->[$i];
951                 if ( ! exists $item->{$name} ) {
952                         $facet->{ _missing }++;
953                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
954                         $facet->{$_}++ foreach @{ $item->{$name} };
955                 } else {
956                         $facet->{ $item->{$name} }++;
957                 }
958         }
959
960         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
961
962         if ( my $code = $self->param('code') ) {
963                 my $out;
964                 foreach my $value ( keys %$facet ) {
965                         my $count = $facet->{$value};
966                         my $checked = $checked_values->{$value};
967                         eval $code;
968                         if ( $@ ) {
969                                 $out = $@;
970                                 warn "ERROR: $@\n$code\n";
971                                 last;
972                         } elsif ( $checked != $checked_values->{$value} ) {
973                                 warn "checked $value $count -> $checked\n";
974                                 $checked_values->{$value} = $checked;
975                         }
976                 }
977                 warn "out ",dump( $out );
978                 $self->stash( out => $out );
979         }
980
981 #       warn "# facet $name ",dump $facet;
982
983         my @facet_names =
984                   $all                      ? keys %$facet
985                 : defined $filters->{$name} ? @{ $filters->{$name} }
986                 : keys %$facet;
987
988         my $numeric = $self->_is_numeric($name);
989
990         my $sort = $self->param('sort');
991         # sort numeric facets with more than 5 values ascending
992         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
993
994         @facet_names = sort {
995                 my $result;
996                 if ( $sort eq 'a' ) {
997                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
998                 } elsif ( $sort eq 'd' ) {
999                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1000                 } elsif ( $sort eq 'c' ) {
1001                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1002                 } else {
1003                         warn "unknown sort: $sort";
1004                         $result = $a cmp $b;
1005                 }
1006                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1007                 $result;
1008         } @facet_names;
1009
1010         $self->render( name => $name, facet => $facet, checked => $checked_values,
1011                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1012         );
1013 }
1014
1015
1016 sub __invalidate_path_column {
1017         my ( $path, $name ) = @_;
1018
1019         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1020                 delete $loaded->{$path}->{sorted}->{$name};
1021                 warn "# invalidate $path sorted $name\n";
1022         }
1023
1024         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1025                 delete $loaded->{$path}->{filtered}->{$_};
1026                 warn "# invalidate $path filtered $_\n";
1027         }
1028 }
1029
1030 sub __path_modified {
1031         my ( $path, $value ) = @_;
1032         $value = 1 unless defined $value;
1033         
1034         $loaded->{$path}->{modified}  = $value;
1035
1036         warn "# __path_modified $path $value\n";
1037 }
1038
1039 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1040
1041 sub _save_change {
1042         my ($self,$change) = @_;
1043
1044         my $change_path = $self->_permanent_path( 'changes' );
1045         mkdir $change_path unless -d $change_path;
1046         $change_path .= '/' . $change->{time};
1047         store $change, $change_path;
1048         utime $change->{time}, $change->{time}, $change_path;
1049         warn "_save_change $change_path ", dump($change);
1050 }
1051
1052 sub edit {
1053         my $self = shift;
1054         my $new_content = $self->param('new_content');
1055         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1056
1057         my $i = $self->param('_row_id');
1058         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1059         my $path = $self->param('path') || die "no path";
1060         my $name = $self->param('name') || die "no name";
1061         my $status = 200; # 200 = OK, 201 = Created
1062
1063         my $data = $self->_loaded('data');
1064
1065         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1066                 $new_content =~ s/^\s+//s;
1067                 $new_content =~ s/\s+$//s;
1068                 my $v;
1069                 if ( $new_content =~ /\xB6/ ) { # para
1070                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1071                 } else {
1072                         $v = [ $new_content ];
1073                 }
1074
1075                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1076                 my $new = dump $v;
1077                 if ( $old ne $new
1078                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1079                 ) {
1080                         $self->_save_change({
1081                                 path => $path,
1082                                 column => $name,
1083                                 pos => $i,
1084                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1085                                 new => $v,
1086                                 time => $self->param('time') || time(),
1087                                 user => $self->param('user') || $ENV{'LOGNAME'},
1088                                 unique => {
1089                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1090                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1091                                         keys %{ $loaded->{$path}->{stats} }
1092                                 },
1093                         });
1094
1095                         warn "# change $path $i $old -> $new\n";
1096                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1097
1098                         __invalidate_path_column( $path, $name );
1099
1100                         $status = 201; # created
1101                         __path_rebuild_stats( $path );
1102         
1103                         $new_content = join("\xB6",@$v);
1104
1105                 } else {
1106                         warn "# unchanged $path $i $old\n";
1107                         $status = 304;
1108                 }
1109         } else {
1110                 $new_content = "$path $i $name doesn't exist\n";
1111                 $status = 404;
1112         }
1113
1114         warn "# edit $status ", dump $new_content;
1115
1116         $self->render(
1117                 status => $status,
1118                 new_content => scalar $new_content,
1119         );
1120 }
1121
1122
1123 sub save {
1124         my $self = shift;
1125         my $path = $self->_param_or_session('path');
1126         my $dump_path = $self->_save( $path );
1127         __path_modified( $path, 0 );
1128
1129         $self->redirect_to( '/data/items' );
1130 }
1131
1132 sub export {
1133         my $self = shift;
1134
1135         my $dir = $self->app->home->rel_dir('public');
1136
1137         if ( my $import = $self->param('import') ) {
1138
1139                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1140                         my $name = $self->_column_from_unac( $1 );
1141
1142                         my @vals = map { chomp; $_ }
1143                                 read_file "$dir/export/$import", binmode => ':utf8';
1144
1145                         $self->_remove_filter( $name );
1146                         $self->_filter_on_data( $name, @vals );
1147                         $self->session( 'offset' => 0 );
1148                         $self->redirect_to('/data/items');
1149                 } else {
1150                         warn "UNKNOWN IMPORT $import";
1151                 }
1152         }
1153
1154         if ( my $remove = $self->param('remove') ) {
1155                 my $path = "$dir/export/$remove";
1156                 unlink $path if -e $path;
1157                 $path .= '.png';
1158                 unlink $path if -e $path;
1159         }
1160
1161         my $path = $self->_export_path || $self->redirect_to('/data/index');
1162
1163         my @files = grep { ! /\.png$/ } glob "$path/*";
1164         my $mtime = { map { $_ => (stat($_))[9] } @files };
1165         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1166         $self->render( export => [ @files ] );
1167 }
1168
1169 sub __loaded_paths {
1170         return
1171                 grep { defined $loaded->{$_}->{data} }
1172                 keys %$loaded;
1173 }
1174
1175 1;