support loading of /tmp/mojofacets.*.storable
[MojoFacets.git] / lib / MojoFacets / Data.pm
1 package MojoFacets::Data;
2
3 use strict;
4 use warnings;
5
6 use base 'Mojolicious::Controller';
7
8 use Data::Dump qw(dump);
9 use File::Slurp;
10 use Encode;
11 use locale;
12 use File::Find;
13 use Storable;
14 use Time::HiRes qw(time);
15 use File::Path qw(mkpath);
16 use Text::Unaccent::PurePerl;
17 use Digest::MD5;
18
19 use MojoFacets::Import::File;
20 use MojoFacets::Import::HTMLTable;
21 use MojoFacets::Import::CSV;
22 use MojoFacets::Import::CouchDB;
23 use MojoFacets::Import::SQL;
24
25 our $loaded;
26 our $filters;
27
28 sub index {
29         my $self = shift;
30
31         my $data_dir = $self->app->home->rel_dir('data');
32         die "no data dir $data_dir" unless -d $data_dir;
33
34         my @files;
35         my $changes;
36         find( sub {
37                 my $file = $File::Find::name;
38                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
39                         $file =~ s/$data_dir\/*//;
40                         push @files, $file;
41                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
42                         push @{ $changes->{$1} }, $2
43                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
44                         $file =~ s/$data_dir\/*//;
45                         push @files, $file;
46                 } elsif ( -f $file && $file =~ m/\.(csv|storable|couchdb|sql)$/i ) {
47                         $file =~ s/$data_dir\/*//;
48                         push @files, $file;
49                 } else {
50                         warn "IGNORE: $file\n";
51                 }
52         }, $data_dir);
53
54         no warnings qw(uninitialized); # mtime
55         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
56                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
57         my $size;
58         $size->{$_} = -s "$data_dir/$_" foreach @files;
59
60         $self->render(
61                 files => [ @files ],
62                 size => $size,
63                 loaded => $loaded,
64                 filters => $filters,
65                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
66                 changes => $changes,
67         );
68 }
69
70 sub _dump_path {
71         my ( $self, $name ) = @_;
72         my $dir = $self->app->home->rel_dir('data');
73         $name =~ s/^$dir//;
74         $name =~ s/\/+/_/g;
75         return '/tmp/mojo_facets.' . $name . '.storable';
76 }
77
78 sub _save {
79         my ( $self, $path ) = @_;
80
81         my $dump_path = $self->_dump_path( $path );
82         my $first_load = ! -e $dump_path;
83         warn "save loaded to $dump_path";
84         my $info = $loaded->{$path};
85         store $info, $dump_path;
86
87         if ( $first_load ) {
88                 my $mtime = $loaded->{$path}->{mtime};
89                 utime $mtime, $mtime, $dump_path;
90                 warn "sync time to $path at $mtime\n";
91         }
92
93         warn $dump_path, ' ', -s $dump_path, " bytes\n";
94         return $dump_path;
95 }
96
97
98 sub __stats {
99
100         my $stats;
101
102         my $nr_items = $#{ $_[0] } + 1;
103
104         warn "__stats $nr_items\n";
105
106 #       foreach my $e ( @{ $_[0] } ) {
107         foreach my $i ( 0 .. $#{$_[0]} ) {
108                 print STDERR " $i" if $i % 5000;
109                 my $e = $_[0]->[$i];
110                 foreach my $n ( keys %$e ) {
111                         $stats->{$n}->{count}++;
112                         my @v;
113                         if ( ref $e->{$n} eq 'ARRAY' ) {
114                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
115                                 @v = @{ $e->{$n} };
116                         } else {
117                                 @v = ( $e->{$n} );
118                         }
119
120                         foreach my $x ( @v ) {
121                                 if ( ! defined $x ) { # FIXME really null
122                                         $stats->{$n}->{empty}++;
123                                         next;
124                                 }
125
126                                 $stats->{$n}->{numeric}++
127                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
128                                 $stats->{$n}->{empty}++
129                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
130                         }
131
132                 }
133         }
134
135         foreach my $n ( keys %$stats ) {
136                 my $s = $stats->{$n};
137                 next unless defined $s->{array};
138                 if ( $s->{array} == $s->{count} ) {
139                         delete $s->{array};
140                         if ( $s->{count} == $nr_items ) {
141                                 warn "check $n for uniqeness\n";
142                                 my $unique;
143                                 foreach my $e ( @{ $_[0] } ) {
144                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
145                                                 $unique = 0;
146                                                 last;
147                                         }
148                                 }
149                                 if ( $unique ) {
150                                         $stats->{$n}->{unique} = 1;
151                                         #warn "# $n unique ",dump( $unique );
152                                 }
153                         }
154                 }
155         }
156
157         warn "# __stats ",dump($stats);
158
159         return $stats;
160 }
161
162 sub _param_or_session {
163         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
164 }
165
166 sub stats {
167         my $self = shift;
168         my $path = $self->_param_or_session('path');
169         warn "stats $path\n";
170         delete $loaded->{$path}->{stats};
171         return $self->redirect_to( '/data/columns' );
172 }
173
174
175 sub _load_path {
176         my ( $self, $path ) = @_;
177
178         return if defined $loaded->{$path}->{generated};
179
180         my $full_path = $self->app->home->rel_file( 'data/' . $path );
181         return $self->redirect_to('/data/index') unless -r $full_path;
182
183         my $dump_path = $self->_dump_path( $path );
184
185         if ( defined $loaded->{$path}->{data} ) {
186                 my $mtime = (stat($full_path))[9];
187                 return if $loaded->{$path}->{mtime} == $mtime;
188                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
189         } elsif ( -e $dump_path ) {
190                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
191                 my $info = retrieve $dump_path;
192                 $loaded->{ $path } = $info;
193                 return;
194         }
195
196         my $data;
197         if ( -f $full_path ) {
198                 if ( $full_path =~ m/.storable$/ ) { # check storable first to catch files copied from /tmp/
199                         warn "open $full_path ", -s $full_path, " bytes";
200                         open(my $pipe, "<", $full_path) || die $!;
201                         while ( my $o = eval { Storable::fd_retrieve $pipe } ) {
202                                 if ( exists $o->{item} ) {
203                                         # stream of storable objects
204                                         push @{ $data->{items} }, $o->{item};
205                                 } elsif ( exists $o->{data}->{items} ) {
206                                         # /tmp/mojofacets.*.storable
207                                         $data = $o->{data};
208                                         $data->{header} = $o->{header};
209                                         last;
210                                 } else {
211                                         warn "SKIP ",dump($o);
212                                 }
213                         }
214                         close($pipe);
215                         warn "loaded ", $#{ $data->{items} } + 1, " items from $full_path\n";
216                         $data->{generated}++;
217                 } elsif ( $full_path =~ m/.csv/i ) {
218                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
219                 } elsif ( $full_path =~ m/.sql/i ) {
220                         $data = MojoFacets::Import::SQL->new( full_path => $full_path )->data;
221                 } elsif ( $full_path =~ m/.couchdb/i ) {
222                         $data = MojoFacets::Import::CouchDB->new( full_path => $full_path )->data;
223                 } else {
224                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
225                 }
226         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
227                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
228         } else {
229                 die "can't load $full_path";
230         }
231
232         my @header;
233
234         if ( defined $data->{header} ) {
235                 if ( ref $data->{header} eq 'ARRAY' ) {
236                         @header = @{ $data->{header} };
237                 } else {
238                         warn "header not array ", dump( $data->{header} );
239                 }
240         }
241
242         my $stats = __stats( $data->{items} );
243
244         @header =
245                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
246                 grep { defined $stats->{$_}->{count} } keys %$stats
247                 unless @header;
248
249         my $info = {
250                 header => [ @header ],
251                 stats  => $stats,
252                 full_path => $full_path,
253                 size => -s $full_path,
254                 mtime => (stat($full_path))[9],
255                 data => $data,
256                 defined $data->{generated} ? ( generated => 1 ) : (),
257         };
258
259         $loaded->{ $path } = $info;
260         $self->_save( $path ) unless $info->{generated};
261
262 }
263
264
265 sub load {
266         my $self = shift;
267
268         my $path = $self->param('path') || return $self->redirect_to( '/data/index' );
269
270         my @paths = $self->param('paths');
271         warn "# paths ", dump @paths;
272
273         foreach my $p ( keys %$loaded ) {
274                 next if grep { /^\Q$p\E$/ } @paths;
275                 warn "remove $p from memory\n";
276                 delete $loaded->{$p};
277         }
278
279         $self->_load_path( $_ ) foreach @paths;
280
281         warn "# path $path\n";
282         $self->_load_path( $path );
283
284         $self->session( 'path' => $path );
285
286         my $redirect_to = '/data/items';
287
288         $self->session( 'header' => $loaded->{$path}->{header} );
289         if ( ! defined $loaded->{$path}->{columns} ) {
290                 my $columns_path = $self->_permanent_path( 'columns' );
291                 if ( -e $columns_path ) {
292                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
293                         $loaded->{$path}->{columns} = [ @columns ];
294                         warn "# columns_path $columns_path ",dump(@columns);
295                 } else {
296                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
297                 }
298
299                 $redirect_to = '/data/columns';
300         }
301         $self->session( 'columns' => $loaded->{$path}->{columns} );
302         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
303         return $self->redirect_to( $redirect_to );
304 }
305
306
307 sub _loaded {
308         my ( $self, $name ) = @_;
309         my $path = $self->session('path') || $self->param('path');
310         return $self->redirect_to('/data/index') unless $path;
311
312         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
313                 my $caller = (caller(1))[3];
314                 if ( $caller =~ m/::edit/ ) {
315                         warn "rebuild stats for $path ignored caller $caller\n";
316                 } else {
317                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
318 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
319                         $loaded->{$path}->{rebuild_stats} = 1;
320                         $loaded->{$path}->{modified} = 1;
321                 }
322         }
323
324         if ( defined $loaded->{$path}->{rebuild_stats} ) {
325                 warn "rebuild_stats $path";
326                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
327                 delete $loaded->{$path}->{rebuild_stats};
328         }
329
330         if ( ! defined $loaded->{$path}->{$name} ) {
331                 warn "$path $name isn't loaded\n";
332                 $self->_load_path( $path );
333                 if ( ! defined $loaded->{$path}->{stats} ) {
334                         warn "rebuild stats for $path\n";
335                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
336                 }
337                 if ( ! defined $loaded->{$path}->{$name} ) {
338                         warn "MISSING $name for $path\n";
339                         return $self->redirect_to('/data/index')
340                 }
341         }
342
343         $self->session( 'modified' => $loaded->{$path}->{modified} );
344
345         return $loaded->{$path}->{$name};
346 }
347
348
349 sub _checked {
350         my $self = shift;
351         my $checked;
352         $checked->{$_}++ foreach @_;
353 #       warn "# _checked ",dump($checked);
354         return $checked;
355 }
356
357 sub _permanent_path {
358         my $self = shift;
359         my $path = $self->_param_or_session('path');
360         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
361 }
362
363 sub __unac {
364         my $n = shift;
365         $n = unac_string($n);
366         $n =~ s/\W+/_/g;
367         return $n;
368 }
369
370 sub _column_from_unac {
371         my ($self,$name) = @_;
372
373         my $stats = $self->_loaded('stats');
374         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
375
376         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
377 }
378
379 sub _export_path {
380         my $self = shift;
381         my $path = $self->_param_or_session('path');
382         if ( ! $path ) {
383                 warn "no path in param or session";
384                 return;
385         }
386         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
387         mkpath $dir unless -e $dir;
388         my $name = __export_path_name( $path, @_ );
389         my $full = $dir . '/' . $name;
390         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
391         return $full;
392 }
393
394 sub __export_path_name {
395         my $max_length = 80;
396
397         my $path = shift;
398         my $name = join('.', map { __unac($_) } @_ );
399         if ( length($name) > $max_length ) {
400                 $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
401         }
402         return $name;
403 }
404
405 sub columns {
406     my $self = shift;
407
408         my $view_path = $self->_permanent_path( 'views' );
409
410         if ( $self->param('columns') ) {
411                 my @columns = $self->_param_array('columns');
412                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
413                 if ( my $view = $self->param('view') ) {
414                         mkdir $view_path unless -e $view_path;
415                         write_file( "$view_path/$view", { binmode => ':utf8' }, map { "$_\n" } @columns );
416                         warn "view $view_path/$view ", -s "$view_path/$view", " bytes\n";
417                 }
418
419                 return $self->redirect_to('/data/items');
420
421         } elsif ( ! $self->session('header') ) {
422                 return $self->redirect_to('/data/load');
423         }
424
425         if ( my $id = $self->param('id') ) {
426                 my $view_full = "$view_path/$id";
427                 if ( -f $view_full ) {
428                         my @columns = map { chomp; $_ } read_file $view_full, binmode => ':utf8';
429                         warn "view $view_full loaded ", dump @columns;
430                         $self->session( 'columns' => [ @columns ] );
431                         return $self->redirect_to('/data/items');
432                 }
433         }
434
435         my $stats = $self->_loaded( 'stats' );
436
437         my @columns;
438         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
439
440         foreach my $c ( @{ $self->session( 'header' ) } ) {
441                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
442         }
443
444         my @views;
445         if ( -d $view_path ) {
446                 @views = map { s{^\Q$view_path\E/*}{}; $_ } glob "$view_path/*";
447                 warn "# views ",dump @views;
448         }
449
450     $self->render(
451                 message => 'Select columns to display',
452                 stats => $stats,
453                 columns => \@columns,
454                 checked => $self->_checked( $self->_param_array('columns') ),
455                 views => \@views,
456         );
457 }
458
459 sub _param_array {
460     my ($self,$name) = @_;
461
462         my @array = $self->param($name);
463         my $path  = $self->session('path');
464
465         if ( @array ) {
466                 $self->session($name => [ @array ]);
467                 $loaded->{$path}->{$name} = [ @array ];
468         } elsif ( my $session = $self->session($name) ) {
469                 if ( ref $session eq 'ARRAY' ) {
470                         @array = @$session;
471                 } else {
472                         die "$name not array ",dump($session);
473                 }
474         }
475         #warn "# $name ",dump @array;
476         return @array;
477 }
478
479 sub _param_scalar {
480     my ($self,$name,$default) = @_;
481
482         my $scalar = $self->param($name);
483
484         if ( defined $scalar ) {
485                 $self->session($name => $scalar);
486         } else {
487                 $scalar = $self->session($name);
488         }
489
490         if ( ! defined $scalar ) {
491                 $scalar = $default;
492                 die "no default for $name" unless defined $scalar;
493                 $self->session($name => $scalar);
494         }
495
496         warn "# _param_scalar $name ",dump $scalar;
497         return $scalar;
498 }
499
500 sub filter {
501         my $self = shift;
502
503         my $name = $self->param('filter_name') || die "name?";
504         my @vals = $self->param('filter_vals');
505
506         $self->_remove_filter( $name );
507         if ( @vals ) {
508                 $self->_filter_on_data( $name, @vals );
509                 if ( my $permanent = $self->param('_permanent') ) {
510                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
511                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
512                         warn "permanent filter $permanent_path ", -s $permanent_path;
513                 }
514         }
515
516         $self->session( 'offset' => 0 );
517
518         return $self->redirect_to('/data/items');
519 }
520
521 sub _filter_on_data {
522         my ( $self, $name, @vals ) = @_;
523
524         my $path = $self->session('path');
525
526         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
527                 warn "filter $name not found in data set";
528                 return;
529         }
530
531         $filters->{$name} = [ @vals ];
532         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
533
534         my $filter_hash;
535         $filter_hash->{$_}++ foreach @vals;
536
537         #warn "# filter_hash ",dump( $filter_hash );
538
539         my $items = $self->_loaded('data')->{items};
540
541         my $include_missing = defined $filter_hash->{_missing};
542         my $filtered_items;
543
544         foreach my $i ( 0 .. $#$items ) {
545
546                 if ( defined $items->[$i]->{$name} ) {
547                         my $row = $items->[$i]->{$name};
548                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
549                         foreach my $v ( @$row ) {
550                                 if ( defined $filter_hash->{ $v } ) {
551                                         $filtered_items->{$i}++;
552                                 }
553                         }
554                 } elsif ( $include_missing ) {
555                         $filtered_items->{$i}++;
556                 }
557         }
558
559         #warn "# filter $name ",dump($filtered_items);
560
561         # invalidate filters on other datasets
562         foreach my $dataset ( grep { exists $loaded->{$_}->{filters}->{$name} } keys %$loaded ) {
563                 delete $loaded->{$dataset}->{filters}->{$name};
564                 delete $loaded->{$dataset}->{filtered};
565         }
566
567         $loaded->{$path}->{filters}->{$name} = $filtered_items;
568         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
569 }
570
571
572 sub _current_filters {
573         my $self = shift;
574         my $current_filters;
575         my $stats = $self->_loaded('stats');
576
577         $current_filters->{ $_ } = $filters->{ $_ }
578                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
579         ;
580         warn "# _current_filters ",dump( keys %$current_filters );
581         return $current_filters;
582 }
583
584 sub _data_sorted_by {
585         my ( $self, $order ) = @_;
586
587         my $path = $self->session('path');
588
589         warn "_data_sorted_by $order from $path";
590
591         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
592                 return $loaded->{$path}->{sorted}->{$order};
593         }
594
595         my $data = $self->_loaded( 'data' );
596         my $numeric = $self->_is_numeric($order);
597         my $missing = $numeric ? 0 : '';
598         no warnings qw(numeric);
599         my $nr = 0;
600         my @sorted = map {
601                 $_->[0]
602         } sort {
603                 if ( $numeric ) {
604                         $a->[1] <=> $b->[1]
605                 } else {
606                         $a->[1] cmp $b->[1]
607                 }
608         } map {
609                 my $v;
610                 if ( ! exists $_->{$order} ) {
611                         $v = $missing;
612                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
613                         $v = join('', @{$_->{$order}});
614                 } else {
615                         $v = $_->{$order};
616                 }
617                 [ $nr++, $v ]
618         } @{ $data->{items} }
619         ;
620
621         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
622         #warn "# sorted ",dump( @sorted );
623
624         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
625 }
626
627
628 sub __all_filters {
629         my $order = pop @_;
630         join(',', sort(@_), 'order', $order);
631 }
632
633 our ($out, $key,$value);
634
635 our $lookup_path_col;
636 our $on;
637
638 sub __commit_begin {
639         warn "__commit_begin";
640         $lookup_path_col = undef;
641         $on = undef;
642 }
643
644 sub __commit_end {
645         warn "__commit_end";
646         $lookup_path_col = undef; # cleanup memory
647         $on = undef;
648 }
649
650 sub lookup {
651         warn "# lookup ",dump @_;
652         my ( $vals, $on_path, $on_col, $code ) = @_;
653         die "code is not sub{ ... } but ", dump $code unless ref $code eq 'CODE';
654         my $items = $loaded->{$on_path}->{data}->{items} || die "no items for $on_path";
655
656         if ( ! exists $lookup_path_col->{$on_path}->{$on_col} ) {
657                 warn "create lookup_path_col $on_path $on_col";
658                 foreach my $i ( 0 .. $#$items ) {
659                         my $item = $items->[$i];
660                         if ( exists $item->{$on_col} ) {
661                                 foreach my $v ( @{ $item->{$on_col} } ) {
662                                         push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
663                                 }
664                         }
665                 }
666                 warn dump $lookup_path_col->{$on_path}->{$on_col};
667         }
668
669         foreach my $v ( @$vals ) {
670                 foreach my $i ( @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} } ) {
671                         $on = $items->[$i];
672                         warn "# lookup code $v $i ",dump $on;
673                         $code->();
674                 }
675         }
676 }
677
678 sub __commit_path_code {
679         my ( $path, $i, $code, $commit_changed ) = @_;
680
681         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
682         my $row = $items->[$i];
683         my $update;
684         eval $code;
685         foreach ( keys %$update ) {
686                 $$commit_changed->{$_}++;
687                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
688         }
689         #warn "__commit_path_code $path $i ",dump( $update );
690 }
691
692 # uses templates/admin.html.ep
693 sub _switch_dataset {
694         my $self = shift;
695
696         my $datasets;
697
698         foreach my $path ( keys %$loaded ) {
699                 next unless exists $loaded->{$path}->{data};
700                 push @$datasets, $path;
701         }
702
703         warn "# datasets ",dump($datasets);
704
705         $self->stash( 'datasets' => $datasets );
706 }
707
708 sub items {
709         my $self = shift;
710
711         $self->_switch_dataset;
712
713         if ( my $show = $self->param('id') ) {
714                 $self->param('show', $show);
715                 warn "show $show\n";
716         }
717
718         my $path = $self->_param_scalar('path');
719
720         my @columns = $self->_param_array('columns');
721         return $self->redirect_to('/data/columns') unless @columns;
722         my $order   = $self->_param_scalar('order', $columns[0]);
723         my $sort    = $self->_param_scalar('sort', 'a');
724         my $offset  = $self->_param_scalar('offset', 0);
725         my $limit   = $self->_param_scalar('limit', 20);
726         $self->_param_scalar('show', 'table');
727
728         # fix offset when changing limit
729         $offset = int( $offset / $limit ) * $limit;
730
731         if ( ! grep { /^\Q$order\E$/ } @columns ) {
732                 $order = $columns[0];
733                 $self->session( order => $order );
734         }
735         my $sorted = $self->_data_sorted_by( $order );
736
737         my @filter_names;
738         if ( $filters ) {
739                 foreach my $name ( keys %$filters ) {
740                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
741                                 warn "skip filter $name not found in $path\n";
742                                 next;
743                         }
744                         push @filter_names, $name;
745                 }
746                 warn "filter_names ",dump( @filter_names );
747                 foreach my $name ( @filter_names ) {
748                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
749                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
750                 }
751         }
752
753         my $all_filters = __all_filters( @filter_names,$order );
754
755 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
756
757         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
758
759                 my $path_filters = $loaded->{$path}->{filters};
760
761                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
762
763                 my @filtered;
764                 foreach my $i ( 0 .. $#$sorted ) {
765                         my $pos = $sorted->[$i];
766
767                         if ( $#filter_names == -1 ) {
768                                 push @filtered, $pos;
769                                 next;
770                         }
771
772                         my $skip = 0;
773                         foreach ( @filter_names ) {
774                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
775                         }
776                         next if $skip;
777
778                         push @filtered, $pos;
779                 }
780
781                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
782         }
783
784         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
785                 if defined $loaded->{$path}->{filtered}->{$all_filters};
786
787         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
788
789         my $data = $self->_loaded('data');
790
791         my $code = $self->_param_scalar('code','');
792         $code =~ s{\r}{}gs;
793         $code =~ s{\n+$}{\n}s;
794
795         my $commit = $self->param('commit');
796         my $test = $self->param('test');
797
798         my $commit_changed;
799         __commit_begin;
800
801         if ( $code && ( $test || $commit ) ) {
802                 # XXX find columns used in code snippet and show them to user
803                 my $order = 0;
804                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
805                         if ( $column =~ s/^(['"])// ) {
806                                 $column =~ s/$1$//;
807                         }
808                         next if $column =~ m/\$/; # hide columns with vars in them
809                         $commit_changed->{$column} = 0;
810                 }
811         }
812
813         my $code_path = $self->app->home->rel_dir('public') . "/code";
814         if ( $commit ) {
815
816                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
817                 ( $key, $value, $out ) = ( 'key', 'value' );
818                 foreach ( 0 .. $#$filtered ) {
819                         my $i = $filtered->[$_];
820                         __commit_path_code( $path, $i, $code, \$commit_changed );
821                 }
822
823                 # this might move before $out to recalculate stats on source dataset?
824                 __path_rebuild_stats( $path );
825                 my $c = { map { $_ => 1 } @columns };
826                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
827                 warn "# added_columns ",dump( @added_columns );
828                 unshift @columns, @added_columns;
829
830                 $loaded->{$path}->{columns} = [ @columns ];
831                 warn "# new columns ",dump( @columns );
832
833                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
834
835                 $self->_save_change({
836                         path => $path,
837                         time => $self->param('time') || time(),
838                         user => $self->param('user') || $ENV{'LOGNAME'},
839                         code => $code,
840                         commit_changed => $commit_changed,
841                 });
842
843                 if ( my $description = $self->param('code_description') ) {
844                         my $depends = $self->param('code_depends') || die "no code_depends?";
845                         my $path = "$code_path/$depends.$description.pl";
846                         if ( -e $path && ! $self->param('overwrite') ) {
847                                 warn "# code $path not saved\n";
848                         } else {
849                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
850                                 warn "code $path ", -s $path, " bytes saved\n";
851                         }
852                 }
853
854                 # remove console
855                 $code = '';
856                 if ( $out ) {
857                         my $commit_dataset = join('.'
858                                 , $self->param('code_depends')
859                                 , $self->param('code_description')
860                                 , time()
861                         );
862                         $key ||= 'key';
863                         $value ||= 'value';
864                         warn "key $key value $value";
865                         my $items;
866                         foreach my $n ( keys %$out ) {
867                                 my $i = { $key => [ $n ] };
868                                 my $ref = ref $out->{$n};
869                                 if ( $ref eq 'HASH' ) {
870                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
871                                 } elsif ( $ref eq 'ARRAY' ) {
872                                         $i->{$_} = $out->{$n};
873                                 } elsif ( ! $ref ) {
874                                         $i->{$value} = [ $out->{$n} ];
875                                 } else {
876                                         $i->{_error} = [ dump($out->{$n}) ];
877                                 }
878                                 push @$items, $i;
879                         };
880                         undef $out;
881                         my $stats = __stats( $items );
882                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
883                         unshift @columns, $key;
884
885                         $loaded->{$commit_dataset} = {
886                                 header => [ @columns ],
887                                 columns => [ @columns ],
888                                 mtime => time(),
889                                 data => { items => $items },
890                                 stats => $stats,
891                                 generated => 1,
892                         };
893                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
894                         $self->session('path', $commit_dataset);
895                         $self->session('columns', [ @columns ]);
896                         $self->session('order', $key);
897                         return $self->redirect_to('/data/items');
898                 }
899
900                 $self->session('columns', [ @columns ]);
901         }
902
903         my $sorted_items;
904         my $from_end = $sort eq 'd' ? $#$filtered : 0;
905         my $test_changed;
906         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
907         foreach ( 0 .. $limit ) {
908                 my $i = $_ + $offset;
909                 last unless defined $filtered->[$i];
910                 $i = $from_end - $i if $from_end;
911                 my $id = $filtered->[$i];
912                 my $row = Storable::dclone $data->{items}->[ $id ];
913                 if ( $code && $test ) {
914                         my $update;
915                         eval $code;
916                         if ( $@ ) {
917                                 warn "ERROR evaling $@", dump($code);
918                                 $self->stash('eval_error', $@) if $@;
919                         } else {
920                                 warn "EVAL ",dump($update);
921                                 foreach ( keys %$update ) {
922                                         $test_changed->{$_}++;
923                                         $row->{$_} = $update->{$_};
924                                 }
925                         }
926                 }
927                 $row->{_row_id} ||= $id;
928                 push @$sorted_items, $row;
929         }
930
931         if ( $self->param('export') ) {
932                 my $export_path = $self->_export_path( 'items', @columns);
933                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
934                 print $fh "#",join("\t",@columns),"\n";
935                 foreach my $f ( 0 .. $#$filtered ) {
936                         print $fh join("\t", map {
937                                 my $i = $data->{items}->[ $filtered->[$f] ];
938                                 my $v = '\N';
939                                 if ( ! defined $i->{$_} ) {
940                                         # nop
941                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
942                                         $v =join(',', @{ $i->{$_} });
943                                         $v = '\N' if length($v) == 0;
944                                 } elsif ( ! ref $i->{$_} ) {
945                                         $v = $i->{$_};
946                                 } else {
947                                         $v = dump $i->{$_};
948                                 }
949                                 $v;
950                         } @columns),"\n";
951                 }
952                 close($fh);
953                 warn "export $export_path ", -s $export_path, " bytes\n";
954         }
955
956         my ( $code_depends, $code_description );
957
958         if ( $test ) {
959
960                 warn "# test_changed ",dump( $test_changed );
961                 my $c = { map { $_ => 1 } @columns };
962                 my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
963                 unshift @columns, @added_columns;
964
965                 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
966
967                 my $depends_on;
968                 my $tmp = $code; $tmp =~ s/\$row->{(['"]?)([\w\s]+)\1/$depends_on->{$2}++/gse;
969                 warn "# depends_on ",dump $depends_on;
970
971                 my $test_added = Storable::dclone $test_changed;
972                 delete $test_added->{$_} foreach keys %$depends_on;
973
974                 $code_depends = $self->param('code_depends')
975                 || join(',', keys %$depends_on);
976
977                 $code_description = $self->param('code_description') ||
978                 join(',', keys %$test_added);
979
980                 $code_depends ||= $code_description; # self-modifing
981                 if ( ! $code_depends && $out ) {
982                         $code_depends = $key;
983                         $code_description = $value;
984                 }
985
986                 warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
987
988         } # test?
989
990         __commit_end;
991
992         $self->render(
993                 order => $order,
994                 offset => $offset,
995                 limit => $limit,
996                 sorted => $sorted_items,
997                 columns => [ @columns ],
998                 rows => $#$filtered + 1,
999                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
1000                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
1001                 filters => $self->_current_filters,
1002                 code => $code,
1003                 cols_changed => $commit ? $commit_changed : $test_changed,
1004                 code_depends => $code_depends,
1005                 code_description => $code_description,
1006                 code_path => $code_path,
1007                 out => $out,
1008         );
1009
1010 }
1011
1012
1013 sub order {
1014         my $self = shift;
1015         $self->session('order', $self->param('order'));
1016         $self->session('sort', $self->param('sort'));
1017         return $self->redirect_to('/data/items');
1018 }
1019
1020 sub _is_numeric {
1021         my ( $self, $name ) = @_;
1022
1023         my $stats = $self->_loaded( 'stats' );
1024
1025         # sort facet numerically if more >50% elements are numeric
1026         my $count = $stats->{$name}->{count};
1027         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
1028         defined $stats->{$name}->{numeric} &&
1029                 $stats->{$name}->{numeric} > $count / 2;
1030 }
1031
1032 sub _is_unique {
1033         my ( $self, $name ) = @_;
1034         my $stats = $self->_loaded( 'stats' );
1035         defined $stats->{$name}->{unique};
1036 }
1037
1038 sub _remove_filter {
1039         my ($self,$name) = @_;
1040         warn "_remove_filter $name\n";
1041
1042         my $path = $self->session('path');
1043
1044         delete $filters->{$name};
1045         delete $loaded->{$path}->{filters}->{$name};
1046         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
1047
1048         foreach (
1049                         grep { /\Q$name\E/ }
1050                         keys %{ $loaded->{$path}->{filtered} }
1051         ) {
1052                 delete $loaded->{$path}->{filtered}->{$_};
1053                 warn "remove filtered cache $_";
1054         }
1055 }
1056
1057 sub facet {
1058         my $self = shift;
1059
1060         my $path = $self->session('path') || return $self->redirect_to( '/data/index' );
1061
1062         if ( my $name = $self->param('remove') ) {
1063                 $self->_remove_filter( $name );
1064                 return $self->redirect_to( '/data/items' );
1065         }
1066
1067         my $facet;
1068         my $name = $self->param('name') || die "no name";
1069
1070         my $all = $self->_param_scalar('all', 1);
1071         my $data = $self->_loaded('data');
1072
1073         my $filters = $self->_current_filters;
1074         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
1075         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
1076                 if defined $loaded->{$path}->{filtered}->{$all_filters};
1077
1078         if ( ! $filtered || $all ) {
1079                 $filtered = [ 0 .. $#{ $data->{items} } ];
1080                 warn "filter all values\n";
1081         } else {
1082                 warn "filter using $all_filters\n";
1083         }
1084
1085         foreach my $i ( @$filtered ) {
1086                 my $item = $data->{items}->[$i];
1087                 if ( ! exists $item->{$name} || ! defined $item->{$name} ) {
1088                         $facet->{ _missing }++;
1089                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
1090                         $facet->{$_}++ foreach @{ $item->{$name} };
1091                 } else {
1092                         $facet->{ $item->{$name} }++;
1093                 }
1094         }
1095
1096         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
1097
1098         if ( my $code = $self->param('code') ) {
1099                 my $out;
1100                 foreach my $value ( keys %$facet ) {
1101                         my $count = $facet->{$value};
1102                         my $checked = $checked_values->{$value};
1103                         eval $code;
1104                         if ( $@ ) {
1105                                 $out = $@;
1106                                 warn "ERROR: $@\n$code\n";
1107                                 last;
1108                         } elsif ( $checked != $checked_values->{$value} ) {
1109                                 warn "checked $value $count -> $checked\n";
1110                                 $checked_values->{$value} = $checked;
1111                         }
1112                 }
1113                 warn "out ",dump( $out );
1114                 $self->stash( out => $out );
1115         }
1116
1117 #       warn "# facet $name ",dump $facet;
1118
1119         my @facet_names =
1120                   $all                      ? keys %$facet
1121                 : defined $filters->{$name} ? @{ $filters->{$name} }
1122                 : keys %$facet;
1123
1124         my $numeric = $self->_is_numeric($name);
1125
1126         my $sort = $self->param('sort');
1127         # sort numeric facets with more than 5 values ascending
1128         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
1129
1130         @facet_names = sort {
1131                 my $result;
1132                 if ( $sort eq 'a' ) {
1133                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
1134                 } elsif ( $sort eq 'd' ) {
1135                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1136                 } elsif ( $sort eq 'c' ) {
1137                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1138                 } else {
1139                         warn "unknown sort: $sort";
1140                         $result = $a cmp $b;
1141                 }
1142                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1143                 $result;
1144         } @facet_names;
1145
1146         $self->render( name => $name, facet => $facet, checked => $checked_values,
1147                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1148         );
1149 }
1150
1151
1152 sub __invalidate_path_column {
1153         my ( $path, $name ) = @_;
1154
1155         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1156                 delete $loaded->{$path}->{sorted}->{$name};
1157                 warn "# invalidate $path sorted $name\n";
1158         }
1159
1160         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1161                 delete $loaded->{$path}->{filtered}->{$_};
1162                 warn "# invalidate $path filtered $_\n";
1163         }
1164 }
1165
1166 sub __path_modified {
1167         my ( $path, $value ) = @_;
1168         $value = 1 unless defined $value;
1169         
1170         $loaded->{$path}->{modified}  = $value;
1171
1172         warn "# __path_modified $path $value\n";
1173 }
1174
1175 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1176
1177 sub _save_change {
1178         my ($self,$change) = @_;
1179
1180         my $change_path = $self->_permanent_path( 'changes' );
1181         mkdir $change_path unless -d $change_path;
1182         $change_path .= '/' . $change->{time};
1183         store $change, $change_path;
1184         utime $change->{time}, $change->{time}, $change_path;
1185         warn "_save_change $change_path ", dump($change);
1186 }
1187
1188 sub edit {
1189         my $self = shift;
1190         my $new_content = $self->param('new_content');
1191         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1192
1193         my $i = $self->param('_row_id');
1194         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1195         my $path = $self->param('path') || die "no path";
1196         my $name = $self->param('name') || die "no name";
1197         my $status = 200; # 200 = OK, 201 = Created
1198
1199         my $data = $self->_loaded('data');
1200
1201         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1202                 $new_content =~ s/^\s+//s;
1203                 $new_content =~ s/\s+$//s;
1204                 my $v;
1205                 if ( $new_content =~ /\xB6/ ) { # para
1206                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1207                 } else {
1208                         $v = [ $new_content ];
1209                 }
1210
1211                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1212                 my $new = dump $v;
1213                 if ( $old ne $new
1214                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1215                 ) {
1216                         $self->_save_change({
1217                                 path => $path,
1218                                 column => $name,
1219                                 pos => $i,
1220                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1221                                 new => $v,
1222                                 time => $self->param('time') || time(),
1223                                 user => $self->param('user') || $ENV{'LOGNAME'},
1224                                 unique => {
1225                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1226                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1227                                         keys %{ $loaded->{$path}->{stats} }
1228                                 },
1229                         });
1230
1231                         warn "# change $path $i $old -> $new\n";
1232                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1233
1234                         __invalidate_path_column( $path, $name );
1235
1236                         $status = 201; # created
1237                         __path_rebuild_stats( $path );
1238         
1239                         $new_content = join("\xB6",@$v);
1240
1241                 } else {
1242                         warn "# unchanged $path $i $old\n";
1243                         $status = 304;
1244                 }
1245         } else {
1246                 $new_content = "$path $i $name doesn't exist\n";
1247                 $status = 404;
1248         }
1249
1250         warn "# edit $status ", dump $new_content;
1251
1252         $self->render(
1253                 status => $status,
1254                 new_content => scalar $new_content,
1255         );
1256 }
1257
1258
1259 sub save {
1260         my $self = shift;
1261         my $path = $self->_param_or_session('path');
1262         my $dump_path = $self->_save( $path );
1263         __path_modified( $path, 0 );
1264
1265         return $self->redirect_to( '/data/items' );
1266 }
1267
1268 sub export {
1269         my $self = shift;
1270
1271         my $dir = $self->app->home->rel_dir('public');
1272
1273         if ( my $import = $self->param('import') ) {
1274
1275                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1276                         my $name = $self->_column_from_unac( $1 );
1277
1278                         my @vals = map { chomp; $_ }
1279                                 read_file "$dir/export/$import", binmode => ':utf8';
1280
1281                         $self->_remove_filter( $name );
1282                         $self->_filter_on_data( $name, @vals );
1283                         $self->session( 'offset' => 0 );
1284                         return $self->redirect_to('/data/items');
1285                 } else {
1286                         warn "UNKNOWN IMPORT $import";
1287                 }
1288         }
1289
1290         if ( my $remove = $self->param('remove') ) {
1291                 my $path = "$dir/export/$remove";
1292                 unlink $path if -e $path;
1293                 $path .= '.png';
1294                 unlink $path if -e $path;
1295         }
1296
1297         my $path = $self->_export_path || return $self->redirect_to('/data/index');
1298
1299         my @files = grep { ! /\.png$/ } glob "$path/*";
1300         my $mtime = { map { $_ => (stat($_))[9] } @files };
1301         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1302         $self->render( export => [ @files ] );
1303 }
1304
1305 sub __loaded_paths {
1306         return
1307                 grep { defined $loaded->{$_}->{data} }
1308                 keys %$loaded;
1309 }
1310
1311 sub remove {
1312         my $self = shift;
1313         my $path = $self->param('path');
1314         if ( $path =~ m{^/tmp/mojo_facets\.} ) {
1315                 unlink $path;
1316                 warn "# unlink $path";
1317                 if ( my $name = $self->param('name') ) {
1318                         delete $loaded->{$name};
1319                         warn "# remove $name from memory";
1320                 }
1321         } else {
1322                 warn "WARNING: $path unlink ignored";
1323         }
1324         return $self->redirect_to( '/data/load' );
1325 }
1326
1327 1;