dump lookup debug with new config option
[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 use Statistics::Descriptive;
19
20 use MojoFacets::Import::File;
21 use MojoFacets::Import::HTMLTable;
22 use MojoFacets::Import::CSV;
23 use MojoFacets::Import::CouchDB;
24 use MojoFacets::Import::SQL;
25
26 our $loaded;
27 our $filters;
28
29 sub index {
30         my $self = shift;
31
32         my $data_dir = $self->app->home->rel_dir('data');
33         die "no data dir $data_dir" unless -d $data_dir;
34
35         my @files;
36         my $changes;
37         find( sub {
38                 my $file = $File::Find::name;
39                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
40                         $file =~ s/$data_dir\/*//;
41                         push @files, $file;
42                 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
43                         push @{ $changes->{$1} }, $2
44                 } elsif ( -d $file && $file =~ m/\.html$/ ) {
45                         $file =~ s/$data_dir\/*//;
46                         push @files, $file;
47                 } elsif ( -f $file && $file =~ m/\.(csv|storable|couchdb|sql)$/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
59         my $size;
60         $size->{$_} = -s "$data_dir/$_" foreach @files;
61
62         $self->render(
63                 files => [ @files ],
64                 size => $size,
65                 loaded => $loaded,
66                 filters => $filters,
67                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
68                 changes => $changes,
69         );
70 }
71
72 sub _dump_path {
73         my ( $self, $name ) = @_;
74         my $dir = $self->app->home->rel_dir('data');
75         $name =~ s/^$dir//;
76         $name =~ s/\/+/_/g;
77         return '/tmp/mojo_facets.' . $name . '.storable';
78 }
79
80 sub _save {
81         my ( $self, $path ) = @_;
82
83         my $dump_path = $self->_dump_path( $path );
84         my $first_load = ! -e $dump_path;
85         warn "save loaded to $dump_path";
86         my $info = $loaded->{$path};
87         store $info, $dump_path;
88
89         if ( $first_load ) {
90                 my $mtime = $loaded->{$path}->{mtime};
91                 utime $mtime, $mtime, $dump_path;
92                 warn "sync time to $path at $mtime\n";
93         }
94
95         warn $dump_path, ' ', -s $dump_path, " bytes\n";
96         return $dump_path;
97 }
98
99
100 sub __stats {
101
102         my $stats;
103
104         my $nr_items = $#{ $_[0] } + 1;
105
106         warn "__stats $nr_items\n";
107
108 #       foreach my $e ( @{ $_[0] } ) {
109         foreach my $i ( 0 .. $#{$_[0]} ) {
110                 print STDERR " $i" if $i % 5000;
111                 my $e = $_[0]->[$i];
112                 foreach my $n ( keys %$e ) {
113                         $stats->{$n}->{count}++;
114                         my @v;
115                         if ( ref $e->{$n} eq 'ARRAY' ) {
116                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
117                                 @v = @{ $e->{$n} };
118                         } else {
119                                 @v = ( $e->{$n} );
120                         }
121
122                         foreach my $x ( @v ) {
123                                 if ( ! defined $x ) { # FIXME really null
124                                         $stats->{$n}->{empty}++;
125                                         next;
126                                 }
127
128                                 $stats->{$n}->{numeric}++
129                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
130                                 $stats->{$n}->{empty}++
131                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
132                         }
133
134                 }
135         }
136
137         foreach my $n ( keys %$stats ) {
138                 my $s = $stats->{$n};
139                 next unless defined $s->{array};
140                 if ( $s->{array} == $s->{count} ) {
141                         delete $s->{array};
142                         if ( $s->{count} == $nr_items ) {
143                                 warn "check $n for uniqeness\n";
144                                 my $unique;
145                                 foreach my $e ( @{ $_[0] } ) {
146                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
147                                                 $unique = 0;
148                                                 last;
149                                         }
150                                 }
151                                 if ( $unique ) {
152                                         $stats->{$n}->{unique} = 1;
153                                         #warn "# $n unique ",dump( $unique );
154                                 }
155                         }
156                 }
157         }
158
159         warn "# __stats ",dump($stats);
160
161         return $stats;
162 }
163
164 sub _param_or_session {
165         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
166 }
167
168 sub stats {
169         my $self = shift;
170         my $path = $self->_param_or_session('path');
171         warn "stats $path\n";
172         delete $loaded->{$path}->{stats};
173         return $self->redirect_to( '/data/columns' );
174 }
175
176
177 sub _load_path {
178         my ( $self, $path ) = @_;
179
180         return if defined $loaded->{$path}->{generated};
181
182         my $full_path = $self->app->home->rel_file( 'data/' . $path );
183         return $self->redirect_to('/data/index') unless -r $full_path;
184
185         my $dump_path = $self->_dump_path( $path );
186
187         if ( defined $loaded->{$path}->{data} ) {
188                 my $mtime = (stat($full_path))[9];
189                 return if $loaded->{$path}->{mtime} == $mtime;
190                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
191         } elsif ( -e $dump_path ) {
192                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
193                 my $info = retrieve $dump_path;
194                 $loaded->{ $path } = $info;
195                 return;
196         }
197
198         my $data;
199         if ( -f $full_path ) {
200                 if ( $full_path =~ m/.storable$/ ) { # check storable first to catch files copied from /tmp/
201                         $data->{generated}++;
202                         warn "open $full_path ", -s $full_path, " bytes";
203                         open(my $pipe, "<", $full_path) || die $!;
204                         while ( my $o = eval { Storable::fd_retrieve $pipe } ) {
205                                 if ( exists $o->{item} ) {
206                                         # stream of storable objects
207                                         push @{ $data->{items} }, $o->{item};
208                                 } elsif ( exists $o->{data}->{items} ) {
209                                         # /tmp/mojofacets.*.storable
210                                         $data->{items} = $o->{data}->{items};
211                                         $data->{header} = $o->{header};
212                                         delete $data->{generated};
213                                 } else {
214                                         warn "SKIP ",dump($o);
215                                 }
216                         }
217                         close($pipe);
218                         warn "loaded ", $#{ $data->{items} } + 1, " items from $full_path\n";
219                 } elsif ( $full_path =~ m/.csv/i ) {
220                         $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
221                 } elsif ( $full_path =~ m/.sql/i ) {
222                         $data = MojoFacets::Import::SQL->new( full_path => $full_path )->data;
223                 } elsif ( $full_path =~ m/.couchdb/i ) {
224                         $data = MojoFacets::Import::CouchDB->new( full_path => $full_path )->data;
225                 } else {
226                         $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
227                 }
228         } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
229                 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
230         } else {
231                 die "can't load $full_path";
232         }
233
234         my @header;
235
236         if ( defined $data->{header} ) {
237                 if ( ref $data->{header} eq 'ARRAY' ) {
238                         @header = @{ $data->{header} };
239                 } else {
240                         warn "header not array ", dump( $data->{header} );
241                 }
242         }
243
244         my $stats = __stats( $data->{items} );
245
246         @header =
247                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
248                 grep { defined $stats->{$_}->{count} } keys %$stats
249                 unless @header;
250
251         my $info = {
252                 header => [ @header ],
253                 stats  => $stats,
254                 full_path => $full_path,
255                 size => -s $full_path,
256                 mtime => (stat($full_path))[9],
257                 data => $data,
258                 defined $data->{generated} ? ( generated => 1 ) : (),
259         };
260
261         $loaded->{ $path } = $info;
262         $self->_save( $path ) unless $info->{generated};
263
264 }
265
266
267 sub load {
268         my $self = shift;
269
270         my $path = $self->param('path') || return $self->redirect_to( '/data/index' );
271
272         my @paths = $self->param('paths');
273         warn "# paths ", dump @paths;
274
275         foreach my $p ( keys %$loaded ) {
276                 next if grep { /^\Q$p\E$/ } @paths;
277                 warn "remove $p from memory\n";
278                 delete $loaded->{$p};
279         }
280
281         $self->_load_path( $_ ) foreach @paths;
282
283         warn "# path $path\n";
284         $self->_load_path( $path );
285
286         $self->session( 'path' => $path );
287
288         my $redirect_to = '/data/items';
289
290         $self->session( 'header' => $loaded->{$path}->{header} );
291         if ( ! defined $loaded->{$path}->{columns} ) {
292                 my $columns_path = $self->_permanent_path( 'columns' );
293                 if ( -e $columns_path ) {
294                         my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path, binmode => ':utf8';
295                         $loaded->{$path}->{columns} = [ @columns ];
296                         warn "# columns_path $columns_path ",dump(@columns);
297                 } else {
298                         $loaded->{$path}->{columns} = $loaded->{$path}->{header}
299                 }
300
301                 $redirect_to = '/data/columns';
302         }
303         $self->session( 'columns' => $loaded->{$path}->{columns} );
304         $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
305         return $self->redirect_to( $redirect_to );
306 }
307
308
309 sub _loaded {
310         my ( $self, $name ) = @_;
311         my $path = $self->session('path') || $self->param('path');
312         return $self->redirect_to('/data/index') unless $path;
313
314         if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
315                 my $caller = (caller(1))[3];
316                 if ( $caller =~ m/::edit/ ) {
317                         warn "rebuild stats for $path ignored caller $caller\n";
318                 } else {
319                         warn "rebuild stats for $path FORCED by modified caller $caller\n";
320 #                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
321                         $loaded->{$path}->{rebuild_stats} = 1;
322                         $loaded->{$path}->{modified} = 1;
323                 }
324         }
325
326         if ( defined $loaded->{$path}->{rebuild_stats} ) {
327                 warn "rebuild_stats $path";
328                 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
329                 delete $loaded->{$path}->{rebuild_stats};
330         }
331
332         if ( ! defined $loaded->{$path}->{$name} ) {
333                 warn "$path $name isn't loaded\n";
334                 $self->_load_path( $path );
335                 if ( ! defined $loaded->{$path}->{stats} ) {
336                         warn "rebuild stats for $path\n";
337                         $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
338                 }
339                 if ( ! defined $loaded->{$path}->{$name} ) {
340                         warn "MISSING $name for $path\n";
341                         return $self->redirect_to('/data/index')
342                 }
343         }
344
345         $self->session( 'modified' => $loaded->{$path}->{modified} );
346
347         return $loaded->{$path}->{$name};
348 }
349
350
351 sub _checked {
352         my $self = shift;
353         my $checked;
354         $checked->{$_}++ foreach @_;
355 #       warn "# _checked ",dump($checked);
356         return $checked;
357 }
358
359 sub _permanent_path {
360         my $self = shift;
361         my $path = $self->_param_or_session('path');
362         $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
363 }
364
365 sub __unac {
366         my $n = shift;
367         $n = unac_string($n);
368         $n =~ s/\W+/_/g;
369         return $n;
370 }
371
372 sub _column_from_unac {
373         my ($self,$name) = @_;
374
375         my $stats = $self->_loaded('stats');
376         my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
377
378         $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
379 }
380
381 sub _export_path {
382         my $self = shift;
383         my $path = $self->_param_or_session('path');
384         if ( ! $path ) {
385                 warn "no path in param or session";
386                 return;
387         }
388         my $dir = $self->app->home->rel_dir('public') . "/export/$path";
389         mkpath $dir unless -e $dir;
390         my $name = __export_path_name( $path, @_ );
391         my $full = $dir . '/' . $name;
392         $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
393         return $full;
394 }
395
396 sub __export_path_name {
397         my $max_length = 80;
398
399         my $path = shift;
400         my $name = join('.', map { __unac($_) } @_ );
401         if ( length($name) > $max_length ) {
402                 $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
403         }
404         return $name;
405 }
406
407 sub columns {
408     my $self = shift;
409
410         my $view_path = $self->_permanent_path( 'views' );
411
412         if ( $self->param('columns') ) {
413                 my @columns = $self->_param_array('columns');
414                 write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
415                 if ( my $view = $self->param('view') ) {
416                         mkdir $view_path unless -e $view_path;
417                         write_file( "$view_path/$view", { binmode => ':utf8' }, map { "$_\n" } @columns );
418                         warn "view $view_path/$view ", -s "$view_path/$view", " bytes\n";
419                 }
420
421                 return $self->redirect_to('/data/items');
422
423         } elsif ( ! $self->session('header') ) {
424                 return $self->redirect_to('/data/load');
425         }
426
427         if ( my $id = $self->param('id') ) {
428                 my $view_full = "$view_path/$id";
429                 if ( -f $view_full ) {
430                         my @columns = map { chomp; $_ } read_file $view_full, binmode => ':utf8';
431                         warn "view $view_full loaded ", dump @columns;
432                         $self->session( 'columns' => [ @columns ] );
433                         return $self->redirect_to('/data/items');
434                 }
435         }
436
437         my $stats = $self->_loaded( 'stats' );
438
439         my @columns;
440         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
441
442         foreach my $c ( @{ $self->session( 'header' ) } ) {
443                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
444         }
445
446         my @views;
447         if ( -d $view_path ) {
448                 @views = map { s{^\Q$view_path\E/*}{}; $_ } glob "$view_path/*";
449                 warn "# views ",dump @views;
450         }
451
452     $self->render(
453                 message => 'Select columns to display',
454                 stats => $stats,
455                 columns => \@columns,
456                 checked => $self->_checked( $self->_param_array('columns') ),
457                 views => \@views,
458         );
459 }
460
461 sub _param_array {
462     my ($self,$name) = @_;
463
464         my @array = $self->param($name);
465         my $path  = $self->session('path');
466
467         if ( @array ) {
468                 $self->session($name => [ @array ]);
469                 $loaded->{$path}->{$name} = [ @array ];
470         } elsif ( my $session = $self->session($name) ) {
471                 if ( ref $session eq 'ARRAY' ) {
472                         @array = @$session;
473                 } else {
474                         die "$name not array ",dump($session);
475                 }
476         }
477         #warn "# $name ",dump @array;
478         return @array;
479 }
480
481 sub _param_scalar {
482     my ($self,$name,$default) = @_;
483
484         my $scalar = $self->param($name);
485
486         if ( defined $scalar ) {
487                 $self->session($name => $scalar);
488         } else {
489                 $scalar = $self->session($name);
490         }
491
492         if ( ! defined $scalar ) {
493                 $scalar = $default;
494                 die "no default for $name" unless defined $scalar;
495                 $self->session($name => $scalar);
496         }
497
498         warn "# _param_scalar $name ",dump $scalar;
499         return $scalar;
500 }
501
502 sub filter {
503         my $self = shift;
504
505         my $name = $self->param('filter_name') || die "name?";
506         my @vals = $self->param('filter_vals');
507
508         $self->_remove_filter( $name );
509         if ( @vals ) {
510                 $self->_filter_on_data( $name, @vals );
511                 if ( my $permanent = $self->param('_permanent') ) {
512                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
513                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
514                         warn "permanent filter $permanent_path ", -s $permanent_path;
515                 }
516         }
517
518         $self->session( 'offset' => 0 );
519
520         return $self->redirect_to('/data/items');
521 }
522
523 sub _filter_on_data {
524         my ( $self, $name, @vals ) = @_;
525
526         my $path = $self->session('path');
527
528         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
529                 warn "filter $name not found in data set";
530                 return;
531         }
532
533         $filters->{$name} = [ @vals ];
534         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
535
536         my $filter_hash;
537         $filter_hash->{$_}++ foreach @vals;
538
539         #warn "# filter_hash ",dump( $filter_hash );
540
541         my $items = $self->_loaded('data')->{items};
542
543         my $include_missing = defined $filter_hash->{_missing};
544         my $filtered_items;
545
546         foreach my $i ( 0 .. $#$items ) {
547
548                 if ( defined $items->[$i]->{$name} ) {
549                         my $row = $items->[$i]->{$name};
550                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
551                         foreach my $v ( @$row ) {
552                                 if ( defined $filter_hash->{ $v } ) {
553                                         $filtered_items->{$i}++;
554                                 }
555                         }
556                 } elsif ( $include_missing ) {
557                         $filtered_items->{$i}++;
558                 }
559         }
560
561         #warn "# filter $name ",dump($filtered_items);
562
563         # invalidate filters on other datasets
564         foreach my $dataset ( grep { exists $loaded->{$_}->{filters}->{$name} } keys %$loaded ) {
565                 delete $loaded->{$dataset}->{filters}->{$name};
566                 delete $loaded->{$dataset}->{filtered};
567         }
568
569         $loaded->{$path}->{filters}->{$name} = $filtered_items;
570         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
571 }
572
573
574 sub _current_filters {
575         my $self = shift;
576         my $current_filters;
577         my $stats = $self->_loaded('stats');
578
579         $current_filters->{ $_ } = $filters->{ $_ }
580                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
581         ;
582         warn "# _current_filters ",dump( keys %$current_filters );
583         return $current_filters;
584 }
585
586 sub _data_sorted_by {
587         my ( $self, $order ) = @_;
588
589         my $path = $self->session('path');
590
591         warn "_data_sorted_by $order from $path";
592
593         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
594                 return $loaded->{$path}->{sorted}->{$order};
595         }
596
597         my $data = $self->_loaded( 'data' );
598         my $numeric = $self->_is_numeric($order);
599         my $missing = $numeric ? 0 : '';
600         no warnings qw(numeric);
601         my $nr = 0;
602         my @sorted = map {
603                 $_->[0]
604         } sort {
605                 if ( $numeric ) {
606                         $a->[1] <=> $b->[1]
607                 } else {
608                         $a->[1] cmp $b->[1]
609                 }
610         } map {
611                 my $v;
612                 if ( ! exists $_->{$order} ) {
613                         $v = $missing;
614                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
615                         $v = join('', @{$_->{$order}});
616                 } else {
617                         $v = $_->{$order};
618                 }
619                 [ $nr++, $v ]
620         } @{ $data->{items} }
621         ;
622
623         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
624         #warn "# sorted ",dump( @sorted );
625
626         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
627 }
628
629
630 sub __all_filters {
631         my $order = pop @_;
632         join(',', sort(@_), 'order', $order);
633 }
634
635 our ($out, $key,$value);
636
637 our $lookup_path_col;
638 our $on;
639
640 sub __commit_begin {
641         warn "__commit_begin";
642         $lookup_path_col = undef;
643         $on = undef;
644 }
645
646 sub __commit_end {
647         warn "__commit_end";
648         $lookup_path_col = undef; # cleanup memory
649         $on = undef;
650 }
651
652 sub lookup {
653         warn "# lookup ",dump @_;
654         my ( $vals, $on_path, $on_col, $code, $stat_code ) = @_;
655         die "code is not sub{ ... } but ", dump $code unless ref $code eq 'CODE';
656
657         if ( ! exists $loaded->{$on_path} ) {
658                 my @possible_paths = grep { /\Q$on_path\E/ } keys %$loaded;
659                 die "more than one dataset available for '$on_path' ",dump @possible_paths if $#possible_paths > 0;
660                 $on_path = shift @possible_paths;
661                 warn "## fuzzy selected path $on_path";
662         }
663
664         my $items = $loaded->{$on_path}->{data}->{items} || die "no items for $on_path";
665
666         if ( ! exists $lookup_path_col->{$on_path}->{$on_col} ) {
667                 warn "create lookup_path_col $on_path $on_col";
668                 foreach my $i ( 0 .. $#$items ) {
669                         my $item = $items->[$i];
670                         if ( exists $item->{$on_col} ) {
671                                 if ( ref $item->{$on_col} eq 'ARRAY' ) {
672                                         foreach my $v ( @{ $item->{$on_col} } ) {
673                                                 push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
674                                         }
675                                 } elsif ( ! ref $item->{$on_col} ) { # scalar
676                                         my $v = $item->{$on_col};
677                                         push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
678                                 } else {
679                                         die "unknown type of ",dump $item->{$on_col};
680                                 }
681                         }
682                 }
683                 warn "XXX ",dump $lookup_path_col->{$on_path}->{$on_col} if $ENV{DEBUG};
684         }
685
686         my $stat;
687         $stat = Statistics::Descriptive::Full->new() if $stat_code;
688
689         foreach my $v ( ref $vals eq 'ARRAY' ? @$vals : ( $vals ) ) {
690                 foreach my $i ( @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} } ) {
691                         $on = $items->[$i];
692                         warn "XXX lookup code $v $i ",dump $on if $ENV{DEBUG};
693                         $code->($stat);
694                 }
695                 $stat_code->( $stat ) if $stat_code;
696         }
697 }
698
699 sub __commit_path_code {
700         my ( $path, $i, $code, $commit_changed ) = @_;
701
702         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
703         my $row = $items->[$i];
704         my $update;
705         eval $code;
706         foreach ( keys %$update ) {
707                 $$commit_changed->{$_}++;
708                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
709         }
710         #warn "__commit_path_code $path $i ",dump( $update );
711 }
712
713 # uses templates/admin.html.ep
714 sub _switch_dataset {
715         my $self = shift;
716
717         my $datasets;
718
719         foreach my $path ( keys %$loaded ) {
720                 next unless exists $loaded->{$path}->{data};
721                 push @$datasets, $path;
722         }
723
724         warn "# datasets ",dump($datasets);
725
726         $self->stash( 'datasets' => $datasets );
727 }
728
729 sub items {
730         my $self = shift;
731
732         $self->_switch_dataset;
733
734         if ( my $show = $self->param('id') ) {
735                 $self->param('show', $show);
736                 warn "show $show\n";
737         }
738
739         my $path = $self->_param_scalar('path');
740
741         my @columns = $self->_param_array('columns');
742         return $self->redirect_to('/data/columns') unless @columns;
743         my $order   = $self->_param_scalar('order', $columns[0]);
744         my $sort    = $self->_param_scalar('sort', 'a');
745         my $offset  = $self->_param_scalar('offset', 0);
746         my $limit   = $self->_param_scalar('limit', 20);
747         $self->_param_scalar('show', 'table');
748
749         # fix offset when changing limit
750         $offset = int( $offset / $limit ) * $limit;
751
752         if ( ! grep { /^\Q$order\E$/ } @columns ) {
753                 $order = $columns[0];
754                 $self->session( order => $order );
755         }
756         my $sorted = $self->_data_sorted_by( $order );
757
758         my @filter_names;
759         if ( $filters ) {
760                 foreach my $name ( keys %$filters ) {
761                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
762                                 warn "skip filter $name not found in $path\n";
763                                 next;
764                         }
765                         push @filter_names, $name;
766                 }
767                 warn "filter_names ",dump( @filter_names );
768                 foreach my $name ( @filter_names ) {
769                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
770                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
771                 }
772         }
773
774         my $all_filters = __all_filters( @filter_names,$order );
775
776 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
777
778         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
779
780                 my $path_filters = $loaded->{$path}->{filters};
781
782                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
783
784                 my @filtered;
785                 foreach my $i ( 0 .. $#$sorted ) {
786                         my $pos = $sorted->[$i];
787
788                         if ( $#filter_names == -1 ) {
789                                 push @filtered, $pos;
790                                 next;
791                         }
792
793                         my $skip = 0;
794                         foreach ( @filter_names ) {
795                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
796                         }
797                         next if $skip;
798
799                         push @filtered, $pos;
800                 }
801
802                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
803         }
804
805         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
806                 if defined $loaded->{$path}->{filtered}->{$all_filters};
807
808         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
809
810         my $data = $self->_loaded('data');
811
812         my $code = $self->_param_scalar('code','');
813         $code =~ s{\r}{}gs;
814         $code =~ s{\n+$}{\n}s;
815
816         my $commit = $self->param('commit');
817         my $test = $self->param('test');
818
819         my $commit_changed;
820         __commit_begin;
821
822         if ( $code && ( $test || $commit ) ) {
823                 # XXX find columns used in code snippet and show them to user
824                 my $order = 0;
825                 foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
826                         if ( $column =~ s/^(['"])// ) {
827                                 $column =~ s/$1$//;
828                         }
829                         next if $column =~ m/\$/; # hide columns with vars in them
830                         $commit_changed->{$column} = 0;
831                 }
832         }
833
834         my $code_path = $self->app->home->rel_dir('public') . "/code";
835         if ( $commit ) {
836
837                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
838                 ( $key, $value, $out ) = ( 'key', 'value' );
839                 foreach ( 0 .. $#$filtered ) {
840                         my $i = $filtered->[$_];
841                         __commit_path_code( $path, $i, $code, \$commit_changed );
842                 }
843
844                 # this might move before $out to recalculate stats on source dataset?
845                 __path_rebuild_stats( $path );
846                 my $c = { map { $_ => 1 } @columns };
847                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
848                 warn "# added_columns ",dump( @added_columns );
849                 unshift @columns, @added_columns;
850
851                 $loaded->{$path}->{columns} = [ @columns ];
852                 warn "# new columns ",dump( @columns );
853
854                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
855
856                 $self->_save_change({
857                         path => $path,
858                         time => $self->param('time') || time(),
859                         user => $self->param('user') || $ENV{'LOGNAME'},
860                         code => $code,
861                         commit_changed => $commit_changed,
862                 });
863
864                 if ( my $description = $self->param('code_description') ) {
865                         my $depends = $self->param('code_depends') || die "no code_depends?";
866                         my $path = "$code_path/$depends.$description.pl";
867                         if ( -e $path && ! $self->param('overwrite') ) {
868                                 warn "# code $path not saved\n";
869                         } else {
870                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
871                                 warn "code $path ", -s $path, " bytes saved\n";
872                         }
873                 }
874
875                 # remove console
876                 $code = '';
877                 if ( $out ) {
878                         my $commit_dataset = join('.'
879                                 , $self->param('code_depends')
880                                 , $self->param('code_description')
881                                 , time()
882                         );
883                         $key ||= 'key';
884                         $value ||= 'value';
885                         warn "key $key value $value";
886                         my $items;
887                         foreach my $n ( keys %$out ) {
888                                 my $i = { $key => [ $n ] };
889                                 my $ref = ref $out->{$n};
890                                 if ( $ref eq 'HASH' ) {
891                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
892                                 } elsif ( $ref eq 'ARRAY' ) {
893                                         $i->{$_} = $out->{$n};
894                                 } elsif ( ! $ref ) {
895                                         $i->{$value} = [ $out->{$n} ];
896                                 } else {
897                                         $i->{_error} = [ dump($out->{$n}) ];
898                                 }
899                                 push @$items, $i;
900                         };
901                         undef $out;
902                         my $stats = __stats( $items );
903                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
904                         unshift @columns, $key;
905
906                         $loaded->{$commit_dataset} = {
907                                 header => [ @columns ],
908                                 columns => [ @columns ],
909                                 mtime => time(),
910                                 data => { items => $items },
911                                 stats => $stats,
912                                 generated => 1,
913                         };
914                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
915                         $self->session('path', $commit_dataset);
916                         $self->session('columns', [ @columns ]);
917                         $self->session('order', $key);
918                         return $self->redirect_to('/data/items');
919                 }
920
921                 $self->session('columns', [ @columns ]);
922         }
923
924         my $sorted_items;
925         my $from_end = $sort eq 'd' ? $#$filtered : 0;
926         my $test_changed;
927         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
928         foreach ( 0 .. $limit ) {
929                 my $i = $_ + $offset;
930                 last unless defined $filtered->[$i];
931                 $i = $from_end - $i if $from_end;
932                 my $id = $filtered->[$i];
933                 my $row = Storable::dclone $data->{items}->[ $id ];
934                 if ( $code && $test ) {
935                         my $update;
936                         eval $code;
937                         if ( $@ ) {
938                                 warn "ERROR evaling $@", dump($code);
939                                 $self->stash('eval_error', $@) if $@;
940                         } else {
941                                 warn "EVAL ",dump($update);
942                                 foreach ( keys %$update ) {
943                                         $test_changed->{$_}++;
944                                         $row->{$_} = $update->{$_};
945                                 }
946                         }
947                 }
948                 $row->{_row_id} ||= $id;
949                 push @$sorted_items, $row;
950         }
951
952         if ( $self->param('export') ) {
953                 my $export_path = $self->_export_path( 'items', @columns);
954                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
955                 print $fh "#",join("\t",@columns),"\n";
956                 foreach my $f ( 0 .. $#$filtered ) {
957                         print $fh join("\t", map {
958                                 my $i = $data->{items}->[ $filtered->[$f] ];
959                                 my $v = '\N';
960                                 if ( ! defined $i->{$_} ) {
961                                         # nop
962                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
963                                         $v =join(',', @{ $i->{$_} });
964                                         $v = '\N' if length($v) == 0;
965                                 } elsif ( ! ref $i->{$_} ) {
966                                         $v = $i->{$_};
967                                 } else {
968                                         $v = dump $i->{$_};
969                                 }
970                                 $v;
971                         } @columns),"\n";
972                 }
973                 close($fh);
974                 warn "export $export_path ", -s $export_path, " bytes\n";
975         }
976
977         my ( $code_depends, $code_description );
978
979         if ( $test ) {
980
981                 warn "# test_changed ",dump( $test_changed );
982                 my $c = { map { $_ => 1 } @columns };
983                 my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
984                 unshift @columns, @added_columns;
985
986                 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
987
988                 my $depends_on;
989                 my $tmp = $code; $tmp =~ s/\$row->{(['"]?)([\w\s]+)\1/$depends_on->{$2}++/gse;
990                 warn "# depends_on ",dump $depends_on;
991
992                 my $test_added = Storable::dclone $test_changed;
993                 delete $test_added->{$_} foreach keys %$depends_on;
994
995                 $code_depends = $self->param('code_depends')
996                 || join(',', keys %$depends_on);
997
998                 $code_description = $self->param('code_description') ||
999                 join(',', keys %$test_added);
1000
1001                 $code_depends ||= $code_description; # self-modifing
1002                 if ( ! $code_depends && $out ) {
1003                         $code_depends = $key;
1004                         $code_description = $value;
1005                 }
1006
1007                 warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
1008
1009         } # test?
1010
1011         __commit_end;
1012
1013         $self->render(
1014                 order => $order,
1015                 offset => $offset,
1016                 limit => $limit,
1017                 sorted => $sorted_items,
1018                 columns => [ @columns ],
1019                 rows => $#$filtered + 1,
1020                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
1021                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
1022                 filters => $self->_current_filters,
1023                 code => $code,
1024                 cols_changed => $commit ? $commit_changed : $test_changed,
1025                 code_depends => $code_depends,
1026                 code_description => $code_description,
1027                 code_path => $code_path,
1028                 out => $out,
1029         );
1030
1031 }
1032
1033
1034 sub order {
1035         my $self = shift;
1036         $self->session('order', $self->param('order'));
1037         $self->session('sort', $self->param('sort'));
1038         return $self->redirect_to('/data/items');
1039 }
1040
1041 sub _is_numeric {
1042         my ( $self, $name ) = @_;
1043
1044         my $stats = $self->_loaded( 'stats' );
1045
1046         # sort facet numerically if more >50% elements are numeric
1047         my $count = $stats->{$name}->{count};
1048         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
1049         defined $stats->{$name}->{numeric} &&
1050                 $stats->{$name}->{numeric} > $count / 2;
1051 }
1052
1053 sub _is_unique {
1054         my ( $self, $name ) = @_;
1055         my $stats = $self->_loaded( 'stats' );
1056         defined $stats->{$name}->{unique};
1057 }
1058
1059 sub _remove_filter {
1060         my ($self,$name) = @_;
1061         warn "_remove_filter $name\n";
1062
1063         my $path = $self->session('path');
1064
1065         delete $filters->{$name};
1066         delete $loaded->{$path}->{filters}->{$name};
1067         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
1068
1069         foreach (
1070                         grep { /\Q$name\E/ }
1071                         keys %{ $loaded->{$path}->{filtered} }
1072         ) {
1073                 delete $loaded->{$path}->{filtered}->{$_};
1074                 warn "remove filtered cache $_";
1075         }
1076 }
1077
1078 sub facet {
1079         my $self = shift;
1080
1081         my $path = $self->session('path') || return $self->redirect_to( '/data/index' );
1082
1083         if ( my $name = $self->param('remove') ) {
1084                 $self->_remove_filter( $name );
1085                 return $self->redirect_to( '/data/items' );
1086         }
1087
1088         my $facet;
1089         my $name = $self->param('name') || die "no name";
1090
1091         my $all = $self->_param_scalar('all', 1);
1092         my $data = $self->_loaded('data');
1093
1094         my $filters = $self->_current_filters;
1095         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
1096         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
1097                 if defined $loaded->{$path}->{filtered}->{$all_filters};
1098
1099         if ( ! $filtered || $all ) {
1100                 $filtered = [ 0 .. $#{ $data->{items} } ];
1101                 warn "filter all values\n";
1102         } else {
1103                 warn "filter using $all_filters\n";
1104         }
1105
1106         foreach my $i ( @$filtered ) {
1107                 my $item = $data->{items}->[$i];
1108                 if ( ! exists $item->{$name} || ! defined $item->{$name} ) {
1109                         $facet->{ _missing }++;
1110                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
1111                         $facet->{$_}++ foreach @{ $item->{$name} };
1112                 } else {
1113                         $facet->{ $item->{$name} }++;
1114                 }
1115         }
1116
1117         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
1118
1119         if ( my $code = $self->param('code') ) {
1120                 my $out;
1121                 foreach my $value ( keys %$facet ) {
1122                         my $count = $facet->{$value};
1123                         my $checked = $checked_values->{$value};
1124                         eval $code;
1125                         if ( $@ ) {
1126                                 $out = $@;
1127                                 warn "ERROR: $@\n$code\n";
1128                                 last;
1129                         } elsif ( $checked != $checked_values->{$value} ) {
1130                                 warn "checked $value $count -> $checked\n";
1131                                 $checked_values->{$value} = $checked;
1132                         }
1133                 }
1134                 warn "out ",dump( $out );
1135                 $self->stash( out => $out );
1136         }
1137
1138 #       warn "# facet $name ",dump $facet;
1139
1140         my @facet_names =
1141                   $all                      ? keys %$facet
1142                 : defined $filters->{$name} ? @{ $filters->{$name} }
1143                 : keys %$facet;
1144
1145         my $numeric = $self->_is_numeric($name);
1146
1147         my $sort = $self->param('sort');
1148         # sort numeric facets with more than 5 values ascending
1149         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
1150
1151         @facet_names = sort {
1152                 my $result;
1153                 if ( $sort eq 'a' ) {
1154                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
1155                 } elsif ( $sort eq 'd' ) {
1156                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1157                 } elsif ( $sort eq 'c' ) {
1158                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1159                 } else {
1160                         warn "unknown sort: $sort";
1161                         $result = $a cmp $b;
1162                 }
1163                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1164                 $result;
1165         } @facet_names;
1166
1167         $self->render( name => $name, facet => $facet, checked => $checked_values,
1168                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1169         );
1170 }
1171
1172
1173 sub __invalidate_path_column {
1174         my ( $path, $name ) = @_;
1175
1176         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1177                 delete $loaded->{$path}->{sorted}->{$name};
1178                 warn "# invalidate $path sorted $name\n";
1179         }
1180
1181         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1182                 delete $loaded->{$path}->{filtered}->{$_};
1183                 warn "# invalidate $path filtered $_\n";
1184         }
1185 }
1186
1187 sub __path_modified {
1188         my ( $path, $value ) = @_;
1189         $value = 1 unless defined $value;
1190         
1191         $loaded->{$path}->{modified}  = $value;
1192
1193         warn "# __path_modified $path $value\n";
1194 }
1195
1196 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1197
1198 sub _save_change {
1199         my ($self,$change) = @_;
1200
1201         my $change_path = $self->_permanent_path( 'changes' );
1202         mkdir $change_path unless -d $change_path;
1203         $change_path .= '/' . $change->{time};
1204         store $change, $change_path;
1205         utime $change->{time}, $change->{time}, $change_path;
1206         warn "_save_change $change_path ", dump($change);
1207 }
1208
1209 sub edit {
1210         my $self = shift;
1211         my $new_content = $self->param('new_content');
1212         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1213
1214         my $i = $self->param('_row_id');
1215         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1216         my $path = $self->param('path') || die "no path";
1217         my $name = $self->param('name') || die "no name";
1218         my $status = 200; # 200 = OK, 201 = Created
1219
1220         my $data = $self->_loaded('data');
1221
1222         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1223                 $new_content =~ s/^\s+//s;
1224                 $new_content =~ s/\s+$//s;
1225                 my $v;
1226                 if ( $new_content =~ /\xB6/ ) { # para
1227                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1228                 } else {
1229                         $v = [ $new_content ];
1230                 }
1231
1232                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1233                 my $new = dump $v;
1234                 if ( $old ne $new
1235                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1236                 ) {
1237                         $self->_save_change({
1238                                 path => $path,
1239                                 column => $name,
1240                                 pos => $i,
1241                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1242                                 new => $v,
1243                                 time => $self->param('time') || time(),
1244                                 user => $self->param('user') || $ENV{'LOGNAME'},
1245                                 unique => {
1246                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1247                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1248                                         keys %{ $loaded->{$path}->{stats} }
1249                                 },
1250                         });
1251
1252                         warn "# change $path $i $old -> $new\n";
1253                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1254
1255                         __invalidate_path_column( $path, $name );
1256
1257                         $status = 201; # created
1258                         __path_rebuild_stats( $path );
1259         
1260                         $new_content = join("\xB6",@$v);
1261
1262                 } else {
1263                         warn "# unchanged $path $i $old\n";
1264                         $status = 304;
1265                 }
1266         } else {
1267                 $new_content = "$path $i $name doesn't exist\n";
1268                 $status = 404;
1269         }
1270
1271         warn "# edit $status ", dump $new_content;
1272
1273         $self->render(
1274                 status => $status,
1275                 new_content => scalar $new_content,
1276         );
1277 }
1278
1279
1280 sub save {
1281         my $self = shift;
1282         my $path = $self->_param_or_session('path');
1283         my $dump_path = $self->_save( $path );
1284         __path_modified( $path, 0 );
1285
1286         return $self->redirect_to( '/data/items' );
1287 }
1288
1289 sub export {
1290         my $self = shift;
1291
1292         my $dir = $self->app->home->rel_dir('public');
1293
1294         if ( my $import = $self->param('import') ) {
1295
1296                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1297                         my $name = $self->_column_from_unac( $1 );
1298
1299                         my @vals = map { chomp; $_ }
1300                                 read_file "$dir/export/$import", binmode => ':utf8';
1301
1302                         $self->_remove_filter( $name );
1303                         $self->_filter_on_data( $name, @vals );
1304                         $self->session( 'offset' => 0 );
1305                         return $self->redirect_to('/data/items');
1306                 } else {
1307                         warn "UNKNOWN IMPORT $import";
1308                 }
1309         }
1310
1311         if ( my $remove = $self->param('remove') ) {
1312                 my $path = "$dir/export/$remove";
1313                 unlink $path if -e $path;
1314                 $path .= '.png';
1315                 unlink $path if -e $path;
1316         }
1317
1318         my $path = $self->_export_path || return $self->redirect_to('/data/index');
1319
1320         my @files = grep { ! /\.png$/ } glob "$path/*";
1321         my $mtime = { map { $_ => (stat($_))[9] } @files };
1322         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1323         $self->render( export => [ @files ] );
1324 }
1325
1326 sub __loaded_paths {
1327         return
1328                 grep { defined $loaded->{$_}->{data} }
1329                 keys %$loaded;
1330 }
1331
1332 sub remove {
1333         my $self = shift;
1334         my $path = $self->param('path');
1335         if ( $path =~ m{^/tmp/mojo_facets\.} ) {
1336                 unlink $path;
1337                 warn "# unlink $path";
1338                 if ( my $name = $self->param('name') ) {
1339                         delete $loaded->{$name};
1340                         warn "# remove $name from memory";
1341                 }
1342         } else {
1343                 warn "WARNING: $path unlink ignored";
1344         }
1345         return $self->redirect_to( '/data/load' );
1346 }
1347
1348 1;