e1dc8ac43df4570b82be518bb8fddf9e6b8da8cf
[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); # broken with Mojo::JSON, see https://rt.cpan.org/Public/Bug/Display.html?id=86592
9 use Data::Dumper;
10 use subs 'dump';
11 sub dump { Dumper(@_) };
12
13 use File::Slurp;
14 use Encode;
15 use locale;
16 use File::Find;
17 use Storable;
18 use Time::HiRes qw(time);
19 use File::Path qw(mkpath);
20 use Text::Unaccent::PurePerl;
21 use Digest::MD5;
22 use Statistics::Descriptive;
23
24 our $imports;
25 foreach my $module ( glob('lib/MojoFacets/Import/*.pm') ) {
26         $module =~ s{lib/(\w+)/(\w+)/(.*)\.pm}{$1::$2::$3};
27         eval "use $module";
28         die "$module: $!" if $!;
29         my ( $ext, $priority ) = $module->ext;
30         $imports->{$priority || 'file'}->{$ext} = $module;
31         warn "# import $ext $module\n";
32 }
33
34 warn "# import loaded ",dump( $imports );
35
36 sub import_module {
37         my $full_path = shift;
38
39 #       warn "# import_module $full_path\n";
40
41         return if $full_path =~ m/\.columns$/;
42
43         foreach my $ext ( keys %{ $imports->{file} } ) {
44                 if ( -f $full_path && $full_path =~ m/$ext/i ) {
45                         return $imports->{file}->{$ext};
46                         last;
47                 }
48         }
49
50         foreach my $ext ( keys %{ $imports->{directory} } ) {
51                 if ( -f $full_path && $full_path =~ m/$ext/i ) {
52                         return $imports->{directory}->{$ext};
53                         last;
54                 }
55         }
56 }
57
58 our $loaded;
59 our $filters;
60
61 sub index {
62         my $self = shift;
63
64         my $data_dir = $self->app->home->rel_file('data');
65         die "no data dir $data_dir" unless -d $data_dir;
66
67         my @files;
68         my $changes;
69
70         find( sub {
71                 my $file = $File::Find::name;
72                 if ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+[\.,]\d+.+)/ ) {
73                         push @{ $changes->{$1} }, $2
74                 } elsif ( import_module( $file ) ) {
75                         my $mtime = (stat($file))[9]; # mtime
76                         $file =~ s/$data_dir\/*//;
77                         push @files, $file;
78                         $loaded->{$file}->{mtime} ||= $mtime;
79                 } else {
80                         #warn "IGNORE: $file\n";
81                 }
82         }, $data_dir);
83
84         no warnings qw(uninitialized); # mtime
85         @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
86                         grep { defined $loaded->{$_}->{generated} } keys %$loaded;
87
88         my $size;
89         $size->{$_} = -s "$data_dir/$_" foreach @files;
90
91         $self->render(
92                 files => [ @files ],
93                 size => $size,
94                 loaded => $loaded,
95                 filters => $filters,
96                 dump_path => { map { $_ => $self->_dump_path($_) } @files },
97                 changes => $changes,
98         );
99 }
100
101 sub _dump_path {
102         my ( $self, $name ) = @_;
103         my $dir = $self->app->home->rel_file('data');
104         $name =~ s/^$dir//;
105         $name =~ s/\/+/_/g;
106         return '/tmp/mojo_facets.' . $name . '.storable';
107 }
108
109 sub _save {
110         my ( $self, $path ) = @_;
111
112         my $dump_path = $self->_dump_path( $path );
113         my $first_load = ! -e $dump_path;
114         warn "save loaded to $dump_path";
115         my $info = $loaded->{$path};
116         store $info, $dump_path;
117
118         if ( $first_load ) {
119                 my $mtime = $loaded->{$path}->{mtime};
120                 utime $mtime, $mtime, $dump_path;
121                 warn "sync time to $path at $mtime\n";
122         }
123
124         warn $dump_path, ' ', -s $dump_path, " bytes\n";
125         return $dump_path;
126 }
127
128
129 sub __stats {
130
131         my $stats;
132
133         my $nr_items = $#{ $_[0] } + 1;
134
135         warn "__stats $nr_items\n";
136
137 #       foreach my $e ( @{ $_[0] } ) {
138         foreach my $i ( 0 .. $#{$_[0]} ) {
139                 print STDERR " $i" if $i % 5000;
140                 my $e = $_[0]->[$i];
141                 foreach my $n ( keys %$e ) {
142                         $stats->{$n}->{count}++;
143                         my @v;
144                         if ( ref $e->{$n} eq 'ARRAY' ) {
145                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
146                                 @v = @{ $e->{$n} };
147                         } else {
148                                 @v = ( $e->{$n} );
149                         }
150
151                         foreach my $x ( @v ) {
152                                 if ( ! defined $x ) { # FIXME really null
153                                         $stats->{$n}->{empty}++;
154                                         next;
155                                 }
156
157                                 $stats->{$n}->{numeric}++
158                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
159                                 $stats->{$n}->{empty}++
160                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
161                         }
162
163                 }
164         }
165
166         foreach my $n ( keys %$stats ) {
167                 my $s = $stats->{$n};
168                 next unless defined $s->{array};
169                 if ( $s->{array} == $s->{count} ) {
170                         delete $s->{array};
171                         if ( $s->{count} == $nr_items ) {
172                                 warn "check $n for uniqeness\n";
173                                 my $unique;
174                                 foreach my $e ( @{ $_[0] } ) {
175                                         if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
176                                                 $unique = 0;
177                                                 last;
178                                         }
179                                 }
180                                 if ( $unique ) {
181                                         $stats->{$n}->{unique} = 1;
182                                         #warn "# $n unique ",dump( $unique );
183                                 }
184                         }
185                 }
186         }
187
188         warn "# __stats ",dump($stats);
189
190         return $stats;
191 }
192
193 sub _param_or_session {
194         $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
195 }
196
197 sub stats {
198         my $self = shift;
199         my $path = $self->_param_or_session('path');
200         warn "stats $path\n";
201         delete $loaded->{$path}->{stats};
202         return $self->redirect_to( '/data/columns' );
203 }
204
205
206 sub _load_path {
207         my ( $self, $path ) = @_;
208
209         return if defined $loaded->{$path}->{generated};
210
211         my $full_path = $self->app->home->rel_file( 'data/' . $path );
212         return $self->redirect_to('/data/index') unless -r $full_path;
213
214         my $dump_path = $self->_dump_path( $path );
215
216         if ( defined $loaded->{$path}->{data} ) {
217                 my $mtime = (stat($full_path))[9];
218                 return if $loaded->{$path}->{mtime} == $mtime;
219                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
220         } elsif ( -e $dump_path ) {
221                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
222                 my $info = retrieve $dump_path;
223                 $loaded->{ $path } = $info;
224                 return;
225         }
226
227         my $data;
228         if ( my $module = import_module( $full_path ) ) {
229                 $data = $module->new( full_path => $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 @paths = @{ $self->every_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         my $path = $self->param('path') || $self->session('path') || $paths[0] || $self->redirect_to('/data/index');
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_file('data') . '/' . join('.', $path, @_);
363 }
364
365 sub __unac {
366         my $n = shift;
367         $n = unac_string('utf-8',$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 = $self->app->home->rel_file('public') . "/export/$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 = @{ $self->every_param($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                 if ( defined $scalar ) {
495                         $self->session($name => $scalar);
496                 } else {
497                         warn "no default for $name";
498                 }
499         }
500
501         warn "# _param_scalar $name ",dump $scalar;
502         return $scalar;
503 }
504
505 sub filter {
506         my $self = shift;
507
508         my $name = $self->param('filter_name') || die "name?";
509         my @vals = @{ $self->every_param('filter_vals') };
510
511         $self->_remove_filter( $name );
512         if ( @vals ) {
513                 $self->_filter_on_data( $name, @vals );
514                 if ( my $permanent = $self->param('_permanent') ) {
515                         my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
516                         write_file $permanent_path, { binmode => ':utf8' }, map { "$_\n" } @vals;
517                         warn "permanent filter $permanent_path ", -s $permanent_path;
518                 }
519         }
520
521         $self->session( 'offset' => 0 );
522
523         return $self->redirect_to('/data/items');
524 }
525
526 sub _filter_on_data {
527         my ( $self, $name, @vals ) = @_;
528
529         my $path = $self->session('path');
530
531         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
532                 warn "filter $name not found in data set";
533                 return;
534         }
535
536         $filters->{$name} = [ @vals ];
537         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
538
539         my $filter_hash;
540         $filter_hash->{$_}++ foreach @vals;
541
542         #warn "# filter_hash ",dump( $filter_hash );
543
544         my $items = $self->_loaded('data')->{items};
545
546         my $include_missing = defined $filter_hash->{_missing};
547         my $filtered_items;
548
549         foreach my $i ( 0 .. $#$items ) {
550
551                 if ( defined $items->[$i]->{$name} ) {
552                         my $row = $items->[$i]->{$name};
553                         $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
554                         foreach my $v ( @$row ) {
555                                 if ( defined $filter_hash->{ $v } ) {
556                                         $filtered_items->{$i}++;
557                                 }
558                         }
559                 } elsif ( $include_missing ) {
560                         $filtered_items->{$i}++;
561                 }
562         }
563
564         #warn "# filter $name ",dump($filtered_items);
565
566         # invalidate filters on other datasets
567         foreach my $dataset ( grep { exists $loaded->{$_}->{filters}->{$name} } keys %$loaded ) {
568                 delete $loaded->{$dataset}->{filters}->{$name};
569                 delete $loaded->{$dataset}->{filtered};
570         }
571
572         $loaded->{$path}->{filters}->{$name} = $filtered_items;
573         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
574 }
575
576
577 sub _current_filters {
578         my $self = shift;
579         my $current_filters;
580         my $stats = $self->_loaded('stats');
581
582         $current_filters->{ $_ } = $filters->{ $_ }
583                 foreach ( grep { defined $filters->{ $_ } } keys %$stats )
584         ;
585         warn "# _current_filters ",dump( keys %$current_filters );
586         return $current_filters;
587 }
588
589 sub _data_sorted_by {
590         my ( $self, $order ) = @_;
591
592         my $path = $self->session('path');
593
594         warn "_data_sorted_by $order from $path";
595
596         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
597                 return $loaded->{$path}->{sorted}->{$order};
598         }
599
600         my $data = $self->_loaded( 'data' );
601         my $numeric = $self->_is_numeric($order);
602         my $missing = $numeric ? 0 : '';
603         no warnings qw(numeric);
604         my $nr = 0;
605         my @sorted = map {
606                 $_->[0]
607         } sort {
608                 if ( $numeric ) {
609                         $a->[1] <=> $b->[1]
610                 } else {
611                         $a->[1] cmp $b->[1]
612                 }
613         } map {
614                 my $v;
615                 if ( ! exists $_->{$order} ) {
616                         $v = $missing;
617                 } elsif ( ref $_->{$order} eq 'ARRAY' ) {
618                         $v = join('', @{$_->{$order}});
619                 } else {
620                         $v = $_->{$order};
621                 }
622                 [ $nr++, $v ]
623         } @{ $data->{items} }
624         ;
625
626         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
627         #warn "# sorted ",dump( @sorted );
628
629         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
630 }
631
632
633 sub __all_filters {
634         my $order = pop @_;
635         join(',', sort(@_), 'order', $order);
636 }
637
638 our ($out, $key,$value);
639
640 our $lookup_path_col;
641 our $on;
642
643 sub __commit_begin {
644         warn "__commit_begin";
645         $lookup_path_col = undef;
646         $on = undef;
647 }
648
649 sub __commit_end {
650         warn "__commit_end";
651         $lookup_path_col = undef; # cleanup memory
652         $on = undef;
653 }
654
655 sub lookup {
656         warn "# lookup ",dump @_;
657         my ( $vals, $on_path, $on_col, $code, $stat_code ) = @_;
658         die "code is not sub{ ... } but ", dump $code unless ref $code eq 'CODE';
659
660         if ( ! exists $loaded->{$on_path} ) {
661                 my @possible_paths = grep { /\Q$on_path\E/ } keys %$loaded;
662                 die "more than one dataset available for '$on_path' ",dump @possible_paths if $#possible_paths > 0;
663                 $on_path = shift @possible_paths;
664                 warn "## fuzzy selected path $on_path";
665         }
666
667         my $items = $loaded->{$on_path}->{data}->{items} || die "no items for $on_path";
668
669         if ( ! exists $lookup_path_col->{$on_path}->{$on_col} ) {
670                 warn "create lookup_path_col $on_path $on_col";
671                 foreach my $i ( 0 .. $#$items ) {
672                         my $item = $items->[$i];
673                         if ( exists $item->{$on_col} ) {
674                                 if ( ref $item->{$on_col} eq 'ARRAY' ) {
675                                         foreach my $v ( @{ $item->{$on_col} } ) {
676                                                 push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
677                                         }
678                                 } elsif ( ! ref $item->{$on_col} ) { # scalar
679                                         my $v = $item->{$on_col};
680                                         push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
681                                 } else {
682                                         die "unknown type of ",dump $item->{$on_col};
683                                 }
684                         }
685                 }
686                 warn "XXX ",dump $lookup_path_col->{$on_path}->{$on_col} if $ENV{DEBUG};
687         }
688
689         my $stat;
690         $stat = Statistics::Descriptive::Full->new() if $stat_code;
691
692         foreach my $v ( ref $vals eq 'ARRAY' ? @$vals : ( $vals ) ) {
693                 foreach my $i ( @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} } ) {
694                         $on = $items->[$i];
695                         warn "XXX lookup code $v $i ",dump $on if $ENV{DEBUG};
696                         $code->($stat);
697                 }
698                 $stat_code->( $stat ) if $stat_code;
699         }
700 }
701
702 sub __commit_path_code {
703         my ( $path, $i, $code, $commit_changed ) = @_;
704
705         my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
706         my $row = $items->[$i];
707         my $update;
708         eval $code;
709         foreach ( keys %$update ) {
710                 $$commit_changed->{$_}++;
711                 $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
712         }
713         #warn "__commit_path_code $path $i ",dump( $update );
714 }
715
716 # uses templates/admin.html.ep
717 sub _switch_dataset {
718         my $self = shift;
719
720         my $datasets;
721
722         foreach my $path ( keys %$loaded ) {
723                 next unless exists $loaded->{$path}->{data};
724                 push @$datasets, $path;
725         }
726
727         warn "# datasets ",dump($datasets);
728
729         $self->stash( 'datasets' => $datasets );
730 }
731
732 sub items {
733         my $self = shift;
734
735         $self->_switch_dataset;
736
737         if ( my $show = $self->param('id') ) {
738                 $self->param('show', $show);
739                 warn "show $show\n";
740         }
741
742         my $path = $self->_param_scalar('path');
743
744         my @columns = $self->_param_array('columns');
745         return $self->redirect_to('/data/columns') unless @columns;
746         my $order   = $self->_param_scalar('order', $columns[0]);
747         my $sort    = $self->_param_scalar('sort', 'a');
748         my $offset  = $self->_param_scalar('offset', 0);
749         my $limit   = $self->_param_scalar('limit', 20);
750         $self->_param_scalar('show', 'table');
751
752         # fix offset when changing limit
753         $offset = int( $offset / $limit ) * $limit;
754
755         if ( ! grep { /^\Q$order\E$/ } @columns ) {
756                 $order = $columns[0];
757                 $self->session( order => $order );
758         }
759         my $sorted = $self->_data_sorted_by( $order );
760
761         my @filter_names;
762         if ( $filters ) {
763                 foreach my $name ( keys %$filters ) {
764                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
765                                 warn "skip filter $name not found in $path\n";
766                                 next;
767                         }
768                         push @filter_names, $name;
769                 }
770                 warn "filter_names ",dump( @filter_names );
771                 foreach my $name ( @filter_names ) {
772                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
773                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
774                 }
775         }
776
777         my $all_filters = __all_filters( @filter_names,$order );
778
779 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
780
781         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
782
783                 my $path_filters = $loaded->{$path}->{filters};
784
785                 warn "create combined filter for $all_filters from ", $#$sorted + 1, " items\n";
786
787                 my @filtered;
788                 foreach my $i ( 0 .. $#$sorted ) {
789                         my $pos = $sorted->[$i];
790
791                         if ( $#filter_names == -1 ) {
792                                 push @filtered, $pos;
793                                 next;
794                         }
795
796                         my $skip = 0;
797                         foreach ( @filter_names ) {
798                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
799                         }
800                         next if $skip;
801
802                         push @filtered, $pos;
803                 }
804
805                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
806         }
807
808         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
809                 if defined $loaded->{$path}->{filtered}->{$all_filters};
810
811         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
812
813         my $data = $self->_loaded('data');
814
815         my $code = $self->_param_scalar('code','');
816         $code =~ s{\r}{}gs;
817         $code =~ s{\n+$}{\n}s;
818
819         # XXX convert @row->{foo} into @{$row->{foo}}
820         $code =~ s|\@(row->\{[^}]+\})|\@{\$$1}|gs;
821
822         my $commit = $self->param('commit');
823         my $test = $self->param('test');
824
825         my $commit_changed;
826         __commit_begin;
827
828         if ( $code && ( $test || $commit ) ) {
829                 # XXX find columns used in code snippet and show them to user
830                 my $order = 0;
831                 foreach my $column ( $code =~ m/\$row->\{([^}]+)\}/g ) {
832                         if ( $column =~ s/^(['"])// ) {
833                                 $column =~ s/$1$//;
834                         }
835                         next if $column =~ m/\$/; # hide columns with vars in them
836                         $commit_changed->{$column} = 0;
837                 }
838         }
839
840         my $code_path = $self->app->home->rel_file('public') . "/code";
841         if ( $commit ) {
842
843                 __path_modified( $path, 'commit' );
844
845                 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
846                 ( $key, $value, $out ) = ( 'key', 'value' );
847                 foreach ( 0 .. $#$filtered ) {
848                         my $i = $filtered->[$_];
849                         __commit_path_code( $path, $i, $code, \$commit_changed );
850                 }
851
852                 # this might move before $out to recalculate stats on source dataset?
853                 __path_rebuild_stats( $path );
854                 my $c = { map { $_ => 1 } @columns };
855                 my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
856                 warn "# added_columns ",dump( @added_columns );
857                 unshift @columns, @added_columns;
858
859                 $loaded->{$path}->{columns} = [ @columns ];
860                 warn "# new columns ",dump( @columns );
861
862                 __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
863
864                 $self->_save_change({
865                         path => $path,
866                         time => $self->param('time') || time(),
867                         user => $self->param('user') || $ENV{'LOGNAME'},
868                         code => $code,
869                         commit_changed => $commit_changed,
870                 });
871
872                 if ( my $description = $self->param('code_description') ) {
873                         my $depends = $self->param('code_depends') || die "no code_depends?";
874                         my $path = "$code_path/$depends.$description.pl";
875                         if ( -e $path && ! $self->param('overwrite') ) {
876                                 warn "# code $path not saved\n";
877                         } else {
878                                 write_file(  $path, { binmode => ':utf8' }, "$code\n" );
879                                 warn "code $path ", -s $path, " bytes saved\n";
880                         }
881                 }
882
883                 # remove console
884                 $code = '';
885                 if ( $out ) {
886                         my $commit_dataset = join('.'
887                                 , $self->param('code_depends')
888                                 , $self->param('code_description')
889                                 , time()
890                         );
891                         $key ||= 'key';
892                         $value ||= 'value';
893                         warn "key $key value $value";
894                         my $items;
895                         foreach my $n ( keys %$out ) {
896                                 my $i = { $key => [ $n ] };
897                                 my $ref = ref $out->{$n};
898                                 if ( $ref eq 'HASH' ) {
899                                         $i->{$_} = [ $out->{$n}->{$_} ] foreach keys %{ $out->{$n} };
900                                 } elsif ( $ref eq 'ARRAY' ) {
901                                         $i->{$_} = $out->{$n};
902                                 } elsif ( ! $ref ) {
903                                         $i->{$value} = [ $out->{$n} ];
904                                 } else {
905                                         $i->{_error} = [ dump($out->{$n}) ];
906                                 }
907                                 push @$items, $i;
908                         };
909                         undef $out;
910                         my $stats = __stats( $items );
911                         my @columns = grep { ! m/^\Q$key\E$/ } sort keys %$stats;
912                         unshift @columns, $key;
913
914                         $loaded->{$commit_dataset} = {
915                                 header => [ @columns ],
916                                 columns => [ @columns ],
917                                 mtime => time(),
918                                 data => { items => $items },
919                                 stats => $stats,
920                                 generated => 1,
921                         };
922                         warn "# loaded out ", dump( $loaded->{$commit_dataset} );
923                         $self->session('path', $commit_dataset);
924                         $self->session('columns', [ @columns ]);
925                         $self->session('order', $key);
926                         return $self->redirect_to('/data/items');
927                 }
928
929                 $self->session('columns', [ @columns ]);
930         }
931
932         my $sorted_items;
933         my $from_end = $sort eq 'd' ? $#$filtered : 0;
934         my $test_changed;
935         my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
936         foreach ( 0 .. $limit ) {
937                 my $i = $_ + $offset;
938                 last unless defined $filtered->[$i];
939                 $i = $from_end - $i if $from_end;
940                 my $id = $filtered->[$i];
941                 my $row = Storable::dclone $data->{items}->[ $id ];
942                 if ( $code && $test ) {
943                         my $update;
944                         eval $code;
945                         if ( $@ ) {
946                                 warn "ERROR evaling $@", dump($code);
947                                 $self->stash('eval_error', $@) if $@;
948                         } else {
949                                 warn "EVAL ",dump($update);
950                                 foreach ( keys %$update ) {
951                                         $test_changed->{$_}++;
952                                         $row->{$_} = $update->{$_};
953                                 }
954                         }
955                 }
956                 $row->{_row_id} ||= $id;
957                 push @$sorted_items, $row;
958         }
959
960         if ( $self->param('export') ) {
961                 my $export_path = $self->_export_path( 'items', @columns);
962                 open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
963                 print $fh "#",join("\t",@columns),"\n";
964                 foreach my $f ( 0 .. $#$filtered ) {
965                         print $fh join("\t", map {
966                                 my $i = $data->{items}->[ $filtered->[$f] ];
967                                 my $v = '\N';
968                                 if ( ! defined $i->{$_} ) {
969                                         # nop
970                                 } elsif ( ref $i->{$_} eq 'ARRAY' ) {
971                                         $v =join(',', @{ $i->{$_} });
972                                         $v = '\N' if length($v) == 0;
973                                 } elsif ( ! ref $i->{$_} ) {
974                                         $v = $i->{$_};
975                                 } else {
976                                         $v = dump $i->{$_};
977                                 }
978                                 $v;
979                         } @columns),"\n";
980                 }
981                 close($fh);
982                 warn "export $export_path ", -s $export_path, " bytes\n";
983         }
984
985         my ( $code_depends, $code_description );
986
987         if ( $test ) {
988
989                 warn "# test_changed ",dump( $test_changed );
990                 my $c = { map { $_ => 1 } @columns };
991                 my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
992                 unshift @columns, @added_columns;
993
994                 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
995
996                 my $depends_on;
997                 my $tmp = $code; $tmp =~ s/\$row->\{(['"]?)([\w\s]+)\1/$depends_on->{$2}++/gse;
998                 warn "# depends_on ",dump $depends_on;
999
1000                 my $test_added = Storable::dclone $test_changed;
1001                 delete $test_added->{$_} foreach keys %$depends_on;
1002
1003                 $code_depends = $self->param('code_depends')
1004                 || join(',', keys %$depends_on);
1005
1006                 $code_description = $self->param('code_description') ||
1007                 join(',', keys %$test_added);
1008
1009                 $code_depends ||= $code_description; # self-modifing
1010                 if ( ! $code_depends && $out ) {
1011                         $code_depends = $key;
1012                         $code_description = $value;
1013                 }
1014
1015                 warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
1016
1017         } # test?
1018
1019         __commit_end;
1020
1021         $self->render(
1022                 order => $order,
1023                 offset => $offset,
1024                 limit => $limit,
1025                 sorted => $sorted_items,
1026                 columns => [ @columns ],
1027                 rows => $#$filtered + 1,
1028                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
1029                 unique  => { map { $_, $self->_is_unique( $_) } @columns },
1030                 filters => $self->_current_filters,
1031                 code => $code,
1032                 cols_changed => $commit ? $commit_changed : $test_changed,
1033                 code_depends => $code_depends,
1034                 code_description => $code_description,
1035                 code_path => $code_path,
1036                 out => $out,
1037         );
1038
1039 }
1040
1041
1042 sub order {
1043         my $self = shift;
1044         $self->session('order', $self->param('order'));
1045         $self->session('sort', $self->param('sort'));
1046         return $self->redirect_to('/data/items');
1047 }
1048
1049 sub _is_numeric {
1050         my ( $self, $name ) = @_;
1051
1052         my $stats = $self->_loaded( 'stats' );
1053
1054         # sort facet numerically if more >50% elements are numeric
1055         my $count = $stats->{$name}->{count};
1056         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
1057         defined $stats->{$name}->{numeric} &&
1058                 $stats->{$name}->{numeric} > $count / 2;
1059 }
1060
1061 sub _is_unique {
1062         my ( $self, $name ) = @_;
1063         my $stats = $self->_loaded( 'stats' );
1064         defined $stats->{$name}->{unique};
1065 }
1066
1067 sub _remove_filter {
1068         my ($self,$name) = @_;
1069         warn "_remove_filter $name\n";
1070
1071         my $path = $self->session('path');
1072
1073         delete $filters->{$name};
1074         delete $loaded->{$path}->{filters}->{$name};
1075         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
1076
1077         foreach (
1078                         grep { /\Q$name\E/ }
1079                         keys %{ $loaded->{$path}->{filtered} }
1080         ) {
1081                 delete $loaded->{$path}->{filtered}->{$_};
1082                 warn "remove filtered cache $_";
1083         }
1084 }
1085
1086 sub facet {
1087         my $self = shift;
1088
1089         my $path = $self->session('path') || return $self->redirect_to( '/data/index' );
1090
1091         if ( my $name = $self->param('remove') ) {
1092                 $self->_remove_filter( $name );
1093                 return $self->redirect_to( '/data/items' );
1094         }
1095
1096         my $facet;
1097         my $name = $self->param('name') || die "no name";
1098
1099         my $all = $self->_param_scalar('all', 1);
1100         my $data = $self->_loaded('data');
1101
1102         my $filters = $self->_current_filters;
1103         my $all_filters = __all_filters( keys %$filters,$self->session('order') );
1104         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
1105                 if defined $loaded->{$path}->{filtered}->{$all_filters};
1106
1107         if ( ! $filtered || $all ) {
1108                 $filtered = [ 0 .. $#{ $data->{items} } ];
1109                 warn "filter all values\n";
1110         } else {
1111                 warn "filter using $all_filters\n";
1112         }
1113
1114         foreach my $i ( @$filtered ) {
1115                 my $item = $data->{items}->[$i];
1116                 if ( ! exists $item->{$name} || ! defined $item->{$name} ) {
1117                         $facet->{ _missing }++;
1118                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
1119                         $facet->{$_}++ foreach @{ $item->{$name} };
1120                 } else {
1121                         $facet->{ $item->{$name} }++;
1122                 }
1123         }
1124
1125         my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
1126
1127         if ( my $code = $self->param('code') ) {
1128                 my $out;
1129                 foreach my $value ( keys %$facet ) {
1130                         my $count = $facet->{$value};
1131                         my $checked = $checked_values->{$value};
1132                         eval $code;
1133                         if ( $@ ) {
1134                                 $out = $@;
1135                                 warn "ERROR: $@\n$code\n";
1136                                 last;
1137                         } elsif ( $checked != $checked_values->{$value} ) {
1138                                 warn "checked $value $count -> $checked\n";
1139                                 $checked_values->{$value} = $checked;
1140                         }
1141                 }
1142                 warn "out ",dump( $out );
1143                 $self->stash( out => $out );
1144         }
1145
1146 #       warn "# facet $name ",dump $facet;
1147
1148         my @facet_names =
1149                   $all                      ? keys %$facet
1150                 : defined $filters->{$name} ? @{ $filters->{$name} }
1151                 : keys %$facet;
1152
1153         my $numeric = $self->_is_numeric($name);
1154
1155         my $sort = $self->param('sort');
1156         # sort numeric facets with more than 5 values ascending
1157         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
1158
1159         @facet_names = sort {
1160                 my $result;
1161                 if ( $sort eq 'a' ) {
1162                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
1163                 } elsif ( $sort eq 'd' ) {
1164                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
1165                 } elsif ( $sort eq 'c' ) {
1166                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
1167                 } else {
1168                         warn "unknown sort: $sort";
1169                         $result = $a cmp $b;
1170                 }
1171                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
1172                 $result;
1173         } @facet_names;
1174
1175         $self->render( name => $name, facet => $facet, checked => $checked_values,
1176                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
1177         );
1178 }
1179
1180
1181 sub __invalidate_path_column {
1182         my ( $path, $name ) = @_;
1183
1184         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
1185                 delete $loaded->{$path}->{sorted}->{$name};
1186                 warn "# invalidate $path sorted $name\n";
1187         }
1188
1189         foreach ( grep { m/\Q$name\E/ } keys %{ $loaded->{$path}->{filtered} } ) {
1190                 delete $loaded->{$path}->{filtered}->{$_};
1191                 warn "# invalidate $path filtered $_\n";
1192         }
1193 }
1194
1195 sub __path_modified {
1196         my ( $path, $value ) = @_;
1197         $value = 1 unless defined $value;
1198         
1199         $loaded->{$path}->{modified}  = $value;
1200
1201         warn "# __path_modified $path $value\n";
1202 }
1203
1204 sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
1205
1206 sub _save_change {
1207         my ($self,$change) = @_;
1208
1209         my $change_path = $self->_permanent_path( 'changes' );
1210         mkdir $change_path unless -d $change_path;
1211         $change_path .= '/' . $change->{time};
1212         store $change, $change_path;
1213         utime $change->{time}, $change->{time}, $change_path;
1214         warn "_save_change $change_path ", dump($change);
1215 }
1216
1217 sub edit {
1218         my $self = shift;
1219         my $new_content = $self->param('new_content');
1220         $new_content  ||= $self->param('content'); # backward compatibility with old actions
1221
1222         my $i = $self->param('_row_id');
1223         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
1224         my $path = $self->param('path') || die "no path";
1225         my $name = $self->param('name') || die "no name";
1226         my $status = 200; # 200 = OK, 201 = Created
1227
1228         my $data = $self->_loaded('data');
1229
1230         if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
1231                 $new_content =~ s/^\s+//s;
1232                 $new_content =~ s/\s+$//s;
1233                 my $v;
1234                 if ( $new_content =~ /\xB6/ ) { # para
1235                         $v = [ split(/\s*\xB6\s*/, $new_content) ];
1236                 } else {
1237                         $v = [ $new_content ];
1238                 }
1239
1240                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
1241                 my $new = dump $v;
1242                 if ( $old ne $new
1243                         && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
1244                 ) {
1245                         $self->_save_change({
1246                                 path => $path,
1247                                 column => $name,
1248                                 pos => $i,
1249                                 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
1250                                 new => $v,
1251                                 time => $self->param('time') || time(),
1252                                 user => $self->param('user') || $ENV{'LOGNAME'},
1253                                 unique => {
1254                                         map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
1255                                         grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
1256                                         keys %{ $loaded->{$path}->{stats} }
1257                                 },
1258                         });
1259
1260                         warn "# change $path $i $old -> $new\n";
1261                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
1262
1263                         __invalidate_path_column( $path, $name );
1264
1265                         $status = 201; # created
1266                         __path_rebuild_stats( $path );
1267         
1268                         $new_content = join("\xB6",@$v);
1269
1270                 } else {
1271                         warn "# unchanged $path $i $old\n";
1272                         $status = 304;
1273                 }
1274         } else {
1275                 $new_content = "$path $i $name doesn't exist\n";
1276                 $status = 404;
1277         }
1278
1279         warn "# edit $status ", dump $new_content;
1280
1281         $self->render(
1282                 status => $status,
1283                 new_content => scalar $new_content,
1284         );
1285 }
1286
1287
1288 sub save {
1289         my $self = shift;
1290         my $path = $self->_param_or_session('path');
1291         my $dump_path = $self->_save( $path );
1292         __path_modified( $path, 0 );
1293
1294         return $self->redirect_to( '/data/items' );
1295 }
1296
1297 sub export {
1298         my $self = shift;
1299
1300         my $dir = $self->app->home->rel_file('public');
1301
1302         if ( my $import = $self->param('import') ) {
1303
1304                 if ( $import =~ m{/filter\.(.+?)\..+} ) {
1305                         my $name = $self->_column_from_unac( $1 );
1306
1307                         my @vals = map { chomp; $_ }
1308                                 read_file "$dir/export/$import", binmode => ':utf8';
1309
1310                         $self->_remove_filter( $name );
1311                         $self->_filter_on_data( $name, @vals );
1312                         $self->session( 'offset' => 0 );
1313                         return $self->redirect_to('/data/items');
1314                 } else {
1315                         warn "UNKNOWN IMPORT $import";
1316                 }
1317         }
1318
1319         if ( my $remove = $self->param('remove') ) {
1320                 my $path = "$dir/export/$remove";
1321                 unlink $path if -e $path;
1322                 $path .= '.png';
1323                 unlink $path if -e $path;
1324         }
1325
1326         my $path = $self->_export_path || return $self->redirect_to('/data/index');
1327
1328         my @files = grep { ! /\.png$/ } glob "$path/*";
1329         my $mtime = { map { $_ => (stat($_))[9] } @files };
1330         @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
1331         $self->render( export => [ @files ] );
1332 }
1333
1334 sub __loaded_paths {
1335         return
1336                 grep { defined $loaded->{$_}->{data} }
1337                 keys %$loaded;
1338 }
1339
1340 sub reload {
1341         my $self = shift;
1342         $self->stash( reload => 1 );
1343         $self->remove;
1344 #       $self->_load_path( $self->param('path') );
1345         $self->redirect_to('/data/load?path=' . $self->param('path') );
1346 }
1347
1348 sub remove {
1349         my $self = shift;
1350         my $path = $self->param('path');
1351         if ( $path =~ m{^/tmp/mojo_facets\.} ) {
1352                 unlink $path;
1353                 warn "# unlink $path";
1354                 if ( my $name = $self->param('name') ) {
1355                         delete $loaded->{$name};
1356                         warn "# remove $name from memory";
1357                 }
1358         } else {
1359                 warn "WARNING: $path unlink ignored";
1360         }
1361         return if $self->stash('reload');
1362         return $self->redirect_to( '/data/load' );
1363 }
1364
1365 1;