reload file if changed
[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 JSON;
11 use Encode;
12 use locale;
13 use File::Find;
14 use Storable;
15
16 our $loaded;
17 our $filters;
18
19 sub index {
20         my $self = shift;
21
22         my $path = $self->app->home->rel_dir('data');
23         die "no data dir $path" unless -d $path;
24
25         my @files;
26         find( sub {
27                 my $file = $File::Find::name;
28                 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
29                         $file =~ s/$path\/*//;
30                         push @files, $file;
31                 } else {
32                         warn "IGNORE: $file\n";
33                 }
34         }, $path);
35
36         @files = sort { lc $a cmp lc $b } @files;
37         my $size;
38         $size->{$_} = -s "$path/$_" foreach @files;
39
40         $self->render(
41                 files => [ @files ],
42                 size => $size,
43                 loaded => $loaded,
44                 filters => $filters,
45         );
46 }
47
48 sub _dump_path {
49         my ( $self, $name ) = @_;
50         my $dir = $self->app->home->rel_dir('data');
51         $name =~ s/^$dir//;
52         $name =~ s/\/+/_/g;
53         return '/tmp/mojo_facets.' . $name . '.storable';
54 }
55
56 sub _save {
57         my ( $self, $path ) = @_;
58
59         my $dump_path = $self->_dump_path( $path );
60         warn "save loaded to $dump_path";
61         my $info = $loaded->{$path};
62         store $info, $dump_path;
63
64         # sync timestamp
65         my $mtime = $loaded->{$path}->{mtime};
66         utime $mtime, $mtime, $dump_path;
67
68         warn $dump_path, ' ', -s $dump_path, " bytes\n";
69         return $dump_path;
70 }
71
72 sub _load_path {
73         my ( $self, $path ) = @_;
74
75         my $full_path = $self->app->home->rel_file( 'data/' . $path );
76         die "$full_path $!" unless -r $full_path;
77
78         my $dump_path = $self->_dump_path( $path );
79
80         if ( defined $loaded->{$path}->{data} ) {
81                 my $mtime = (stat($full_path))[9];
82                 return if $loaded->{$path}->{mtime} == $mtime;
83                 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
84         } elsif ( -e $dump_path ) {
85                 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
86                 my $info = retrieve $dump_path;
87                 $loaded->{ $path } = $info;
88                 return;
89         }
90
91         # we could use Mojo::JSON here, but it's too slow
92 #       $data = from_json read_file $path;
93         my $data = read_file $full_path;
94         warn "# data snippet: ", substr($data,0,200);
95         my @header;
96         if ( $path =~ m/\.js/ ) {
97                 Encode::_utf8_on($data);
98                 $data = from_json $data;
99         } elsif ( $path =~ m/\.txt/ ) {
100                 my @lines = split(/\r?\n/, $data);
101                 $data = { items => [] };
102
103                 my $header_line = shift @lines;
104                 my $multiline = $header_line =~ s/\^//g;
105                 @header = split(/\|/, $header_line );
106                 warn "# header ", dump( @header );
107                 while ( my $line = shift @lines ) {
108                         $line =~ s/\^//g;
109                         chomp $line;
110                         my @v = split(/\|/, $line);
111                         while ( @lines && $#v < $#header ) {
112                                 $line = $lines[0];
113                                 $line =~ s/\^//g;
114                                 chomp $line;
115                                 my @more_v = split(/\|/, $line);
116                                 if ( $#v + $#more_v > $#header ) {
117                                         warn "short line: ",dump( @v );
118                                         last;
119                                 }
120                                 shift @lines;
121                                 $v[ $#v ] .= shift @more_v if @more_v;
122                                 push @v, @more_v if @more_v;
123
124                                 if ( $#v > $#header ) {
125                                         die "# splice $#header ", dump( @v );
126                                         @v = splice @v, 0, $#header;
127                                 }
128                         }
129                         my $item;
130                         foreach my $i ( 0 .. $#v ) {
131                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
132                         }
133                         push @{ $data->{items} }, $item;
134                 }
135         } else {
136                 warn "file format unknown $path";
137         }
138
139         my $stats;
140
141         foreach my $e ( @{ $data->{items} } ) {
142                 foreach my $n ( keys %$e ) {
143                         $stats->{$n}->{count}++;
144                         my @v;
145                         if ( ref $e->{$n} eq 'ARRAY' ) {
146                                 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
147                                 @v = @{ $e->{$n} };
148                         } else {
149                                 @v = ( $e->{$n} );
150                         }
151
152                         foreach my $x ( @v ) {
153                                 $stats->{$n}->{numeric}++
154                                         if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
155                                 $stats->{$n}->{empty}++
156                                         if length $x == 0; # faster than $x =~ m/^\s*$/;
157                         }
158
159                 }
160         }
161
162         foreach my $n ( keys %$stats ) {
163                 next unless defined $stats->{$n}->{array};
164                 delete $stats->{$n}->{array}
165                         if $stats->{$n}->{array} == $stats->{$n}->{count};
166         }
167
168         if ( ! @header ) {
169                 if ( defined $data->{header} ) {
170                         if ( ref $data->{header} eq 'ARRAY' ) {
171                                 @header = @{ $data->{header} };
172                         } else {
173                                 warn "header not array ", dump( $data->{header} );
174                         }
175                 }
176         }
177
178         @header =
179                 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
180                 grep { defined $stats->{$_}->{count} } keys %$stats
181                 unless @header;
182
183         warn dump($stats);
184
185         my $info = {
186                 header => [ @header ],
187                 stats  => $stats,
188                 full_path => $full_path,
189                 size => -s $full_path,
190                 mtime => (stat($full_path))[9],
191                 data => $data,
192         };
193
194         $loaded->{ $path } = $info;
195         $self->_save( $path );
196
197 }
198
199
200 sub load {
201         my $self = shift;
202
203         my @paths = $self->param('paths');
204         warn "# paths ", dump @paths;
205         $self->_load_path( $_ ) foreach @paths;
206
207         my $path = $self->param('path') || $self->redirect_to( '/data/index' );
208         warn "# path $path\n";
209         $self->session('path' => $path);
210         $self->_load_path( $path );
211
212         $self->session( 'header' => $loaded->{$path}->{header} );
213         if ( ! defined $loaded->{$path}->{columns} ) {
214                 $self->session( 'columns' => $loaded->{$path}->{header} );
215                 $self->session( 'order'   => $loaded->{$path}->{header}->[0] );
216                 $self->redirect_to( '/data/columns' );
217         } else {
218                 $self->session( 'columns' => $loaded->{$path}->{columns} );
219                 $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
220                 $self->redirect_to( '/data/items' );
221         }
222 }
223
224
225 sub _loaded {
226         my ( $self, $name ) = @_;
227         my $path = $self->session('path');
228         $self->redirect_to('/data/index') unless $path;
229         if ( ! defined $loaded->{$path}->{$name} ) {
230                 warn "$path $name doesn't exist in loaded ",dump( $loaded );
231                 $self->redirect_to('/data/index');
232         }
233         return $loaded->{$path}->{$name};
234 }
235
236
237 sub _checked {
238         my $self = shift;
239         my $checked;
240         $checked->{$_}++ foreach @_;
241 #       warn "# _checked ",dump($checked);
242         return $checked;
243 }
244
245
246 sub columns {
247     my $self = shift;
248
249         if ( $self->param('columns') ) {
250                 $self->_perm_array('columns');
251                 $self->redirect_to('/data/items');
252         }
253
254         my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
255
256         my @columns;
257         @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
258
259         foreach my $c ( @{ $self->session( 'header' ) } ) {
260                 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
261         }
262
263     $self->render(
264                 message => 'Select columns to display',
265                 stats => $stats,
266                 columns => \@columns,
267                 checked => $self->_checked( $self->_perm_array('columns') ),
268         );
269 }
270
271 sub _perm_array {
272     my ($self,$name) = @_;
273
274         my @array = $self->param($name);
275         my $path  = $self->session('path');
276
277         if ( @array ) {
278                 $self->session($name => [ @array ]);
279                 $loaded->{$path}->{$name} = [ @array ];
280         } elsif ( my $session = $self->session($name) ) {
281                 if ( ref $session eq 'ARRAY' ) {
282                         @array = @$session;
283                 } else {
284                         die "$name not array ",dump($session);
285                 }
286         }
287         #warn "# $name ",dump @array;
288         return @array;
289 }
290
291 sub _perm_scalar {
292     my ($self,$name,$default) = @_;
293
294         my $scalar = $self->param($name);
295
296         if ( defined $scalar ) {
297                 $self->session($name => $scalar);
298         } else {
299                 $scalar = $self->session($name);
300         }
301
302         if ( ! defined $scalar ) {
303                 $scalar = $default;
304                 die "no default for $name" unless defined $scalar;
305                 $self->session($name => $scalar);
306         }
307
308         warn "# _perm_scalar $name ",dump $scalar;
309         return $scalar;
310 }
311
312 sub filter {
313         my $self = shift;
314
315         my $name = $self->param('filter_name') || die "name?";
316         my @vals = $self->param('filter_vals');
317
318         $self->_remove_filter( $name );
319         $self->_filter_on_data( $name, @vals ) if @vals;
320
321         $self->session( 'offset' => 0 );
322
323         $self->redirect_to('/data/items');
324 }
325
326 sub _filter_on_data {
327         my ( $self, $name, @vals ) = @_;
328
329         my $path = $self->session('path');
330
331         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
332                 warn "filter $name not found in data set";
333                 return;
334         }
335
336         $filters->{$name} = [ @vals ];
337         warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
338
339         my $filter_hash;
340         $filter_hash->{$_}++ foreach @vals;
341
342         #warn "# filter_hash ",dump( $filter_hash );
343
344         my $items = $self->_loaded('data')->{items};
345
346         my $include_missing = defined $filter_hash->{_missing};
347         my $filtered_items;
348
349         foreach my $i ( 0 .. $#$items ) {
350
351                 if ( defined $items->[$i]->{$name} ) {
352                         foreach my $v ( @{ $items->[$i]->{$name} } ) {
353                                 if ( defined $filter_hash->{ $v } ) {
354                                         $filtered_items->{$i}++;
355                                 }
356                         }
357                 } elsif ( $include_missing ) {
358                         $filtered_items->{$i}++;
359                 }
360         }
361
362         #warn "# filter $name ",dump($filtered_items);
363
364         $loaded->{$path}->{filters}->{$name} = $filtered_items;
365         warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
366 }
367
368 sub _data_items {
369         my ( $self, $all ) = @_;
370         my $data = $self->_loaded( 'data' );
371
372         return @{ $data->{items} } if $all == 1;
373
374         my $filters = $self->_current_filters;
375         my $filter_value;
376         foreach my $f ( keys %$filters ) {
377                 foreach my $n ( @{ $filters->{$f} } ) {
378                         $filter_value->{$f}->{$n} = 1;
379                 }
380         }
381         my @items = @{ $data->{items} };
382         @items = grep {
383                 my $i = $_;
384                 my $pass = 1;
385                 foreach my $n ( keys %$filter_value ) {
386                         if ( ! exists $i->{$n} ) {
387                                 if ( defined $filter_value->{$n}->{_missing} ) {
388                                         $pass = 1;
389                                         next;
390                                 } else {
391                                         $pass = 0;
392                                         last;
393                                 }
394                         }
395                         # and match any of values in element
396                         my $have_values = 0;
397                         foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
398                                 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
399                         }
400                         if ( ! $have_values ) {
401                                 $pass = 0;
402                                 last;
403                         }
404                 }
405                 $pass;
406         } @items if $filter_value;
407         return @items;
408 }
409
410
411 sub _current_filters {
412         my $self = shift;
413         my $current_filters;
414         $current_filters->{ $_ } = $filters->{ $_ }
415                 foreach (
416                         grep { defined $filters->{ $_ } }
417                         @{ $self->_loaded('header') }
418                 );
419         #warn "# current_filters ",dump($current_filters);
420         return $current_filters;
421 }
422
423 sub _data_sorted_by {
424         my ( $self, $order ) = @_;
425
426         my $path = $self->session('path');
427
428         warn "_data_sorted_by $order from $path";
429
430         if ( defined $loaded->{$path}->{sorted}->{$order} ) {
431                 return $loaded->{$path}->{sorted}->{$order};
432         }
433
434         my $data = $self->_loaded( 'data' );
435         my $numeric = $self->_is_numeric($order);
436         my $missing = $numeric ? 0 : '';
437         no warnings qw(numeric);
438         my $nr = 0;
439         my @sorted = map {
440                 $_->[0]
441         } sort {
442                 if ( $numeric ) {
443                         $a->[1] <=> $b->[1]
444                 } else {
445                         $a->[1] cmp $b->[1]
446                 }
447         } map {
448                 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
449         } @{ $data->{items} }
450         ;
451
452         warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
453         #warn "# sorted ",dump( @sorted );
454
455         $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
456 }
457
458
459 sub items {
460         my $self = shift;
461
462         my $path = $self->session('path');
463         $self->redirect_to('/data/index') unless defined $loaded->{ $path };
464
465         my @columns = $self->_perm_array('columns');
466         $self->redirect_to('/data/columns') unless @columns;
467         my $order   = $self->_perm_scalar('order', $columns[0]);
468         my $sort    = $self->_perm_scalar('sort', 'a');
469         my $offset  = $self->_perm_scalar('offset', 0);
470         my $limit   = $self->_perm_scalar('limit', 20);
471         $self->_perm_scalar('show', 'table');
472
473         # fix offset when changing limit
474         $offset = int( $offset / $limit ) * $limit;
475
476         my $sorted = $self->_data_sorted_by( $order );
477
478         my @filter_names;
479         if ( $filters ) {
480                 foreach my $name ( keys %$filters ) {
481                         if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
482                                 warn "skip filter $name not found in $path\n";
483                                 next;
484                         }
485                         push @filter_names, $name;
486                 }
487                 warn "filter_names ",dump( @filter_names );
488                 foreach my $name ( @filter_names ) {
489                         next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
490                         $self->_filter_on_data( $name, @{ $filters->{$name} } );
491                 }
492         }
493
494         my $all_filters = join(' ',sort @filter_names,'order:',$order);
495
496 #       warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
497
498         if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
499
500                 my $path_filters = $loaded->{$path}->{filters};
501
502                 warn "create combined filter for $all_filters\n";
503
504                 my @filtered;
505                 foreach my $i ( 0 .. $#$sorted ) {
506                         my $pos = $sorted->[$i];
507
508                         if ( $#filter_names == -1 ) {
509                                 push @filtered, $pos;
510                                 next;
511                         }
512
513                         my $skip = 0;
514                         foreach ( @filter_names ) {
515                                 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
516                         }
517                         next if $skip;
518
519                         push @filtered, $pos;
520                 }
521
522                 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
523         }
524
525         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
526                 if defined $loaded->{$path}->{filtered}->{$all_filters};
527
528         warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
529
530         my $sorted_items;
531         my $data = $self->_loaded('data');
532         my $from_end = $sort eq 'd' ? $#$filtered : 0;
533         foreach ( 0 .. $limit ) {
534                 my $i = $_ + $offset;
535                 last unless defined $filtered->[$i];
536                 $i = $from_end - $i if $from_end;
537                 my $id = $filtered->[$i];
538                 push @$sorted_items,
539                 my $item = $data->{items}->[ $id ];
540                 $item->{_row_id} ||= $id;
541         }
542
543         warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
544
545         $self->render(
546                 order => $order,
547                 offset => $offset,
548                 limit => $limit,
549                 sorted => $sorted_items,
550                 columns => [ @columns ],
551                 rows => $#$filtered + 1,
552                 numeric => { map { $_, $self->_is_numeric($_) } @columns },
553                 filters => $self->_current_filters,
554         );
555
556 }
557
558
559 sub order {
560         my $self = shift;
561         $self->session('order', $self->param('order'));
562         $self->session('sort', $self->param('sort'));
563         $self->redirect_to('/data/items');
564 }
565
566 sub _is_numeric {
567         my ( $self, $name ) = @_;
568
569         my $stats = $self->_loaded( 'stats' );
570
571         # sort facet numerically if more >50% elements are numeric
572         my $count = $stats->{$name}->{count};
573         $count   -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
574         defined $stats->{$name}->{numeric} &&
575                 $stats->{$name}->{numeric} > $count / 2;
576 }
577
578 sub _remove_filter {
579         my ($self,$name) = @_;
580         warn "_remove_filter $name\n";
581
582         my $path = $self->session('path');
583
584         delete $filters->{$name};
585         delete $loaded->{$path}->{filters}->{$name};
586         warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
587
588         foreach (
589                         grep { /\b$name\b/ }
590                         keys %{ $loaded->{$path}->{filtered} }
591         ) {
592                 delete $loaded->{$path}->{filtered}->{$_};
593                 warn "remove filtered cache $_";
594         }
595 }
596
597 sub facet {
598         my $self = shift;
599
600         my $path = $self->session('path') || $self->redirect_to( '/data/index' );
601
602         if ( my $name = $self->param('remove') ) {
603                 $self->_remove_filter( $name );
604                 $self->redirect_to( '/data/items' );
605         }
606
607         my $facet;
608         my $name = $self->param('name') || die "no name";
609
610         my $all = $self->_perm_scalar('all', 1);
611         my $data = $self->_loaded('data');
612
613         my $filters = $self->_current_filters;
614         my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
615         my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
616                 if defined $loaded->{$path}->{filtered}->{$all_filters};
617
618         if ( ! $filtered || $all ) {
619                 $filtered = [ 0 .. $#{ $data->{items} } ];
620                 warn "filter all values\n";
621         } else {
622                 warn "filter using $all_filters\n";
623         }
624
625         foreach my $i ( @$filtered ) {
626                 my $item = $data->{items}->[$i];
627                 if ( ! exists $item->{$name} ) {
628                         $facet->{ _missing }++;
629                 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
630                         $facet->{$_}++ foreach @{ $item->{$name} };
631                 } else {
632                         $facet->{ $item->{$name} }++;
633                 }
634         }
635
636 #       warn "# facet $name ",dump $facet;
637
638         my $checked;
639         my @facet_names =
640                   $all                      ? keys %$facet
641                 : defined $filters->{$name} ? @{ $filters->{$name} }
642                 : keys %$facet;
643
644         $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
645
646         my $numeric = $self->_is_numeric($name);
647
648         my $sort = $self->param('sort');
649         # sort numeric facets with more than 5 values ascending
650         $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
651
652         @facet_names = sort {
653                 my $result;
654                 if ( $sort eq 'a' ) {
655                         $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
656                 } elsif ( $sort eq 'd' ) {
657                         $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
658                 } elsif ( $sort eq 'c' ) {
659                         $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
660                 } else {
661                         warn "unknown sort: $sort";
662                         $result = $a cmp $b;
663                 }
664                 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
665                 $result;
666         } @facet_names;
667
668         $self->render( name => $name, facet => $facet, checked => $checked,
669                 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
670         );
671 }
672
673 sub edit {
674         my $self = shift;
675         my $content = $self->param('content');
676
677         my $i = $self->param('_row_id');
678         die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
679         my $path = $self->param('path') || die "no path";
680         my $name = $self->param('name') || die "no name";
681         my $status = 200; # 200 = OK, 201 = Created
682
683         if ( defined $loaded->{$path}->{data}->{items}->[$i]->{$name} ) {
684                 $content =~ s/^\s+//s;
685                 $content =~ s/\s+$//s;
686                 my $v;
687                 if ( $content =~ /\xB6/ ) {     # para
688                         $v = [ split(/\s*\xB6\s*/, $content) ];
689                 } else {
690                         $v = [ $content ];
691                 }
692
693                 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
694                 my $new = dump $v;
695                 if ( $old ne $new ) {
696                         warn "# update $path $i $old -> $new\n";
697                         $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
698
699                         if ( defined $loaded->{$path}->{sorted}->{$name} ) {
700                             delete $loaded->{$path}->{sorted}->{$name};
701                                 warn "# invalidate $path sorted $name\n";
702                         }
703
704                         foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
705                             delete $loaded->{$path}->{filtered}->{$_};
706                                 warn "# invalidate $path filtered $_\n";
707                         }
708
709                         $status = 201; # created
710                         $self->session('save_path' => $path);
711
712                 } else {
713                         warn "# unchanged $path $i $old\n";
714                         $status = 304;
715                 }
716         } else {
717                 $content = "$path $i $name doesn't exist\n";
718                 $status = 404;
719                 warn "# $content\n";
720         }
721
722         $self->render(
723                 status => $status,
724                 content => $content,
725         );
726 }
727
728
729 sub save {
730         my $self = shift;
731         my $path = $self->param('path');
732         $path  ||= $self->session('path');
733
734         my $dump_path = $self->_save( $path );
735         $self->session('save_path' => 0);
736
737         $self->redirect_to( '/data/items' );
738 }
739
740 1;