1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
14 use Time::HiRes qw(time);
16 use MojoFacets::Import::File;
17 use MojoFacets::Import::HTMLTable;
25 my $data_dir = $self->app->home->rel_dir('data');
26 die "no data dir $data_dir" unless -d $data_dir;
31 my $file = $File::Find::name;
32 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
33 $file =~ s/$data_dir\/*//;
35 } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
36 push @{ $changes->{$1} }, $2
37 } elsif ( -d $file && $file =~ m/\.html$/ ) {
38 $file =~ s/$data_dir\/*//;
41 warn "IGNORE: $file\n";
45 @files = sort { lc $a cmp lc $b } @files;
47 $size->{$_} = -s "$data_dir/$_" foreach @files;
54 dump_path => { map { $_ => $self->_dump_path($_) } @files },
60 my ( $self, $name ) = @_;
61 my $dir = $self->app->home->rel_dir('data');
64 return '/tmp/mojo_facets.' . $name . '.storable';
68 my ( $self, $path ) = @_;
70 my $dump_path = $self->_dump_path( $path );
71 my $first_load = ! -e $dump_path;
72 warn "save loaded to $dump_path";
73 my $info = $loaded->{$path};
74 store $info, $dump_path;
77 my $mtime = $loaded->{$path}->{mtime};
78 utime $mtime, $mtime, $dump_path;
79 warn "sync time to $path at $mtime\n";
82 warn $dump_path, ' ', -s $dump_path, " bytes\n";
91 my $nr_items = $#{ $_[0] } + 1;
93 warn "__stats $nr_items\n";
95 foreach my $e ( @{ $_[0] } ) {
96 foreach my $n ( keys %$e ) {
97 $stats->{$n}->{count}++;
99 if ( ref $e->{$n} eq 'ARRAY' ) {
100 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
106 foreach my $x ( @v ) {
107 $stats->{$n}->{numeric}++
108 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
109 $stats->{$n}->{empty}++
110 if length $x == 0; # faster than $x =~ m/^\s*$/;
116 foreach my $n ( keys %$stats ) {
117 my $s = $stats->{$n};
118 next unless defined $s->{array};
119 if ( $s->{array} == $s->{count} ) {
121 if ( $s->{count} == $nr_items ) {
122 warn "check $n for uniqeness\n";
124 foreach my $e ( @{ $_[0] } ) {
125 if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
131 $stats->{$n}->{unique} = 1;
132 warn "# $n unique ",dump( $unique );
138 warn "# __stats ",dump($stats);
143 sub _param_or_session {
144 $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
149 my $path = $self->_param_or_session('path');
150 warn "stats $path\n";
151 delete $loaded->{$path}->{stats};
152 $self->redirect_to( '/data/columns' );
157 my ( $self, $path ) = @_;
159 my $full_path = $self->app->home->rel_file( 'data/' . $path );
160 die "$full_path $!" unless -r $full_path;
162 my $dump_path = $self->_dump_path( $path );
164 if ( defined $loaded->{$path}->{data} ) {
165 my $mtime = (stat($full_path))[9];
166 return if $loaded->{$path}->{mtime} == $mtime;
167 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
168 } elsif ( -e $dump_path ) {
169 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
170 my $info = retrieve $dump_path;
171 $loaded->{ $path } = $info;
176 if ( -f $full_path ) {
177 $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
178 } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
179 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
181 die "can't load $full_path";
186 if ( defined $data->{header} ) {
187 if ( ref $data->{header} eq 'ARRAY' ) {
188 @header = @{ $data->{header} };
190 warn "header not array ", dump( $data->{header} );
194 my $stats = __stats( $data->{items} );
197 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
198 grep { defined $stats->{$_}->{count} } keys %$stats
202 header => [ @header ],
204 full_path => $full_path,
205 size => -s $full_path,
206 mtime => (stat($full_path))[9],
210 $loaded->{ $path } = $info;
211 $self->_save( $path );
219 my @paths = $self->param('paths');
220 warn "# paths ", dump @paths;
221 $self->_load_path( $_ ) foreach @paths;
223 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
224 warn "# path $path\n";
225 $self->_load_path( $path );
227 $self->session( 'path' => $path );
228 $self->session( 'modified' => $loaded->{$path}->{modified} );
230 my $redirect_to = '/data/items';
232 $self->session( 'header' => $loaded->{$path}->{header} );
233 if ( ! defined $loaded->{$path}->{columns} ) {
234 my $columns_path = $self->_permanent_path( 'columns' );
235 if ( -e $columns_path ) {
236 my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path;
237 $loaded->{$path}->{columns} = [ @columns ];
238 warn "# columns_path $columns_path ",dump(@columns);
240 $loaded->{$path}->{columns} = $loaded->{$path}->{header}
243 $redirect_to = '/data/columns';
245 $self->session( 'columns' => $loaded->{$path}->{columns} );
246 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
247 $self->redirect_to( $redirect_to );
252 my ( $self, $name ) = @_;
253 my $path = $self->session('path') || $self->param('path');
254 $self->redirect_to('/data/index') unless $path;
256 if ( $loaded->{$path}->{modified} > 1 ) {
257 warn "rebuild stats for $path forced by modified\n";
258 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
259 $loaded->{$path}->{modified} = 1;
262 if ( ! defined $loaded->{$path}->{$name} ) {
263 warn "$path $name isn't loaded\n";
264 $self->_load_path( $path );
265 $self->redirect_to('/data/index')
266 unless defined $loaded->{$path}->{$name};
267 if ( ! defined $loaded->{$path}->{stats} ) {
268 warn "rebuild stats for $path\n";
269 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
272 return $loaded->{$path}->{$name};
279 $checked->{$_}++ foreach @_;
280 # warn "# _checked ",dump($checked);
284 sub _permanent_path {
286 my $path = $self->_param_or_session('path');
287 $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
292 my $path = $self->_param_or_session('path');
293 my $dir = $self->app->home->rel_dir('public') . '/export/';
294 mkdir $dir unless -e $dir;
296 mkdir $dir unless -e $dir;
297 $dir . '/' . join('.', @_);
303 if ( $self->param('columns') ) {
304 my @columns = $self->_param_array('columns');
305 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
306 $self->redirect_to('/data/items');
309 my $stats = $self->_loaded( 'stats' );
312 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
314 foreach my $c ( @{ $self->session( 'header' ) } ) {
315 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
319 message => 'Select columns to display',
321 columns => \@columns,
322 checked => $self->_checked( $self->_param_array('columns') ),
327 my ($self,$name) = @_;
329 my @array = $self->param($name);
330 my $path = $self->session('path');
333 $self->session($name => [ @array ]);
334 $loaded->{$path}->{$name} = [ @array ];
335 } elsif ( my $session = $self->session($name) ) {
336 if ( ref $session eq 'ARRAY' ) {
339 die "$name not array ",dump($session);
342 #warn "# $name ",dump @array;
347 my ($self,$name,$default) = @_;
349 my $scalar = $self->param($name);
351 if ( defined $scalar ) {
352 $self->session($name => $scalar);
354 $scalar = $self->session($name);
357 if ( ! defined $scalar ) {
359 die "no default for $name" unless defined $scalar;
360 $self->session($name => $scalar);
363 warn "# _perm_scalar $name ",dump $scalar;
370 my $name = $self->param('filter_name') || die "name?";
371 my @vals = $self->param('filter_vals');
373 $self->_remove_filter( $name );
375 $self->_filter_on_data( $name, @vals );
376 if ( my $permanent = $self->param('_permanent') ) {
377 my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
378 write_file $permanent_path, map { "$_\n" } @vals;
379 warn "permanent filter $permanent_path ", -s $permanent_path;
383 $self->session( 'offset' => 0 );
385 $self->redirect_to('/data/items');
388 sub _filter_on_data {
389 my ( $self, $name, @vals ) = @_;
391 my $path = $self->session('path');
393 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
394 warn "filter $name not found in data set";
398 $filters->{$name} = [ @vals ];
399 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
402 $filter_hash->{$_}++ foreach @vals;
404 #warn "# filter_hash ",dump( $filter_hash );
406 my $items = $self->_loaded('data')->{items};
408 my $include_missing = defined $filter_hash->{_missing};
411 foreach my $i ( 0 .. $#$items ) {
413 if ( defined $items->[$i]->{$name} ) {
414 foreach my $v ( @{ $items->[$i]->{$name} } ) {
415 if ( defined $filter_hash->{ $v } ) {
416 $filtered_items->{$i}++;
419 } elsif ( $include_missing ) {
420 $filtered_items->{$i}++;
424 #warn "# filter $name ",dump($filtered_items);
426 $loaded->{$path}->{filters}->{$name} = $filtered_items;
427 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
431 my ( $self, $all ) = @_;
432 my $data = $self->_loaded( 'data' );
434 return @{ $data->{items} } if $all == 1;
436 my $filters = $self->_current_filters;
438 foreach my $f ( keys %$filters ) {
439 foreach my $n ( @{ $filters->{$f} } ) {
440 $filter_value->{$f}->{$n} = 1;
443 my @items = @{ $data->{items} };
447 foreach my $n ( keys %$filter_value ) {
448 if ( ! exists $i->{$n} ) {
449 if ( defined $filter_value->{$n}->{_missing} ) {
457 # and match any of values in element
459 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
460 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
462 if ( ! $have_values ) {
468 } @items if $filter_value;
473 sub _current_filters {
476 $current_filters->{ $_ } = $filters->{ $_ }
478 grep { defined $filters->{ $_ } }
479 @{ $self->_loaded('header') }
481 #warn "# current_filters ",dump($current_filters);
482 return $current_filters;
485 sub _data_sorted_by {
486 my ( $self, $order ) = @_;
488 my $path = $self->session('path');
490 warn "_data_sorted_by $order from $path";
492 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
493 return $loaded->{$path}->{sorted}->{$order};
496 my $data = $self->_loaded( 'data' );
497 my $numeric = $self->_is_numeric($order);
498 my $missing = $numeric ? 0 : '';
499 no warnings qw(numeric);
510 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
511 } grep { ref $_->{$order} eq 'ARRAY' } @{ $data->{items} }
514 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
515 #warn "# sorted ",dump( @sorted );
517 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
524 my $path = $self->session('path');
525 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
527 my @columns = $self->_param_array('columns');
528 $self->redirect_to('/data/columns') unless @columns;
529 my $order = $self->_param_scalar('order', $columns[0]);
530 my $sort = $self->_param_scalar('sort', 'a');
531 my $offset = $self->_param_scalar('offset', 0);
532 my $limit = $self->_param_scalar('limit', 20);
533 $self->_param_scalar('show', 'table');
535 # fix offset when changing limit
536 $offset = int( $offset / $limit ) * $limit;
538 my $sorted = $self->_data_sorted_by( $order );
542 foreach my $name ( keys %$filters ) {
543 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
544 warn "skip filter $name not found in $path\n";
547 push @filter_names, $name;
549 warn "filter_names ",dump( @filter_names );
550 foreach my $name ( @filter_names ) {
551 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
552 $self->_filter_on_data( $name, @{ $filters->{$name} } );
556 my $all_filters = join(' ',sort @filter_names,'order:',$order);
558 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
560 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
562 my $path_filters = $loaded->{$path}->{filters};
564 warn "create combined filter for $all_filters\n";
567 foreach my $i ( 0 .. $#$sorted ) {
568 my $pos = $sorted->[$i];
570 if ( $#filter_names == -1 ) {
571 push @filtered, $pos;
576 foreach ( @filter_names ) {
577 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
581 push @filtered, $pos;
584 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
587 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
588 if defined $loaded->{$path}->{filtered}->{$all_filters};
590 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
593 my $data = $self->_loaded('data');
594 my $from_end = $sort eq 'd' ? $#$filtered : 0;
595 foreach ( 0 .. $limit ) {
596 my $i = $_ + $offset;
597 last unless defined $filtered->[$i];
598 $i = $from_end - $i if $from_end;
599 my $id = $filtered->[$i];
601 my $item = $data->{items}->[ $id ];
602 $item->{_row_id} ||= $id;
605 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
611 sorted => $sorted_items,
612 columns => [ @columns ],
613 rows => $#$filtered + 1,
614 numeric => { map { $_, $self->_is_numeric($_) } @columns },
615 filters => $self->_current_filters,
623 $self->session('order', $self->param('order'));
624 $self->session('sort', $self->param('sort'));
625 $self->redirect_to('/data/items');
629 my ( $self, $name ) = @_;
631 my $stats = $self->_loaded( 'stats' );
633 # sort facet numerically if more >50% elements are numeric
634 my $count = $stats->{$name}->{count};
635 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
636 defined $stats->{$name}->{numeric} &&
637 $stats->{$name}->{numeric} > $count / 2;
641 my ($self,$name) = @_;
642 warn "_remove_filter $name\n";
644 my $path = $self->session('path');
646 delete $filters->{$name};
647 delete $loaded->{$path}->{filters}->{$name};
648 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
652 keys %{ $loaded->{$path}->{filtered} }
654 delete $loaded->{$path}->{filtered}->{$_};
655 warn "remove filtered cache $_";
662 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
664 if ( my $name = $self->param('remove') ) {
665 $self->_remove_filter( $name );
666 $self->redirect_to( '/data/items' );
670 my $name = $self->param('name') || die "no name";
672 my $all = $self->_param_scalar('all', 1);
673 my $data = $self->_loaded('data');
675 my $filters = $self->_current_filters;
676 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
677 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
678 if defined $loaded->{$path}->{filtered}->{$all_filters};
680 if ( ! $filtered || $all ) {
681 $filtered = [ 0 .. $#{ $data->{items} } ];
682 warn "filter all values\n";
684 warn "filter using $all_filters\n";
687 foreach my $i ( @$filtered ) {
688 my $item = $data->{items}->[$i];
689 if ( ! exists $item->{$name} ) {
690 $facet->{ _missing }++;
691 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
692 $facet->{$_}++ foreach @{ $item->{$name} };
694 $facet->{ $item->{$name} }++;
698 # warn "# facet $name ",dump $facet;
703 : defined $filters->{$name} ? @{ $filters->{$name} }
706 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
708 my $numeric = $self->_is_numeric($name);
710 my $sort = $self->param('sort');
711 # sort numeric facets with more than 5 values ascending
712 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
714 @facet_names = sort {
716 if ( $sort eq 'a' ) {
717 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
718 } elsif ( $sort eq 'd' ) {
719 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
720 } elsif ( $sort eq 'c' ) {
721 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
723 warn "unknown sort: $sort";
726 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
730 $self->render( name => $name, facet => $facet, checked => $checked,
731 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
737 my $new_content = $self->param('new_content');
738 $new_content ||= $self->param('content'); # backward compatibility with old actions
740 my $i = $self->param('_row_id');
741 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
742 my $path = $self->param('path') || die "no path";
743 my $name = $self->param('name') || die "no name";
744 my $status = 200; # 200 = OK, 201 = Created
746 my $data = $self->_loaded('data');
748 if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
749 $new_content =~ s/^\s+//s;
750 $new_content =~ s/\s+$//s;
752 if ( $new_content =~ /\xB6/ ) { # para
753 $v = [ split(/\s*\xB6\s*/, $new_content) ];
755 $v = [ $new_content ];
758 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
761 && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
767 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
769 time => $self->param('time') || time(),
770 user => $self->param('user') || $ENV{'LOGNAME'},
772 map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
773 grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
774 keys %{ $loaded->{$path}->{stats} }
777 my $change_path = $self->_permanent_path( 'changes' );
778 mkdir $change_path unless -d $change_path;
779 $change_path .= '/' . $change->{time};
780 store $change, $change_path;
781 utime $change->{time}, $change->{time}, $change_path;
782 warn "# $change_path ", dump($change);
784 warn "# change $path $i $old -> $new\n";
785 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
787 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
788 delete $loaded->{$path}->{sorted}->{$name};
789 warn "# invalidate $path sorted $name\n";
792 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
793 delete $loaded->{$path}->{filtered}->{$_};
794 warn "# invalidate $path filtered $_\n";
797 $status = 201; # created
798 # modified = 2 -- force rebuild of stats
799 $loaded->{$path}->{modified} = 2;
800 $self->session( 'modified' => 2 );
802 $new_content = join("\xB6",@$v);
805 warn "# unchanged $path $i $old\n";
809 $new_content = "$path $i $name doesn't exist\n";
813 warn "# edit $status ", dump $new_content;
817 new_content => scalar $new_content,
824 my $path = $self->_param_or_session('path');
825 my $dump_path = $self->_save( $path );
826 $loaded->{$path}->{modified} = 0;
827 $self->session( 'modified' => 0 );
829 $self->redirect_to( '/data/items' );
834 $self->render( export => [
835 map { s{^.+/public/export/}{}; $_ }
836 glob( $self->_export_path . '*' )
842 grep { defined $loaded->{$_}->{data} }