1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
15 use Time::HiRes qw(time);
17 use MojoFacets::Import::File;
18 use MojoFacets::Import::HTMLTable;
26 my $data_dir = $self->app->home->rel_dir('data');
27 die "no data dir $data_dir" unless -d $data_dir;
32 my $file = $File::Find::name;
33 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
34 $file =~ s/$data_dir\/*//;
36 } elsif ( -f $file && $file =~ m/([^\/]+)\.edits\/(\d+\.\d+.+)/ ) {
37 push @{ $edits->{$1} }, $2
38 } elsif ( -d $file && $file =~ m/\.html$/ ) {
39 $file =~ s/$data_dir\/*//;
42 warn "IGNORE: $file\n";
46 @files = sort { lc $a cmp lc $b } @files;
48 $size->{$_} = -s "$data_dir/$_" foreach @files;
55 dump_path => { map { $_ => $self->_dump_path($_) } @files },
61 my ( $self, $name ) = @_;
62 my $dir = $self->app->home->rel_dir('data');
65 return '/tmp/mojo_facets.' . $name . '.storable';
69 my ( $self, $path ) = @_;
71 my $dump_path = $self->_dump_path( $path );
72 my $first_load = ! -e $dump_path;
73 warn "save loaded to $dump_path";
74 my $info = $loaded->{$path};
75 store $info, $dump_path;
78 my $mtime = $loaded->{$path}->{mtime};
79 utime $mtime, $mtime, $dump_path;
80 warn "sync time to $path at $mtime\n";
83 warn $dump_path, ' ', -s $dump_path, " bytes\n";
92 my $nr_items = $#{ $_[0] } + 1;
94 warn "__stats $nr_items\n";
96 foreach my $e ( @{ $_[0] } ) {
97 foreach my $n ( keys %$e ) {
98 $stats->{$n}->{count}++;
100 if ( ref $e->{$n} eq 'ARRAY' ) {
101 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
107 foreach my $x ( @v ) {
108 $stats->{$n}->{numeric}++
109 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
110 $stats->{$n}->{empty}++
111 if length $x == 0; # faster than $x =~ m/^\s*$/;
117 foreach my $n ( keys %$stats ) {
118 my $s = $stats->{$n};
119 next unless defined $s->{array};
120 if ( $s->{array} == $s->{count} ) {
122 if ( $s->{count} == $nr_items ) {
123 warn "check $n for uniqeness\n";
125 foreach my $e ( @{ $_[0] } ) {
126 if ( ++$unique->{ $e->{$n}->[0] } == 2 ) {
132 $stats->{$n}->{unique} = 1;
133 warn "# $n unique ",dump( $unique );
139 warn "# __stats ",dump($stats);
144 sub _param_or_session {
145 $_[0]->param( $_[1] ) || $_[0]->session( $_[1] )
150 my $path = $self->_param_or_session('path');
151 warn "stats $path\n";
152 delete $loaded->{$path}->{stats};
153 $self->redirect_to( '/data/columns' );
158 my ( $self, $path ) = @_;
160 my $full_path = $self->app->home->rel_file( 'data/' . $path );
161 die "$full_path $!" unless -r $full_path;
163 my $dump_path = $self->_dump_path( $path );
165 if ( defined $loaded->{$path}->{data} ) {
166 my $mtime = (stat($full_path))[9];
167 return if $loaded->{$path}->{mtime} == $mtime;
168 warn "reload $full_path, modified ", time() - $mtime, " seconds ago\n";
169 } elsif ( -e $dump_path ) {
170 warn "dump_path $dump_path ", -s $dump_path, " bytes loading...\n";
171 my $info = retrieve $dump_path;
172 $loaded->{ $path } = $info;
177 if ( -f $full_path ) {
178 $data = MojoFacets::Import::File->new( path => $full_path )->data;
179 } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
180 $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
182 die "can't load $full_path";
187 if ( defined $data->{header} ) {
188 if ( ref $data->{header} eq 'ARRAY' ) {
189 @header = @{ $data->{header} };
191 warn "header not array ", dump( $data->{header} );
195 my $stats = __stats( $data->{items} );
198 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
199 grep { defined $stats->{$_}->{count} } keys %$stats
203 header => [ @header ],
205 full_path => $full_path,
206 size => -s $full_path,
207 mtime => (stat($full_path))[9],
211 $loaded->{ $path } = $info;
212 $self->_save( $path );
220 my @paths = $self->param('paths');
221 warn "# paths ", dump @paths;
222 $self->_load_path( $_ ) foreach @paths;
224 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
225 warn "# path $path\n";
226 $self->_load_path( $path );
228 $self->session( 'path' => $path );
229 $self->session( 'modified' => $loaded->{$path}->{modified} );
231 my $redirect_to = '/data/items';
233 $self->session( 'header' => $loaded->{$path}->{header} );
234 if ( ! defined $loaded->{$path}->{columns} ) {
235 my $columns_path = $self->_permanent_path( 'columns' );
236 if ( -e $columns_path ) {
237 my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path;
238 $loaded->{$path}->{columns} = [ @columns ];
239 warn "# columns_path $columns_path ",dump(@columns);
241 $loaded->{$path}->{columns} = $loaded->{$path}->{header}
244 $redirect_to = '/data/columns';
246 $self->session( 'columns' => $loaded->{$path}->{columns} );
247 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
248 $self->redirect_to( $redirect_to );
253 my ( $self, $name ) = @_;
254 my $path = $self->session('path') || $self->param('path');
255 $self->redirect_to('/data/index') unless $path;
256 if ( ! defined $loaded->{$path}->{$name} ) {
257 warn "$path $name isn't loaded\n";
258 $self->_load_path( $path );
259 $self->redirect_to('/data/index')
260 unless defined $loaded->{$path}->{$name};
261 if ( ! defined $loaded->{$path}->{stats} ) {
262 warn "rebuild stats for $path\n";
263 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
266 return $loaded->{$path}->{$name};
273 $checked->{$_}++ foreach @_;
274 # warn "# _checked ",dump($checked);
278 sub _permanent_path {
280 my $path = $self->_param_or_session('path');
281 $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
286 my $path = $self->_param_or_session('path');
287 my $dir = $self->app->home->rel_dir('public') . '/export/';
288 mkdir $dir unless -e $dir;
290 mkdir $dir unless -e $dir;
291 $dir . '/' . join('.', @_);
297 if ( $self->param('columns') ) {
298 my @columns = $self->_param_array('columns');
299 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
300 $self->redirect_to('/data/items');
303 my $stats = $self->_loaded( 'stats' );
306 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
308 foreach my $c ( @{ $self->session( 'header' ) } ) {
309 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
313 message => 'Select columns to display',
315 columns => \@columns,
316 checked => $self->_checked( $self->_param_array('columns') ),
321 my ($self,$name) = @_;
323 my @array = $self->param($name);
324 my $path = $self->session('path');
327 $self->session($name => [ @array ]);
328 $loaded->{$path}->{$name} = [ @array ];
329 } elsif ( my $session = $self->session($name) ) {
330 if ( ref $session eq 'ARRAY' ) {
333 die "$name not array ",dump($session);
336 #warn "# $name ",dump @array;
341 my ($self,$name,$default) = @_;
343 my $scalar = $self->param($name);
345 if ( defined $scalar ) {
346 $self->session($name => $scalar);
348 $scalar = $self->session($name);
351 if ( ! defined $scalar ) {
353 die "no default for $name" unless defined $scalar;
354 $self->session($name => $scalar);
357 warn "# _perm_scalar $name ",dump $scalar;
364 my $name = $self->param('filter_name') || die "name?";
365 my @vals = $self->param('filter_vals');
367 $self->_remove_filter( $name );
369 $self->_filter_on_data( $name, @vals );
370 if ( my $permanent = $self->param('_permanent') ) {
371 my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
372 write_file $permanent_path, map { "$_\n" } @vals;
373 warn "permanent filter $permanent_path ", -s $permanent_path;
377 $self->session( 'offset' => 0 );
379 $self->redirect_to('/data/items');
382 sub _filter_on_data {
383 my ( $self, $name, @vals ) = @_;
385 my $path = $self->session('path');
387 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
388 warn "filter $name not found in data set";
392 $filters->{$name} = [ @vals ];
393 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
396 $filter_hash->{$_}++ foreach @vals;
398 #warn "# filter_hash ",dump( $filter_hash );
400 my $items = $self->_loaded('data')->{items};
402 my $include_missing = defined $filter_hash->{_missing};
405 foreach my $i ( 0 .. $#$items ) {
407 if ( defined $items->[$i]->{$name} ) {
408 foreach my $v ( @{ $items->[$i]->{$name} } ) {
409 if ( defined $filter_hash->{ $v } ) {
410 $filtered_items->{$i}++;
413 } elsif ( $include_missing ) {
414 $filtered_items->{$i}++;
418 #warn "# filter $name ",dump($filtered_items);
420 $loaded->{$path}->{filters}->{$name} = $filtered_items;
421 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
425 my ( $self, $all ) = @_;
426 my $data = $self->_loaded( 'data' );
428 return @{ $data->{items} } if $all == 1;
430 my $filters = $self->_current_filters;
432 foreach my $f ( keys %$filters ) {
433 foreach my $n ( @{ $filters->{$f} } ) {
434 $filter_value->{$f}->{$n} = 1;
437 my @items = @{ $data->{items} };
441 foreach my $n ( keys %$filter_value ) {
442 if ( ! exists $i->{$n} ) {
443 if ( defined $filter_value->{$n}->{_missing} ) {
451 # and match any of values in element
453 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
454 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
456 if ( ! $have_values ) {
462 } @items if $filter_value;
467 sub _current_filters {
470 $current_filters->{ $_ } = $filters->{ $_ }
472 grep { defined $filters->{ $_ } }
473 @{ $self->_loaded('header') }
475 #warn "# current_filters ",dump($current_filters);
476 return $current_filters;
479 sub _data_sorted_by {
480 my ( $self, $order ) = @_;
482 my $path = $self->session('path');
484 warn "_data_sorted_by $order from $path";
486 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
487 return $loaded->{$path}->{sorted}->{$order};
490 my $data = $self->_loaded( 'data' );
491 my $numeric = $self->_is_numeric($order);
492 my $missing = $numeric ? 0 : '';
493 no warnings qw(numeric);
504 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
505 } @{ $data->{items} }
508 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
509 #warn "# sorted ",dump( @sorted );
511 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
518 my $path = $self->session('path');
519 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
521 my @columns = $self->_param_array('columns');
522 $self->redirect_to('/data/columns') unless @columns;
523 my $order = $self->_param_scalar('order', $columns[0]);
524 my $sort = $self->_param_scalar('sort', 'a');
525 my $offset = $self->_param_scalar('offset', 0);
526 my $limit = $self->_param_scalar('limit', 20);
527 $self->_param_scalar('show', 'table');
529 # fix offset when changing limit
530 $offset = int( $offset / $limit ) * $limit;
532 my $sorted = $self->_data_sorted_by( $order );
536 foreach my $name ( keys %$filters ) {
537 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
538 warn "skip filter $name not found in $path\n";
541 push @filter_names, $name;
543 warn "filter_names ",dump( @filter_names );
544 foreach my $name ( @filter_names ) {
545 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
546 $self->_filter_on_data( $name, @{ $filters->{$name} } );
550 my $all_filters = join(' ',sort @filter_names,'order:',$order);
552 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
554 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
556 my $path_filters = $loaded->{$path}->{filters};
558 warn "create combined filter for $all_filters\n";
561 foreach my $i ( 0 .. $#$sorted ) {
562 my $pos = $sorted->[$i];
564 if ( $#filter_names == -1 ) {
565 push @filtered, $pos;
570 foreach ( @filter_names ) {
571 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
575 push @filtered, $pos;
578 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
581 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
582 if defined $loaded->{$path}->{filtered}->{$all_filters};
584 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
587 my $data = $self->_loaded('data');
588 my $from_end = $sort eq 'd' ? $#$filtered : 0;
589 foreach ( 0 .. $limit ) {
590 my $i = $_ + $offset;
591 last unless defined $filtered->[$i];
592 $i = $from_end - $i if $from_end;
593 my $id = $filtered->[$i];
595 my $item = $data->{items}->[ $id ];
596 $item->{_row_id} ||= $id;
599 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
605 sorted => $sorted_items,
606 columns => [ @columns ],
607 rows => $#$filtered + 1,
608 numeric => { map { $_, $self->_is_numeric($_) } @columns },
609 filters => $self->_current_filters,
617 $self->session('order', $self->param('order'));
618 $self->session('sort', $self->param('sort'));
619 $self->redirect_to('/data/items');
623 my ( $self, $name ) = @_;
625 my $stats = $self->_loaded( 'stats' );
627 # sort facet numerically if more >50% elements are numeric
628 my $count = $stats->{$name}->{count};
629 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
630 defined $stats->{$name}->{numeric} &&
631 $stats->{$name}->{numeric} > $count / 2;
635 my ($self,$name) = @_;
636 warn "_remove_filter $name\n";
638 my $path = $self->session('path');
640 delete $filters->{$name};
641 delete $loaded->{$path}->{filters}->{$name};
642 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
646 keys %{ $loaded->{$path}->{filtered} }
648 delete $loaded->{$path}->{filtered}->{$_};
649 warn "remove filtered cache $_";
656 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
658 if ( my $name = $self->param('remove') ) {
659 $self->_remove_filter( $name );
660 $self->redirect_to( '/data/items' );
664 my $name = $self->param('name') || die "no name";
666 my $all = $self->_param_scalar('all', 1);
667 my $data = $self->_loaded('data');
669 my $filters = $self->_current_filters;
670 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
671 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
672 if defined $loaded->{$path}->{filtered}->{$all_filters};
674 if ( ! $filtered || $all ) {
675 $filtered = [ 0 .. $#{ $data->{items} } ];
676 warn "filter all values\n";
678 warn "filter using $all_filters\n";
681 foreach my $i ( @$filtered ) {
682 my $item = $data->{items}->[$i];
683 if ( ! exists $item->{$name} ) {
684 $facet->{ _missing }++;
685 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
686 $facet->{$_}++ foreach @{ $item->{$name} };
688 $facet->{ $item->{$name} }++;
692 # warn "# facet $name ",dump $facet;
697 : defined $filters->{$name} ? @{ $filters->{$name} }
700 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
702 my $numeric = $self->_is_numeric($name);
704 my $sort = $self->param('sort');
705 # sort numeric facets with more than 5 values ascending
706 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
708 @facet_names = sort {
710 if ( $sort eq 'a' ) {
711 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
712 } elsif ( $sort eq 'd' ) {
713 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
714 } elsif ( $sort eq 'c' ) {
715 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
717 warn "unknown sort: $sort";
720 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
724 $self->render( name => $name, facet => $facet, checked => $checked,
725 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
731 my $new_content = $self->param('new_content');
733 my $i = $self->param('_row_id');
734 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
735 my $path = $self->param('path') || die "no path";
736 my $name = $self->param('name') || die "no name";
737 my $status = 200; # 200 = OK, 201 = Created
739 my $data = $self->_loaded('data');
741 if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
742 $new_content =~ s/^\s+//s;
743 $new_content =~ s/\s+$//s;
745 if ( $new_content =~ /\xB6/ ) { # para
746 $v = [ split(/\s*\xB6\s*/, $new_content) ];
748 $v = [ $new_content ];
751 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
754 && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
760 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
762 time => $self->param('time') || time(),
763 user => $self->param('user') || $ENV{'LOGNAME'},
765 map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
766 grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
767 keys %{ $loaded->{$path}->{stats} }
770 my $edit_path = $self->_permanent_path( 'edits' );
771 mkdir $edit_path unless -d $edit_path;
772 $edit_path .= '/' . $edit->{time};
773 store $edit, $edit_path;
774 utime $edit->{time}, $edit->{time}, $edit_path;
775 warn "# $edit_path ", dump($edit);
777 warn "# edit $path $i $old -> $new\n";
778 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
780 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
781 delete $loaded->{$path}->{sorted}->{$name};
782 warn "# invalidate $path sorted $name\n";
785 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
786 delete $loaded->{$path}->{filtered}->{$_};
787 warn "# invalidate $path filtered $_\n";
790 $status = 201; # created
791 $loaded->{$path}->{modified} = 1;
792 $self->session( 'modified' => 1 );
794 $new_content = join("\xB6",@$v);
797 warn "# unchanged $path $i $old\n";
801 $new_content = "$path $i $name doesn't exist\n";
805 warn "# edit $status ", dump $new_content;
809 new_content => scalar $new_content,
816 my $path = $self->_param_or_session('path');
817 my $dump_path = $self->_save( $path );
818 $loaded->{$path}->{modified} = 0;
819 $self->session( 'modified' => 0 );
821 $self->redirect_to( '/data/items' );
826 $self->render( export => [
827 map { s{^.+/public/export/}{}; $_ }
828 glob( $self->_export_path . '*' )
834 grep { defined $loaded->{$_}->{data} }