1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
14 use Time::HiRes qw(time);
15 use File::Path qw(mkpath);
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/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
37 push @{ $changes->{$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( full_path => $full_path, path => $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 );
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 my $caller = (caller(1))[3];
258 if ( $caller =~ m/::edit/ ) {
259 warn "rebuild stats for $path ignored caller $caller\n";
261 warn "rebuild stats for $path FORCED by modified caller $caller\n";
262 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
263 $loaded->{$path}->{modified} = 1;
267 if ( ! defined $loaded->{$path}->{$name} ) {
268 warn "$path $name isn't loaded\n";
269 $self->_load_path( $path );
270 $self->redirect_to('/data/index')
271 unless defined $loaded->{$path}->{$name};
272 if ( ! defined $loaded->{$path}->{stats} ) {
273 warn "rebuild stats for $path\n";
274 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
278 $self->session( 'modified' => $loaded->{$path}->{modified} );
280 return $loaded->{$path}->{$name};
287 $checked->{$_}++ foreach @_;
288 # warn "# _checked ",dump($checked);
292 sub _permanent_path {
294 my $path = $self->_param_or_session('path');
295 $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
300 my $path = $self->_param_or_session('path');
302 warn "no path in param or session";
305 my $dir = $self->app->home->rel_dir('public') . "/export/$path";
306 mkpath $dir unless -e $dir;
307 $dir . '/' . join('.', @_);
313 if ( $self->param('columns') ) {
314 my @columns = $self->_param_array('columns');
315 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
316 $self->redirect_to('/data/items');
319 my $stats = $self->_loaded( 'stats' );
322 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
324 foreach my $c ( @{ $self->session( 'header' ) } ) {
325 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
329 message => 'Select columns to display',
331 columns => \@columns,
332 checked => $self->_checked( $self->_param_array('columns') ),
337 my ($self,$name) = @_;
339 my @array = $self->param($name);
340 my $path = $self->session('path');
343 $self->session($name => [ @array ]);
344 $loaded->{$path}->{$name} = [ @array ];
345 } elsif ( my $session = $self->session($name) ) {
346 if ( ref $session eq 'ARRAY' ) {
349 die "$name not array ",dump($session);
352 #warn "# $name ",dump @array;
357 my ($self,$name,$default) = @_;
359 my $scalar = $self->param($name);
361 if ( defined $scalar ) {
362 $self->session($name => $scalar);
364 $scalar = $self->session($name);
367 if ( ! defined $scalar ) {
369 die "no default for $name" unless defined $scalar;
370 $self->session($name => $scalar);
373 warn "# _perm_scalar $name ",dump $scalar;
380 my $name = $self->param('filter_name') || die "name?";
381 my @vals = $self->param('filter_vals');
383 $self->_remove_filter( $name );
385 $self->_filter_on_data( $name, @vals );
386 if ( my $permanent = $self->param('_permanent') ) {
387 my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
388 write_file $permanent_path, map { "$_\n" } @vals;
389 warn "permanent filter $permanent_path ", -s $permanent_path;
393 $self->session( 'offset' => 0 );
395 $self->redirect_to('/data/items');
398 sub _filter_on_data {
399 my ( $self, $name, @vals ) = @_;
401 my $path = $self->session('path');
403 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
404 warn "filter $name not found in data set";
408 $filters->{$name} = [ @vals ];
409 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
412 $filter_hash->{$_}++ foreach @vals;
414 #warn "# filter_hash ",dump( $filter_hash );
416 my $items = $self->_loaded('data')->{items};
418 my $include_missing = defined $filter_hash->{_missing};
421 foreach my $i ( 0 .. $#$items ) {
423 if ( defined $items->[$i]->{$name} ) {
424 foreach my $v ( @{ $items->[$i]->{$name} } ) {
425 if ( defined $filter_hash->{ $v } ) {
426 $filtered_items->{$i}++;
429 } elsif ( $include_missing ) {
430 $filtered_items->{$i}++;
434 #warn "# filter $name ",dump($filtered_items);
436 $loaded->{$path}->{filters}->{$name} = $filtered_items;
437 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
441 my ( $self, $all ) = @_;
442 my $data = $self->_loaded( 'data' );
444 return @{ $data->{items} } if $all == 1;
446 my $filters = $self->_current_filters;
448 foreach my $f ( keys %$filters ) {
449 foreach my $n ( @{ $filters->{$f} } ) {
450 $filter_value->{$f}->{$n} = 1;
453 my @items = @{ $data->{items} };
457 foreach my $n ( keys %$filter_value ) {
458 if ( ! exists $i->{$n} ) {
459 if ( defined $filter_value->{$n}->{_missing} ) {
467 # and match any of values in element
469 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
470 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
472 if ( ! $have_values ) {
478 } @items if $filter_value;
483 sub _current_filters {
486 $current_filters->{ $_ } = $filters->{ $_ }
488 grep { defined $filters->{ $_ } }
489 @{ $self->_loaded('header') }
491 #warn "# current_filters ",dump($current_filters);
492 return $current_filters;
495 sub _data_sorted_by {
496 my ( $self, $order ) = @_;
498 my $path = $self->session('path');
500 warn "_data_sorted_by $order from $path";
502 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
503 return $loaded->{$path}->{sorted}->{$order};
506 my $data = $self->_loaded( 'data' );
507 my $numeric = $self->_is_numeric($order);
508 my $missing = $numeric ? 0 : '';
509 no warnings qw(numeric);
520 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
521 } grep { ref $_->{$order} eq 'ARRAY' } @{ $data->{items} }
524 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
525 #warn "# sorted ",dump( @sorted );
527 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
534 if ( my $show = $self->param('id') ) {
535 $self->param('show', $show);
539 my $path = $self->session('path');
540 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
542 my @columns = $self->_param_array('columns');
543 $self->redirect_to('/data/columns') unless @columns;
544 my $order = $self->_param_scalar('order', $columns[0]);
545 my $sort = $self->_param_scalar('sort', 'a');
546 my $offset = $self->_param_scalar('offset', 0);
547 my $limit = $self->_param_scalar('limit', 20);
548 $self->_param_scalar('show', 'table');
550 # fix offset when changing limit
551 $offset = int( $offset / $limit ) * $limit;
553 my $sorted = $self->_data_sorted_by( $order );
557 foreach my $name ( keys %$filters ) {
558 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
559 warn "skip filter $name not found in $path\n";
562 push @filter_names, $name;
564 warn "filter_names ",dump( @filter_names );
565 foreach my $name ( @filter_names ) {
566 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
567 $self->_filter_on_data( $name, @{ $filters->{$name} } );
571 my $all_filters = join(' ',sort @filter_names,'order:',$order);
573 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
575 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
577 my $path_filters = $loaded->{$path}->{filters};
579 warn "create combined filter for $all_filters\n";
582 foreach my $i ( 0 .. $#$sorted ) {
583 my $pos = $sorted->[$i];
585 if ( $#filter_names == -1 ) {
586 push @filtered, $pos;
591 foreach ( @filter_names ) {
592 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
596 push @filtered, $pos;
599 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
602 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
603 if defined $loaded->{$path}->{filtered}->{$all_filters};
605 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
608 my $data = $self->_loaded('data');
609 my $from_end = $sort eq 'd' ? $#$filtered : 0;
610 foreach ( 0 .. $limit ) {
611 my $i = $_ + $offset;
612 last unless defined $filtered->[$i];
613 $i = $from_end - $i if $from_end;
614 my $id = $filtered->[$i];
616 my $item = $data->{items}->[ $id ];
617 $item->{_row_id} ||= $id;
620 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
626 sorted => $sorted_items,
627 columns => [ @columns ],
628 rows => $#$filtered + 1,
629 numeric => { map { $_, $self->_is_numeric($_) } @columns },
630 filters => $self->_current_filters,
638 $self->session('order', $self->param('order'));
639 $self->session('sort', $self->param('sort'));
640 $self->redirect_to('/data/items');
644 my ( $self, $name ) = @_;
646 my $stats = $self->_loaded( 'stats' );
648 # sort facet numerically if more >50% elements are numeric
649 my $count = $stats->{$name}->{count};
650 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
651 defined $stats->{$name}->{numeric} &&
652 $stats->{$name}->{numeric} > $count / 2;
656 my ($self,$name) = @_;
657 warn "_remove_filter $name\n";
659 my $path = $self->session('path');
661 delete $filters->{$name};
662 delete $loaded->{$path}->{filters}->{$name};
663 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
667 keys %{ $loaded->{$path}->{filtered} }
669 delete $loaded->{$path}->{filtered}->{$_};
670 warn "remove filtered cache $_";
677 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
679 if ( my $name = $self->param('remove') ) {
680 $self->_remove_filter( $name );
681 $self->redirect_to( '/data/items' );
685 my $name = $self->param('name') || die "no name";
687 my $all = $self->_param_scalar('all', 1);
688 my $data = $self->_loaded('data');
690 my $filters = $self->_current_filters;
691 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
692 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
693 if defined $loaded->{$path}->{filtered}->{$all_filters};
695 if ( ! $filtered || $all ) {
696 $filtered = [ 0 .. $#{ $data->{items} } ];
697 warn "filter all values\n";
699 warn "filter using $all_filters\n";
702 foreach my $i ( @$filtered ) {
703 my $item = $data->{items}->[$i];
704 if ( ! exists $item->{$name} ) {
705 $facet->{ _missing }++;
706 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
707 $facet->{$_}++ foreach @{ $item->{$name} };
709 $facet->{ $item->{$name} }++;
713 # warn "# facet $name ",dump $facet;
718 : defined $filters->{$name} ? @{ $filters->{$name} }
721 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
723 my $numeric = $self->_is_numeric($name);
725 my $sort = $self->param('sort');
726 # sort numeric facets with more than 5 values ascending
727 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
729 @facet_names = sort {
731 if ( $sort eq 'a' ) {
732 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
733 } elsif ( $sort eq 'd' ) {
734 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
735 } elsif ( $sort eq 'c' ) {
736 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
738 warn "unknown sort: $sort";
741 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
745 $self->render( name => $name, facet => $facet, checked => $checked,
746 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
751 sub __invalidate_path_column {
752 my ( $path, $name ) = @_;
754 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
755 delete $loaded->{$path}->{sorted}->{$name};
756 warn "# invalidate $path sorted $name\n";
759 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
760 delete $loaded->{$path}->{filtered}->{$_};
761 warn "# invalidate $path filtered $_\n";
765 sub __path_modified {
766 my ( $path, $value ) = @_;
767 $value = 1 unless defined $value;
769 $loaded->{$path}->{modified} = $value;
771 warn "# __path_modified $path $value\n";
776 my $new_content = $self->param('new_content');
777 $new_content ||= $self->param('content'); # backward compatibility with old actions
779 my $i = $self->param('_row_id');
780 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
781 my $path = $self->param('path') || die "no path";
782 my $name = $self->param('name') || die "no name";
783 my $status = 200; # 200 = OK, 201 = Created
785 my $data = $self->_loaded('data');
787 if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
788 $new_content =~ s/^\s+//s;
789 $new_content =~ s/\s+$//s;
791 if ( $new_content =~ /\xB6/ ) { # para
792 $v = [ split(/\s*\xB6\s*/, $new_content) ];
794 $v = [ $new_content ];
797 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
800 && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
806 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
808 time => $self->param('time') || time(),
809 user => $self->param('user') || $ENV{'LOGNAME'},
811 map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
812 grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
813 keys %{ $loaded->{$path}->{stats} }
816 my $change_path = $self->_permanent_path( 'changes' );
817 mkdir $change_path unless -d $change_path;
818 $change_path .= '/' . $change->{time};
819 store $change, $change_path;
820 utime $change->{time}, $change->{time}, $change_path;
821 warn "# $change_path ", dump($change);
823 warn "# change $path $i $old -> $new\n";
824 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
826 __invalidate_path_column( $path, $name );
828 $status = 201; # created
829 # modified = 2 -- force rebuild of stats
830 __path_modified( $path, 2 );
832 $new_content = join("\xB6",@$v);
835 warn "# unchanged $path $i $old\n";
839 $new_content = "$path $i $name doesn't exist\n";
843 warn "# edit $status ", dump $new_content;
847 new_content => scalar $new_content,
854 my $path = $self->_param_or_session('path');
855 my $dump_path = $self->_save( $path );
856 __path_modified( $path, 0 );
858 $self->redirect_to( '/data/items' );
864 if ( my $import = $self->param('import') ) {
866 if ( $import =~ m{/filter\.(.+?)\..+} ) {
868 my @vals = map { chomp; $_ }
869 read_file $self->app->home->rel_dir('public') . "/export/$import";
870 $self->_remove_filter( $name );
871 $self->_filter_on_data( $name, @vals );
872 $self->session( 'offset' => 0 );
873 $self->redirect_to('/data/items');
875 warn "UNKNOWN IMPORT $import";
879 $self->render( export => [
880 glob( $self->_export_path . '*' )
886 grep { defined $loaded->{$_}->{data} }