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 $path = $self->param('path') || $self->redirect_to( '/data/index' );
222 my @paths = $self->param('paths');
223 warn "# paths ", dump @paths;
225 foreach my $p ( keys %$loaded ) {
226 next if grep { /^\Q$p\E$/ } @paths;
227 warn "remove $p from memory\n";
228 delete $loaded->{$p};
231 $self->_load_path( $_ ) foreach @paths;
233 warn "# path $path\n";
234 $self->_load_path( $path );
236 $self->session( 'path' => $path );
238 my $redirect_to = '/data/items';
240 $self->session( 'header' => $loaded->{$path}->{header} );
241 if ( ! defined $loaded->{$path}->{columns} ) {
242 my $columns_path = $self->_permanent_path( 'columns' );
243 if ( -e $columns_path ) {
244 my @columns = map { s/[\r\n]+$//; $_ } read_file $columns_path;
245 $loaded->{$path}->{columns} = [ @columns ];
246 warn "# columns_path $columns_path ",dump(@columns);
248 $loaded->{$path}->{columns} = $loaded->{$path}->{header}
251 $redirect_to = '/data/columns';
253 $self->session( 'columns' => $loaded->{$path}->{columns} );
254 $self->session( 'order' => $loaded->{$path}->{columns}->[0] );
255 $self->redirect_to( $redirect_to );
260 my ( $self, $name ) = @_;
261 my $path = $self->session('path') || $self->param('path');
262 $self->redirect_to('/data/index') unless $path;
264 if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
265 my $caller = (caller(1))[3];
266 if ( $caller =~ m/::edit/ ) {
267 warn "rebuild stats for $path ignored caller $caller\n";
269 warn "rebuild stats for $path FORCED by modified caller $caller\n";
270 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
271 $loaded->{$path}->{modified} = 1;
275 if ( ! defined $loaded->{$path}->{$name} ) {
276 warn "$path $name isn't loaded\n";
277 $self->_load_path( $path );
278 if ( ! defined $loaded->{$path}->{stats} ) {
279 warn "rebuild stats for $path\n";
280 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
282 if ( ! defined $loaded->{$path}->{$name} ) {
283 warn "MISSING $name for $path\n";
284 $self->redirect_to('/data/index')
288 $self->session( 'modified' => $loaded->{$path}->{modified} );
290 return $loaded->{$path}->{$name};
297 $checked->{$_}++ foreach @_;
298 # warn "# _checked ",dump($checked);
302 sub _permanent_path {
304 my $path = $self->_param_or_session('path');
305 $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
310 my $path = $self->_param_or_session('path');
312 warn "no path in param or session";
315 my $dir = $self->app->home->rel_dir('public') . "/export/$path";
316 mkpath $dir unless -e $dir;
317 $dir . '/' . join('.', @_);
323 if ( $self->param('columns') ) {
324 my @columns = $self->_param_array('columns');
325 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
326 $self->redirect_to('/data/items');
329 my $stats = $self->_loaded( 'stats' );
332 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
334 foreach my $c ( @{ $self->session( 'header' ) } ) {
335 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
339 message => 'Select columns to display',
341 columns => \@columns,
342 checked => $self->_checked( $self->_param_array('columns') ),
347 my ($self,$name) = @_;
349 my @array = $self->param($name);
350 my $path = $self->session('path');
353 $self->session($name => [ @array ]);
354 $loaded->{$path}->{$name} = [ @array ];
355 } elsif ( my $session = $self->session($name) ) {
356 if ( ref $session eq 'ARRAY' ) {
359 die "$name not array ",dump($session);
362 #warn "# $name ",dump @array;
367 my ($self,$name,$default) = @_;
369 my $scalar = $self->param($name);
371 if ( defined $scalar ) {
372 $self->session($name => $scalar);
374 $scalar = $self->session($name);
377 if ( ! defined $scalar ) {
379 die "no default for $name" unless defined $scalar;
380 $self->session($name => $scalar);
383 warn "# _param_scalar $name ",dump $scalar;
390 my $name = $self->param('filter_name') || die "name?";
391 my @vals = $self->param('filter_vals');
393 $self->_remove_filter( $name );
395 $self->_filter_on_data( $name, @vals );
396 if ( my $permanent = $self->param('_permanent') ) {
397 my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
398 write_file $permanent_path, map { "$_\n" } @vals;
399 warn "permanent filter $permanent_path ", -s $permanent_path;
403 $self->session( 'offset' => 0 );
405 $self->redirect_to('/data/items');
408 sub _filter_on_data {
409 my ( $self, $name, @vals ) = @_;
411 my $path = $self->session('path');
413 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
414 warn "filter $name not found in data set";
418 $filters->{$name} = [ @vals ];
419 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
422 $filter_hash->{$_}++ foreach @vals;
424 #warn "# filter_hash ",dump( $filter_hash );
426 my $items = $self->_loaded('data')->{items};
428 my $include_missing = defined $filter_hash->{_missing};
431 foreach my $i ( 0 .. $#$items ) {
433 if ( defined $items->[$i]->{$name} ) {
434 my $row = $items->[$i]->{$name};
435 $row = [ $row ] unless ref $row eq 'ARRAY'; # FIXME probably wrong place
436 foreach my $v ( @$row ) {
437 if ( defined $filter_hash->{ $v } ) {
438 $filtered_items->{$i}++;
441 } elsif ( $include_missing ) {
442 $filtered_items->{$i}++;
446 #warn "# filter $name ",dump($filtered_items);
448 $loaded->{$path}->{filters}->{$name} = $filtered_items;
449 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
453 sub _current_filters {
456 $current_filters->{ $_ } = $filters->{ $_ }
458 grep { defined $filters->{ $_ } }
459 @{ $self->_loaded('columns') }
461 #warn "# current_filters ",dump($current_filters);
462 return $current_filters;
465 sub _data_sorted_by {
466 my ( $self, $order ) = @_;
468 my $path = $self->session('path');
470 warn "_data_sorted_by $order from $path";
472 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
473 return $loaded->{$path}->{sorted}->{$order};
476 my $data = $self->_loaded( 'data' );
477 my $numeric = $self->_is_numeric($order);
478 my $missing = $numeric ? 0 : '';
479 no warnings qw(numeric);
490 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
491 } grep { ref $_->{$order} eq 'ARRAY' } @{ $data->{items} }
494 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
495 #warn "# sorted ",dump( @sorted );
497 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
504 if ( my $show = $self->param('id') ) {
505 $self->param('show', $show);
509 my $path = $self->session('path');
511 my @columns = $self->_param_array('columns');
512 $self->redirect_to('/data/columns') unless @columns;
513 my $order = $self->_param_scalar('order', $columns[0]);
514 my $sort = $self->_param_scalar('sort', 'a');
515 my $offset = $self->_param_scalar('offset', 0);
516 my $limit = $self->_param_scalar('limit', 20);
517 $self->_param_scalar('show', 'table');
519 # fix offset when changing limit
520 $offset = int( $offset / $limit ) * $limit;
522 if ( ! grep { /^\Q$order\E$/ } @columns ) {
523 $order = $columns[0];
524 $self->session( order => $order );
526 my $sorted = $self->_data_sorted_by( $order );
530 foreach my $name ( keys %$filters ) {
531 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
532 warn "skip filter $name not found in $path\n";
535 push @filter_names, $name;
537 warn "filter_names ",dump( @filter_names );
538 foreach my $name ( @filter_names ) {
539 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
540 $self->_filter_on_data( $name, @{ $filters->{$name} } );
544 my $all_filters = join(' ',sort @filter_names,'order:',$order);
546 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
548 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
550 my $path_filters = $loaded->{$path}->{filters};
552 warn "create combined filter for $all_filters\n";
555 foreach my $i ( 0 .. $#$sorted ) {
556 my $pos = $sorted->[$i];
558 if ( $#filter_names == -1 ) {
559 push @filtered, $pos;
564 foreach ( @filter_names ) {
565 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
569 push @filtered, $pos;
572 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
575 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
576 if defined $loaded->{$path}->{filtered}->{$all_filters};
578 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
580 my $data = $self->_loaded('data');
582 my $code = $self->_param_scalar('code','');
583 $code =~ s{[\r\n]+$}{}s;
585 my $commit = $self->param('commit');
586 my $test = $self->param('test');
590 if ( $code && ( $test || $commit ) ) {
591 # XXX find columns used in code snippet and show them to user
592 foreach my $column ( $code =~ m/\$row->{(.+?)}/g ) {
593 if ( $column =~ s/^(['"])// ) {
596 $cols_changed->{$column}++;
597 next if grep { /$column/ } @columns;
598 $cols_changed->{$column}++;
599 unshift @columns, $column;
601 $self->session('columns', [ @columns ]);
602 $loaded->{$path}->{columns} = [ @columns ];
603 __path_modified( $path, 2 );
609 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
610 foreach ( 0 .. $#$filtered ) {
611 my $i = $filtered->[$_];
612 my $row = $data->{items}->[$i];
615 if ( my $description = $self->param('code_description') ) {
616 my $depends = $self->param('code_depends') || die "no code_depends?";
617 my $path = $self->app->home->rel_dir('public') . "/code/$depends.$description.pl";
618 write_file $path, $code;
619 warn "code $path ", -s $path, " bytes saved\n";
625 my $from_end = $sort eq 'd' ? $#$filtered : 0;
626 foreach ( 0 .. $limit ) {
627 my $i = $_ + $offset;
628 last unless defined $filtered->[$i];
629 $i = $from_end - $i if $from_end;
630 my $id = $filtered->[$i];
631 my $row = $data->{items}->[ $id ];
632 if ( $code && $test ) {
633 $row = Storable::dclone $row;
636 warn "ERROR evaling\n$code\n$@";
637 $self->stash('eval_error', $@) if $@;
639 warn "EVAL ",dump($row);
642 $row->{_row_id} ||= $id;
643 push @$sorted_items, $row;
646 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
652 sorted => $sorted_items,
653 columns => [ @columns ],
654 rows => $#$filtered + 1,
655 numeric => { map { $_, $self->_is_numeric($_) } @columns },
656 filters => $self->_current_filters,
658 cols_changed => $cols_changed,
660 $self->param('code_depends') ||
661 join(',', grep { defined $cols_changed->{$_} && $cols_changed->{$_} == 1 } @columns ),
663 $self->param('code_description') ||
664 join(',', grep { defined $cols_changed->{$_} && $cols_changed->{$_} == 2 } @columns ),
672 $self->session('order', $self->param('order'));
673 $self->session('sort', $self->param('sort'));
674 $self->redirect_to('/data/items');
678 my ( $self, $name ) = @_;
680 my $stats = $self->_loaded( 'stats' );
682 # sort facet numerically if more >50% elements are numeric
683 my $count = $stats->{$name}->{count};
684 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
685 defined $stats->{$name}->{numeric} &&
686 $stats->{$name}->{numeric} > $count / 2;
690 my ($self,$name) = @_;
691 warn "_remove_filter $name\n";
693 my $path = $self->session('path');
695 delete $filters->{$name};
696 delete $loaded->{$path}->{filters}->{$name};
697 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
701 keys %{ $loaded->{$path}->{filtered} }
703 delete $loaded->{$path}->{filtered}->{$_};
704 warn "remove filtered cache $_";
711 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
713 if ( my $name = $self->param('remove') ) {
714 $self->_remove_filter( $name );
715 $self->redirect_to( '/data/items' );
719 my $name = $self->param('name') || die "no name";
721 my $all = $self->_param_scalar('all', 1);
722 my $data = $self->_loaded('data');
724 my $filters = $self->_current_filters;
725 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
726 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
727 if defined $loaded->{$path}->{filtered}->{$all_filters};
729 if ( ! $filtered || $all ) {
730 $filtered = [ 0 .. $#{ $data->{items} } ];
731 warn "filter all values\n";
733 warn "filter using $all_filters\n";
736 foreach my $i ( @$filtered ) {
737 my $item = $data->{items}->[$i];
738 if ( ! exists $item->{$name} ) {
739 $facet->{ _missing }++;
740 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
741 $facet->{$_}++ foreach @{ $item->{$name} };
743 $facet->{ $item->{$name} }++;
747 # warn "# facet $name ",dump $facet;
752 : defined $filters->{$name} ? @{ $filters->{$name} }
755 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
757 my $numeric = $self->_is_numeric($name);
759 my $sort = $self->param('sort');
760 # sort numeric facets with more than 5 values ascending
761 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
763 @facet_names = sort {
765 if ( $sort eq 'a' ) {
766 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
767 } elsif ( $sort eq 'd' ) {
768 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
769 } elsif ( $sort eq 'c' ) {
770 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
772 warn "unknown sort: $sort";
775 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
779 $self->render( name => $name, facet => $facet, checked => $checked,
780 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
785 sub __invalidate_path_column {
786 my ( $path, $name ) = @_;
788 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
789 delete $loaded->{$path}->{sorted}->{$name};
790 warn "# invalidate $path sorted $name\n";
793 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
794 delete $loaded->{$path}->{filtered}->{$_};
795 warn "# invalidate $path filtered $_\n";
799 sub __path_modified {
800 my ( $path, $value ) = @_;
801 $value = 1 unless defined $value;
803 $loaded->{$path}->{modified} = $value;
805 warn "# __path_modified $path $value\n";
810 my $new_content = $self->param('new_content');
811 $new_content ||= $self->param('content'); # backward compatibility with old actions
813 my $i = $self->param('_row_id');
814 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
815 my $path = $self->param('path') || die "no path";
816 my $name = $self->param('name') || die "no name";
817 my $status = 200; # 200 = OK, 201 = Created
819 my $data = $self->_loaded('data');
821 if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
822 $new_content =~ s/^\s+//s;
823 $new_content =~ s/\s+$//s;
825 if ( $new_content =~ /\xB6/ ) { # para
826 $v = [ split(/\s*\xB6\s*/, $new_content) ];
828 $v = [ $new_content ];
831 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
834 && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
840 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
842 time => $self->param('time') || time(),
843 user => $self->param('user') || $ENV{'LOGNAME'},
845 map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
846 grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
847 keys %{ $loaded->{$path}->{stats} }
850 my $change_path = $self->_permanent_path( 'changes' );
851 mkdir $change_path unless -d $change_path;
852 $change_path .= '/' . $change->{time};
853 store $change, $change_path;
854 utime $change->{time}, $change->{time}, $change_path;
855 warn "# $change_path ", dump($change);
857 warn "# change $path $i $old -> $new\n";
858 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
860 __invalidate_path_column( $path, $name );
862 $status = 201; # created
863 # modified = 2 -- force rebuild of stats
864 __path_modified( $path, 2 );
866 $new_content = join("\xB6",@$v);
869 warn "# unchanged $path $i $old\n";
873 $new_content = "$path $i $name doesn't exist\n";
877 warn "# edit $status ", dump $new_content;
881 new_content => scalar $new_content,
888 my $path = $self->_param_or_session('path');
889 my $dump_path = $self->_save( $path );
890 __path_modified( $path, 0 );
892 $self->redirect_to( '/data/items' );
898 if ( my $import = $self->param('import') ) {
900 if ( $import =~ m{/filter\.(.+?)\..+} ) {
902 my @vals = map { chomp; $_ }
903 read_file $self->app->home->rel_dir('public') . "/export/$import";
904 $self->_remove_filter( $name );
905 $self->_filter_on_data( $name, @vals );
906 $self->session( 'offset' => 0 );
907 $self->redirect_to('/data/items');
909 warn "UNKNOWN IMPORT $import";
913 $self->render( export => [
914 glob( $self->_export_path . '*' )
920 grep { defined $loaded->{$_}->{data} }