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 ( defined $loaded->{$path}->{modified} && $loaded->{$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 if ( ! defined $loaded->{$path}->{stats} ) {
271 warn "rebuild stats for $path\n";
272 $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
274 if ( ! defined $loaded->{$path}->{$name} ) {
275 warn "MISSING $name for $path\n";
276 $self->redirect_to('/data/index')
280 $self->session( 'modified' => $loaded->{$path}->{modified} );
282 return $loaded->{$path}->{$name};
289 $checked->{$_}++ foreach @_;
290 # warn "# _checked ",dump($checked);
294 sub _permanent_path {
296 my $path = $self->_param_or_session('path');
297 $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
302 my $path = $self->_param_or_session('path');
304 warn "no path in param or session";
307 my $dir = $self->app->home->rel_dir('public') . "/export/$path";
308 mkpath $dir unless -e $dir;
309 $dir . '/' . join('.', @_);
315 if ( $self->param('columns') ) {
316 my @columns = $self->_param_array('columns');
317 write_file( $self->_permanent_path( 'columns' ), map { "$_\n" } @columns );
318 $self->redirect_to('/data/items');
321 my $stats = $self->_loaded( 'stats' );
324 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
326 foreach my $c ( @{ $self->session( 'header' ) } ) {
327 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
331 message => 'Select columns to display',
333 columns => \@columns,
334 checked => $self->_checked( $self->_param_array('columns') ),
339 my ($self,$name) = @_;
341 my @array = $self->param($name);
342 my $path = $self->session('path');
345 $self->session($name => [ @array ]);
346 $loaded->{$path}->{$name} = [ @array ];
347 } elsif ( my $session = $self->session($name) ) {
348 if ( ref $session eq 'ARRAY' ) {
351 die "$name not array ",dump($session);
354 #warn "# $name ",dump @array;
359 my ($self,$name,$default) = @_;
361 my $scalar = $self->param($name);
363 if ( defined $scalar ) {
364 $self->session($name => $scalar);
366 $scalar = $self->session($name);
369 if ( ! defined $scalar ) {
371 die "no default for $name" unless defined $scalar;
372 $self->session($name => $scalar);
375 warn "# _param_scalar $name ",dump $scalar;
382 my $name = $self->param('filter_name') || die "name?";
383 my @vals = $self->param('filter_vals');
385 $self->_remove_filter( $name );
387 $self->_filter_on_data( $name, @vals );
388 if ( my $permanent = $self->param('_permanent') ) {
389 my $permanent_path = $self->_export_path( 'filter', $name, $permanent );
390 write_file $permanent_path, map { "$_\n" } @vals;
391 warn "permanent filter $permanent_path ", -s $permanent_path;
395 $self->session( 'offset' => 0 );
397 $self->redirect_to('/data/items');
400 sub _filter_on_data {
401 my ( $self, $name, @vals ) = @_;
403 my $path = $self->session('path');
405 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
406 warn "filter $name not found in data set";
410 $filters->{$name} = [ @vals ];
411 warn "_filter_on_data $name ", $#vals + 1, " values on $path\n";
414 $filter_hash->{$_}++ foreach @vals;
416 #warn "# filter_hash ",dump( $filter_hash );
418 my $items = $self->_loaded('data')->{items};
420 my $include_missing = defined $filter_hash->{_missing};
423 foreach my $i ( 0 .. $#$items ) {
425 if ( defined $items->[$i]->{$name} ) {
426 foreach my $v ( @{ $items->[$i]->{$name} } ) {
427 if ( defined $filter_hash->{ $v } ) {
428 $filtered_items->{$i}++;
431 } elsif ( $include_missing ) {
432 $filtered_items->{$i}++;
436 #warn "# filter $name ",dump($filtered_items);
438 $loaded->{$path}->{filters}->{$name} = $filtered_items;
439 warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
443 my ( $self, $all ) = @_;
444 my $data = $self->_loaded( 'data' );
446 return @{ $data->{items} } if $all == 1;
448 my $filters = $self->_current_filters;
450 foreach my $f ( keys %$filters ) {
451 foreach my $n ( @{ $filters->{$f} } ) {
452 $filter_value->{$f}->{$n} = 1;
455 my @items = @{ $data->{items} };
459 foreach my $n ( keys %$filter_value ) {
460 if ( ! exists $i->{$n} ) {
461 if ( defined $filter_value->{$n}->{_missing} ) {
469 # and match any of values in element
471 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
472 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
474 if ( ! $have_values ) {
480 } @items if $filter_value;
485 sub _current_filters {
488 $current_filters->{ $_ } = $filters->{ $_ }
490 grep { defined $filters->{ $_ } }
491 @{ $self->_loaded('header') }
493 #warn "# current_filters ",dump($current_filters);
494 return $current_filters;
497 sub _data_sorted_by {
498 my ( $self, $order ) = @_;
500 my $path = $self->session('path');
502 warn "_data_sorted_by $order from $path";
504 if ( defined $loaded->{$path}->{sorted}->{$order} ) {
505 return $loaded->{$path}->{sorted}->{$order};
508 my $data = $self->_loaded( 'data' );
509 my $numeric = $self->_is_numeric($order);
510 my $missing = $numeric ? 0 : '';
511 no warnings qw(numeric);
522 [ $nr++, exists $_->{$order} ? join('', @{$_->{$order}}) : $missing ]
523 } grep { ref $_->{$order} eq 'ARRAY' } @{ $data->{items} }
526 warn "sorted: $order numeric: $numeric items: ", $#sorted + 1, "\n";
527 #warn "# sorted ",dump( @sorted );
529 $loaded->{$path}->{sorted}->{$order} = [ @sorted ];
536 if ( my $show = $self->param('id') ) {
537 $self->param('show', $show);
541 my $path = $self->session('path');
543 my @columns = $self->_param_array('columns');
544 $self->redirect_to('/data/columns') unless @columns;
545 my $order = $self->_param_scalar('order', $columns[0]);
546 my $sort = $self->_param_scalar('sort', 'a');
547 my $offset = $self->_param_scalar('offset', 0);
548 my $limit = $self->_param_scalar('limit', 20);
549 $self->_param_scalar('show', 'table');
551 # fix offset when changing limit
552 $offset = int( $offset / $limit ) * $limit;
554 my $sorted = $self->_data_sorted_by( $order );
558 foreach my $name ( keys %$filters ) {
559 if ( ! defined $loaded->{$path}->{stats}->{ $name } ) {
560 warn "skip filter $name not found in $path\n";
563 push @filter_names, $name;
565 warn "filter_names ",dump( @filter_names );
566 foreach my $name ( @filter_names ) {
567 next if ref $loaded->{$path}->{filters}->{$name} eq 'ARRAY';
568 $self->_filter_on_data( $name, @{ $filters->{$name} } );
572 my $all_filters = join(' ',sort @filter_names,'order:',$order);
574 # warn "# all_filters $all_filters ", dump( $loaded->{$path}->{filtered}->{$all_filters} );
576 if ( ! defined $loaded->{$path}->{filtered}->{$all_filters} ) {
578 my $path_filters = $loaded->{$path}->{filters};
580 warn "create combined filter for $all_filters\n";
583 foreach my $i ( 0 .. $#$sorted ) {
584 my $pos = $sorted->[$i];
586 if ( $#filter_names == -1 ) {
587 push @filtered, $pos;
592 foreach ( @filter_names ) {
593 $skip ||= 1 if ! defined $path_filters->{$_}->{$pos};
597 push @filtered, $pos;
600 $loaded->{$path}->{filtered}->{$all_filters} = [ @filtered ];
603 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
604 if defined $loaded->{$path}->{filtered}->{$all_filters};
606 warn "all_filters $all_filters produced ", $#$filtered + 1, " items\n" if $filtered;
608 my $data = $self->_loaded('data');
610 my $code = $self->_param_scalar('code','');
613 my $commit = $self->param('commit');
614 my $test = $self->param('test');
616 if ( $code && ( $test || $commit ) ) {
617 # XXX find columns used in code snippet and show them to user
618 foreach my $column ( $code =~ m/\$rec->{(.+?)}/g ) {
619 if ( $column =~ s/^(['"])// ) {
622 next if grep { /$column/ } @columns;
623 unshift @columns, $column;
625 $self->session('columns', [ @columns ]);
626 $loaded->{$path}->{columns} = [ @columns ];
632 warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
633 foreach ( 0 .. $#$filtered ) {
634 my $i = $filtered->[$_];
635 my $rec = $data->{items}->[$i];
641 my $from_end = $sort eq 'd' ? $#$filtered : 0;
642 foreach ( 0 .. $limit ) {
643 my $i = $_ + $offset;
644 last unless defined $filtered->[$i];
645 $i = $from_end - $i if $from_end;
646 my $id = $filtered->[$i];
647 my $rec = $data->{items}->[ $id ];
648 $rec->{_row_id} ||= $id;
649 if ( $code && $test ) {
650 $rec = Storable::dclone $rec;
653 warn "ERROR evaling\n$code\n$@";
654 $self->stash('eval_error', $@) if $@;
656 warn "EVAL ",dump($rec);
659 push @$sorted_items, $rec;
662 warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
668 sorted => $sorted_items,
669 columns => [ @columns ],
670 rows => $#$filtered + 1,
671 numeric => { map { $_, $self->_is_numeric($_) } @columns },
672 filters => $self->_current_filters,
681 $self->session('order', $self->param('order'));
682 $self->session('sort', $self->param('sort'));
683 $self->redirect_to('/data/items');
687 my ( $self, $name ) = @_;
689 my $stats = $self->_loaded( 'stats' );
691 # sort facet numerically if more >50% elements are numeric
692 my $count = $stats->{$name}->{count};
693 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
694 defined $stats->{$name}->{numeric} &&
695 $stats->{$name}->{numeric} > $count / 2;
699 my ($self,$name) = @_;
700 warn "_remove_filter $name\n";
702 my $path = $self->session('path');
704 delete $filters->{$name};
705 delete $loaded->{$path}->{filters}->{$name};
706 warn "filters left: ", keys %{ $loaded->{$path}->{filters} };
710 keys %{ $loaded->{$path}->{filtered} }
712 delete $loaded->{$path}->{filtered}->{$_};
713 warn "remove filtered cache $_";
720 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
722 if ( my $name = $self->param('remove') ) {
723 $self->_remove_filter( $name );
724 $self->redirect_to( '/data/items' );
728 my $name = $self->param('name') || die "no name";
730 my $all = $self->_param_scalar('all', 1);
731 my $data = $self->_loaded('data');
733 my $filters = $self->_current_filters;
734 my $all_filters = join(' ',sort keys %$filters,'order:',$self->session('order'));
735 my $filtered = $loaded->{$path}->{filtered}->{$all_filters}
736 if defined $loaded->{$path}->{filtered}->{$all_filters};
738 if ( ! $filtered || $all ) {
739 $filtered = [ 0 .. $#{ $data->{items} } ];
740 warn "filter all values\n";
742 warn "filter using $all_filters\n";
745 foreach my $i ( @$filtered ) {
746 my $item = $data->{items}->[$i];
747 if ( ! exists $item->{$name} ) {
748 $facet->{ _missing }++;
749 } elsif ( ref $item->{$name} eq 'ARRAY' ) {
750 $facet->{$_}++ foreach @{ $item->{$name} };
752 $facet->{ $item->{$name} }++;
756 # warn "# facet $name ",dump $facet;
761 : defined $filters->{$name} ? @{ $filters->{$name} }
764 $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
766 my $numeric = $self->_is_numeric($name);
768 my $sort = $self->param('sort');
769 # sort numeric facets with more than 5 values ascending
770 $sort ||= $numeric && $#facet_names > 4 ? 'a' : 'c';
772 @facet_names = sort {
774 if ( $sort eq 'a' ) {
775 $result = $numeric ? $a <=> $b : lc $a cmp lc $b;
776 } elsif ( $sort eq 'd' ) {
777 $result = $numeric ? $b <=> $a : lc $b cmp lc $a;
778 } elsif ( $sort eq 'c' ) {
779 $result = ( $facet->{$b} || -1 ) <=> ( $facet->{$a} || -1 )
781 warn "unknown sort: $sort";
784 $result = $a cmp $b unless defined $result; # FIXME cludge for numeric facets with invalid data
788 $self->render( name => $name, facet => $facet, checked => $checked,
789 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
794 sub __invalidate_path_column {
795 my ( $path, $name ) = @_;
797 if ( defined $loaded->{$path}->{sorted}->{$name} ) {
798 delete $loaded->{$path}->{sorted}->{$name};
799 warn "# invalidate $path sorted $name\n";
802 foreach ( grep { m/$name/ } keys %{ $loaded->{$path}->{filtered} } ) {
803 delete $loaded->{$path}->{filtered}->{$_};
804 warn "# invalidate $path filtered $_\n";
808 sub __path_modified {
809 my ( $path, $value ) = @_;
810 $value = 1 unless defined $value;
812 $loaded->{$path}->{modified} = $value;
814 warn "# __path_modified $path $value\n";
819 my $new_content = $self->param('new_content');
820 $new_content ||= $self->param('content'); # backward compatibility with old actions
822 my $i = $self->param('_row_id');
823 die "invalid _row_id ",dump($i) unless $i =~ m/^\d+$/;
824 my $path = $self->param('path') || die "no path";
825 my $name = $self->param('name') || die "no name";
826 my $status = 200; # 200 = OK, 201 = Created
828 my $data = $self->_loaded('data');
830 if ( defined $loaded->{$path}->{data}->{items}->[$i] ) {
831 $new_content =~ s/^\s+//s;
832 $new_content =~ s/\s+$//s;
834 if ( $new_content =~ /\xB6/ ) { # para
835 $v = [ split(/\s*\xB6\s*/, $new_content) ];
837 $v = [ $new_content ];
840 my $old = dump $loaded->{$path}->{data}->{items}->[$i]->{$name};
843 && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
849 old => $loaded->{$path}->{data}->{items}->[$i]->{$name},
851 time => $self->param('time') || time(),
852 user => $self->param('user') || $ENV{'LOGNAME'},
854 map { $_ => $loaded->{$path}->{data}->{items}->[$i]->{$_}->[0] }
855 grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
856 keys %{ $loaded->{$path}->{stats} }
859 my $change_path = $self->_permanent_path( 'changes' );
860 mkdir $change_path unless -d $change_path;
861 $change_path .= '/' . $change->{time};
862 store $change, $change_path;
863 utime $change->{time}, $change->{time}, $change_path;
864 warn "# $change_path ", dump($change);
866 warn "# change $path $i $old -> $new\n";
867 $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
869 __invalidate_path_column( $path, $name );
871 $status = 201; # created
872 # modified = 2 -- force rebuild of stats
873 __path_modified( $path, 2 );
875 $new_content = join("\xB6",@$v);
878 warn "# unchanged $path $i $old\n";
882 $new_content = "$path $i $name doesn't exist\n";
886 warn "# edit $status ", dump $new_content;
890 new_content => scalar $new_content,
897 my $path = $self->_param_or_session('path');
898 my $dump_path = $self->_save( $path );
899 __path_modified( $path, 0 );
901 $self->redirect_to( '/data/items' );
907 if ( my $import = $self->param('import') ) {
909 if ( $import =~ m{/filter\.(.+?)\..+} ) {
911 my @vals = map { chomp; $_ }
912 read_file $self->app->home->rel_dir('public') . "/export/$import";
913 $self->_remove_filter( $name );
914 $self->_filter_on_data( $name, @vals );
915 $self->session( 'offset' => 0 );
916 $self->redirect_to('/data/items');
918 warn "UNKNOWN IMPORT $import";
922 $self->render( export => [
923 glob( $self->_export_path . '*' )
929 grep { defined $loaded->{$_}->{data} }