1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
21 my $path = $self->app->home->rel_dir('data');
22 die "no data dir $path" unless -d $path;
26 my $file = $File::Find::name;
27 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
28 $file =~ s/$path\/*//;
31 warn "IGNORE: $file\n";
35 @files = sort { lc $a cmp lc $b } @files;
37 $size->{$_} = -s "$path/$_" foreach @files;
48 my ( $self, $path ) = @_;
50 return if defined $loaded->{$path}->{data};
52 my $full_path = $self->app->home->rel_file( 'data/' . $path );
53 die "$full_path $!" unless -r $full_path;
55 # we could use Mojo::JSON here, but it's too slow
56 # $data = from_json read_file $path;
57 my $data = read_file $full_path;
58 warn "# data snippet: ", substr($data,0,200);
60 if ( $path =~ m/\.js/ ) {
61 Encode::_utf8_on($data);
62 $data = from_json $data;
63 } elsif ( $path =~ m/\.txt/ ) {
64 my @lines = split(/\r?\n/, $data);
65 $data = { items => [] };
67 my $header_line = shift @lines;
68 my $multiline = $header_line =~ s/\^//g;
69 @header = split(/\|/, $header_line );
70 warn "# header ", dump( @header );
71 while ( my $line = shift @lines ) {
74 my @v = split(/\|/, $line);
75 while ( @lines && $#v < $#header ) {
79 my @more_v = split(/\|/, $line);
80 if ( $#v + $#more_v > $#header ) {
81 warn "short line: ",dump( @v );
85 $v[ $#v ] .= shift @more_v if @more_v;
86 push @v, @more_v if @more_v;
88 if ( $#v > $#header ) {
89 die "# splice $#header ", dump( @v );
90 @v = splice @v, 0, $#header;
94 $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
95 push @{ $data->{items} }, $item;
98 warn "file format unknown $path";
103 foreach my $e ( @{ $data->{items} } ) {
104 foreach my $n ( keys %$e ) {
105 $stats->{$n}->{count}++;
107 if ( ref $e->{$n} eq 'ARRAY' ) {
108 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
114 foreach my $x ( @v ) {
115 $stats->{$n}->{numeric}++
116 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
117 $stats->{$n}->{empty}++
124 foreach my $n ( keys %$stats ) {
125 next unless defined $stats->{$n}->{array};
126 delete $stats->{$n}->{array}
127 if $stats->{$n}->{array} == $stats->{$n}->{count};
131 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
132 grep { defined $stats->{$_}->{count} } keys %$stats
137 $loaded->{ $path } = {
138 header => [ @header ],
140 full_path => $full_path,
141 size => -s $full_path,
151 my @paths = $self->param('paths');
152 warn "# paths ", dump @paths;
153 $self->_load_path( $_ ) foreach @paths;
155 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
156 warn "# path $path\n";
157 $self->session('path' => $path);
158 $self->_load_path( $path );
160 $self->session( 'header' => $loaded->{$path}->{header} );
161 if ( ! defined $loaded->{$path}->{columns} ) {
162 $self->session( 'columns' => $loaded->{$path}->{header} );
163 $self->redirect_to( '/data/columns' );
165 $self->session( 'columns' => $loaded->{$path}->{columns} );
166 $self->redirect_to( '/data/items' );
173 my ( $self, $name ) = @_;
174 my $path = $self->session('path');
175 die "$path $name doesn't exist in loaded ",dump( $loaded )
176 unless defined $loaded->{$path}->{$name};
177 return $loaded->{$path}->{$name};
184 $checked->{$_}++ foreach @_;
185 warn "# _checked ",dump($checked);
193 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
196 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
198 foreach my $c ( @{ $self->session( 'header' ) } ) {
199 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
203 message => 'Select columns to display',
205 columns => \@columns,
206 checked => $self->_checked( $self->_perm_array('columns') ),
211 my ($self,$name) = @_;
213 my @array = $self->param($name);
214 my $path = $self->session('path');
217 $self->session($name => [ @array ]);
218 $loaded->{$path}->{$name} = [ @array ];
219 } elsif ( my $session = $self->session($name) ) {
220 if ( ref $session eq 'ARRAY' ) {
223 die "$name not array ",dump($session);
226 warn "# $name ",dump @array;
231 my ($self,$name,$default) = @_;
233 my $scalar = $self->param($name);
235 if ( defined $scalar ) {
236 $self->session($name => $scalar);
238 $scalar = $self->session($name);
241 if ( ! defined $scalar ) {
243 die "no default for $name" unless defined $scalar;
244 $self->session($name => $scalar);
247 warn "# $name ",dump $scalar;
254 my $name = $self->param('filter_name') || die "name?";
255 my @vals = $self->param('filter_vals');
257 warn "# filter $name vals ",dump(@vals);
260 $filters->{$name} = [ @vals ];
261 warn "# filter + $name $#vals\n";
263 warn "# filter - $name\n";
264 delete $filters->{$name};
267 warn "# filters ",dump($filters);
269 $self->session( 'offset' => 0 );
271 $self->redirect_to('/data/items');
277 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
278 my $data = $self->_loaded( 'data' );
280 my $filters = $self->_current_filters;
282 foreach my $f ( keys %$filters ) {
283 foreach my $n ( @{ $filters->{$f} } ) {
284 $filter_value->{$f}->{$n} = 1;
287 my @items = @{ $data->{items} };
291 foreach my $n ( keys %$filter_value ) {
292 if ( ! exists $i->{$n} ) {
293 if ( defined $filter_value->{$n}->{_missing} ) {
301 # and match any of values in element
303 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
304 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
306 if ( ! $have_values ) {
312 } @items if $filter_value;
317 sub _current_filters {
320 $current_filters->{ $_ } = $filters->{ $_ }
322 grep { defined $filters->{ $_ } }
323 @{ $self->_loaded('header') }
325 warn "# current_filters ",dump($current_filters);
326 return $current_filters;
333 my $path = $self->session('path');
334 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
336 my @columns = $self->_perm_array('columns');
337 $self->redirect_to('/data/columns') unless @columns;
338 my $order = $self->_perm_scalar('order', $columns[0]);
339 my $sort = $self->_perm_scalar('sort', 'a');
340 my $offset = $self->_perm_scalar('offset', 0);
341 my $limit = $self->_perm_scalar('limit', 20);
342 $self->_perm_scalar('show', 'table');
344 # fix offset when changing limit
345 $offset = int( $offset / $limit ) * $limit;
347 # FIXME - multi-level sort
348 my $numeric = $self->_is_numeric($order);
349 my $missing = $numeric ? 0 : '';
350 no warnings qw(numeric);
352 my $v1 = exists $a->{$order} ? join('', @{$a->{$order}}) : $missing;
353 my $v2 = exists $b->{$order} ? join('', @{$b->{$order}}) : $missing;
354 ($v1,$v2) = ($v2,$v1) if $sort eq 'd';
355 $numeric ? $v1 <=> $v2 : $v1 cmp $v2 ;
356 } $self->_data_items;
358 # warn "# sorted ", dump @sorted;
360 my $rows = $#sorted + 1;
366 sorted => [ splice @sorted, $offset, $limit ],
367 columns => [ @columns ],
369 numeric => { map { $_, $self->_is_numeric($_) } @columns },
370 filters => $self->_current_filters,
378 $self->session('order', $self->param('order'));
379 $self->session('sort', $self->param('sort'));
380 $self->redirect_to('/data/items');
384 my ( $self, $name ) = @_;
386 my $stats = $self->_loaded( 'stats' );
388 # sort facet numerically if more >50% elements are numeric
389 my $count = $stats->{$name}->{count};
390 $count -= $stats->{$name}->{empty} if defined $stats->{$name}->{empty};
391 defined $stats->{$name}->{numeric} &&
392 $stats->{$name}->{numeric} > $count / 2;
398 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
400 if ( my $remove = $self->param('remove') ) {
401 delete $filters->{$remove};
402 $self->redirect_to( '/data/items' );
406 my $name = $self->param('name') || die "no name";
408 foreach my $i ( $self->_data_items ) {
409 if ( ! exists $i->{$name} ) {
410 $facet->{ _missing }++;
411 } elsif ( ref $i->{$name} eq 'ARRAY' ) {
412 $facet->{$_}++ foreach @{ $i->{$name} };
414 $facet->{ $i->{$name} }++;
418 # warn "# facet $name ",dump $facet;
421 if ( defined $filters->{$name} ) {
422 $checked = $self->_checked( @{ $filters->{$name} } );
425 my $sort = $self->param('sort') || 'c';
427 # sort facet numerically if more >50% elements are numeric
428 my $numeric = $self->_is_numeric($name);
430 my @facet_names = sort {
431 if ( $sort =~ m/a/i ) {
432 $numeric ? $a <=> $b : lc $a cmp lc $b;
433 } elsif ( $sort =~ m/d/i ) {
434 $numeric ? $b <=> $a : lc $b cmp lc $a;
435 } elsif ( $sort =~ m/c/i ) {
436 $facet->{$b} <=> $facet->{$a};
438 warn "unknown sort: $sort";
443 $self->render( name => $name, facet => $facet, checked => $checked,
444 facet_names => \@facet_names, sort => $sort, numeric => $numeric,