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}++;
106 if ( ref $e->{$n} eq 'ARRAY' ) {
108 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
110 foreach my $x ( @{$e->{$n}} ) {
111 $stats->{$n}->{numeric}++
112 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
116 $stats->{$n}->{numeric}++
117 if $e->{$n} =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
122 foreach my $n ( keys %$stats ) {
123 next unless defined $stats->{$n}->{array};
124 delete $stats->{$n}->{array}
125 if $stats->{$n}->{array} == $stats->{$n}->{count};
129 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
130 grep { defined $stats->{$_}->{count} } keys %$stats
135 $loaded->{ $path } = {
136 header => [ @header ],
138 full_path => $full_path,
139 size => -s $full_path,
149 my @paths = $self->param('paths');
150 warn "# paths ", dump @paths;
151 $self->_load_path( $_ ) foreach @paths;
153 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
154 warn "# path $path\n";
155 $self->session('path' => $path);
156 $self->_load_path( $path );
158 $self->session( 'header' => $loaded->{$path}->{header} );
159 if ( ! defined $loaded->{$path}->{columns} ) {
160 $self->session( 'columns' => $loaded->{$path}->{header} );
161 $self->redirect_to( '/data/columns' );
163 $self->session( 'columns' => $loaded->{$path}->{columns} );
164 $self->redirect_to( '/data/items' );
171 my ( $self, $name ) = @_;
172 my $path = $self->session('path');
173 die "$path $name doesn't exist in loaded ",dump( $loaded )
174 unless defined $loaded->{$path}->{$name};
175 return $loaded->{$path}->{$name};
182 $checked->{$_}++ foreach @_;
183 warn "# _checked ",dump($checked);
191 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
194 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
196 foreach my $c ( @{ $self->session( 'header' ) } ) {
197 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
201 message => 'Select columns to display',
203 columns => \@columns,
204 checked => $self->_checked( $self->_perm_array('columns') ),
209 my ($self,$name) = @_;
211 my @array = $self->param($name);
212 my $path = $self->session('path');
215 $self->session($name => [ @array ]);
216 $loaded->{$path}->{$name} = [ @array ];
217 } elsif ( my $session = $self->session($name) ) {
218 if ( ref $session eq 'ARRAY' ) {
221 die "$name not array ",dump($session);
224 warn "# $name ",dump @array;
229 my ($self,$name,$default) = @_;
231 my $scalar = $self->param($name);
233 if ( defined $scalar ) {
234 $self->session($name => $scalar);
236 $scalar = $self->session($name);
239 if ( ! defined $scalar ) {
241 die "no default for $name" unless defined $scalar;
242 $self->session($name => $scalar);
245 warn "# $name ",dump $scalar;
252 my $name = $self->param('filter_name') || die "name?";
253 my @vals = $self->param('filter_vals');
255 warn "# filter $name vals ",dump(@vals);
258 $filters->{$name} = [ @vals ];
260 delete $filters->{$name};
263 warn "# filters ",dump($filters);
265 $self->session( 'offset' => 0 );
267 $self->redirect_to('/data/items');
273 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
274 my $data = $self->_loaded( 'data' );
276 my $filters = $self->_current_filters;
278 foreach my $f ( keys %$filters ) {
279 foreach my $n ( @{ $filters->{$f} } ) {
280 $filter_value->{$f}->{$n} = 1;
283 my @items = @{ $data->{items} };
287 foreach my $n ( keys %$filter_value ) {
288 if ( ! exists $i->{$n} ) {
289 if ( defined $filter_value->{$n}->{_missing} ) {
297 # and match any of values in element
299 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
300 $have_values ||= 1 if defined $filter_value->{$n}->{$v};
302 if ( ! $have_values ) {
308 } @items if $filter_value;
313 sub _current_filters {
316 $current_filters->{ $_ } = $filters->{ $_ }
318 grep { defined $filters->{ $_ } }
319 @{ $self->_loaded('header') }
321 warn "# current_filters ",dump($current_filters);
322 return $current_filters;
329 my $path = $self->session('path');
330 $self->redirect_to('/data/index') unless defined $loaded->{ $path };
332 my @columns = $self->_perm_array('columns');
333 $self->redirect_to('/data/columns') unless @columns;
334 my $order = $self->_perm_scalar('order', $columns[0]);
335 my $sort = $self->_perm_scalar('sort', 'a');
336 my $offset = $self->_perm_scalar('offset', 0);
337 my $limit = $self->_perm_scalar('limit', 20);
338 $self->_perm_scalar('show', 'table');
340 # fix offset when changing limit
341 $offset = int( $offset / $limit ) * $limit;
343 # FIXME - multi-level sort
344 my $numeric = $self->_is_numeric($order);
345 my $missing = $numeric ? 0 : '';
346 no warnings qw(numeric);
348 my $v1 = exists $a->{$order} ? join('', @{$a->{$order}}) : $missing;
349 my $v2 = exists $b->{$order} ? join('', @{$b->{$order}}) : $missing;
350 ($v1,$v2) = ($v2,$v1) if $sort eq 'd';
351 $numeric ? $v1 <=> $v2 : $v1 cmp $v2 ;
352 } $self->_data_items;
354 # warn "# sorted ", dump @sorted;
356 my $rows = $#sorted + 1;
362 sorted => [ splice @sorted, $offset, $limit ],
363 columns => [ @columns ],
365 numeric => { map { $_, $self->_is_numeric($_) } @columns },
366 filters => $self->_current_filters,
374 $self->session('order', $self->param('order'));
375 $self->session('sort', $self->param('sort'));
376 $self->redirect_to('/data/items');
380 my ( $self, $name ) = @_;
382 my $stats = $self->_loaded( 'stats' );
384 # sort facet numerically if more >50% elements are numeric
385 defined $stats->{$name}->{numeric} &&
386 $stats->{$name}->{numeric} > $stats->{$name}->{count} / 2;
392 my $path = $self->session('path') || $self->redirect_to( '/data/index' );
394 if ( my $remove = $self->param('remove') ) {
395 delete $filters->{$remove};
396 $self->redirect_to( '/data/items' );
400 my $name = $self->param('name') || die "no name";
402 foreach my $i ( $self->_data_items ) {
403 if ( ! exists $i->{$name} ) {
404 $facet->{ _missing }++;
405 } elsif ( ref $i->{$name} eq 'ARRAY' ) {
406 $facet->{$_}++ foreach @{ $i->{$name} };
408 $facet->{ $i->{$name} }++;
412 # warn "# facet $name ",dump $facet;
415 if ( defined $filters->{$name} ) {
416 $checked = $self->_checked( @{ $filters->{$name} } );
419 my $sort = $self->param('sort') || 'c';
421 # sort facet numerically if more >50% elements are numeric
422 my $numeric = $self->_is_numeric($name);
424 my @facet_names = sort {
425 if ( $sort =~ m/a/i ) {
426 $numeric ? $a <=> $b : lc $a cmp lc $b;
427 } elsif ( $sort =~ m/d/i ) {
428 $numeric ? $b <=> $a : lc $b cmp lc $a;
429 } elsif ( $sort =~ m/c/i ) {
430 $facet->{$b} <=> $facet->{$a};
432 warn "unknown sort: $sort";
437 $self->render( name => $name, facet => $facet, checked => $checked,
438 facet_names => \@facet_names, sort => $sort, numeric => $numeric,