1 package MojoFacets::Data;
6 use base 'Mojolicious::Controller';
8 use Data::Dump qw(dump);
20 my $path = $self->app->home->rel_dir('data');
21 die "no data dir $path" unless -d $path;
25 my $file = $File::Find::name;
26 if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
27 $file =~ s/$path\/*//;
30 warn "IGNORE: $file\n";
34 @files = sort { lc $a cmp lc $b } @files;
36 $size->{$_} = -s "$path/$_" foreach @files;
46 my ( $self, $path ) = @_;
48 return if defined $loaded->{$path}->{data};
50 my $full_path = $self->app->home->rel_file( 'data/' . $path );
51 die "$full_path $!" unless -r $full_path;
53 # we could use Mojo::JSON here, but it's too slow
54 # $data = from_json read_file $path;
55 my $data = read_file $full_path;
56 warn "# data snippet: ", substr($data,0,200);
58 if ( $path =~ m/\.js/ ) {
59 Encode::_utf8_on($data);
60 $data = from_json $data;
61 } elsif ( $path =~ m/\.txt/ ) {
62 my @lines = split(/\r?\n/, $data);
63 $data = { items => [] };
65 my $header_line = shift @lines;
66 my $multiline = $header_line =~ s/\^//g;
67 @header = split(/\|/, $header_line );
68 warn "# header ", dump( @header );
69 while ( my $line = shift @lines ) {
72 my @v = split(/\|/, $line);
73 while ( @lines && $#v < $#header ) {
77 my @more_v = split(/\|/, $line);
78 if ( $#v + $#more_v > $#header ) {
79 warn "short line: ",dump( @v );
83 $v[ $#v ] .= shift @more_v;
84 push @v, @more_v if @more_v;
86 if ( $#v > $#header ) {
87 die "# splice $#header ", dump( @v );
88 @v = splice @v, 0, $#header;
92 $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
93 push @{ $data->{items} }, $item;
96 warn "file format unknown $path";
101 foreach my $e ( @{ $data->{items} } ) {
102 foreach my $n ( keys %$e ) {
103 $stats->{$n}->{count}++;
104 if ( ref $e->{$n} eq 'ARRAY' ) {
106 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
108 foreach my $x ( @{$e->{$n}} ) {
109 $stats->{$n}->{numeric}++
110 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
114 $stats->{$n}->{numeric}++
115 if $e->{$n} =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
120 foreach my $n ( keys %$stats ) {
121 next unless defined $stats->{$n}->{array};
122 delete $stats->{$n}->{array}
123 if $stats->{$n}->{array} == $stats->{$n}->{count};
127 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
128 grep { defined $stats->{$_}->{count} } keys %$stats
133 $loaded->{ $path } = {
134 header => [ @header ],
136 full_path => $full_path,
137 size => -s $full_path,
147 my @paths = $self->param('paths');
148 warn "# paths ", dump @paths;
149 $self->_load_path( $_ ) foreach @paths;
151 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
152 warn "# path $path\n";
153 $self->session('path' => $path);
154 $self->_load_path( $path );
156 $self->session( 'header' => $loaded->{$path}->{header} );
157 $self->session( 'columns' => $loaded->{$path}->{header} );
159 $self->session( 'filters' => {} );
161 $self->redirect_to( '/data/columns' );
166 my ( $self, $name ) = @_;
167 my $path = $self->session('path');
168 die "$path $name doesn't exist in loaded ",dump( $loaded )
169 unless defined $loaded->{$path}->{$name};
170 return $loaded->{$path}->{$name};
179 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
182 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
184 foreach my $c ( @{ $self->session( 'header' ) } ) {
185 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
189 message => 'Select columns to display',
191 columns => \@columns,
192 checked => $self->_checked( $self->_perm_array('columns') ),
197 my ($self,$name) = @_;
199 my @array = $self->param($name);
202 $self->session($name => [ @array ]);
203 } elsif ( my $session = $self->session($name) ) {
204 if ( ref $session eq 'ARRAY' ) {
207 die "$name not array ",dump($session);
210 warn "# $name ",dump @array;
215 my ($self,$name,$default) = @_;
217 my $scalar = $self->param($name);
219 if ( defined $scalar ) {
220 $self->session($name => $scalar);
222 $scalar = $self->session($name);
225 if ( ! defined $scalar ) {
227 die "no default for $name" unless defined $scalar;
228 $self->session($name => $scalar);
231 warn "# $name ",dump $scalar;
238 my $name = $self->param('filter_name') || die "name?";
239 my @vals = $self->param('filter_vals');
241 warn "# filter $name vals ",dump(@vals);
243 my $filters = $self->session('filters');
245 $filters->{$name} = [ @vals ];
247 delete $filters->{$name};
249 $self->session( 'filters' => $filters );
251 warn "# filters ",dump($self->session('filters'));
253 $self->session( 'offset' => 0 );
255 $self->redirect_to('/data/items');
259 my ( $self, $filters, $i ) = @_;
261 foreach my $n ( keys %$filters ) {
262 my @filter_values = @{ $filters->{$n} };
263 my $include_missing = grep { /^_missing/ } @filter_values;
264 if ( ! exists $i->{$n} ) {
265 if ( $include_missing ) {
273 # and match any of values in element
275 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
276 $have_values ||= 1 if grep { m/^\Q$v\E$/ } @filter_values;
278 if ( ! $have_values ) {
288 my $filters = $self->session('filters');
289 my $data = $self->_loaded( 'data' );
291 $filters ? $self->_filter_item( $filters, $_ ) : 1;
292 } @{ $data->{items} };
298 $self->redirect_to('/data/index') unless defined $loaded->{ $self->session('path') };
300 my @columns = $self->_perm_array('columns');
301 $self->redirect_to('/data/columns') unless @columns;
302 my $order = $self->_perm_scalar('order', $columns[0]);
303 my $sort = $self->_perm_scalar('sort', 'a');
304 my $offset = $self->_perm_scalar('offset', 0);
305 my $limit = $self->_perm_scalar('limit', 20);
306 $self->_perm_scalar('show', 'table');
308 # fix offset when changing limit
309 $offset = int( $offset / $limit ) * $limit;
311 # FIXME - multi-level sort
312 my $numeric = $self->_is_numeric($order);
313 my $missing = $numeric ? 0 : '';
314 no warnings qw(numeric);
316 my $v1 = exists $a->{$order} ? join('', @{$a->{$order}}) : $missing;
317 my $v2 = exists $b->{$order} ? join('', @{$b->{$order}}) : $missing;
318 ($v1,$v2) = ($v2,$v1) if $sort eq 'd';
319 $numeric ? $v1 <=> $v2 : $v1 cmp $v2 ;
320 } $self->_data_items;
322 # warn "# sorted ", dump @sorted;
324 my $rows = $#sorted + 1;
330 sorted => [ splice @sorted, $offset, $limit ],
331 columns => [ @columns ],
333 numeric => { map { $_, $self->_is_numeric($_) } @columns },
341 $self->session('order', $self->param('order'));
342 $self->session('sort', $self->param('sort'));
343 $self->redirect_to('/data/items');
347 my ( $self, $name ) = @_;
349 my $stats = $self->_loaded( 'stats' );
351 # sort facet numerically if more >50% elements are numeric
352 defined $stats->{$name}->{numeric} &&
353 $stats->{$name}->{numeric} > $stats->{$name}->{count} / 2;
359 if ( my $remove = $self->param('remove') ) {
360 my $f = $self->session('filters');
361 delete $f->{$remove};
362 $self->session( 'filters' => $f );
363 $self->redirect_to( '/data/items' );
367 my $name = $self->param('name') || die "no name";
369 foreach my $i ( $self->_data_items ) {
370 if ( ! exists $i->{$name} ) {
371 $facet->{ _missing }++;
372 } elsif ( ref $i->{$name} eq 'ARRAY' ) {
373 $facet->{$_}++ foreach @{ $i->{$name} };
375 $facet->{ $i->{$name} }++;
379 # warn "# facet $name ",dump $facet;
382 if ( my $f = $self->session('filters') ) {
383 if ( defined $f->{$name} ) {
384 $checked = $self->_checked( @{ $f->{$name} } );
388 my $sort = $self->param('sort') || 'c';
390 # sort facet numerically if more >50% elements are numeric
391 my $numeric = $self->_is_numeric($name);
393 my @facet_names = sort {
394 if ( $sort =~ m/a/i ) {
395 $numeric ? $a <=> $b : lc $a cmp lc $b;
396 } elsif ( $sort =~ m/d/i ) {
397 $numeric ? $b <=> $a : lc $b cmp lc $a;
398 } elsif ( $sort =~ m/c/i ) {
399 $facet->{$b} <=> $facet->{$a};
401 warn "unknown sort: $sort";
406 $self->render( name => $name, facet => $facet, checked => $checked,
407 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
414 $checked->{$_}++ foreach @_;
415 warn "# _checked ",dump($checked);