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 ) {
71 my @v = split(/\|/, $line);
72 while ( @lines && $#v < $#header ) {
76 my @more_v = split(/\|/, $line);
77 if ( $#v + $#more_v > $#header ) {
78 warn "short line: ",dump( @v );
82 $v[ $#v ] .= shift @more_v;
83 push @v, @more_v if @more_v;
85 if ( $#v > $#header ) {
86 die "# splice $#header ", dump( @v );
87 @v = splice @v, 0, $#header;
91 $item->{ $header[$_] || "f_$_" } = [ $v[$_] ] foreach ( 0 .. $#v );
92 push @{ $data->{items} }, $item;
95 warn "file format unknown $path";
100 foreach my $e ( @{ $data->{items} } ) {
101 foreach my $n ( keys %$e ) {
102 $stats->{$n}->{count}++;
103 if ( ref $e->{$n} eq 'ARRAY' ) {
105 $stats->{$n}->{array} += $#{ $e->{$n} } + 1;
107 foreach my $x ( @{$e->{$n}} ) {
108 $stats->{$n}->{numeric}++
109 if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
113 $stats->{$n}->{numeric}++
114 if $e->{$n} =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
119 foreach my $n ( keys %$stats ) {
120 next unless defined $stats->{$n}->{array};
121 delete $stats->{$n}->{array}
122 if $stats->{$n}->{array} == $stats->{$n}->{count};
126 sort { $stats->{$b}->{count} <=> $stats->{$a}->{count} }
127 grep { defined $stats->{$_}->{count} } keys %$stats
132 $loaded->{ $path } = {
133 header => [ @header ],
135 full_path => $full_path,
136 size => -s $full_path,
146 my @paths = $self->param('paths');
147 warn "# paths ", dump @paths;
148 $self->_load_path( $_ ) foreach @paths;
150 my $path = $self->param('path') || $self->redirect_to( '/data/index' );
151 warn "# path $path\n";
152 $self->session('path' => $path);
153 $self->_load_path( $path );
155 $self->session( 'header' => $loaded->{$path}->{header} );
156 $self->session( 'columns' => $loaded->{$path}->{header} );
158 $self->session( 'filters' => {} );
160 $self->redirect_to( '/data/columns' );
165 my ( $self, $name ) = @_;
166 my $path = $self->session('path');
167 die "$path $name doesn't exist in loaded ",dump( $loaded )
168 unless defined $loaded->{$path}->{$name};
169 return $loaded->{$path}->{$name};
178 my $stats = $self->_loaded( 'stats' ); # || $self->redirect_to( '/data/index' );
181 @columns = grep { defined $stats->{$_}->{count} } @{ $self->session('columns') } if $self->session('columns');
183 foreach my $c ( @{ $self->session( 'header' ) } ) {
184 push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
188 message => 'Select columns to display',
190 columns => \@columns,
191 checked => $self->_checked( $self->_perm_array('columns') ),
196 my ($self,$name) = @_;
198 my @array = $self->param($name);
201 $self->session($name => [ @array ]);
202 } elsif ( my $session = $self->session($name) ) {
203 if ( ref $session eq 'ARRAY' ) {
206 die "$name not array ",dump($session);
209 warn "# $name ",dump @array;
214 my ($self,$name,$default) = @_;
216 my $scalar = $self->param($name);
218 if ( defined $scalar ) {
219 $self->session($name => $scalar);
221 $scalar = $self->session($name);
224 if ( ! defined $scalar ) {
226 die "no default for $name" unless defined $scalar;
227 $self->session($name => $scalar);
230 warn "# $name ",dump $scalar;
237 my $name = $self->param('filter_name') || die "name?";
238 my @vals = $self->param('filter_vals');
240 warn "# filter $name vals ",dump(@vals);
242 my $filters = $self->session('filters');
244 $filters->{$name} = [ @vals ];
246 delete $filters->{$name};
248 $self->session( 'filters' => $filters );
250 warn "# filters ",dump($self->session('filters'));
252 $self->session( 'offset' => 0 );
254 $self->redirect_to('/data/items');
258 my ( $self, $filters, $i ) = @_;
260 foreach my $n ( keys %$filters ) {
261 my @filter_values = @{ $filters->{$n} };
262 my $include_missing = grep { /^_missing/ } @filter_values;
263 if ( ! exists $i->{$n} ) {
264 if ( $include_missing ) {
272 # and match any of values in element
274 foreach my $v ( @{ $i->{$n} } ) { # FIXME not array?
275 $have_values ||= 1 if grep { m/^\Q$v\E$/ } @filter_values;
277 if ( ! $have_values ) {
287 my $filters = $self->session('filters');
288 my $data = $self->_loaded( 'data' );
290 $filters ? $self->_filter_item( $filters, $_ ) : 1;
291 } @{ $data->{items} };
297 $self->redirect_to('/data/index') unless defined $loaded->{ $self->session('path') };
299 my @columns = $self->_perm_array('columns');
300 $self->redirect_to('/data/columns') unless @columns;
301 my $order = $self->_perm_scalar('order', $columns[0]);
302 my $sort = $self->_perm_scalar('sort', 'a');
303 my $offset = $self->_perm_scalar('offset', 0);
304 my $limit = $self->_perm_scalar('limit', 20);
305 $self->_perm_scalar('show', 'table');
307 # fix offset when changing limit
308 $offset = int( $offset / $limit ) * $limit;
310 # FIXME - multi-level sort
311 my $numeric = $self->_is_numeric($order);
312 my $missing = $numeric ? 0 : '';
313 no warnings qw(numeric);
315 my $v1 = exists $a->{$order} ? join('', @{$a->{$order}}) : $missing;
316 my $v2 = exists $b->{$order} ? join('', @{$b->{$order}}) : $missing;
317 ($v1,$v2) = ($v2,$v1) if $sort eq 'd';
318 $numeric ? $v1 <=> $v2 : $v1 cmp $v2 ;
319 } $self->_data_items;
321 # warn "# sorted ", dump @sorted;
323 my $rows = $#sorted + 1;
329 sorted => [ splice @sorted, $offset, $limit ],
330 columns => [ @columns ],
332 numeric => { map { $_, $self->_is_numeric($_) } @columns },
340 $self->session('order', $self->param('order'));
341 $self->session('sort', $self->param('sort'));
342 $self->redirect_to('/data/items');
346 my ( $self, $name ) = @_;
348 my $stats = $self->_loaded( 'stats' );
350 # sort facet numerically if more >50% elements are numeric
351 defined $stats->{$name}->{numeric} &&
352 $stats->{$name}->{numeric} > $stats->{$name}->{count} / 2;
358 if ( my $remove = $self->param('remove') ) {
359 my $f = $self->session('filters');
360 delete $f->{$remove};
361 $self->session( 'filters' => $f );
362 $self->redirect_to( '/data/items' );
366 my $name = $self->param('name') || die "no name";
368 foreach my $i ( $self->_data_items ) {
369 if ( ! exists $i->{$name} ) {
370 $facet->{ _missing }++;
371 } elsif ( ref $i->{$name} eq 'ARRAY' ) {
372 $facet->{$_}++ foreach @{ $i->{$name} };
374 $facet->{ $i->{$name} }++;
378 # warn "# facet $name ",dump $facet;
381 if ( my $f = $self->session('filters') ) {
382 if ( defined $f->{$name} ) {
383 $checked = $self->_checked( @{ $f->{$name} } );
387 my $sort = $self->param('sort') || 'c';
389 # sort facet numerically if more >50% elements are numeric
390 my $numeric = $self->_is_numeric($name);
392 my @facet_names = sort {
393 if ( $sort =~ m/a/i ) {
394 $numeric ? $a <=> $b : lc $a cmp lc $b;
395 } elsif ( $sort =~ m/d/i ) {
396 $numeric ? $b <=> $a : lc $b cmp lc $a;
397 } elsif ( $sort =~ m/c/i ) {
398 $facet->{$b} <=> $facet->{$a};
400 warn "unknown sort: $sort";
405 $self->render( name => $name, facet => $facet, checked => $checked,
406 facet_names => \@facet_names, sort => $sort, numeric => $numeric,
413 $checked->{$_}++ foreach @_;
414 warn "# _checked ",dump($checked);