sort data files by mtime
[MojoFacets.git] / lib / MojoFacets / Data.pm
index a8dc773..e1dc8ac 100644 (file)
@@ -5,7 +5,11 @@ use warnings;
 
 use base 'Mojolicious::Controller';
 
-use Data::Dump qw(dump);
+#use Data::Dump qw(dump); # broken with Mojo::JSON, see https://rt.cpan.org/Public/Bug/Display.html?id=86592
+use Data::Dumper;
+use subs 'dump';
+sub dump { Dumper(@_) };
+
 use File::Slurp;
 use Encode;
 use locale;
@@ -13,10 +17,43 @@ use File::Find;
 use Storable;
 use Time::HiRes qw(time);
 use File::Path qw(mkpath);
+use Text::Unaccent::PurePerl;
+use Digest::MD5;
+use Statistics::Descriptive;
+
+our $imports;
+foreach my $module ( glob('lib/MojoFacets/Import/*.pm') ) {
+       $module =~ s{lib/(\w+)/(\w+)/(.*)\.pm}{$1::$2::$3};
+       eval "use $module";
+       die "$module: $!" if $!;
+       my ( $ext, $priority ) = $module->ext;
+       $imports->{$priority || 'file'}->{$ext} = $module;
+       warn "# import $ext $module\n";
+}
+
+warn "# import loaded ",dump( $imports );
+
+sub import_module {
+       my $full_path = shift;
+
+#      warn "# import_module $full_path\n";
 
-use MojoFacets::Import::File;
-use MojoFacets::Import::HTMLTable;
-use MojoFacets::Import::CSV;
+       return if $full_path =~ m/\.columns$/;
+
+       foreach my $ext ( keys %{ $imports->{file} } ) {
+               if ( -f $full_path && $full_path =~ m/$ext/i ) {
+                       return $imports->{file}->{$ext};
+                       last;
+               }
+       }
+
+       foreach my $ext ( keys %{ $imports->{directory} } ) {
+               if ( -f $full_path && $full_path =~ m/$ext/i ) {
+                       return $imports->{directory}->{$ext};
+                       last;
+               }
+       }
+}
 
 our $loaded;
 our $filters;
@@ -24,31 +61,30 @@ our $filters;
 sub index {
        my $self = shift;
 
-       my $data_dir = $self->app->home->rel_dir('data');
+       my $data_dir = $self->app->home->rel_file('data');
        die "no data dir $data_dir" unless -d $data_dir;
 
        my @files;
        my $changes;
+
        find( sub {
                my $file = $File::Find::name;
-               if ( -f $file && $file =~ m/\.(js(on)?|txt)$/ ) {
-                       $file =~ s/$data_dir\/*//;
-                       push @files, $file;
-               } elsif ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+\.\d+.+)/ ) {
+               if ( -f $file && $file =~ m/([^\/]+)\.changes\/(\d+[\.,]\d+.+)/ ) {
                        push @{ $changes->{$1} }, $2
-               } elsif ( -d $file && $file =~ m/\.html$/ ) {
-                       $file =~ s/$data_dir\/*//;
-                       push @files, $file;
-               } elsif ( -f $file && $file =~ m/\.csv$/i ) {
+               } elsif ( import_module( $file ) ) {
+                       my $mtime = (stat($file))[9]; # mtime
                        $file =~ s/$data_dir\/*//;
                        push @files, $file;
+                       $loaded->{$file}->{mtime} ||= $mtime;
                } else {
                        #warn "IGNORE: $file\n";
                }
        }, $data_dir);
 
+       no warnings qw(uninitialized); # mtime
        @files = sort { $loaded->{$b}->{mtime} <=> $loaded->{$a}->{mtime} || lc $a cmp lc $b } @files,
-                       grep { defined $loaded->{$_}->{generated} } keys %$loaded;
+                       grep { defined $loaded->{$_}->{generated} } keys %$loaded;
+
        my $size;
        $size->{$_} = -s "$data_dir/$_" foreach @files;
 
@@ -64,7 +100,7 @@ sub index {
 
 sub _dump_path {
        my ( $self, $name ) = @_;
-       my $dir = $self->app->home->rel_dir('data');
+       my $dir = $self->app->home->rel_file('data');
        $name =~ s/^$dir//;
        $name =~ s/\/+/_/g;
        return '/tmp/mojo_facets.' . $name . '.storable';
@@ -98,7 +134,10 @@ sub __stats {
 
        warn "__stats $nr_items\n";
 
-       foreach my $e ( @{ $_[0] } ) {
+#      foreach my $e ( @{ $_[0] } ) {
+       foreach my $i ( 0 .. $#{$_[0]} ) {
+               print STDERR " $i" if $i % 5000;
+               my $e = $_[0]->[$i];
                foreach my $n ( keys %$e ) {
                        $stats->{$n}->{count}++;
                        my @v;
@@ -110,6 +149,11 @@ sub __stats {
                        }
 
                        foreach my $x ( @v ) {
+                               if ( ! defined $x ) { # FIXME really null
+                                       $stats->{$n}->{empty}++;
+                                       next;
+                               }
+
                                $stats->{$n}->{numeric}++
                                        if $x =~ m/^[-+]?([0-9]*\.[0-9]+|[0-9]+)$/;
                                $stats->{$n}->{empty}++
@@ -155,17 +199,17 @@ sub stats {
        my $path = $self->_param_or_session('path');
        warn "stats $path\n";
        delete $loaded->{$path}->{stats};
-       $self->redirect_to( '/data/columns' );
+       return $self->redirect_to( '/data/columns' );
 }
 
 
 sub _load_path {
        my ( $self, $path ) = @_;
 
-       return if defined $loaded->{$path}->{'generated'};
+       return if defined $loaded->{$path}->{generated};
 
        my $full_path = $self->app->home->rel_file( 'data/' . $path );
-       $self->redirect_to('/data/index') unless -r $full_path;
+       return $self->redirect_to('/data/index') unless -r $full_path;
 
        my $dump_path = $self->_dump_path( $path );
 
@@ -181,14 +225,8 @@ sub _load_path {
        }
 
        my $data;
-       if ( -f $full_path ) {
-               if ( $full_path =~ m/.csv/i ) {
-                       $data = MojoFacets::Import::CSV->new( full_path => $full_path )->data;
-               } else {
-                       $data = MojoFacets::Import::File->new( full_path => $full_path, path => $path )->data;
-               }
-       } elsif ( -d $full_path && $full_path =~ m/.html/ ) {
-               $data = MojoFacets::Import::HTMLTable->new( dir => $full_path )->data;
+       if ( my $module = import_module( $full_path ) ) {
+               $data = $module->new( full_path => $full_path )->data;
        } else {
                die "can't load $full_path";
        }
@@ -217,10 +255,11 @@ sub _load_path {
                size => -s $full_path,
                mtime => (stat($full_path))[9],
                data => $data,
+               defined $data->{generated} ? ( generated => 1 ) : (),
        };
 
        $loaded->{ $path } = $info;
-       $self->_save( $path );
+       $self->_save( $path ) unless $info->{generated};
 
 }
 
@@ -228,9 +267,7 @@ sub _load_path {
 sub load {
        my $self = shift;
 
-       my $path = $self->param('path') || $self->redirect_to( '/data/index' );
-
-       my @paths = $self->param('paths');
+       my @paths = @{ $self->every_param('paths') };
        warn "# paths ", dump @paths;
 
        foreach my $p ( keys %$loaded ) {
@@ -241,6 +278,8 @@ sub load {
 
        $self->_load_path( $_ ) foreach @paths;
 
+       my $path = $self->param('path') || $self->session('path') || $paths[0] || $self->redirect_to('/data/index');
+
        warn "# path $path\n";
        $self->_load_path( $path );
 
@@ -263,14 +302,14 @@ sub load {
        }
        $self->session( 'columns' => $loaded->{$path}->{columns} );
        $self->session( 'order'   => $loaded->{$path}->{columns}->[0] );
-       $self->redirect_to( $redirect_to );
+       return $self->redirect_to( $redirect_to );
 }
 
 
 sub _loaded {
        my ( $self, $name ) = @_;
        my $path = $self->session('path') || $self->param('path');
-       $self->redirect_to('/data/index') unless $path;
+       return $self->redirect_to('/data/index') unless $path;
 
        if ( defined $loaded->{$path}->{modified} && $loaded->{$path}->{modified} > 1 ) {
                my $caller = (caller(1))[3];
@@ -278,11 +317,18 @@ sub _loaded {
                        warn "rebuild stats for $path ignored caller $caller\n";
                } else {
                        warn "rebuild stats for $path FORCED by modified caller $caller\n";
-                       $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
+#                      $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
+                       $loaded->{$path}->{rebuild_stats} = 1;
                        $loaded->{$path}->{modified} = 1;
                }
        }
 
+       if ( defined $loaded->{$path}->{rebuild_stats} ) {
+               warn "rebuild_stats $path";
+               $loaded->{$path}->{stats} = __stats( $loaded->{$path}->{data}->{items} );
+               delete $loaded->{$path}->{rebuild_stats};
+       }
+
        if ( ! defined $loaded->{$path}->{$name} ) {
                warn "$path $name isn't loaded\n";
                $self->_load_path( $path );
@@ -292,7 +338,7 @@ sub _loaded {
                }
                if ( ! defined $loaded->{$path}->{$name} ) {
                        warn "MISSING $name for $path\n";
-                       $self->redirect_to('/data/index')
+                       return $self->redirect_to('/data/index')
                }
        }
 
@@ -313,7 +359,23 @@ sub _checked {
 sub _permanent_path {
        my $self = shift;
        my $path = $self->_param_or_session('path');
-       $self->app->home->rel_dir('data') . '/' . join('.', $path, @_);
+       $self->app->home->rel_file('data') . '/' . join('.', $path, @_);
+}
+
+sub __unac {
+       my $n = shift;
+       $n = unac_string('utf-8',$n);
+       $n =~ s/\W+/_/g;
+       return $n;
+}
+
+sub _column_from_unac {
+       my ($self,$name) = @_;
+
+       my $stats = $self->_loaded('stats');
+       my $cols_norm = { map { __unac( $_ ) => $_ } keys %$stats };
+
+       $cols_norm->{$name} || die "can't find column $name in ", dump($cols_norm);
 }
 
 sub _export_path {
@@ -323,18 +385,53 @@ sub _export_path {
                warn "no path in param or session";
                return;
        }
-       my $dir = $self->app->home->rel_dir('public') . "/export/$path";
+       my $dir = $self->app->home->rel_file('public') . "/export/$path";
        mkpath $dir unless -e $dir;
-       $dir . '/' . join('.', @_);
+       my $name = __export_path_name( $path, @_ );
+       my $full = $dir . '/' . $name;
+       $full =~ s/\/+$// if -d $full; # strip trailing slash for dirs
+       return $full;
+}
+
+sub __export_path_name {
+       my $max_length = 80;
+
+       my $path = shift;
+       my $name = join('.', map { __unac($_) } @_ );
+       if ( length($name) > $max_length ) {
+               $name = substr($name,0,$max_length) . Digest::MD5::md5_hex substr($name,$max_length);
+       }
+       return $name;
 }
 
 sub columns {
     my $self = shift;
 
+       my $view_path = $self->_permanent_path( 'views' );
+
        if ( $self->param('columns') ) {
                my @columns = $self->_param_array('columns');
                write_file( $self->_permanent_path( 'columns' ), { binmode => ':utf8' }, map { "$_\n" } @columns );
-               $self->redirect_to('/data/items');
+               if ( my $view = $self->param('view') ) {
+                       mkdir $view_path unless -e $view_path;
+                       write_file( "$view_path/$view", { binmode => ':utf8' }, map { "$_\n" } @columns );
+                       warn "view $view_path/$view ", -s "$view_path/$view", " bytes\n";
+               }
+
+               return $self->redirect_to('/data/items');
+
+       } elsif ( ! $self->session('header') ) {
+               return $self->redirect_to('/data/load');
+       }
+
+       if ( my $id = $self->param('id') ) {
+               my $view_full = "$view_path/$id";
+               if ( -f $view_full ) {
+                       my @columns = map { chomp; $_ } read_file $view_full, binmode => ':utf8';
+                       warn "view $view_full loaded ", dump @columns;
+                       $self->session( 'columns' => [ @columns ] );
+                       return $self->redirect_to('/data/items');
+               }
        }
 
        my $stats = $self->_loaded( 'stats' );
@@ -346,18 +443,25 @@ sub columns {
                push @columns, $c unless grep { /^\Q$c\E$/ } @columns;
        }
 
+       my @views;
+       if ( -d $view_path ) {
+               @views = map { s{^\Q$view_path\E/*}{}; $_ } glob "$view_path/*";
+               warn "# views ",dump @views;
+       }
+
     $self->render(
                message => 'Select columns to display',
                stats => $stats,
                columns => \@columns,
                checked => $self->_checked( $self->_param_array('columns') ),
+               views => \@views,
        );
 }
 
 sub _param_array {
     my ($self,$name) = @_;
 
-       my @array = $self->param($name);
+       my @array = @{ $self->every_param($name) };
        my $path  = $self->session('path');
 
        if ( @array ) {
@@ -387,8 +491,11 @@ sub _param_scalar {
 
        if ( ! defined $scalar ) {
                $scalar = $default;
-               die "no default for $name" unless defined $scalar;
-               $self->session($name => $scalar);
+               if ( defined $scalar ) {
+                       $self->session($name => $scalar);
+               } else {
+                       warn "no default for $name";
+               }
        }
 
        warn "# _param_scalar $name ",dump $scalar;
@@ -399,7 +506,7 @@ sub filter {
        my $self = shift;
 
        my $name = $self->param('filter_name') || die "name?";
-       my @vals = $self->param('filter_vals');
+       my @vals = @{ $self->every_param('filter_vals') };
 
        $self->_remove_filter( $name );
        if ( @vals ) {
@@ -413,7 +520,7 @@ sub filter {
 
        $self->session( 'offset' => 0 );
 
-       $self->redirect_to('/data/items');
+       return $self->redirect_to('/data/items');
 }
 
 sub _filter_on_data {
@@ -456,6 +563,12 @@ sub _filter_on_data {
 
        #warn "# filter $name ",dump($filtered_items);
 
+       # invalidate filters on other datasets
+       foreach my $dataset ( grep { exists $loaded->{$_}->{filters}->{$name} } keys %$loaded ) {
+               delete $loaded->{$dataset}->{filters}->{$name};
+               delete $loaded->{$dataset}->{filtered};
+       }
+
        $loaded->{$path}->{filters}->{$name} = $filtered_items;
        warn "filter $name with ", scalar keys %$filtered_items, " items created\n";
 }
@@ -464,15 +577,12 @@ sub _filter_on_data {
 sub _current_filters {
        my $self = shift;
        my $current_filters;
-       my $columns = $self->_loaded('header');
-       if ( my $sc = $self->session('columns') ) {
-               $columns = $sc;
-       }
+       my $stats = $self->_loaded('stats');
 
        $current_filters->{ $_ } = $filters->{ $_ }
-               foreach ( grep { defined $filters->{ $_ } } @$columns )
+               foreach ( grep { defined $filters->{ $_ } } keys %$stats )
        ;
-       warn "# _current_filters ",dump($columns);
+       warn "# _current_filters ",dump( keys %$current_filters );
        return $current_filters;
 }
 
@@ -525,18 +635,114 @@ sub __all_filters {
        join(',', sort(@_), 'order', $order);
 }
 
+our ($out, $key,$value);
+
+our $lookup_path_col;
+our $on;
+
+sub __commit_begin {
+       warn "__commit_begin";
+       $lookup_path_col = undef;
+       $on = undef;
+}
+
+sub __commit_end {
+       warn "__commit_end";
+       $lookup_path_col = undef; # cleanup memory
+       $on = undef;
+}
+
+sub lookup {
+       warn "# lookup ",dump @_;
+       my ( $vals, $on_path, $on_col, $code, $stat_code ) = @_;
+       die "code is not sub{ ... } but ", dump $code unless ref $code eq 'CODE';
+
+       if ( ! exists $loaded->{$on_path} ) {
+               my @possible_paths = grep { /\Q$on_path\E/ } keys %$loaded;
+               die "more than one dataset available for '$on_path' ",dump @possible_paths if $#possible_paths > 0;
+               $on_path = shift @possible_paths;
+               warn "## fuzzy selected path $on_path";
+       }
+
+       my $items = $loaded->{$on_path}->{data}->{items} || die "no items for $on_path";
+
+       if ( ! exists $lookup_path_col->{$on_path}->{$on_col} ) {
+               warn "create lookup_path_col $on_path $on_col";
+               foreach my $i ( 0 .. $#$items ) {
+                       my $item = $items->[$i];
+                       if ( exists $item->{$on_col} ) {
+                               if ( ref $item->{$on_col} eq 'ARRAY' ) {
+                                       foreach my $v ( @{ $item->{$on_col} } ) {
+                                               push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
+                                       }
+                               } elsif ( ! ref $item->{$on_col} ) { # scalar
+                                       my $v = $item->{$on_col};
+                                       push @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} }, $i;
+                               } else {
+                                       die "unknown type of ",dump $item->{$on_col};
+                               }
+                       }
+               }
+               warn "XXX ",dump $lookup_path_col->{$on_path}->{$on_col} if $ENV{DEBUG};
+       }
+
+       my $stat;
+       $stat = Statistics::Descriptive::Full->new() if $stat_code;
+
+       foreach my $v ( ref $vals eq 'ARRAY' ? @$vals : ( $vals ) ) {
+               foreach my $i ( @{ $lookup_path_col->{$on_path}->{$on_col}->{$v} } ) {
+                       $on = $items->[$i];
+                       warn "XXX lookup code $v $i ",dump $on if $ENV{DEBUG};
+                       $code->($stat);
+               }
+               $stat_code->( $stat ) if $stat_code;
+       }
+}
+
+sub __commit_path_code {
+       my ( $path, $i, $code, $commit_changed ) = @_;
+
+       my $items = $loaded->{$path}->{data}->{items} || die "no items for $path";
+       my $row = $items->[$i];
+       my $update;
+       eval $code;
+       foreach ( keys %$update ) {
+               $$commit_changed->{$_}++;
+               $loaded->{$path}->{data}->{items}->[$i]->{$_} = $update->{$_};
+       }
+       #warn "__commit_path_code $path $i ",dump( $update );
+}
+
+# uses templates/admin.html.ep
+sub _switch_dataset {
+       my $self = shift;
+
+       my $datasets;
+
+       foreach my $path ( keys %$loaded ) {
+               next unless exists $loaded->{$path}->{data};
+               push @$datasets, $path;
+       }
+
+       warn "# datasets ",dump($datasets);
+
+       $self->stash( 'datasets' => $datasets );
+}
+
 sub items {
        my $self = shift;
 
+       $self->_switch_dataset;
+
        if ( my $show = $self->param('id') ) {
                $self->param('show', $show);
                warn "show $show\n";
        }
 
-       my $path = $self->session('path');
+       my $path = $self->_param_scalar('path');
 
        my @columns = $self->_param_array('columns');
-       $self->redirect_to('/data/columns') unless @columns;
+       return $self->redirect_to('/data/columns') unless @columns;
        my $order   = $self->_param_scalar('order', $columns[0]);
        my $sort    = $self->_param_scalar('sort', 'a');
        my $offset  = $self->_param_scalar('offset', 0);
@@ -607,17 +813,22 @@ sub items {
        my $data = $self->_loaded('data');
 
        my $code = $self->_param_scalar('code','');
-       $code =~ s{[\r\n]+$}{}s;
+       $code =~ s{\r}{}gs;
+       $code =~ s{\n+$}{\n}s;
+
+       # XXX convert @row->{foo} into @{$row->{foo}}
+       $code =~ s|\@(row->\{[^}]+\})|\@{\$$1}|gs;
 
        my $commit = $self->param('commit');
        my $test = $self->param('test');
 
        my $commit_changed;
+       __commit_begin;
 
        if ( $code && ( $test || $commit ) ) {
                # XXX find columns used in code snippet and show them to user
                my $order = 0;
-               foreach my $column ( $code =~ m/\$row->{([^}]+)}/g ) {
+               foreach my $column ( $code =~ m/\$row->\{([^}]+)\}/g ) {
                        if ( $column =~ s/^(['"])// ) {
                                $column =~ s/$1$//;
                        }
@@ -626,31 +837,50 @@ sub items {
                }
        }
 
-       my $code_path = $self->app->home->rel_dir('public') . "/code";
+       my $code_path = $self->app->home->rel_file('public') . "/code";
        if ( $commit ) {
 
+               __path_modified( $path, 'commit' );
+
                warn "# commit on ", $#$filtered + 1, " items:\n$code\n";
-               my $out;
+               ( $key, $value, $out ) = ( 'key', 'value' );
                foreach ( 0 .. $#$filtered ) {
                        my $i = $filtered->[$_];
-                       my $row = $data->{items}->[$i];
-                       my $update;
-                       eval $code;
-                       foreach ( keys %$update ) {
-                               $commit_changed->{$_}++;
-                               $row->{$_} = $update->{$_};
-                       }
+                       __commit_path_code( $path, $i, $code, \$commit_changed );
                }
+
+               # this might move before $out to recalculate stats on source dataset?
+               __path_rebuild_stats( $path );
+               my $c = { map { $_ => 1 } @columns };
+               my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
+               warn "# added_columns ",dump( @added_columns );
+               unshift @columns, @added_columns;
+
+               $loaded->{$path}->{columns} = [ @columns ];
+               warn "# new columns ",dump( @columns );
+
+               __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
+
+               $self->_save_change({
+                       path => $path,
+                       time => $self->param('time') || time(),
+                       user => $self->param('user') || $ENV{'LOGNAME'},
+                       code => $code,
+                       commit_changed => $commit_changed,
+               });
+
                if ( my $description = $self->param('code_description') ) {
                        my $depends = $self->param('code_depends') || die "no code_depends?";
                        my $path = "$code_path/$depends.$description.pl";
                        if ( -e $path && ! $self->param('overwrite') ) {
                                warn "# code $path not saved\n";
                        } else {
-                               write_file(  $path, { binmode => ':utf8' }, $code );
+                               write_file(  $path, { binmode => ':utf8' }, "$code\n" );
                                warn "code $path ", -s $path, " bytes saved\n";
                        }
                }
+
+               # remove console
                $code = '';
                if ( $out ) {
                        my $commit_dataset = join('.'
@@ -658,9 +888,9 @@ sub items {
                                , $self->param('code_description')
                                , time()
                        );
-                       my $key = $self->param('code_depends');
-                       $key =~ s/,.+$//;
                        $key ||= 'key';
+                       $value ||= 'value';
+                       warn "key $key value $value";
                        my $items;
                        foreach my $n ( keys %$out ) {
                                my $i = { $key => [ $n ] };
@@ -670,7 +900,7 @@ sub items {
                                } elsif ( $ref eq 'ARRAY' ) {
                                        $i->{$_} = $out->{$n};
                                } elsif ( ! $ref ) {
-                                       $i->{value} = [ $out->{$n} ];
+                                       $i->{$value} = [ $out->{$n} ];
                                } else {
                                        $i->{_error} = [ dump($out->{$n}) ];
                                }
@@ -693,28 +923,16 @@ sub items {
                        $self->session('path', $commit_dataset);
                        $self->session('columns', [ @columns ]);
                        $self->session('order', $key);
-                       $self->redirect_to('/data/items');
-                       return; # FIXME needed to correctly show columns
+                       return $self->redirect_to('/data/items');
                }
 
-               # this might move before $out to recalculate stats on source dataset?
-               __path_modified( $path, 2 );
-               my $c = { map { $_ => 1 } @columns };
-               my @added_columns = sort grep { ! $c->{$_} } keys %$commit_changed;
-               warn "# added_columns ",dump( @added_columns );
-               unshift @columns, @added_columns;
-
                $self->session('columns', [ @columns ]);
-               $loaded->{$path}->{columns} = [ @columns ];
-               warn "# new columns ",dump( @columns );
-
-               __invalidate_path_column( $path, $_ ) foreach keys %$commit_changed;
        }
 
        my $sorted_items;
        my $from_end = $sort eq 'd' ? $#$filtered : 0;
        my $test_changed;
-       my $out;
+       my ( $key, $value, $out ) = ( 'key', 'value' ); # XXX make local
        foreach ( 0 .. $limit ) {
                my $i = $_ + $offset;
                last unless defined $filtered->[$i];
@@ -725,7 +943,7 @@ sub items {
                        my $update;
                        eval $code;
                        if ( $@ ) {
-                               warn "ERROR evaling\n$code\n$@";
+                               warn "ERROR evaling $@", dump($code);
                                $self->stash('eval_error', $@) if $@;
                        } else {
                                warn "EVAL ",dump($update);
@@ -740,37 +958,65 @@ sub items {
        }
 
        if ( $self->param('export') ) {
-               my $export_path = "public/export/$path/" . join('.', @columns);
-               open(my $fh, '>', $export_path) || warn "ERROR: can't open $export_path: $!";
+               my $export_path = $self->_export_path( 'items', @columns);
+               open(my $fh, '>', $export_path) || die "ERROR: can't open $export_path: $!";
+               print $fh "#",join("\t",@columns),"\n";
                foreach my $f ( 0 .. $#$filtered ) {
                        print $fh join("\t", map {
                                my $i = $data->{items}->[ $filtered->[$f] ];
-                               if ( ref $i->{$_} eq 'ARRAY' ) {
-                                       join(',', @{ $i->{$_} });
+                               my $v = '\N';
+                               if ( ! defined $i->{$_} ) {
+                                       # nop
+                               } elsif ( ref $i->{$_} eq 'ARRAY' ) {
+                                       $v =join(',', @{ $i->{$_} });
+                                       $v = '\N' if length($v) == 0;
+                               } elsif ( ! ref $i->{$_} ) {
+                                       $v = $i->{$_};
                                } else {
-                                       dump $i->{$_};
+                                       $v = dump $i->{$_};
                                }
+                               $v;
                        } @columns),"\n";
                }
                close($fh);
                warn "export $export_path ", -s $export_path, " bytes\n";
        }
 
-       warn "# test_changed ",dump( $test_changed );
-       my $c = { map { $_ => 1 } @columns };
-       my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
-       unshift @columns, @added_columns;
+       my ( $code_depends, $code_description );
 
-       warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
+       if ( $test ) {
 
-       my $code_depends = $self->param('code_depends')||
-       join(',', sort grep { $test_changed->{$_} == 0 } keys %$test_changed );
-       my $code_description = $self->param('code_description') ||
-       join(',', @added_columns);
+               warn "# test_changed ",dump( $test_changed );
+               my $c = { map { $_ => 1 } @columns };
+               my @added_columns = sort grep { ! $c->{$_} } keys %$test_changed;
+               unshift @columns, @added_columns;
+
+               warn "# sorted_items ", $#$sorted_items + 1, " offset $offset limit $limit order $sort";
+
+               my $depends_on;
+               my $tmp = $code; $tmp =~ s/\$row->\{(['"]?)([\w\s]+)\1/$depends_on->{$2}++/gse;
+               warn "# depends_on ",dump $depends_on;
+
+               my $test_added = Storable::dclone $test_changed;
+               delete $test_added->{$_} foreach keys %$depends_on;
+
+               $code_depends = $self->param('code_depends')
+               || join(',', keys %$depends_on);
 
-       $code_depends ||= $code_description; # self-modifing
+               $code_description = $self->param('code_description') ||
+               join(',', keys %$test_added);
 
-       warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
+               $code_depends ||= $code_description; # self-modifing
+               if ( ! $code_depends && $out ) {
+                       $code_depends = $key;
+                       $code_description = $value;
+               }
+
+               warn "# test_changed ",dump( $test_changed, $code_depends, $code_description );
+
+       } # test?
+
+       __commit_end;
 
        $self->render(
                order => $order,
@@ -797,7 +1043,7 @@ sub order {
        my $self = shift;
        $self->session('order', $self->param('order'));
        $self->session('sort', $self->param('sort'));
-       $self->redirect_to('/data/items');
+       return $self->redirect_to('/data/items');
 }
 
 sub _is_numeric {
@@ -840,11 +1086,11 @@ sub _remove_filter {
 sub facet {
        my $self = shift;
 
-       my $path = $self->session('path') || $self->redirect_to( '/data/index' );
+       my $path = $self->session('path') || return $self->redirect_to( '/data/index' );
 
        if ( my $name = $self->param('remove') ) {
                $self->_remove_filter( $name );
-               $self->redirect_to( '/data/items' );
+               return $self->redirect_to( '/data/items' );
        }
 
        my $facet;
@@ -867,7 +1113,7 @@ sub facet {
 
        foreach my $i ( @$filtered ) {
                my $item = $data->{items}->[$i];
-               if ( ! exists $item->{$name} ) {
+               if ( ! exists $item->{$name} || ! defined $item->{$name} ) {
                        $facet->{ _missing }++;
                } elsif ( ref $item->{$name} eq 'ARRAY' ) {
                        $facet->{$_}++ foreach @{ $item->{$name} };
@@ -876,16 +1122,34 @@ sub facet {
                }
        }
 
+       my $checked_values = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
+
+       if ( my $code = $self->param('code') ) {
+               my $out;
+               foreach my $value ( keys %$facet ) {
+                       my $count = $facet->{$value};
+                       my $checked = $checked_values->{$value};
+                       eval $code;
+                       if ( $@ ) {
+                               $out = $@;
+                               warn "ERROR: $@\n$code\n";
+                               last;
+                       } elsif ( $checked != $checked_values->{$value} ) {
+                               warn "checked $value $count -> $checked\n";
+                               $checked_values->{$value} = $checked;
+                       }
+               }
+               warn "out ",dump( $out );
+               $self->stash( out => $out );
+       }
+
 #      warn "# facet $name ",dump $facet;
 
-       my $checked;
        my @facet_names =
                  $all                      ? keys %$facet
                : defined $filters->{$name} ? @{ $filters->{$name} }
                : keys %$facet;
 
-       $checked = $self->_checked( @{ $filters->{$name} } ) if defined $filters->{$name};
-
        my $numeric = $self->_is_numeric($name);
 
        my $sort = $self->param('sort');
@@ -908,7 +1172,7 @@ sub facet {
                $result;
        } @facet_names;
 
-       $self->render( name => $name, facet => $facet, checked => $checked,
+       $self->render( name => $name, facet => $facet, checked => $checked_values,
                facet_names => \@facet_names, sort => $sort, numeric => $numeric,
        );
 }
@@ -937,6 +1201,19 @@ sub __path_modified {
        warn "# __path_modified $path $value\n";
 }
 
+sub __path_rebuild_stats { $loaded->{ $_[0] }->{rebuild_stats} = 1 };
+
+sub _save_change {
+       my ($self,$change) = @_;
+
+       my $change_path = $self->_permanent_path( 'changes' );
+       mkdir $change_path unless -d $change_path;
+       $change_path .= '/' . $change->{time};
+       store $change, $change_path;
+       utime $change->{time}, $change->{time}, $change_path;
+       warn "_save_change $change_path ", dump($change);
+}
+
 sub edit {
        my $self = shift;
        my $new_content = $self->param('new_content');
@@ -965,7 +1242,7 @@ sub edit {
                if ( $old ne $new
                        && ! ( $old eq 'undef' && length($new_content) == 0 ) # new value empty, previous undef
                ) {
-                       my $change = {
+                       $self->_save_change({
                                path => $path,
                                column => $name,
                                pos => $i,
@@ -978,13 +1255,7 @@ sub edit {
                                        grep { defined $loaded->{$path}->{stats}->{$_}->{unique} }
                                        keys %{ $loaded->{$path}->{stats} }
                                },
-                       };
-                       my $change_path = $self->_permanent_path( 'changes' );
-                       mkdir $change_path unless -d $change_path;
-                       $change_path .= '/' . $change->{time};
-                       store $change, $change_path;
-                       utime $change->{time}, $change->{time}, $change_path;
-                       warn "# $change_path ", dump($change);
+                       });
 
                        warn "# change $path $i $old -> $new\n";
                        $loaded->{$path}->{data}->{items}->[$i]->{$name} = $v;
@@ -992,8 +1263,7 @@ sub edit {
                        __invalidate_path_column( $path, $name );
 
                        $status = 201; # created
-                       # modified = 2 -- force rebuild of stats
-                       __path_modified( $path, 2 );
+                       __path_rebuild_stats( $path );
        
                        $new_content = join("\xB6",@$v);
 
@@ -1021,30 +1291,44 @@ sub save {
        my $dump_path = $self->_save( $path );
        __path_modified( $path, 0 );
 
-       $self->redirect_to( '/data/items' );
+       return $self->redirect_to( '/data/items' );
 }
 
 sub export {
        my $self = shift;
 
+       my $dir = $self->app->home->rel_file('public');
+
        if ( my $import = $self->param('import') ) {
 
                if ( $import =~ m{/filter\.(.+?)\..+} ) {
-                       my $name = $1;
+                       my $name = $self->_column_from_unac( $1 );
+
                        my @vals = map { chomp; $_ }
-                               read_file $self->app->home->rel_dir('public') . "/export/$import", binmode => ':utf8';
+                               read_file "$dir/export/$import", binmode => ':utf8';
+
                        $self->_remove_filter( $name );
                        $self->_filter_on_data( $name, @vals );
                        $self->session( 'offset' => 0 );
-                       $self->redirect_to('/data/items');
+                       return $self->redirect_to('/data/items');
                } else {
                        warn "UNKNOWN IMPORT $import";
                }
        }
 
-       $self->render( export => [
-               glob( $self->_export_path . '*' )
-       ] );
+       if ( my $remove = $self->param('remove') ) {
+               my $path = "$dir/export/$remove";
+               unlink $path if -e $path;
+               $path .= '.png';
+               unlink $path if -e $path;
+       }
+
+       my $path = $self->_export_path || return $self->redirect_to('/data/index');
+
+       my @files = grep { ! /\.png$/ } glob "$path/*";
+       my $mtime = { map { $_ => (stat($_))[9] } @files };
+       @files = sort { $mtime->{$b} <=> $mtime->{$a} } @files;
+       $self->render( export => [ @files ] );
 }
 
 sub __loaded_paths {
@@ -1053,4 +1337,29 @@ sub __loaded_paths {
                keys %$loaded;
 }
 
+sub reload {
+       my $self = shift;
+       $self->stash( reload => 1 );
+       $self->remove;
+#      $self->_load_path( $self->param('path') );
+       $self->redirect_to('/data/load?path=' . $self->param('path') );
+}
+
+sub remove {
+       my $self = shift;
+       my $path = $self->param('path');
+       if ( $path =~ m{^/tmp/mojo_facets\.} ) {
+               unlink $path;
+               warn "# unlink $path";
+               if ( my $name = $self->param('name') ) {
+                       delete $loaded->{$name};
+                       warn "# remove $name from memory";
+               }
+       } else {
+               warn "WARNING: $path unlink ignored";
+       }
+       return if $self->stash('reload');
+       return $self->redirect_to( '/data/load' );
+}
+
 1;