| Akron | baade94 | 2017-09-25 18:19:02 +0200 | [diff] [blame] | 1 | package Krawfish::Koral::Result::Group::Fields; |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 2 | use Krawfish::Util::PatternList qw/pattern_list/; |
| 3 | use Krawfish::Log; |
| 4 | use strict; |
| 5 | use warnings; |
| 6 | |
| Akron | 0f8fb08 | 2017-10-09 15:54:41 +0200 | [diff] [blame] | 7 | use constant DEBUG => 0; |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 8 | |
| Akron | 09ab24b | 2017-08-24 12:45:39 +0200 | [diff] [blame] | 9 | # TODO: |
| 10 | # In addition to the group name |
| 11 | # create a signature that is universal for each group |
| 12 | |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 13 | sub new { |
| 14 | my $class = shift; |
| 15 | bless { |
| 16 | field_keys => shift, |
| 17 | cache => undef, |
| 18 | group => {}, |
| 19 | freq => 0 |
| 20 | }, $class; |
| 21 | }; |
| 22 | |
| 23 | |
| 24 | # Increment on a pattern, which is an ordered list of fields |
| 25 | sub incr_doc { |
| 26 | my ($self, $pattern) = @_; |
| 27 | |
| 28 | # Store all patterns in a cache |
| 29 | $self->{cache} = [ |
| 30 | pattern_list(@$pattern) |
| 31 | ]; |
| 32 | }; |
| 33 | |
| 34 | |
| 35 | # increment on match |
| 36 | sub incr_match { |
| 37 | $_[0]->{freq}++; |
| 38 | |
| 39 | if (DEBUG) { |
| 40 | print_log('p_g_fields', 'Increment match frequency'); |
| 41 | }; |
| 42 | }; |
| 43 | |
| 44 | |
| 45 | # Flush cache |
| 46 | sub flush { |
| 47 | my $self = shift; |
| 48 | |
| 49 | if ($self->{freq}) { |
| 50 | |
| 51 | # Iterate over all lists |
| 52 | foreach my $group (@{$self->{cache}}) { |
| 53 | |
| 54 | # This uses an array as a key ... probably should be realized differently |
| 55 | my $group_name = join('_',@$group); |
| 56 | my $freq = ($self->{group}->{$group_name} //= [0,0]); |
| 57 | |
| 58 | if (DEBUG) { |
| 59 | print_log('p_g_fields', 'Group on name ' . $group_name); |
| 60 | }; |
| 61 | |
| 62 | $freq->[0]++; |
| 63 | $freq->[1] += $self->{freq}; |
| 64 | }; |
| 65 | |
| 66 | $self->{cache} = undef; |
| 67 | $self->{freq} = 0; |
| 68 | }; |
| 69 | }; |
| 70 | |
| 71 | |
| 72 | sub inflate { |
| 73 | my ($self, $dict) = @_; |
| 74 | my $field_keys = $self->{field_keys}; |
| 75 | |
| 76 | $self->{field_terms} = []; |
| 77 | $self->{group_terms} = []; |
| 78 | |
| 79 | # Inflate head line |
| 80 | foreach my $field_id (@{$field_keys}) { |
| 81 | my $field_term = $dict->term_by_term_id($field_id); |
| 82 | |
| Akron | 0a57b2c | 2017-09-23 13:29:51 +0200 | [diff] [blame] | 83 | # $field_term =~ s/^!//; |
| 84 | # TODO: |
| 85 | # This may be a direct feature of the dictionary |
| 86 | $field_term = substr($field_term, 1); |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 87 | |
| 88 | push @{$self->{field_terms}}, $field_term; |
| 89 | }; |
| 90 | |
| 91 | # Inflate groups |
| 92 | foreach my $group (sort keys %{$self->{group}}) { |
| 93 | my @group = (); |
| 94 | my $i = 0; |
| 95 | foreach my $term_id (split('_', $group)) { |
| 96 | |
| 97 | # Term is defined at the position |
| 98 | if ($term_id != 0) { |
| 99 | |
| 100 | # Retrieve term |
| 101 | my $term = $dict->term_by_term_id($term_id); |
| 102 | my $field_term = $self->{field_terms}->[$i]; |
| Akron | 0a57b2c | 2017-09-23 13:29:51 +0200 | [diff] [blame] | 103 | |
| 104 | # TODO: |
| 105 | # This may be a direct feature of the dictionary |
| 106 | $term =~ s/^.$field_term://; |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 107 | |
| 108 | push @group, $term; |
| 109 | } |
| 110 | else { |
| 111 | push @group, ''; |
| 112 | }; |
| 113 | |
| 114 | # Move to next field |
| 115 | $i++; |
| 116 | }; |
| 117 | |
| 118 | push @{$self->{group_terms}}, [\@group, @{$self->{group}->{$group}}]; |
| 119 | }; |
| 120 | |
| 121 | return $self; |
| 122 | }; |
| 123 | |
| 124 | |
| 125 | sub to_string { |
| 126 | my $self = shift; |
| 127 | my $str = 'gFields:['; |
| 128 | |
| 129 | if ($self->{field_terms}) { |
| 130 | $str .= join(',', @{$self->{field_terms}}) . ':['; |
| 131 | foreach my $group (@{$self->{group_terms}}) { |
| 132 | $str .= join('|', @{$group->[0]}) . ':' . $group->[1] . ',' . $group->[2] . ';'; |
| 133 | }; |
| 134 | } |
| 135 | else { |
| 136 | $str .= join(',', map { '#' . $_ } @{$self->{field_keys}}) . ':['; |
| 137 | foreach my $group (sort keys %{$self->{group}}) { |
| 138 | $str .= $group .'=' . join(',', @{$self->{group}->{$group}}) . ';'; |
| 139 | }; |
| 140 | }; |
| 141 | chop $str; |
| Akron | 4040621 | 2017-09-18 11:51:27 +0200 | [diff] [blame] | 142 | return $str . ']'; |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 143 | }; |
| 144 | |
| 145 | |
| Akron | 09ab24b | 2017-08-24 12:45:39 +0200 | [diff] [blame] | 146 | sub to_koral_query { |
| 147 | # Create groups like |
| 148 | # { |
| 149 | # "@type":"koral:collection", |
| 150 | # "groupedBy":"groupedBy:fields", # or "aggregatedBy, "sortedBy" |
| 151 | # "labels":[...], |
| 152 | # "items":[ |
| 153 | # { |
| 154 | # "@type":"koral:item", |
| 155 | # // "signature":"ab47mhjhjgfjuizgtzurzt", |
| 156 | # "cols":[...] |
| 157 | # } |
| 158 | # ] |
| 159 | # } |
| 160 | ... |
| 161 | }; |
| 162 | |
| Akron | dc8dceb | 2017-08-22 20:25:39 +0200 | [diff] [blame] | 163 | 1; |