| Akron | 0a0e924 | 2016-10-28 14:42:29 +0200 | [diff] [blame] | 1 | package Krawfish::Koral::Query; |
| Akron | 6621e11 | 2016-11-05 17:21:39 +0100 | [diff] [blame] | 2 | use parent 'Krawfish::Info'; |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 3 | use Krawfish::Koral::Query::Builder; |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 4 | use Krawfish::Koral::Query::Importer; |
| Akron | 965f5d9 | 2017-01-20 18:38:08 +0100 | [diff] [blame] | 5 | use Mojo::Util qw/md5_sum/; |
| Akron | 0a0e924 | 2016-10-28 14:42:29 +0200 | [diff] [blame] | 6 | use warnings; |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 7 | use strict; |
| Akron | 0a0e924 | 2016-10-28 14:42:29 +0200 | [diff] [blame] | 8 | |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 9 | sub new { |
| 10 | my $class = shift; |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 11 | my $self = bless { |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 12 | any => 0, |
| 13 | optional => 0, |
| 14 | null => 0, |
| 15 | negative => 0, |
| 16 | extended => 0, |
| Akron | ddf077a | 2016-11-05 15:00:00 +0100 | [diff] [blame] | 17 | extended_left => 0, |
| Akron | 6621e11 | 2016-11-05 17:21:39 +0100 | [diff] [blame] | 18 | extended_right => 0 |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 19 | }, $class; |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 20 | |
| 21 | if ($_[0]) { |
| 22 | return $self->from_koral(shift); |
| 23 | }; |
| 24 | |
| 25 | $self; |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 26 | }; |
| 27 | |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 28 | ######################################### |
| 29 | # Query Planning methods and attributes # |
| 30 | ######################################### |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 31 | |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 32 | # Prepare a query for an index |
| Akron | 6a74973 | 2017-02-14 14:43:06 +0100 | [diff] [blame] | 33 | # TODO: Rename to compile() |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 34 | sub prepare_for { |
| 35 | my ($self, $index) = @_; |
| Akron | c3657bf | 2016-10-31 00:15:43 +0100 | [diff] [blame] | 36 | |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 37 | my $query = $self; |
| 38 | |
| 39 | # There is a possible 'any' extension, |
| 40 | # that may exceed the text |
| 41 | if ($self->is_extended_right) { |
| 42 | my $builder = $self->builder; |
| 43 | |
| 44 | # Wrap query in a text element |
| 45 | $query = $builder->position( |
| 46 | ['endsWith', 'isAround', 'startsWith', 'matches'], |
| 47 | $builder->span('base/s=t'), |
| 48 | $self |
| 49 | ); |
| 50 | }; |
| 51 | |
| 52 | # Return the planned query |
| 53 | # TODO: Check for serialization errors |
| 54 | $query->plan_for($index); |
| 55 | }; |
| 56 | |
| 57 | # Plan a query for an index (to be overwritten) |
| Akron | 349747d | 2016-12-05 11:05:53 +0100 | [diff] [blame] | 58 | # TODO: Rename to_primitive(index) |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 59 | sub plan_for; |
| 60 | |
| Akron | 22b6858 | 2017-01-19 12:05:21 +0100 | [diff] [blame] | 61 | # Filter a query based on a document query |
| 62 | sub filter_by; |
| 63 | |
| Akron | dc9f116 | 2016-11-05 15:31:40 +0100 | [diff] [blame] | 64 | sub is_any { $_[0]->{any} // 0 }; |
| 65 | sub is_optional { $_[0]->{optional} // 0 }; |
| 66 | sub is_null { $_[0]->{null} // 0 }; |
| 67 | sub is_negative { $_[0]->{negative} // 0 }; |
| Akron | dc9f116 | 2016-11-05 15:31:40 +0100 | [diff] [blame] | 68 | sub is_extended_right { $_[0]->{extended_right} // 0 }; |
| 69 | sub is_extended_left { $_[0]->{extended_left} // 0 }; |
| Akron | 84b8b75 | 2016-11-19 15:55:12 +0100 | [diff] [blame] | 70 | sub is_extended { $_[0]->is_extended_right || $_[0]->is_extended_left // 0 }; |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 71 | sub freq { -1 }; |
| Akron | 774c5db | 2016-11-09 16:11:38 +0100 | [diff] [blame] | 72 | sub type { '' }; |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 73 | |
| Akron | 774c5db | 2016-11-09 16:11:38 +0100 | [diff] [blame] | 74 | # TODO: Probably better to be renamed "potential_anchor" |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 75 | sub maybe_anchor { |
| 76 | my $self = shift; |
| 77 | return if $self->is_negative; |
| 78 | return if $self->is_optional; |
| 79 | return if $self->is_any; |
| 80 | return 1; |
| 81 | }; |
| 82 | |
| 83 | # Check if the wrapped query may need to be sorted |
| 84 | # on focussing on a specific class. |
| 85 | # Normally spans are always sorted, but in case of |
| 86 | # a wrapped relation query, classed operands may |
| 87 | # be in arbitrary order. When focussing on these |
| 88 | # classes, the span has to me reordered. |
| Akron | 1b09c5b | 2016-11-20 15:59:34 +0100 | [diff] [blame] | 89 | sub maybe_unsorted { $_[0]->{maybe_unsorted} // 0 }; |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 90 | |
| Akron | 965f5d9 | 2017-01-20 18:38:08 +0100 | [diff] [blame] | 91 | |
| 92 | # Iterate over all subqueries and possibly replace them |
| 93 | sub subqueries; |
| 94 | |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 95 | ############################# |
| 96 | # Query Application methods # |
| 97 | ############################# |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 98 | |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 99 | # Deserialization of KoralQuery |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 100 | # TODO: export this method from Importer |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 101 | sub from_koral { |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 102 | my ($class, $kq) = @_; |
| 103 | my $importer = Krawfish::Koral::Query::Importer->new; |
| 104 | |
| 105 | my $type = $kq->{'@type'}; |
| 106 | if ($type eq 'koral:group') { |
| 107 | my $op = $kq->{operation}; |
| 108 | if ($op eq 'operation:sequence') { |
| 109 | return $importer->seq($kq); |
| 110 | } |
| 111 | |
| 112 | elsif ($op eq 'operation:class') { |
| 113 | return $importer->class($kq); |
| 114 | } |
| 115 | else { |
| 116 | warn 'Operation ' . $op . ' not supported'; |
| 117 | }; |
| 118 | } |
| 119 | |
| 120 | elsif ($type eq 'koral:token') { |
| 121 | return $importer->token($kq); |
| 122 | } |
| 123 | else { |
| 124 | warn $type . ' unknown'; |
| 125 | }; |
| 126 | |
| 127 | return; |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 128 | }; |
| 129 | |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 130 | # Overwritten |
| 131 | sub to_koral_fragment; |
| 132 | |
| Akron | c3657bf | 2016-10-31 00:15:43 +0100 | [diff] [blame] | 133 | # Overwritten |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 134 | sub to_string; |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 135 | |
| Akron | 965f5d9 | 2017-01-20 18:38:08 +0100 | [diff] [blame] | 136 | # TODO: This may be optimizable and |
| 137 | # implemented in all query and corpus wrappers |
| 138 | sub to_signature { |
| 139 | md5_sum $_[0]->to_string; |
| 140 | }; |
| 141 | |
| 142 | # TODO: Returns a value of complexity of the query, |
| 143 | # that can be used to decide, if a query should be cached. |
| 144 | sub complexity; |
| 145 | |
| Akron | 573e7ec | 2016-11-05 19:03:01 +0100 | [diff] [blame] | 146 | # Clone the query |
| 147 | # sub clone; |
| Akron | c3657bf | 2016-10-31 00:15:43 +0100 | [diff] [blame] | 148 | |
| Akron | 4763ea6 | 2016-11-02 19:36:18 +0100 | [diff] [blame] | 149 | # Create KoralQuery builder |
| 150 | sub builder { |
| 151 | return Krawfish::Koral::Query::Builder->new; |
| 152 | }; |
| 153 | |
| Akron | 944091b | 2016-11-24 16:40:58 +0100 | [diff] [blame] | 154 | # Create KoralQuery builder |
| 155 | sub importer { |
| 156 | return Krawfish::Koral::Query::Importer->new; |
| 157 | }; |
| 158 | |
| Akron | 169ede4 | 2017-02-05 12:52:22 +0100 | [diff] [blame] | 159 | |
| 160 | # Serialization helper |
| 161 | sub boundary { |
| 162 | my $self = shift; |
| 163 | my %hash = ( |
| 164 | '@type' => 'koral:boundary' |
| 165 | ); |
| 166 | $hash{min} = $self->{min} if defined $self->{min}; |
| 167 | $hash{max} = $self->{max} if defined $self->{max}; |
| 168 | return \%hash; |
| 169 | } |
| 170 | |
| 171 | |
| Akron | 0a0e924 | 2016-10-28 14:42:29 +0200 | [diff] [blame] | 172 | 1; |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 173 | |
| Akron | a211bf5 | 2016-10-29 18:03:29 +0200 | [diff] [blame] | 174 | |
| Akron | 33f1dcb | 2016-10-29 17:27:23 +0200 | [diff] [blame] | 175 | __END__ |
| 176 | |