|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Elastica\Suggest; |
|
4
|
|
|
|
|
5
|
|
|
use Elastica\Suggest\CandidateGenerator\AbstractCandidateGenerator; |
|
6
|
|
|
use Elastica\Suggest\CandidateGenerator\DirectGenerator; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class Phrase. |
|
10
|
|
|
* |
|
11
|
|
|
* @see https://www.elastic.co/guide/en/elasticsearch/reference/current/search-suggesters-phrase.html |
|
12
|
|
|
*/ |
|
13
|
|
|
class Phrase extends AbstractSuggest |
|
14
|
|
|
{ |
|
15
|
|
|
public const DEFAULT_REAL_WORD_ERROR_LIKELIHOOD = 0.95; |
|
16
|
|
|
public const DEFAULT_CONFIDENCE = 1.0; |
|
17
|
|
|
public const DEFAULT_MAX_ERRORS = 1.0; |
|
18
|
|
|
public const DEFAULT_STUPID_BACKOFF_DISCOUNT = 0.4; |
|
19
|
|
|
public const DEFAULT_LAPLACE_SMOOTHING_ALPHA = 0.5; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @return $this |
|
23
|
|
|
*/ |
|
24
|
|
|
public function setAnalyzer(string $analyzer): Phrase |
|
25
|
|
|
{ |
|
26
|
|
|
return $this->setParam('analyzer', $analyzer); |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Set the max size of the n-grams (shingles) in the field. |
|
31
|
|
|
* |
|
32
|
|
|
* @return $this |
|
33
|
|
|
*/ |
|
34
|
|
|
public function setGramSize(int $size): Phrase |
|
35
|
|
|
{ |
|
36
|
|
|
return $this->setParam('gram_size', $size); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Set the likelihood of a term being misspelled even if the term exists in the dictionary. |
|
41
|
|
|
* |
|
42
|
|
|
* @param float $likelihood Defaults to 0.95, meaning 5% of the words are misspelled. |
|
43
|
|
|
* |
|
44
|
|
|
* @return $this |
|
45
|
|
|
*/ |
|
46
|
|
|
public function setRealWordErrorLikelihood(float $likelihood): Phrase |
|
47
|
|
|
{ |
|
48
|
|
|
return $this->setParam('real_word_error_likelihood', $likelihood); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Set the factor applied to the input phrases score to be used as a threshold for other suggestion candidates. |
|
53
|
|
|
* Only candidates which score higher than this threshold will be included in the result. |
|
54
|
|
|
* |
|
55
|
|
|
* @param float $confidence Defaults to 1.0. |
|
56
|
|
|
* |
|
57
|
|
|
* @return $this |
|
58
|
|
|
*/ |
|
59
|
|
|
public function setConfidence(float $confidence): Phrase |
|
60
|
|
|
{ |
|
61
|
|
|
return $this->setParam('confidence', $confidence); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* Set the maximum percentage of the terms considered to be misspellings in order to form a correction. |
|
66
|
|
|
* |
|
67
|
|
|
* @return $this |
|
68
|
|
|
*/ |
|
69
|
|
|
public function setMaxErrors(float $max): Phrase |
|
70
|
|
|
{ |
|
71
|
|
|
return $this->setParam('max_errors', $max); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @return $this |
|
76
|
|
|
*/ |
|
77
|
|
|
public function setSeparator(string $separator): Phrase |
|
78
|
|
|
{ |
|
79
|
|
|
return $this->setParam('separator', $separator); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Set suggestion highlighting. |
|
84
|
|
|
* |
|
85
|
|
|
* @return $this |
|
86
|
|
|
*/ |
|
87
|
|
|
public function setHighlight(string $preTag, string $postTag): Phrase |
|
88
|
|
|
{ |
|
89
|
|
|
return $this->setParam('highlight', [ |
|
90
|
|
|
'pre_tag' => $preTag, |
|
91
|
|
|
'post_tag' => $postTag, |
|
92
|
|
|
]); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @return $this |
|
97
|
|
|
*/ |
|
98
|
|
|
public function setStupidBackoffSmoothing(float $discount): Phrase |
|
99
|
|
|
{ |
|
100
|
|
|
return $this->setSmoothingModel('stupid_backoff', [ |
|
101
|
|
|
'discount' => $discount, |
|
102
|
|
|
]); |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* @return $this |
|
107
|
|
|
*/ |
|
108
|
|
|
public function setLaplaceSmoothing(float $alpha): Phrase |
|
109
|
|
|
{ |
|
110
|
|
|
return $this->setSmoothingModel('laplace', [ |
|
111
|
|
|
'alpha' => $alpha, |
|
112
|
|
|
]); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
/** |
|
116
|
|
|
* @return $this |
|
117
|
|
|
*/ |
|
118
|
|
|
public function setLinearInterpolationSmoothing(float $trigramLambda, float $bigramLambda, float $unigramLambda): Phrase |
|
119
|
|
|
{ |
|
120
|
|
|
return $this->setSmoothingModel('linear_interpolation', [ |
|
121
|
|
|
'trigram_lambda' => $trigramLambda, |
|
122
|
|
|
'bigram_lambda' => $bigramLambda, |
|
123
|
|
|
'unigram_lambda' => $unigramLambda, |
|
124
|
|
|
]); |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
/** |
|
128
|
|
|
* @param string $model the name of the smoothing model |
|
129
|
|
|
* |
|
130
|
|
|
* @return $this |
|
131
|
|
|
*/ |
|
132
|
|
|
public function setSmoothingModel(string $model, array $params): Phrase |
|
133
|
|
|
{ |
|
134
|
|
|
return $this->setParam('smoothing', [ |
|
135
|
|
|
$model => $params, |
|
136
|
|
|
]); |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* @return $this |
|
141
|
|
|
*/ |
|
142
|
|
|
public function addDirectGenerator(DirectGenerator $generator): self |
|
143
|
|
|
{ |
|
144
|
|
|
return $this->addParam('candidate_generator', $generator); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* @deprecated since version 7.2.0, use the "addDirectGenerator()" method instead. |
|
149
|
|
|
* |
|
150
|
|
|
* @return $this |
|
151
|
|
|
*/ |
|
152
|
|
|
public function addCandidateGenerator(AbstractCandidateGenerator $generator): Phrase |
|
153
|
|
|
{ |
|
154
|
|
|
\trigger_deprecation('ruflin/elastica', '7.2.0', 'The "%s()" method is deprecated, use the "addDirectGenerator()" method instead. It will be removed in 8.0.', __METHOD__); |
|
155
|
|
|
|
|
156
|
|
|
return $this->addParam('candidate_generator', $generator); |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
/** |
|
160
|
|
|
* {@inheritdoc} |
|
161
|
|
|
*/ |
|
162
|
|
|
public function toArray(): array |
|
163
|
|
|
{ |
|
164
|
|
|
$array = parent::toArray(); |
|
165
|
|
|
|
|
166
|
|
|
$baseName = $this->_getBaseName(); |
|
167
|
|
|
|
|
168
|
|
|
if (isset($array[$baseName]['candidate_generator'])) { |
|
169
|
|
|
$generators = $array[$baseName]['candidate_generator']; |
|
170
|
|
|
unset($array[$baseName]['candidate_generator']); |
|
171
|
|
|
|
|
172
|
|
|
foreach ($generators as $generator) { |
|
173
|
|
|
$keys = \array_keys($generator); |
|
174
|
|
|
$values = \array_values($generator); |
|
175
|
|
|
|
|
176
|
|
|
$array[$baseName][$keys[0]][] = $values[0]; |
|
177
|
|
|
} |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
return $array; |
|
181
|
|
|
} |
|
182
|
|
|
} |
|
183
|
|
|
|
ruflin /
Elastica
