|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace TreeHouse\Model\Config\Matcher; |
|
4
|
|
|
|
|
5
|
|
|
use TreeHouse\Model\Config\Matcher\Normalizer\NormalizerInterface; |
|
6
|
|
|
use TreeHouse\Model\Config\Matcher\Stemmer\StemmerInterface; |
|
7
|
|
|
|
|
8
|
|
|
class SynonymMatcher implements MatcherInterface |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* @var NormalizerInterface |
|
12
|
|
|
*/ |
|
13
|
|
|
protected $normalizer; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var StemmerInterface |
|
17
|
|
|
*/ |
|
18
|
|
|
protected $stemmer; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var array |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $synonyms; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var array |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $stopwords; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @param NormalizerInterface $normalizer |
|
32
|
|
|
* @param StemmerInterface $stemmer |
|
33
|
|
|
* @param array $synonyms |
|
34
|
|
|
* @param array $stopwords |
|
35
|
|
|
*/ |
|
36
|
|
|
public function __construct(NormalizerInterface $normalizer, StemmerInterface $stemmer, array $synonyms, array $stopwords) |
|
37
|
|
|
{ |
|
38
|
|
|
$this->normalizer = $normalizer; |
|
39
|
|
|
$this->stemmer = $stemmer; |
|
40
|
|
|
$this->synonyms = $synonyms; |
|
41
|
|
|
$this->stopwords = $stopwords; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* @param string $field |
|
46
|
|
|
* @param string $value |
|
47
|
|
|
* |
|
48
|
|
|
* @return integer|null |
|
49
|
|
|
*/ |
|
50
|
|
|
public function match($field, $value) |
|
51
|
|
|
{ |
|
52
|
|
|
$matches = $this->getMatches($field, $value); |
|
53
|
|
|
|
|
54
|
|
|
return empty($matches) ? null : $matches[min(array_keys($matches))]; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @param string $field |
|
59
|
|
|
* @param string $value |
|
60
|
|
|
* |
|
61
|
|
|
* @throws \OutOfBoundsException |
|
62
|
|
|
* @throws \InvalidArgumentException |
|
63
|
|
|
* @throws \BadMethodCallException |
|
64
|
|
|
* |
|
65
|
|
|
* @return array |
|
66
|
|
|
*/ |
|
67
|
|
|
public function getMatches($field, $value) |
|
68
|
|
|
{ |
|
69
|
|
|
if (!$this->supports($field)) { |
|
70
|
|
|
throw new \BadMethodCallException(sprintf('Config field "%s" is not supported', $field)); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
// don't try to match null value |
|
74
|
|
|
if (is_null($value)) { |
|
75
|
|
|
return null; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
// we need something we can cast to string |
|
79
|
|
|
if (!is_scalar($value)) { |
|
80
|
|
|
throw new \InvalidArgumentException( |
|
81
|
|
|
sprintf( |
|
82
|
|
|
'Expecting scalar value for matching "%s", got "%s" instead', |
|
83
|
|
|
$field, |
|
84
|
|
|
gettype($value) |
|
85
|
|
|
) |
|
86
|
|
|
); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
$synonyms = $this->getSynonyms($field); |
|
90
|
|
|
$stopwords = $this->getStopwords($field); |
|
91
|
|
|
|
|
92
|
|
|
$value = (string) $value; |
|
93
|
|
|
|
|
94
|
|
|
if ($value === '') { |
|
95
|
|
|
return null; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
// compare in lowercase |
|
99
|
|
|
$value = $this->normalizer->normalize($value); |
|
100
|
|
|
|
|
101
|
|
|
// check for synonyms |
|
102
|
|
|
$matches = []; |
|
103
|
|
|
foreach ($synonyms as $key => $keySynonyms) { |
|
104
|
|
|
foreach ($keySynonyms as $synonym) { |
|
105
|
|
|
$distance = $this->getDistance($value, $synonym); |
|
106
|
|
|
if ($distance === 0) { |
|
107
|
|
|
// exact match, use this |
|
108
|
|
|
return [0 => $key]; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
// it has to be somewhat similar |
|
112
|
|
|
if ($distance <= strlen($value) / 3) { |
|
113
|
|
|
$matches[$distance] = $key; |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
// try for all stopwords |
|
117
|
|
|
foreach ($stopwords as $stopword) { |
|
118
|
|
|
if (($stopword === $value) || false === strpos($value, $stopword)) { |
|
119
|
|
|
// exact stopword match or no match |
|
120
|
|
|
continue; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$distance = $this->getDistance(str_replace($stopword, '', $value), $synonym); |
|
124
|
|
|
if ($distance === 0) { |
|
125
|
|
|
return [0 => $key]; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
// it has to be somewhat similar |
|
129
|
|
|
if ($distance <= strlen($value) / 3) { |
|
130
|
|
|
$matches[$distance] = $key; |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
return $matches; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* @param string $field |
|
141
|
|
|
* |
|
142
|
|
|
* @return boolean |
|
143
|
|
|
*/ |
|
144
|
|
|
public function supports($field) |
|
145
|
|
|
{ |
|
146
|
|
|
return array_key_exists($field, $this->synonyms); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* @param string $field |
|
151
|
|
|
* |
|
152
|
|
|
* @return array |
|
153
|
|
|
*/ |
|
154
|
|
|
public function getSynonyms($field) |
|
155
|
|
|
{ |
|
156
|
|
|
return array_key_exists($field, $this->synonyms) ? $this->synonyms[$field] : []; |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
/** |
|
160
|
|
|
* @param string $field |
|
161
|
|
|
* |
|
162
|
|
|
* @return array |
|
163
|
|
|
*/ |
|
164
|
|
|
public function getStopwords($field) |
|
165
|
|
|
{ |
|
166
|
|
|
return array_key_exists($field, $this->stopwords) ? $this->stopwords[$field] : []; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* @param string $value |
|
171
|
|
|
* @param string $synonym |
|
172
|
|
|
* |
|
173
|
|
|
* @return integer |
|
174
|
|
|
*/ |
|
175
|
|
|
protected function getDistance($value, $synonym) |
|
176
|
|
|
{ |
|
177
|
|
|
// first try an exact match |
|
178
|
|
|
if ($value === $synonym) { |
|
179
|
|
|
return 0; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
// try the slugified versions and stem them |
|
183
|
|
|
$synonymSlug = $this->stemmer->stem($value); |
|
184
|
|
|
$valueSlug = $this->stemmer->stem($synonym); |
|
185
|
|
|
|
|
186
|
|
|
return levenshtein($synonymSlug, $valueSlug); |
|
187
|
|
|
} |
|
188
|
|
|
} |
|
189
|
|
|
|
treehouselabs /
model-config
