1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace TheIconic\NameParser; |
4
|
|
|
|
5
|
|
|
use TheIconic\NameParser\Language\English; |
6
|
|
|
use TheIconic\NameParser\Mapper\NicknameMapper; |
7
|
|
|
use TheIconic\NameParser\Mapper\SalutationMapper; |
8
|
|
|
use TheIconic\NameParser\Mapper\SuffixMapper; |
9
|
|
|
use TheIconic\NameParser\Mapper\InitialMapper; |
10
|
|
|
use TheIconic\NameParser\Mapper\LastnameMapper; |
11
|
|
|
use TheIconic\NameParser\Mapper\FirstnameMapper; |
12
|
|
|
use TheIconic\NameParser\Mapper\MiddlenameMapper; |
13
|
|
|
|
14
|
|
|
class Parser |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* @var string |
18
|
|
|
*/ |
19
|
|
|
protected $whitespace = " \r\n\t"; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var array |
23
|
|
|
*/ |
24
|
|
|
protected $mappers = []; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var array |
28
|
|
|
*/ |
29
|
|
|
protected $languages = []; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var array |
33
|
|
|
*/ |
34
|
|
|
protected $nicknameDelimiters = []; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* @var int |
38
|
|
|
*/ |
39
|
|
|
protected $maxSalutationIndex = 0; |
40
|
|
|
|
41
|
|
|
public function __construct(array $languages = []) |
42
|
|
|
{ |
43
|
|
|
if (empty($languages)) { |
44
|
|
|
$languages = [new English()]; |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
$this->languages = $languages; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* split full names into the following parts: |
52
|
|
|
* - prefix / salutation (Mr., Mrs., etc) |
53
|
|
|
* - given name / first name |
54
|
|
|
* - middle initials |
55
|
|
|
* - surname / last name |
56
|
|
|
* - suffix (II, Phd, Jr, etc) |
57
|
|
|
* |
58
|
|
|
* @param string $name |
59
|
|
|
* @return Name |
60
|
|
|
*/ |
61
|
|
|
public function parse($name): Name |
62
|
|
|
{ |
63
|
|
|
$name = $this->normalize($name); |
64
|
|
|
|
65
|
|
|
$segments = explode(',', $name); |
66
|
|
|
|
67
|
|
|
if (1 < count($segments)) { |
68
|
|
|
return $this->parseSplitName($segments[0], $segments[1], $segments[2] ?? ''); |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
$parts = explode(' ', $name); |
72
|
|
|
|
73
|
|
|
foreach ($this->getMappers() as $mapper) { |
74
|
|
|
$parts = $mapper->map($parts); |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
return new Name($parts); |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* handles split-parsing of comma-separated name parts |
82
|
|
|
* |
83
|
|
|
* @param $left - the name part left of the comma |
84
|
|
|
* @param $right - the name part right of the comma |
85
|
|
|
* |
86
|
|
|
* @return Name |
87
|
|
|
*/ |
88
|
|
|
protected function parseSplitName($first, $second, $third): Name |
89
|
|
|
{ |
90
|
|
|
$parts = array_merge( |
91
|
|
|
$this->getFirstSegmentParser()->parse($first)->getParts(), |
92
|
|
|
$this->getSecondSegmentParser()->parse($second)->getParts(), |
93
|
|
|
$this->getThirdSegmentParser()->parse($third)->getParts() |
94
|
|
|
); |
95
|
|
|
|
96
|
|
|
return new Name($parts); |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* @return Parser |
101
|
|
|
*/ |
102
|
|
|
protected function getFirstSegmentParser(): Parser |
103
|
|
|
{ |
104
|
|
|
$parser = new Parser(); |
105
|
|
|
|
106
|
|
|
$parser->setMappers([ |
107
|
|
|
new SalutationMapper($this->getSalutations(), $this->getMaxSalutationIndex()), |
108
|
|
|
new SuffixMapper($this->getSuffixes()), |
109
|
|
|
new LastnameMapper($this->getPrefixes(), true), |
110
|
|
|
new FirstnameMapper(), |
111
|
|
|
new MiddlenameMapper(), |
112
|
|
|
]); |
113
|
|
|
|
114
|
|
|
return $parser; |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* @return Parser |
119
|
|
|
*/ |
120
|
|
|
protected function getSecondSegmentParser(): Parser |
121
|
|
|
{ |
122
|
|
|
$parser = new Parser(); |
123
|
|
|
|
124
|
|
|
$parser->setMappers([ |
125
|
|
|
new SalutationMapper($this->getSalutations(), $this->getMaxSalutationIndex()), |
126
|
|
|
new SuffixMapper($this->getSuffixes(), true), |
127
|
|
|
new NicknameMapper($this->getNicknameDelimiters()), |
128
|
|
|
new InitialMapper(true), |
129
|
|
|
new FirstnameMapper(), |
130
|
|
|
new MiddlenameMapper(true), |
131
|
|
|
]); |
132
|
|
|
|
133
|
|
|
return $parser; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
protected function getThirdSegmentParser(): Parser |
137
|
|
|
{ |
138
|
|
|
$parser = new Parser(); |
139
|
|
|
|
140
|
|
|
$parser->setMappers([ |
141
|
|
|
new SuffixMapper($this->getSuffixes(), true), |
142
|
|
|
]); |
143
|
|
|
|
144
|
|
|
return $parser; |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* get the mappers for this parser |
149
|
|
|
* |
150
|
|
|
* @return array |
151
|
|
|
*/ |
152
|
|
|
public function getMappers(): array |
153
|
|
|
{ |
154
|
|
|
if (empty($this->mappers)) { |
155
|
|
|
$this->setMappers([ |
156
|
|
|
new NicknameMapper($this->getNicknameDelimiters()), |
157
|
|
|
new SalutationMapper($this->getSalutations(), $this->getMaxSalutationIndex()), |
158
|
|
|
new SuffixMapper($this->getSuffixes()), |
159
|
|
|
new InitialMapper(), |
160
|
|
|
new LastnameMapper($this->getPrefixes()), |
161
|
|
|
new FirstnameMapper(), |
162
|
|
|
new MiddlenameMapper(), |
163
|
|
|
]); |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
return $this->mappers; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
/** |
170
|
|
|
* set the mappers for this parser |
171
|
|
|
* |
172
|
|
|
* @param array $mappers |
173
|
|
|
* @return Parser |
174
|
|
|
*/ |
175
|
|
|
public function setMappers(array $mappers): Parser |
176
|
|
|
{ |
177
|
|
|
$this->mappers = $mappers; |
178
|
|
|
|
179
|
|
|
return $this; |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* normalize the name |
184
|
|
|
* |
185
|
|
|
* @param string $name |
186
|
|
|
* @return string |
187
|
|
|
*/ |
188
|
|
|
protected function normalize(string $name): string |
189
|
|
|
{ |
190
|
|
|
$whitespace = $this->getWhitespace(); |
191
|
|
|
|
192
|
|
|
$name = trim($name); |
193
|
|
|
|
194
|
|
|
return preg_replace('/[' . preg_quote($whitespace) . ']+/', ' ', $name); |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* get a string of characters that are supposed to be treated as whitespace |
199
|
|
|
* |
200
|
|
|
* @return string |
201
|
|
|
*/ |
202
|
|
|
public function getWhitespace(): string |
203
|
|
|
{ |
204
|
|
|
return $this->whitespace; |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
/** |
208
|
|
|
* set the string of characters that are supposed to be treated as whitespace |
209
|
|
|
* |
210
|
|
|
* @param $whitespace |
211
|
|
|
* @return Parser |
212
|
|
|
*/ |
213
|
|
|
public function setWhitespace($whitespace): Parser |
214
|
|
|
{ |
215
|
|
|
$this->whitespace = $whitespace; |
216
|
|
|
|
217
|
|
|
return $this; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
/** |
221
|
|
|
* @return array |
222
|
|
|
*/ |
223
|
|
|
protected function getPrefixes() |
224
|
|
|
{ |
225
|
|
|
$prefixes = []; |
226
|
|
|
|
227
|
|
|
/** @var LanguageInterface $language */ |
228
|
|
|
foreach ($this->languages as $language) { |
229
|
|
|
$prefixes += $language->getLastnamePrefixes(); |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
return $prefixes; |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
/** |
236
|
|
|
* @return array |
237
|
|
|
*/ |
238
|
|
|
protected function getSuffixes() |
239
|
|
|
{ |
240
|
|
|
$suffixes = []; |
241
|
|
|
|
242
|
|
|
/** @var LanguageInterface $language */ |
243
|
|
|
foreach ($this->languages as $language) { |
244
|
|
|
$suffixes += $language->getSuffixes(); |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
return $suffixes; |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* @return array |
252
|
|
|
*/ |
253
|
|
|
protected function getSalutations() |
254
|
|
|
{ |
255
|
|
|
$salutations = []; |
256
|
|
|
|
257
|
|
|
/** @var LanguageInterface $language */ |
258
|
|
|
foreach ($this->languages as $language) { |
259
|
|
|
$salutations += $language->getSalutations(); |
260
|
|
|
} |
261
|
|
|
|
262
|
|
|
return $salutations; |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
/** |
266
|
|
|
* @return array |
267
|
|
|
*/ |
268
|
|
|
public function getNicknameDelimiters(): array |
269
|
|
|
{ |
270
|
|
|
return $this->nicknameDelimiters; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* @param array $nicknameDelimiters |
275
|
|
|
* @return Parser |
276
|
|
|
*/ |
277
|
|
|
public function setNicknameDelimiters(array $nicknameDelimiters): Parser |
278
|
|
|
{ |
279
|
|
|
$this->nicknameDelimiters = $nicknameDelimiters; |
280
|
|
|
|
281
|
|
|
return $this; |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
/** |
285
|
|
|
* @return int |
286
|
|
|
*/ |
287
|
|
|
public function getMaxSalutationIndex(): int |
288
|
|
|
{ |
289
|
|
|
return $this->maxSalutationIndex; |
290
|
|
|
} |
291
|
|
|
|
292
|
|
|
/** |
293
|
|
|
* @param int $maxSalutationIndex |
294
|
|
|
* @return Parser |
295
|
|
|
*/ |
296
|
|
|
public function setMaxSalutationIndex(int $maxSalutationIndex): Parser |
297
|
|
|
{ |
298
|
|
|
$this->maxSalutationIndex = $maxSalutationIndex; |
299
|
|
|
|
300
|
|
|
return $this; |
301
|
|
|
} |
302
|
|
|
} |
303
|
|
|
|