|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace TheIconic\NameParser\Mapper; |
|
4
|
|
|
|
|
5
|
|
|
use TheIconic\NameParser\LanguageInterface; |
|
6
|
|
|
use TheIconic\NameParser\Part\AbstractPart; |
|
7
|
|
|
use TheIconic\NameParser\Part\Lastname; |
|
8
|
|
|
use TheIconic\NameParser\Part\LastnamePrefix; |
|
9
|
|
|
use TheIconic\NameParser\Part\Nickname; |
|
10
|
|
|
use TheIconic\NameParser\Part\Salutation; |
|
11
|
|
|
use TheIconic\NameParser\Part\Suffix; |
|
12
|
|
|
|
|
13
|
|
|
class LastnameMapper extends AbstractMapper |
|
14
|
|
|
{ |
|
15
|
|
|
protected $prefixes = []; |
|
16
|
|
|
|
|
17
|
|
|
protected $matchSinglePart = false; |
|
18
|
|
|
|
|
19
|
|
|
public function __construct(array $prefixes, bool $matchSinglePart = false) |
|
20
|
|
|
{ |
|
21
|
|
|
$this->prefixes = $prefixes; |
|
22
|
|
|
$this->matchSinglePart = $matchSinglePart; |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* map lastnames in the parts array |
|
27
|
|
|
* |
|
28
|
|
|
* @param array $parts the name parts |
|
29
|
|
|
* @return array the mapped parts |
|
30
|
|
|
*/ |
|
31
|
|
|
public function map(array $parts): array |
|
32
|
|
|
{ |
|
33
|
|
|
if (!$this->matchSinglePart && count($parts) < 2) { |
|
34
|
|
|
return $parts; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
return $this->mapParts($parts); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* we map the parts in reverse order because it makes more |
|
42
|
|
|
* sense to parse for the lastname starting from the end |
|
43
|
|
|
* |
|
44
|
|
|
* @param array $parts |
|
45
|
|
|
* @return array |
|
46
|
|
|
*/ |
|
47
|
|
|
protected function mapParts(array $parts): array |
|
48
|
|
|
{ |
|
49
|
|
|
$k = $this->skipIgnoredParts($parts) + 1; |
|
50
|
|
|
$remapIgnored = true; |
|
51
|
|
|
|
|
52
|
|
|
while (--$k >= 0) { |
|
53
|
|
|
$part = $parts[$k]; |
|
54
|
|
|
|
|
55
|
|
|
if ($part instanceof AbstractPart) { |
|
56
|
|
|
break; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
if ($this->isFollowedByLastnamePart($parts, $k)) { |
|
60
|
|
|
if ($mapped = $this->mapAsPrefixIfPossible($parts, $k)) { |
|
61
|
|
|
$parts[$k] = $mapped; |
|
62
|
|
|
continue; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
if ($this->shouldStopMapping($parts, $k)) { |
|
66
|
|
|
break; |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
$parts[$k] = new Lastname($part); |
|
71
|
|
|
$remapIgnored = false; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
if ($remapIgnored) { |
|
75
|
|
|
$parts = $this->remapIgnored($parts); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
return $parts; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* try to map this part as a lastname prefix or as a combined |
|
83
|
|
|
* lastname part containing a prefix |
|
84
|
|
|
* |
|
85
|
|
|
* @param array $parts |
|
86
|
|
|
* @param int $k |
|
87
|
|
|
* @return Lastname|null |
|
88
|
|
|
*/ |
|
89
|
|
|
private function mapAsPrefixIfPossible(array $parts, int $k): ?Lastname |
|
90
|
|
|
{ |
|
91
|
|
|
if ($this->isApplicablePrefix($parts, $k)) { |
|
92
|
|
|
return new LastnamePrefix($parts[$k], $this->prefixes[$this->getKey($parts[$k])]); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
if ($this->isCombinedWithPrefix($parts[$k])) { |
|
96
|
|
|
return new Lastname($parts[$k]); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
return null; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* check if the given part is a combined lastname part |
|
104
|
|
|
* that ends in a lastname prefix |
|
105
|
|
|
* |
|
106
|
|
|
* @param string $part |
|
107
|
|
|
* @return bool |
|
108
|
|
|
*/ |
|
109
|
|
|
private function isCombinedWithPrefix(string $part): bool |
|
110
|
|
|
{ |
|
111
|
|
|
$pos = strpos($part, '-'); |
|
112
|
|
|
|
|
113
|
|
|
if (false === $pos) { |
|
114
|
|
|
return false; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
return $this->isPrefix(substr($part, $pos + 1)); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* skip through the parts we want to ignore and return the start index |
|
122
|
|
|
* |
|
123
|
|
|
* @param array $parts |
|
124
|
|
|
* @return int |
|
125
|
|
|
*/ |
|
126
|
|
|
protected function skipIgnoredParts(array $parts): int |
|
127
|
|
|
{ |
|
128
|
|
|
$k = count($parts); |
|
129
|
|
|
|
|
130
|
|
|
while (--$k >= 0) { |
|
131
|
|
|
if (!$this->isIgnoredPart($parts[$k])) { |
|
132
|
|
|
break; |
|
133
|
|
|
} |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
return $k; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* indicates if we should stop mapping at the given index $k |
|
141
|
|
|
* |
|
142
|
|
|
* the assumption is that lastname parts have already been found |
|
143
|
|
|
* but we want to see if we should add more parts |
|
144
|
|
|
* |
|
145
|
|
|
* @param array $parts |
|
146
|
|
|
* @param int $k |
|
147
|
|
|
* @return bool |
|
148
|
|
|
*/ |
|
149
|
|
|
protected function shouldStopMapping(array $parts, int $k): bool |
|
150
|
|
|
{ |
|
151
|
|
|
if ($k < 1) { |
|
152
|
|
|
return true; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
$lastPart = $parts[$k + 1]; |
|
156
|
|
|
|
|
157
|
|
|
if ($lastPart instanceof LastnamePrefix) { |
|
158
|
|
|
return true; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
|
|
162
|
|
|
|
|
163
|
|
|
return strlen($lastPart->getValue()) >= 3; |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
/** |
|
167
|
|
|
* indicates if the given part should be ignored (skipped) during mapping |
|
168
|
|
|
* |
|
169
|
|
|
* @param $part |
|
170
|
|
|
* @return bool |
|
171
|
|
|
*/ |
|
172
|
|
|
protected function isIgnoredPart($part) { |
|
173
|
|
|
return $part instanceof Suffix || $part instanceof Nickname || $part instanceof Salutation; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
/** |
|
177
|
|
|
* remap ignored parts as lastname |
|
178
|
|
|
* |
|
179
|
|
|
* if the mapping did not derive any lastname this is called to transform |
|
180
|
|
|
* any previously ignored parts into lastname parts |
|
181
|
|
|
* |
|
182
|
|
|
* @param array $parts |
|
183
|
|
|
* @return array |
|
184
|
|
|
*/ |
|
185
|
|
|
protected function remapIgnored(array $parts): array |
|
186
|
|
|
{ |
|
187
|
|
|
$k = count($parts); |
|
188
|
|
|
|
|
189
|
|
|
while (--$k >= 0) { |
|
190
|
|
|
$part = $parts[$k]; |
|
191
|
|
|
|
|
192
|
|
|
if (!$this->isIgnoredPart($part)) { |
|
193
|
|
|
break; |
|
194
|
|
|
} |
|
195
|
|
|
|
|
196
|
|
|
$parts[$k] = new Lastname($part); |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
return $parts; |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
/** |
|
203
|
|
|
* @param array $parts |
|
204
|
|
|
* @param int $index |
|
205
|
|
|
* @return bool |
|
206
|
|
|
*/ |
|
207
|
|
|
protected function isFollowedByLastnamePart(array $parts, int $index): bool |
|
208
|
|
|
{ |
|
209
|
|
|
$next = $this->skipNicknameParts($parts, $index + 1); |
|
210
|
|
|
|
|
211
|
|
|
return (isset($parts[$next]) && $parts[$next] instanceof Lastname); |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
/** |
|
215
|
|
|
* Assuming that the part at the given index is matched as a prefix, |
|
216
|
|
|
* determines if the prefix should be applied to the lastname. |
|
217
|
|
|
* |
|
218
|
|
|
* We only apply it to the lastname if we already have at least one |
|
219
|
|
|
* lastname part and there are other parts left in |
|
220
|
|
|
* the name (this effectively prioritises firstname over prefix matching). |
|
221
|
|
|
* |
|
222
|
|
|
* This expects the parts array and index to be in the original order. |
|
223
|
|
|
* |
|
224
|
|
|
* @param array $parts |
|
225
|
|
|
* @param int $index |
|
226
|
|
|
* @return bool |
|
227
|
|
|
*/ |
|
228
|
|
|
protected function isApplicablePrefix(array $parts, int $index): bool |
|
229
|
|
|
{ |
|
230
|
|
|
if (!$this->isPrefix($parts[$index])) { |
|
231
|
|
|
return false; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
return $this->hasUnmappedPartsBefore($parts, $index); |
|
235
|
|
|
} |
|
236
|
|
|
|
|
237
|
|
|
/** |
|
238
|
|
|
* check if the given word is a lastname prefix |
|
239
|
|
|
* |
|
240
|
|
|
* @param string $word the word to check |
|
241
|
|
|
* @return bool |
|
242
|
|
|
*/ |
|
243
|
|
|
protected function isPrefix($word): bool |
|
244
|
|
|
{ |
|
245
|
|
|
return (array_key_exists($this->getKey($word), $this->prefixes)); |
|
246
|
|
|
} |
|
247
|
|
|
|
|
248
|
|
|
/** |
|
249
|
|
|
* find the next non-nickname index in parts |
|
250
|
|
|
* |
|
251
|
|
|
* @param $parts |
|
252
|
|
|
* @param $startIndex |
|
253
|
|
|
* @return int|void |
|
254
|
|
|
*/ |
|
255
|
|
|
protected function skipNicknameParts($parts, $startIndex) |
|
256
|
|
|
{ |
|
257
|
|
|
$total = count($parts); |
|
258
|
|
|
|
|
259
|
|
|
for ($i = $startIndex; $i < $total; $i++) { |
|
260
|
|
|
if (!($parts[$i] instanceof Nickname)) { |
|
261
|
|
|
return $i; |
|
262
|
|
|
} |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
return $total - 1; |
|
266
|
|
|
} |
|
267
|
|
|
} |
|
268
|
|
|
|