|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
namespace CultuurNet\UDB3; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* Class CulturefeedSlugger |
|
8
|
|
|
* |
|
9
|
|
|
* Taken from the culturefeed Drupal module. |
|
10
|
|
|
*/ |
|
11
|
|
|
class CulturefeedSlugger implements SluggerInterface |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* The maximum length the slug can have. |
|
15
|
|
|
* |
|
16
|
|
|
* @var int |
|
17
|
|
|
*/ |
|
18
|
|
|
protected $length; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* The separator that will be used instead of whitespaces. |
|
22
|
|
|
* |
|
23
|
|
|
* @var string |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $separator; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @param int $length |
|
29
|
|
|
* The maximum length the slug can have. |
|
30
|
|
|
* @param string $separator |
|
31
|
|
|
* The separator that will be used instead of whitespaces. |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct($length = 50, $separator = '-') |
|
34
|
|
|
{ |
|
35
|
|
|
$this->length = $length; |
|
36
|
|
|
$this->separator = $separator; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @inheritdoc |
|
41
|
|
|
*/ |
|
42
|
|
|
public function slug($string) |
|
43
|
|
|
{ |
|
44
|
|
|
// transliterate |
|
45
|
|
|
$string = $this->transliterate($string); |
|
46
|
|
|
|
|
47
|
|
|
// lowercase |
|
48
|
|
|
$string = strtolower($string); |
|
49
|
|
|
|
|
50
|
|
|
// replace non alphanumeric and non underscore characters by separator |
|
51
|
|
|
$string = preg_replace('/[^a-z0-9]/i', $this->separator, $string); |
|
52
|
|
|
|
|
53
|
|
|
// replace multiple occurrences of separator by one instance |
|
54
|
|
|
$string = preg_replace( |
|
55
|
|
|
'/' . preg_quote($this->separator) . '[' . preg_quote($this->separator) . ']*/', |
|
56
|
|
|
$this->separator, |
|
57
|
|
|
$string |
|
58
|
|
|
); |
|
59
|
|
|
|
|
60
|
|
|
// cut off to maximum length |
|
61
|
|
|
if ($this->length > -1 && strlen($string) > $this->length) { |
|
62
|
|
|
$string = substr($string, 0, $this->length); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
// remove separator from start and end of string |
|
66
|
|
|
$string = preg_replace( |
|
67
|
|
|
'/' . preg_quote($this->separator) . '$/', |
|
68
|
|
|
'', |
|
69
|
|
|
$string |
|
70
|
|
|
); |
|
71
|
|
|
$string = preg_replace( |
|
72
|
|
|
'/^' . preg_quote($this->separator) . '/', |
|
73
|
|
|
'', |
|
74
|
|
|
$string |
|
75
|
|
|
); |
|
76
|
|
|
|
|
77
|
|
|
return $string; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
private function transliterate(string $string): string |
|
81
|
|
|
{ |
|
82
|
|
|
static $charMap; |
|
83
|
|
|
if (!$charMap) { |
|
84
|
|
|
$charMap = array( |
|
85
|
|
|
// Decompositions for Latin-1 Supplement |
|
86
|
|
|
chr(195) . chr(128) => 'A', |
|
87
|
|
|
chr(195) . chr(129) => 'A', |
|
88
|
|
|
chr(195) . chr(130) => 'A', |
|
89
|
|
|
chr(195) . chr(131) => 'A', |
|
90
|
|
|
chr(195) . chr(132) => 'A', |
|
91
|
|
|
chr(195) . chr(133) => 'A', |
|
92
|
|
|
chr(195) . chr(135) => 'C', |
|
93
|
|
|
chr(195) . chr(136) => 'E', |
|
94
|
|
|
chr(195) . chr(137) => 'E', |
|
95
|
|
|
chr(195) . chr(138) => 'E', |
|
96
|
|
|
chr(195) . chr(139) => 'E', |
|
97
|
|
|
chr(195) . chr(140) => 'I', |
|
98
|
|
|
chr(195) . chr(141) => 'I', |
|
99
|
|
|
chr(195) . chr(142) => 'I', |
|
100
|
|
|
chr(195) . chr(143) => 'I', |
|
101
|
|
|
chr(195) . chr(145) => 'N', |
|
102
|
|
|
chr(195) . chr(146) => 'O', |
|
103
|
|
|
chr(195) . chr(147) => 'O', |
|
104
|
|
|
chr(195) . chr(148) => 'O', |
|
105
|
|
|
chr(195) . chr(149) => 'O', |
|
106
|
|
|
chr(195) . chr(150) => 'O', |
|
107
|
|
|
chr(195) . chr(153) => 'U', |
|
108
|
|
|
chr(195) . chr(154) => 'U', |
|
109
|
|
|
chr(195) . chr(155) => 'U', |
|
110
|
|
|
chr(195) . chr(156) => 'U', |
|
111
|
|
|
chr(195) . chr(157) => 'Y', |
|
112
|
|
|
chr(195) . chr(159) => 's', |
|
113
|
|
|
chr(195) . chr(160) => 'a', |
|
114
|
|
|
chr(195) . chr(161) => 'a', |
|
115
|
|
|
chr(195) . chr(162) => 'a', |
|
116
|
|
|
chr(195) . chr(163) => 'a', |
|
117
|
|
|
chr(195) . chr(164) => 'a', |
|
118
|
|
|
chr(195) . chr(165) => 'a', |
|
119
|
|
|
chr(195) . chr(167) => 'c', |
|
120
|
|
|
chr(195) . chr(168) => 'e', |
|
121
|
|
|
chr(195) . chr(169) => 'e', |
|
122
|
|
|
chr(195) . chr(170) => 'e', |
|
123
|
|
|
chr(195) . chr(171) => 'e', |
|
124
|
|
|
chr(195) . chr(172) => 'i', |
|
125
|
|
|
chr(195) . chr(173) => 'i', |
|
126
|
|
|
chr(195) . chr(174) => 'i', |
|
127
|
|
|
chr(195) . chr(175) => 'i', |
|
128
|
|
|
chr(195) . chr(177) => 'n', |
|
129
|
|
|
chr(195) . chr(178) => 'o', |
|
130
|
|
|
chr(195) . chr(179) => 'o', |
|
131
|
|
|
chr(195) . chr(180) => 'o', |
|
132
|
|
|
chr(195) . chr(181) => 'o', |
|
133
|
|
|
chr(195) . chr(182) => 'o', |
|
134
|
|
|
chr(195) . chr(182) => 'o', |
|
135
|
|
|
chr(195) . chr(185) => 'u', |
|
136
|
|
|
chr(195) . chr(186) => 'u', |
|
137
|
|
|
chr(195) . chr(187) => 'u', |
|
138
|
|
|
chr(195) . chr(188) => 'u', |
|
139
|
|
|
chr(195) . chr(189) => 'y', |
|
140
|
|
|
chr(195) . chr(191) => 'y', |
|
141
|
|
|
// Decompositions for Latin Extended-A |
|
142
|
|
|
chr(196) . chr(128) => 'A', |
|
143
|
|
|
chr(196) . chr(129) => 'a', |
|
144
|
|
|
chr(196) . chr(130) => 'A', |
|
145
|
|
|
chr(196) . chr(131) => 'a', |
|
146
|
|
|
chr(196) . chr(132) => 'A', |
|
147
|
|
|
chr(196) . chr(133) => 'a', |
|
148
|
|
|
chr(196) . chr(134) => 'C', |
|
149
|
|
|
chr(196) . chr(135) => 'c', |
|
150
|
|
|
chr(196) . chr(136) => 'C', |
|
151
|
|
|
chr(196) . chr(137) => 'c', |
|
152
|
|
|
chr(196) . chr(138) => 'C', |
|
153
|
|
|
chr(196) . chr(139) => 'c', |
|
154
|
|
|
chr(196) . chr(140) => 'C', |
|
155
|
|
|
chr(196) . chr(141) => 'c', |
|
156
|
|
|
chr(196) . chr(142) => 'D', |
|
157
|
|
|
chr(196) . chr(143) => 'd', |
|
158
|
|
|
chr(196) . chr(144) => 'D', |
|
159
|
|
|
chr(196) . chr(145) => 'd', |
|
160
|
|
|
chr(196) . chr(146) => 'E', |
|
161
|
|
|
chr(196) . chr(147) => 'e', |
|
162
|
|
|
chr(196) . chr(148) => 'E', |
|
163
|
|
|
chr(196) . chr(149) => 'e', |
|
164
|
|
|
chr(196) . chr(150) => 'E', |
|
165
|
|
|
chr(196) . chr(151) => 'e', |
|
166
|
|
|
chr(196) . chr(152) => 'E', |
|
167
|
|
|
chr(196) . chr(153) => 'e', |
|
168
|
|
|
chr(196) . chr(154) => 'E', |
|
169
|
|
|
chr(196) . chr(155) => 'e', |
|
170
|
|
|
chr(196) . chr(156) => 'G', |
|
171
|
|
|
chr(196) . chr(157) => 'g', |
|
172
|
|
|
chr(196) . chr(158) => 'G', |
|
173
|
|
|
chr(196) . chr(159) => 'g', |
|
174
|
|
|
chr(196) . chr(160) => 'G', |
|
175
|
|
|
chr(196) . chr(161) => 'g', |
|
176
|
|
|
chr(196) . chr(162) => 'G', |
|
177
|
|
|
chr(196) . chr(163) => 'g', |
|
178
|
|
|
chr(196) . chr(164) => 'H', |
|
179
|
|
|
chr(196) . chr(165) => 'h', |
|
180
|
|
|
chr(196) . chr(166) => 'H', |
|
181
|
|
|
chr(196) . chr(167) => 'h', |
|
182
|
|
|
chr(196) . chr(168) => 'I', |
|
183
|
|
|
chr(196) . chr(169) => 'i', |
|
184
|
|
|
chr(196) . chr(170) => 'I', |
|
185
|
|
|
chr(196) . chr(171) => 'i', |
|
186
|
|
|
chr(196) . chr(172) => 'I', |
|
187
|
|
|
chr(196) . chr(173) => 'i', |
|
188
|
|
|
chr(196) . chr(174) => 'I', |
|
189
|
|
|
chr(196) . chr(175) => 'i', |
|
190
|
|
|
chr(196) . chr(176) => 'I', |
|
191
|
|
|
chr(196) . chr(177) => 'i', |
|
192
|
|
|
chr(196) . chr(178) => 'IJ', |
|
193
|
|
|
chr(196) . chr(179) => 'ij', |
|
194
|
|
|
chr(196) . chr(180) => 'J', |
|
195
|
|
|
chr(196) . chr(181) => 'j', |
|
196
|
|
|
chr(196) . chr(182) => 'K', |
|
197
|
|
|
chr(196) . chr(183) => 'k', |
|
198
|
|
|
chr(196) . chr(184) => 'k', |
|
199
|
|
|
chr(196) . chr(185) => 'L', |
|
200
|
|
|
chr(196) . chr(186) => 'l', |
|
201
|
|
|
chr(196) . chr(187) => 'L', |
|
202
|
|
|
chr(196) . chr(188) => 'l', |
|
203
|
|
|
chr(196) . chr(189) => 'L', |
|
204
|
|
|
chr(196) . chr(190) => 'l', |
|
205
|
|
|
chr(196) . chr(191) => 'L', |
|
206
|
|
|
chr(197) . chr(128) => 'l', |
|
207
|
|
|
chr(197) . chr(129) => 'L', |
|
208
|
|
|
chr(197) . chr(130) => 'l', |
|
209
|
|
|
chr(197) . chr(131) => 'N', |
|
210
|
|
|
chr(197) . chr(132) => 'n', |
|
211
|
|
|
chr(197) . chr(133) => 'N', |
|
212
|
|
|
chr(197) . chr(134) => 'n', |
|
213
|
|
|
chr(197) . chr(135) => 'N', |
|
214
|
|
|
chr(197) . chr(136) => 'n', |
|
215
|
|
|
chr(197) . chr(137) => 'N', |
|
216
|
|
|
chr(197) . chr(138) => 'n', |
|
217
|
|
|
chr(197) . chr(139) => 'N', |
|
218
|
|
|
chr(197) . chr(140) => 'O', |
|
219
|
|
|
chr(197) . chr(141) => 'o', |
|
220
|
|
|
chr(197) . chr(142) => 'O', |
|
221
|
|
|
chr(197) . chr(143) => 'o', |
|
222
|
|
|
chr(197) . chr(144) => 'O', |
|
223
|
|
|
chr(197) . chr(145) => 'o', |
|
224
|
|
|
chr(197) . chr(146) => 'OE', |
|
225
|
|
|
chr(197) . chr(147) => 'oe', |
|
226
|
|
|
chr(197) . chr(148) => 'R', |
|
227
|
|
|
chr(197) . chr(149) => 'r', |
|
228
|
|
|
chr(197) . chr(150) => 'R', |
|
229
|
|
|
chr(197) . chr(151) => 'r', |
|
230
|
|
|
chr(197) . chr(152) => 'R', |
|
231
|
|
|
chr(197) . chr(153) => 'r', |
|
232
|
|
|
chr(197) . chr(154) => 'S', |
|
233
|
|
|
chr(197) . chr(155) => 's', |
|
234
|
|
|
chr(197) . chr(156) => 'S', |
|
235
|
|
|
chr(197) . chr(157) => 's', |
|
236
|
|
|
chr(197) . chr(158) => 'S', |
|
237
|
|
|
chr(197) . chr(159) => 's', |
|
238
|
|
|
chr(197) . chr(160) => 'S', |
|
239
|
|
|
chr(197) . chr(161) => 's', |
|
240
|
|
|
chr(197) . chr(162) => 'T', |
|
241
|
|
|
chr(197) . chr(163) => 't', |
|
242
|
|
|
chr(197) . chr(164) => 'T', |
|
243
|
|
|
chr(197) . chr(165) => 't', |
|
244
|
|
|
chr(197) . chr(166) => 'T', |
|
245
|
|
|
chr(197) . chr(167) => 't', |
|
246
|
|
|
chr(197) . chr(168) => 'U', |
|
247
|
|
|
chr(197) . chr(169) => 'u', |
|
248
|
|
|
chr(197) . chr(170) => 'U', |
|
249
|
|
|
chr(197) . chr(171) => 'u', |
|
250
|
|
|
chr(197) . chr(172) => 'U', |
|
251
|
|
|
chr(197) . chr(173) => 'u', |
|
252
|
|
|
chr(197) . chr(174) => 'U', |
|
253
|
|
|
chr(197) . chr(175) => 'u', |
|
254
|
|
|
chr(197) . chr(176) => 'U', |
|
255
|
|
|
chr(197) . chr(177) => 'u', |
|
256
|
|
|
chr(197) . chr(178) => 'U', |
|
257
|
|
|
chr(197) . chr(179) => 'u', |
|
258
|
|
|
chr(197) . chr(180) => 'W', |
|
259
|
|
|
chr(197) . chr(181) => 'w', |
|
260
|
|
|
chr(197) . chr(182) => 'Y', |
|
261
|
|
|
chr(197) . chr(183) => 'y', |
|
262
|
|
|
chr(197) . chr(184) => 'Y', |
|
263
|
|
|
chr(197) . chr(185) => 'Z', |
|
264
|
|
|
chr(197) . chr(186) => 'z', |
|
265
|
|
|
chr(197) . chr(187) => 'Z', |
|
266
|
|
|
chr(197) . chr(188) => 'z', |
|
267
|
|
|
chr(197) . chr(189) => 'Z', |
|
268
|
|
|
chr(197) . chr(190) => 'z', |
|
269
|
|
|
chr(197) . chr(191) => 's', |
|
270
|
|
|
// Euro Sign |
|
271
|
|
|
chr(226) . chr(130) . chr(172) => 'E', |
|
272
|
|
|
); |
|
273
|
|
|
} |
|
274
|
|
|
|
|
275
|
|
|
// transliterate |
|
276
|
|
|
return strtr($string, $charMap); |
|
277
|
|
|
} |
|
278
|
|
|
} |
|
279
|
|
|
|