1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Value object class for a collation |
4
|
|
|
*/ |
5
|
|
|
|
6
|
|
|
declare(strict_types=1); |
7
|
|
|
|
8
|
|
|
namespace PhpMyAdmin\Charsets; |
9
|
|
|
|
10
|
|
|
use function __; |
11
|
|
|
use function _pgettext; |
12
|
|
|
use function explode; |
13
|
|
|
use function implode; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Value object class for a collation |
17
|
|
|
*/ |
18
|
|
|
final class Collation |
19
|
|
|
{ |
20
|
|
|
/** |
21
|
|
|
* A description of the collation |
22
|
|
|
*/ |
23
|
|
|
private string $description; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* @param string $name The collation name |
27
|
|
|
* @param string $charset The name of the character set with which the collation is associated |
28
|
|
|
* @param int $id The collation ID |
29
|
|
|
* @param bool $isDefault Whether the collation is the default for its character set |
30
|
|
|
* @param bool $isCompiled Whether the character set is compiled into the server |
31
|
|
|
* @param int $sortLength Used for determining the memory used to sort strings in this collation |
32
|
|
|
* @param string $padAttribute The collation pad attribute |
33
|
|
|
*/ |
34
|
392 |
|
private function __construct( |
35
|
|
|
private string $name, |
36
|
|
|
private string $charset, |
37
|
|
|
private int $id, |
38
|
|
|
private bool $isDefault, |
39
|
|
|
private bool $isCompiled, |
40
|
|
|
private int $sortLength, |
41
|
|
|
private string $padAttribute, |
42
|
|
|
) { |
43
|
392 |
|
$this->description = $this->buildDescription(); |
44
|
|
|
} |
45
|
|
|
|
46
|
|
|
/** @param string[] $state State obtained from the database server */ |
47
|
392 |
|
public static function fromServer(array $state): self |
48
|
|
|
{ |
49
|
392 |
|
return new self( |
50
|
392 |
|
$state['Collation'] ?? '', |
51
|
392 |
|
$state['Charset'] ?? '', |
52
|
392 |
|
(int) ($state['Id'] ?? 0), |
53
|
392 |
|
isset($state['Default']) && ($state['Default'] === 'Yes' || $state['Default'] === '1'), |
54
|
392 |
|
isset($state['Compiled']) && ($state['Compiled'] === 'Yes' || $state['Compiled'] === '1'), |
55
|
392 |
|
(int) ($state['Sortlen'] ?? 0), |
56
|
392 |
|
$state['Pad_attribute'] ?? '', |
57
|
392 |
|
); |
58
|
|
|
} |
59
|
|
|
|
60
|
392 |
|
public function getName(): string |
61
|
|
|
{ |
62
|
392 |
|
return $this->name; |
63
|
|
|
} |
64
|
|
|
|
65
|
392 |
|
public function getDescription(): string |
66
|
|
|
{ |
67
|
392 |
|
return $this->description; |
68
|
|
|
} |
69
|
|
|
|
70
|
4 |
|
public function getCharset(): string |
71
|
|
|
{ |
72
|
4 |
|
return $this->charset; |
73
|
|
|
} |
74
|
|
|
|
75
|
4 |
|
public function getId(): int |
76
|
|
|
{ |
77
|
4 |
|
return $this->id; |
78
|
|
|
} |
79
|
|
|
|
80
|
4 |
|
public function isDefault(): bool |
81
|
|
|
{ |
82
|
4 |
|
return $this->isDefault; |
83
|
|
|
} |
84
|
|
|
|
85
|
4 |
|
public function isCompiled(): bool |
86
|
|
|
{ |
87
|
4 |
|
return $this->isCompiled; |
88
|
|
|
} |
89
|
|
|
|
90
|
4 |
|
public function getSortLength(): int |
91
|
|
|
{ |
92
|
4 |
|
return $this->sortLength; |
93
|
|
|
} |
94
|
|
|
|
95
|
4 |
|
public function getPadAttribute(): string |
96
|
|
|
{ |
97
|
4 |
|
return $this->padAttribute; |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* Returns description for given collation |
102
|
|
|
* |
103
|
|
|
* @return string collation description |
104
|
|
|
*/ |
105
|
392 |
|
private function buildDescription(): string |
106
|
|
|
{ |
107
|
392 |
|
$parts = explode('_', $this->getName()); |
108
|
|
|
|
109
|
392 |
|
$name = __('Unknown'); |
110
|
392 |
|
$variant = null; |
111
|
392 |
|
$suffixes = []; |
112
|
392 |
|
$unicode = false; |
113
|
392 |
|
$unknown = false; |
114
|
|
|
|
115
|
392 |
|
$level = 0; |
116
|
392 |
|
foreach ($parts as $part) { |
117
|
392 |
|
if ($level === 0) { |
118
|
|
|
/* Next will be language */ |
119
|
392 |
|
$level = 1; |
120
|
|
|
/* First should be charset */ |
121
|
392 |
|
[$name, $unicode, $unknown, $variant] = $this->getNameForLevel0($unicode, $unknown, $part, $variant); |
122
|
392 |
|
continue; |
123
|
|
|
} |
124
|
|
|
|
125
|
244 |
|
if ($level === 1) { |
126
|
|
|
/* Next will be variant unless changed later */ |
127
|
244 |
|
$level = 4; |
128
|
|
|
/* Locale name or code */ |
129
|
244 |
|
[$name, $level, $found] = $this->getNameForLevel1($unicode, $unknown, $part, $name, $level); |
130
|
244 |
|
if ($found) { |
131
|
224 |
|
continue; |
132
|
|
|
} |
133
|
|
|
// Not parsed token, fall to next level |
134
|
|
|
} |
135
|
|
|
|
136
|
232 |
|
if ($level === 2) { |
137
|
|
|
/* Next will be variant */ |
138
|
8 |
|
$level = 4; |
139
|
|
|
/* Germal variant */ |
140
|
8 |
|
if ($part === 'pb') { |
141
|
4 |
|
$name = _pgettext('Collation', 'German (phone book order)'); |
142
|
4 |
|
continue; |
143
|
|
|
} |
144
|
|
|
|
145
|
4 |
|
$name = _pgettext('Collation', 'German (dictionary order)'); |
146
|
|
|
// Not parsed token, fall to next level |
147
|
|
|
} |
148
|
|
|
|
149
|
232 |
|
if ($level === 3) { |
150
|
|
|
/* Next will be variant */ |
151
|
8 |
|
$level = 4; |
152
|
|
|
/* Spanish variant */ |
153
|
8 |
|
if ($part === 'trad') { |
154
|
4 |
|
$name = _pgettext('Collation', 'Spanish (traditional)'); |
155
|
4 |
|
continue; |
156
|
|
|
} |
157
|
|
|
|
158
|
4 |
|
$name = _pgettext('Collation', 'Spanish (modern)'); |
159
|
|
|
// Not parsed token, fall to next level |
160
|
|
|
} |
161
|
|
|
|
162
|
232 |
|
if ($level === 4) { |
163
|
|
|
/* Next will be suffix */ |
164
|
232 |
|
$level = 5; |
165
|
|
|
/* Variant */ |
166
|
232 |
|
$variantFound = $this->getVariant($part); |
167
|
232 |
|
if ($variantFound !== null) { |
168
|
56 |
|
$variant = $variantFound; |
169
|
56 |
|
continue; |
170
|
|
|
} |
171
|
|
|
// Not parsed token, fall to next level |
172
|
|
|
} |
173
|
|
|
|
174
|
232 |
|
if ($level < 5) { |
175
|
|
|
continue; |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
/* Suffixes */ |
179
|
232 |
|
$suffix = $this->addSuffixes($part); |
180
|
232 |
|
if ($suffix === null) { |
181
|
124 |
|
continue; |
182
|
|
|
} |
183
|
|
|
|
184
|
116 |
|
$suffixes[] = $suffix; |
185
|
|
|
} |
186
|
|
|
|
187
|
392 |
|
return $this->buildName($name, $variant, $suffixes); |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
/** @param string[] $suffixes */ |
191
|
392 |
|
private function buildName(string $result, string|null $variant, array $suffixes): string |
192
|
|
|
{ |
193
|
392 |
|
if ($variant !== null) { |
194
|
72 |
|
$result .= ' (' . $variant . ')'; |
195
|
|
|
} |
196
|
|
|
|
197
|
392 |
|
if ($suffixes !== []) { |
198
|
116 |
|
$result .= ', ' . implode(', ', $suffixes); |
199
|
|
|
} |
200
|
|
|
|
201
|
392 |
|
return $result; |
202
|
|
|
} |
203
|
|
|
|
204
|
232 |
|
private function getVariant(string $part): string|null |
205
|
|
|
{ |
206
|
232 |
|
return match ($part) { |
207
|
232 |
|
'0900' => 'UCA 9.0.0', |
208
|
232 |
|
'520' => 'UCA 5.2.0', |
209
|
232 |
|
'mysql561' => 'MySQL 5.6.1', |
210
|
232 |
|
'mysql500' => 'MySQL 5.0.0', |
211
|
232 |
|
default => null, |
212
|
232 |
|
}; |
213
|
|
|
} |
214
|
|
|
|
215
|
232 |
|
private function addSuffixes(string $part): string|null |
216
|
|
|
{ |
217
|
232 |
|
return match ($part) { |
218
|
232 |
|
'ci' => _pgettext('Collation variant', 'case-insensitive'), |
219
|
232 |
|
'cs' => _pgettext('Collation variant', 'case-sensitive'), |
220
|
232 |
|
'ai' => _pgettext('Collation variant', 'accent-insensitive'), |
221
|
232 |
|
'as' => _pgettext('Collation variant', 'accent-sensitive'), |
222
|
232 |
|
'ks' => _pgettext('Collation variant', 'kana-sensitive'), |
223
|
232 |
|
'w2','l2' => _pgettext('Collation variant', 'multi-level'), |
224
|
232 |
|
'bin' => _pgettext('Collation variant', 'binary'), |
225
|
232 |
|
'nopad' => _pgettext('Collation variant', 'no-pad'), |
226
|
232 |
|
default => null, |
227
|
232 |
|
}; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* @return array<int, bool|string|null> |
232
|
|
|
* @psalm-return array{string, bool, bool, string|null} |
233
|
|
|
*/ |
234
|
392 |
|
private function getNameForLevel0( |
235
|
|
|
bool $unicode, |
236
|
|
|
bool $unknown, |
237
|
|
|
string $part, |
238
|
|
|
string|null $variant, |
239
|
|
|
): array { |
240
|
|
|
switch ($part) { |
241
|
392 |
|
case 'binary': |
242
|
4 |
|
$name = _pgettext('Collation', 'Binary'); |
243
|
4 |
|
break; |
244
|
|
|
// Unicode charsets |
245
|
388 |
|
case 'utf8mb4': |
246
|
64 |
|
$variant = 'UCA 4.0.0'; |
247
|
|
|
// Fall through to other unicode |
248
|
324 |
|
case 'ucs2': |
249
|
316 |
|
case 'utf8': |
250
|
296 |
|
case 'utf8mb3': |
251
|
296 |
|
case 'utf16': |
252
|
296 |
|
case 'utf16le': |
253
|
296 |
|
case 'utf16be': |
254
|
296 |
|
case 'utf32': |
255
|
100 |
|
$name = _pgettext('Collation', 'Unicode'); |
256
|
100 |
|
$unicode = true; |
257
|
100 |
|
break; |
258
|
|
|
// West European charsets |
259
|
288 |
|
case 'ascii': |
260
|
284 |
|
case 'cp850': |
261
|
280 |
|
case 'dec8': |
262
|
276 |
|
case 'hp8': |
263
|
272 |
|
case 'latin1': |
264
|
264 |
|
case 'macroman': |
265
|
24 |
|
$name = _pgettext('Collation', 'West European'); |
266
|
24 |
|
break; |
267
|
|
|
// Central European charsets |
268
|
264 |
|
case 'cp1250': |
269
|
256 |
|
case 'cp852': |
270
|
252 |
|
case 'latin2': |
271
|
248 |
|
case 'macce': |
272
|
20 |
|
$name = _pgettext('Collation', 'Central European'); |
273
|
20 |
|
break; |
274
|
|
|
// Russian charsets |
275
|
244 |
|
case 'cp866': |
276
|
240 |
|
case 'koi8r': |
277
|
8 |
|
$name = _pgettext('Collation', 'Russian'); |
278
|
8 |
|
break; |
279
|
|
|
// Chinese charsets |
280
|
236 |
|
case 'gb2312': |
281
|
228 |
|
case 'gbk': |
282
|
16 |
|
$name = _pgettext('Collation', 'Simplified Chinese'); |
283
|
16 |
|
break; |
284
|
220 |
|
case 'big5': |
285
|
12 |
|
$name = _pgettext('Collation', 'Traditional Chinese'); |
286
|
12 |
|
break; |
287
|
208 |
|
case 'gb18030': |
288
|
4 |
|
$name = _pgettext('Collation', 'Chinese'); |
289
|
4 |
|
$unicode = true; |
290
|
4 |
|
break; |
291
|
|
|
// Japanese charsets |
292
|
204 |
|
case 'sjis': |
293
|
200 |
|
case 'ujis': |
294
|
196 |
|
case 'cp932': |
295
|
192 |
|
case 'eucjpms': |
296
|
16 |
|
$name = _pgettext('Collation', 'Japanese'); |
297
|
16 |
|
break; |
298
|
|
|
// Baltic charsets |
299
|
188 |
|
case 'cp1257': |
300
|
184 |
|
case 'latin7': |
301
|
8 |
|
$name = _pgettext('Collation', 'Baltic'); |
302
|
8 |
|
break; |
303
|
|
|
// Other |
304
|
180 |
|
case 'armscii8': |
305
|
176 |
|
case 'armscii': |
306
|
8 |
|
$name = _pgettext('Collation', 'Armenian'); |
307
|
8 |
|
break; |
308
|
172 |
|
case 'cp1251': |
309
|
4 |
|
$name = _pgettext('Collation', 'Cyrillic'); |
310
|
4 |
|
break; |
311
|
168 |
|
case 'cp1256': |
312
|
4 |
|
$name = _pgettext('Collation', 'Arabic'); |
313
|
4 |
|
break; |
314
|
164 |
|
case 'euckr': |
315
|
4 |
|
$name = _pgettext('Collation', 'Korean'); |
316
|
4 |
|
break; |
317
|
160 |
|
case 'hebrew': |
318
|
4 |
|
$name = _pgettext('Collation', 'Hebrew'); |
319
|
4 |
|
break; |
320
|
156 |
|
case 'geostd8': |
321
|
4 |
|
$name = _pgettext('Collation', 'Georgian'); |
322
|
4 |
|
break; |
323
|
152 |
|
case 'greek': |
324
|
4 |
|
$name = _pgettext('Collation', 'Greek'); |
325
|
4 |
|
break; |
326
|
148 |
|
case 'keybcs2': |
327
|
4 |
|
$name = _pgettext('Collation', 'Czech-Slovak'); |
328
|
4 |
|
break; |
329
|
144 |
|
case 'koi8u': |
330
|
4 |
|
$name = _pgettext('Collation', 'Ukrainian'); |
331
|
4 |
|
break; |
332
|
140 |
|
case 'latin5': |
333
|
4 |
|
$name = _pgettext('Collation', 'Turkish'); |
334
|
4 |
|
break; |
335
|
136 |
|
case 'swe7': |
336
|
4 |
|
$name = _pgettext('Collation', 'Swedish'); |
337
|
4 |
|
break; |
338
|
132 |
|
case 'tis620': |
339
|
4 |
|
$name = _pgettext('Collation', 'Thai'); |
340
|
4 |
|
break; |
341
|
|
|
default: |
342
|
128 |
|
$name = _pgettext('Collation', 'Unknown'); |
343
|
128 |
|
$unknown = true; |
344
|
128 |
|
break; |
345
|
|
|
} |
346
|
|
|
|
347
|
392 |
|
return [$name, $unicode, $unknown, $variant]; |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
/** |
351
|
|
|
* @return array<int, bool|int|string> |
352
|
|
|
* @psalm-return array{string, int, bool} |
353
|
|
|
*/ |
354
|
244 |
|
private function getNameForLevel1( |
355
|
|
|
bool $unicode, |
356
|
|
|
bool $unknown, |
357
|
|
|
string $part, |
358
|
|
|
string $name, |
359
|
|
|
int $level, |
360
|
|
|
): array { |
361
|
244 |
|
$found = true; |
362
|
|
|
|
363
|
|
|
switch ($part) { |
364
|
244 |
|
case 'general': |
365
|
16 |
|
break; |
366
|
228 |
|
case 'bulgarian': |
367
|
224 |
|
case 'bg': |
368
|
4 |
|
$name = _pgettext('Collation', 'Bulgarian'); |
369
|
4 |
|
break; |
370
|
224 |
|
case 'chinese': |
371
|
204 |
|
case 'cn': |
372
|
204 |
|
case 'zh': |
373
|
20 |
|
if ($unicode) { |
374
|
4 |
|
$name = _pgettext('Collation', 'Chinese'); |
375
|
|
|
} |
376
|
|
|
|
377
|
20 |
|
break; |
378
|
204 |
|
case 'croatian': |
379
|
196 |
|
case 'hr': |
380
|
8 |
|
$name = _pgettext('Collation', 'Croatian'); |
381
|
8 |
|
break; |
382
|
196 |
|
case 'czech': |
383
|
184 |
|
case 'cs': |
384
|
12 |
|
$name = _pgettext('Collation', 'Czech'); |
385
|
12 |
|
break; |
386
|
184 |
|
case 'danish': |
387
|
180 |
|
case 'da': |
388
|
4 |
|
$name = _pgettext('Collation', 'Danish'); |
389
|
4 |
|
break; |
390
|
180 |
|
case 'english': |
391
|
176 |
|
case 'en': |
392
|
4 |
|
$name = _pgettext('Collation', 'English'); |
393
|
4 |
|
break; |
394
|
176 |
|
case 'esperanto': |
395
|
172 |
|
case 'eo': |
396
|
4 |
|
$name = _pgettext('Collation', 'Esperanto'); |
397
|
4 |
|
break; |
398
|
172 |
|
case 'estonian': |
399
|
168 |
|
case 'et': |
400
|
4 |
|
$name = _pgettext('Collation', 'Estonian'); |
401
|
4 |
|
break; |
402
|
168 |
|
case 'german1': |
403
|
4 |
|
$name = _pgettext('Collation', 'German (dictionary order)'); |
404
|
4 |
|
break; |
405
|
164 |
|
case 'german2': |
406
|
4 |
|
$name = _pgettext('Collation', 'German (phone book order)'); |
407
|
4 |
|
break; |
408
|
160 |
|
case 'german': |
409
|
160 |
|
case 'de': |
410
|
|
|
/* Name is set later */ |
411
|
8 |
|
$level = 2; |
412
|
8 |
|
break; |
413
|
152 |
|
case 'hungarian': |
414
|
148 |
|
case 'hu': |
415
|
4 |
|
$name = _pgettext('Collation', 'Hungarian'); |
416
|
4 |
|
break; |
417
|
148 |
|
case 'icelandic': |
418
|
144 |
|
case 'is': |
419
|
4 |
|
$name = _pgettext('Collation', 'Icelandic'); |
420
|
4 |
|
break; |
421
|
144 |
|
case 'japanese': |
422
|
140 |
|
case 'ja': |
423
|
8 |
|
$name = _pgettext('Collation', 'Japanese'); |
424
|
8 |
|
break; |
425
|
136 |
|
case 'la': |
426
|
4 |
|
$name = _pgettext('Collation', 'Classical Latin'); |
427
|
4 |
|
break; |
428
|
132 |
|
case 'latvian': |
429
|
128 |
|
case 'lv': |
430
|
4 |
|
$name = _pgettext('Collation', 'Latvian'); |
431
|
4 |
|
break; |
432
|
128 |
|
case 'lithuanian': |
433
|
124 |
|
case 'lt': |
434
|
4 |
|
$name = _pgettext('Collation', 'Lithuanian'); |
435
|
4 |
|
break; |
436
|
124 |
|
case 'korean': |
437
|
120 |
|
case 'ko': |
438
|
4 |
|
$name = _pgettext('Collation', 'Korean'); |
439
|
4 |
|
break; |
440
|
120 |
|
case 'myanmar': |
441
|
116 |
|
case 'my': |
442
|
4 |
|
$name = _pgettext('Collation', 'Burmese'); |
443
|
4 |
|
break; |
444
|
116 |
|
case 'persian': |
445
|
4 |
|
$name = _pgettext('Collation', 'Persian'); |
446
|
4 |
|
break; |
447
|
112 |
|
case 'polish': |
448
|
108 |
|
case 'pl': |
449
|
4 |
|
$name = _pgettext('Collation', 'Polish'); |
450
|
4 |
|
break; |
451
|
108 |
|
case 'roman': |
452
|
4 |
|
$name = _pgettext('Collation', 'West European'); |
453
|
4 |
|
break; |
454
|
104 |
|
case 'romanian': |
455
|
100 |
|
case 'ro': |
456
|
4 |
|
$name = _pgettext('Collation', 'Romanian'); |
457
|
4 |
|
break; |
458
|
100 |
|
case 'ru': |
459
|
4 |
|
$name = _pgettext('Collation', 'Russian'); |
460
|
4 |
|
break; |
461
|
96 |
|
case 'si': |
462
|
96 |
|
case 'sinhala': |
463
|
4 |
|
$name = _pgettext('Collation', 'Sinhalese'); |
464
|
4 |
|
break; |
465
|
92 |
|
case 'slovak': |
466
|
88 |
|
case 'sk': |
467
|
4 |
|
$name = _pgettext('Collation', 'Slovak'); |
468
|
4 |
|
break; |
469
|
88 |
|
case 'slovenian': |
470
|
84 |
|
case 'sl': |
471
|
4 |
|
$name = _pgettext('Collation', 'Slovenian'); |
472
|
4 |
|
break; |
473
|
84 |
|
case 'spanish': |
474
|
4 |
|
$name = _pgettext('Collation', 'Spanish (modern)'); |
475
|
4 |
|
break; |
476
|
80 |
|
case 'es': |
477
|
|
|
/* Name is set later */ |
478
|
8 |
|
$level = 3; |
479
|
8 |
|
break; |
480
|
72 |
|
case 'spanish2': |
481
|
4 |
|
$name = _pgettext('Collation', 'Spanish (traditional)'); |
482
|
4 |
|
break; |
483
|
68 |
|
case 'swedish': |
484
|
64 |
|
case 'sv': |
485
|
4 |
|
$name = _pgettext('Collation', 'Swedish'); |
486
|
4 |
|
break; |
487
|
64 |
|
case 'thai': |
488
|
56 |
|
case 'th': |
489
|
8 |
|
$name = _pgettext('Collation', 'Thai'); |
490
|
8 |
|
break; |
491
|
56 |
|
case 'turkish': |
492
|
48 |
|
case 'tr': |
493
|
12 |
|
$name = _pgettext('Collation', 'Turkish'); |
494
|
12 |
|
break; |
495
|
44 |
|
case 'ukrainian': |
496
|
40 |
|
case 'uk': |
497
|
4 |
|
$name = _pgettext('Collation', 'Ukrainian'); |
498
|
4 |
|
break; |
499
|
40 |
|
case 'vietnamese': |
500
|
36 |
|
case 'vi': |
501
|
4 |
|
$name = _pgettext('Collation', 'Vietnamese'); |
502
|
4 |
|
break; |
503
|
36 |
|
case 'unicode': |
504
|
16 |
|
if ($unknown) { |
505
|
4 |
|
$name = _pgettext('Collation', 'Unicode'); |
506
|
|
|
} |
507
|
|
|
|
508
|
16 |
|
break; |
509
|
|
|
default: |
510
|
20 |
|
$found = false; |
511
|
|
|
} |
512
|
|
|
|
513
|
244 |
|
return [$name, $level, $found]; |
514
|
|
|
} |
515
|
|
|
} |
516
|
|
|
|