1 /*
2 +----------------------------------------------------------------------+
3 | Copyright (c) The PHP Group |
4 +----------------------------------------------------------------------+
5 | This source file is subject to version 3.01 of the PHP license, |
6 | that is bundled with this package in the file LICENSE, and is |
7 | available through the world-wide-web at the following url: |
8 | https://www.php.net/license/3_01.txt |
9 | If you did not receive a copy of the PHP license and are unable to |
10 | obtain it through the world-wide-web, please send a note to |
11 | license@php.net so we can mail you a copy immediately. |
12 +----------------------------------------------------------------------+
13 | Authors: Andrey Hristov <andrey@php.net> |
14 | Ulf Wendel <uw@php.net> |
15 | Georg Richter <georg@php.net> |
16 +----------------------------------------------------------------------+
17 */
18
19 #include "php.h"
20 #include "mysqlnd.h"
21 #include "mysqlnd_priv.h"
22 #include "mysqlnd_debug.h"
23 #include "mysqlnd_charset.h"
24
25 #define ENUMERATE_ENCODINGS_CHARLEN(ENUMERATOR) \
26 ENUMERATOR(mysqlnd_mbcharlen_null) \
27 ENUMERATOR(mysqlnd_mbcharlen_big5) \
28 ENUMERATOR(mysqlnd_mbcharlen_ujis) \
29 ENUMERATOR(mysqlnd_mbcharlen_sjis) \
30 ENUMERATOR(mysqlnd_mbcharlen_euckr) \
31 ENUMERATOR(mysqlnd_mbcharlen_gb2312) \
32 ENUMERATOR(mysqlnd_mbcharlen_gbk) \
33 ENUMERATOR(mysqlnd_mbcharlen_utf8mb3) \
34 ENUMERATOR(mysqlnd_mbcharlen_ucs2) \
35 ENUMERATOR(mysqlnd_mbcharlen_eucjpms) \
36 ENUMERATOR(mysqlnd_mbcharlen_utf8) \
37 ENUMERATOR(mysqlnd_mbcharlen_utf16) \
38 ENUMERATOR(mysqlnd_mbcharlen_utf32) \
39 ENUMERATOR(mysqlnd_mbcharlen_cp932) \
40 ENUMERATOR(mysqlnd_mbcharlen_gb18030)
41
42 #define ENUMERATE_ENCODINGS_VALID(ENUMERATOR) \
43 ENUMERATOR(check_null) \
44 ENUMERATOR(check_mb_big5) \
45 ENUMERATOR(check_mb_ujis) \
46 ENUMERATOR(check_mb_sjis) \
47 ENUMERATOR(check_mb_euckr) \
48 ENUMERATOR(check_mb_gb2312) \
49 ENUMERATOR(check_mb_gbk) \
50 ENUMERATOR(check_mb_utf8mb3_valid) \
51 ENUMERATOR(check_mb_ucs2) \
52 ENUMERATOR(check_mb_eucjpms) \
53 ENUMERATOR(check_mb_utf8_valid) \
54 ENUMERATOR(check_mb_utf16) \
55 ENUMERATOR(check_mb_utf32) \
56 ENUMERATOR(check_mb_cp932) \
57 ENUMERATOR(my_ismbchar_gb18030)
58
59 #define LOWEST_MB_BIG5 0xA1
60 #define LOWEST_MB_UJIS 0x80
61 #define LOWEST_MB_SJIS 0x80
62 #define LOWEST_MB_EUCKR 0x80
63 #define LOWEST_MB_GB2312 0xA1
64 #define LOWEST_MB_GBK 0x80
65 #define LOWEST_MB_UTF8MB3 0x80
66 #define LOWEST_MB_UCS2 0x00
67 #define LOWEST_MB_EUCJPMS 0x80
68 #define LOWEST_MB_UTF8 0x80
69 #define LOWEST_MB_UTF16 0x00
70 #define LOWEST_MB_UTF32 0x00
71 #define LOWEST_MB_CP932 0x80
72 #define LOWEST_MB_GB18030 0x00
73
74 #define ENUMERATOR_ENUM(x) x##_id,
75 enum mysqlnd_encoding_charlen {
76 ENUMERATE_ENCODINGS_CHARLEN(ENUMERATOR_ENUM)
77 };
78 enum mysqlnd_encoding_valid {
79 ENUMERATE_ENCODINGS_VALID(ENUMERATOR_ENUM)
80 };
81 #undef ENUMERATOR_ENUM
82
83 static unsigned int mysqlnd_mbcharlen_dispatch(enum mysqlnd_encoding_charlen encoding, const unsigned int c);
84 static unsigned int mysqlnd_mbvalid_dispatch(enum mysqlnd_encoding_valid encoding, const char * const start, const char * const end);
85
86 /* {{{ utf8 functions */
check_mb_utf8mb3_sequence(const char * const start,const char * const end)87 static unsigned int check_mb_utf8mb3_sequence(const char * const start, const char * const end)
88 {
89 zend_uchar c;
90
91 if (UNEXPECTED(start >= end)) {
92 return 0;
93 }
94
95 c = (zend_uchar) start[0];
96
97 if (c < 0x80) {
98 return 1; /* single byte character */
99 }
100 if (c < 0xC2) {
101 return 0; /* invalid mb character */
102 }
103 if (c < 0xE0) {
104 if (start + 2 > end) {
105 return 0; /* too small */
106 }
107 if (!(((zend_uchar)start[1] ^ 0x80) < 0x40)) {
108 return 0;
109 }
110 return 2;
111 }
112 if (c < 0xF0) {
113 if (start + 3 > end) {
114 return 0; /* too small */
115 }
116 if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 && ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
117 (c >= 0xE1 || (zend_uchar)start[1] >= 0xA0))) {
118 return 0; /* invalid utf8 character */
119 }
120 return 3;
121 }
122 return 0;
123 }
124
125
check_mb_utf8_sequence(const char * const start,const char * const end)126 static unsigned int check_mb_utf8_sequence(const char * const start, const char * const end)
127 {
128 zend_uchar c;
129
130 if (UNEXPECTED(start >= end)) {
131 return 0;
132 }
133
134 c = (zend_uchar) start[0];
135
136 if (c < 0x80) {
137 return 1; /* single byte character */
138 }
139 if (c < 0xC2) {
140 return 0; /* invalid mb character */
141 }
142 if (c < 0xE0) {
143 if (start + 2 > end) {
144 return 0; /* too small */
145 }
146 if (!(((zend_uchar)start[1] ^ 0x80) < 0x40)) {
147 return 0;
148 }
149 return 2;
150 }
151 if (c < 0xF0) {
152 if (start + 3 > end) {
153 return 0; /* too small */
154 }
155 if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 && ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
156 (c >= 0xE1 || (zend_uchar)start[1] >= 0xA0))) {
157 return 0; /* invalid utf8 character */
158 }
159 return 3;
160 }
161 if (c < 0xF5) {
162 if (start + 4 > end) { /* We need 4 characters */
163 return 0; /* too small */
164 }
165
166 /*
167 UTF-8 quick four-byte mask:
168 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
169 Encoding allows to encode U+00010000..U+001FFFFF
170
171 The maximum character defined in the Unicode standard is U+0010FFFF.
172 Higher characters U+00110000..U+001FFFFF are not used.
173
174 11110000.10010000.10xxxxxx.10xxxxxx == F0.90.80.80 == U+00010000 (min)
175 11110100.10001111.10111111.10111111 == F4.8F.BF.BF == U+0010FFFF (max)
176
177 Valid codes:
178 [F0][90..BF][80..BF][80..BF]
179 [F1][80..BF][80..BF][80..BF]
180 [F2][80..BF][80..BF][80..BF]
181 [F3][80..BF][80..BF][80..BF]
182 [F4][80..8F][80..BF][80..BF]
183 */
184
185 if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 &&
186 ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
187 ((zend_uchar)start[3] ^ 0x80) < 0x40 &&
188 (c >= 0xf1 || (zend_uchar)start[1] >= 0x90) &&
189 (c <= 0xf3 || (zend_uchar)start[1] <= 0x8F)))
190 {
191 return 0; /* invalid utf8 character */
192 }
193 return 4;
194 }
195 return 0;
196 }
197
check_mb_utf8mb3_valid(const char * const start,const char * const end)198 static unsigned int check_mb_utf8mb3_valid(const char * const start, const char * const end)
199 {
200 unsigned int len = check_mb_utf8mb3_sequence(start, end);
201 return (len > 1)? len:0;
202 }
203
check_mb_utf8_valid(const char * const start,const char * const end)204 static unsigned int check_mb_utf8_valid(const char * const start, const char * const end)
205 {
206 unsigned int len = check_mb_utf8_sequence(start, end);
207 return (len > 1)? len:0;
208 }
209
210
mysqlnd_mbcharlen_utf8mb3(const unsigned int utf8)211 static unsigned int mysqlnd_mbcharlen_utf8mb3(const unsigned int utf8)
212 {
213 if (utf8 < 0x80) {
214 return 1; /* single byte character */
215 }
216 if (utf8 < 0xC2) {
217 return 0; /* invalid multibyte header */
218 }
219 if (utf8 < 0xE0) {
220 return 2; /* double byte character */
221 }
222 if (utf8 < 0xF0) {
223 return 3; /* triple byte character */
224 }
225 return 0;
226 }
227
228
mysqlnd_mbcharlen_utf8(const unsigned int utf8)229 static unsigned int mysqlnd_mbcharlen_utf8(const unsigned int utf8)
230 {
231 if (utf8 < 0x80) {
232 return 1; /* single byte character */
233 }
234 if (utf8 < 0xC2) {
235 return 0; /* invalid multibyte header */
236 }
237 if (utf8 < 0xE0) {
238 return 2; /* double byte character */
239 }
240 if (utf8 < 0xF0) {
241 return 3; /* triple byte character */
242 }
243 if (utf8 < 0xF8) {
244 return 4; /* four byte character */
245 }
246 return 0;
247 }
248 /* }}} */
249
250
251 /* {{{ big5 functions */
252 #define valid_big5head(c) (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xF9)
253 #define valid_big5tail(c) ((0x40 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0x7E) || \
254 (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xFE))
255
256 #define isbig5code(c,d) (isbig5head(c) && isbig5tail(d))
257
check_mb_big5(const char * const start,const char * const end)258 static unsigned int check_mb_big5(const char * const start, const char * const end)
259 {
260 return (valid_big5head(*(start)) && (end - start) > 1 && valid_big5tail(*(start + 1)) ? 2 : 0);
261 }
262
263
mysqlnd_mbcharlen_big5(const unsigned int big5)264 static unsigned int mysqlnd_mbcharlen_big5(const unsigned int big5)
265 {
266 return (valid_big5head(big5)) ? 2 : 1;
267 }
268 /* }}} */
269
270
271 /* {{{ cp932 functions */
272 #define valid_cp932head(c) ((0x81 <= (c) && (c) <= 0x9F) || (0xE0 <= (c) && c <= 0xFC))
273 #define valid_cp932tail(c) ((0x40 <= (c) && (c) <= 0x7E) || (0x80 <= (c) && c <= 0xFC))
274
275
check_mb_cp932(const char * const start,const char * const end)276 static unsigned int check_mb_cp932(const char * const start, const char * const end)
277 {
278 return (valid_cp932head((zend_uchar)start[0]) && (end - start > 1) &&
279 valid_cp932tail((zend_uchar)start[1])) ? 2 : 0;
280 }
281
282
mysqlnd_mbcharlen_cp932(const unsigned int cp932)283 static unsigned int mysqlnd_mbcharlen_cp932(const unsigned int cp932)
284 {
285 return (valid_cp932head((zend_uchar)cp932)) ? 2 : 1;
286 }
287 /* }}} */
288
289
290 /* {{{ euckr functions */
291 #define valid_euckr(c) ((0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xFE))
292
check_mb_euckr(const char * const start,const char * const end)293 static unsigned int check_mb_euckr(const char * const start, const char * const end)
294 {
295 if (end - start <= 1) {
296 return 0; /* invalid length */
297 }
298 if (*(zend_uchar *)start < 0x80) {
299 return 0; /* invalid euckr character */
300 }
301 if (valid_euckr(start[1])) {
302 return 2;
303 }
304 return 0;
305 }
306
307
mysqlnd_mbcharlen_euckr(const unsigned int kr)308 static unsigned int mysqlnd_mbcharlen_euckr(const unsigned int kr)
309 {
310 return (valid_euckr(kr)) ? 2 : 1;
311 }
312 /* }}} */
313
314
315 /* {{{ eucjpms functions */
316 #define valid_eucjpms(c) (((c) & 0xFF) >= 0xA1 && ((c) & 0xFF) <= 0xFE)
317 #define valid_eucjpms_kata(c) (((c) & 0xFF) >= 0xA1 && ((c) & 0xFF) <= 0xDF)
318 #define valid_eucjpms_ss2(c) (((c) & 0xFF) == 0x8E)
319 #define valid_eucjpms_ss3(c) (((c) & 0xFF) == 0x8F)
320
check_mb_eucjpms(const char * const start,const char * const end)321 static unsigned int check_mb_eucjpms(const char * const start, const char * const end)
322 {
323 if (*((zend_uchar *)start) < 0x80) {
324 return 0; /* invalid eucjpms character */
325 }
326 if (valid_eucjpms(start[0]) && (end - start) > 1 && valid_eucjpms(start[1])) {
327 return 2;
328 }
329 if (valid_eucjpms_ss2(start[0]) && (end - start) > 1 && valid_eucjpms_kata(start[1])) {
330 return 2;
331 }
332 if (valid_eucjpms_ss3(start[0]) && (end - start) > 2 && valid_eucjpms(start[1]) &&
333 valid_eucjpms(start[2])) {
334 return 3;
335 }
336 return 0;
337 }
338
339
mysqlnd_mbcharlen_eucjpms(const unsigned int jpms)340 static unsigned int mysqlnd_mbcharlen_eucjpms(const unsigned int jpms)
341 {
342 if (valid_eucjpms(jpms) || valid_eucjpms_ss2(jpms)) {
343 return 2;
344 }
345 if (valid_eucjpms_ss3(jpms)) {
346 return 3;
347 }
348 return 1;
349 }
350 /* }}} */
351
352
353 /* {{{ gb2312 functions */
354 #define valid_gb2312_head(c) (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xF7)
355 #define valid_gb2312_tail(c) (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xFE)
356
357
check_mb_gb2312(const char * const start,const char * const end)358 static unsigned int check_mb_gb2312(const char * const start, const char * const end)
359 {
360 return (valid_gb2312_head((unsigned int)start[0]) && end - start > 1 &&
361 valid_gb2312_tail((unsigned int)start[1])) ? 2 : 0;
362 }
363
364
mysqlnd_mbcharlen_gb2312(const unsigned int gb)365 static unsigned int mysqlnd_mbcharlen_gb2312(const unsigned int gb)
366 {
367 return (valid_gb2312_head(gb)) ? 2 : 1;
368 }
369 /* }}} */
370
371
372 /* {{{ gbk functions */
373 #define valid_gbk_head(c) (0x81<=(zend_uchar)(c) && (zend_uchar)(c)<=0xFE)
374 #define valid_gbk_tail(c) ((0x40<=(zend_uchar)(c) && (zend_uchar)(c)<=0x7E) || (0x80<=(zend_uchar)(c) && (zend_uchar)(c)<=0xFE))
375
check_mb_gbk(const char * const start,const char * const end)376 static unsigned int check_mb_gbk(const char * const start, const char * const end)
377 {
378 return (valid_gbk_head(start[0]) && (end) - (start) > 1 && valid_gbk_tail(start[1])) ? 2 : 0;
379 }
380
mysqlnd_mbcharlen_gbk(const unsigned int gbk)381 static unsigned int mysqlnd_mbcharlen_gbk(const unsigned int gbk)
382 {
383 return (valid_gbk_head(gbk) ? 2 : 1);
384 }
385 /* }}} */
386
387
388 /* {{{ sjis functions */
389 #define valid_sjis_head(c) ((0x81 <= (c) && (c) <= 0x9F) || (0xE0 <= (c) && (c) <= 0xFC))
390 #define valid_sjis_tail(c) ((0x40 <= (c) && (c) <= 0x7E) || (0x80 <= (c) && (c) <= 0xFC))
391
392
check_mb_sjis(const char * const start,const char * const end)393 static unsigned int check_mb_sjis(const char * const start, const char * const end)
394 {
395 return (valid_sjis_head((zend_uchar)start[0]) && (end - start) > 1 && valid_sjis_tail((zend_uchar)start[1])) ? 2 : 0;
396 }
397
398
mysqlnd_mbcharlen_sjis(const unsigned int sjis)399 static unsigned int mysqlnd_mbcharlen_sjis(const unsigned int sjis)
400 {
401 return (valid_sjis_head((zend_uchar)sjis)) ? 2 : 1;
402 }
403 /* }}} */
404
405
406 /* {{{ ucs2 functions */
check_mb_ucs2(const char * const start __attribute ((unused)),const char * const end __attribute ((unused)))407 static unsigned int check_mb_ucs2(const char * const start __attribute((unused)), const char * const end __attribute((unused)))
408 {
409 return 2; /* always 2 */
410 }
411
mysqlnd_mbcharlen_ucs2(const unsigned int ucs2 __attribute ((unused)))412 static unsigned int mysqlnd_mbcharlen_ucs2(const unsigned int ucs2 __attribute((unused)))
413 {
414 return 2; /* always 2 */
415 }
416 /* }}} */
417
418
419 /* {{{ ujis functions */
420 #define valid_ujis(c) ((0xA1 <= ((c)&0xFF) && ((c)&0xFF) <= 0xFE))
421 #define valid_ujis_kata(c) ((0xA1 <= ((c)&0xFF) && ((c)&0xFF) <= 0xDF))
422 #define valid_ujis_ss2(c) (((c)&0xFF) == 0x8E)
423 #define valid_ujis_ss3(c) (((c)&0xFF) == 0x8F)
424
check_mb_ujis(const char * const start,const char * const end)425 static unsigned int check_mb_ujis(const char * const start, const char * const end)
426 {
427 if (*(zend_uchar*)start < 0x80) {
428 return 0; /* invalid ujis character */
429 }
430 if (valid_ujis(*(start)) && valid_ujis(*((start)+1))) {
431 return 2;
432 }
433 if (valid_ujis_ss2(*(start)) && valid_ujis_kata(*((start)+1))) {
434 return 2;
435 }
436 if (valid_ujis_ss3(*(start)) && (end-start) > 2 && valid_ujis(*((start)+1)) && valid_ujis(*((start)+2))) {
437 return 3;
438 }
439 return 0;
440 }
441
442
mysqlnd_mbcharlen_ujis(const unsigned int ujis)443 static unsigned int mysqlnd_mbcharlen_ujis(const unsigned int ujis)
444 {
445 return (valid_ujis(ujis)? 2: valid_ujis_ss2(ujis)? 2: valid_ujis_ss3(ujis)? 3: 1);
446 }
447 /* }}} */
448
449
450
451 /* {{{ utf16 functions */
452 #define UTF16_HIGH_HEAD(x) ((((zend_uchar) (x)) & 0xFC) == 0xD8)
453 #define UTF16_LOW_HEAD(x) ((((zend_uchar) (x)) & 0xFC) == 0xDC)
454
check_mb_utf16(const char * const start,const char * const end)455 static unsigned int check_mb_utf16(const char * const start, const char * const end)
456 {
457 if (start + 2 > end) {
458 return 0;
459 }
460
461 if (UTF16_HIGH_HEAD(*start)) {
462 return (start + 4 <= end) && UTF16_LOW_HEAD(start[2]) ? 4 : 0;
463 }
464
465 if (UTF16_LOW_HEAD(*start)) {
466 return 0;
467 }
468 return 2;
469 }
470
471
mysqlnd_mbcharlen_utf16(const unsigned int utf16)472 static uint32_t mysqlnd_mbcharlen_utf16(const unsigned int utf16)
473 {
474 return UTF16_HIGH_HEAD(utf16) ? 4 : 2;
475 }
476 /* }}} */
477
478
479 /* {{{ utf32 functions */
check_mb_utf32(const char * const start __attribute ((unused)),const char * const end __attribute ((unused)))480 static unsigned int check_mb_utf32(const char * const start __attribute((unused)), const char * const end __attribute((unused)))
481 {
482 return 4;
483 }
484
485
mysqlnd_mbcharlen_utf32(const unsigned int utf32 __attribute ((unused)))486 static unsigned int mysqlnd_mbcharlen_utf32(const unsigned int utf32 __attribute((unused)))
487 {
488 return 4;
489 }
490 /* }}} */
491
492
493 /* {{{ gb18030 functions */
494 #define is_gb18030_odd(c) (0x81 <= (zend_uchar) (c) && (zend_uchar) (c) <= 0xFE)
495 #define is_gb18030_even_2(c) ((0x40 <= (zend_uchar) (c) && (zend_uchar) (c) <= 0x7E) || (0x80 <= (zend_uchar) (c) && (zend_uchar) (c) <= 0xFE))
496 #define is_gb18030_even_4(c) (0x30 <= (zend_uchar) (c) && (zend_uchar) (c) <= 0x39)
497
498
mysqlnd_mbcharlen_gb18030(const unsigned int c)499 static unsigned int mysqlnd_mbcharlen_gb18030(const unsigned int c)
500 {
501 if (c <= 0xFF) {
502 return !is_gb18030_odd(c);
503 }
504 if (c > 0xFFFF || !is_gb18030_odd((c >> 8) & 0xFF)) {
505 return 0;
506 }
507 if (is_gb18030_even_2((c & 0xFF))) {
508 return 2;
509 }
510 if (is_gb18030_even_4((c & 0xFF))) {
511 return 4;
512 }
513
514 return 0;
515 }
516
517
my_ismbchar_gb18030(const char * start,const char * end)518 static unsigned int my_ismbchar_gb18030(const char * start, const char * end)
519 {
520 if (end - start <= 1 || !is_gb18030_odd(start[0])) {
521 return 0;
522 }
523
524 if (is_gb18030_even_2(start[1])) {
525 return 2;
526 } else if (end - start > 3 && is_gb18030_even_4(start[1]) && is_gb18030_odd(start[2]) && is_gb18030_even_4(start[3])) {
527 return 4;
528 }
529
530 return 0;
531 }
532 /* }}} */
533
534 /*
535 The server compiles sometimes the full utf-8 (the mb4) as utf8m4, and the old as utf8,
536 for BC reasons. Sometimes, utf8mb4 is just utf8 but the old charsets are utf8mb3.
537 Change easily now, with a macro, could be made compilastion dependable.
538 */
539
540 #define UTF8_MB4 "utf8mb4"
541 #define UTF8_MB3 "utf8"
542
mysqlnd_mbcharlen_null(const unsigned int c)543 static zend_always_inline unsigned int mysqlnd_mbcharlen_null(const unsigned int c)
544 {
545 ZEND_UNREACHABLE();
546 return 0;
547 }
548
check_null(const char * const start,const char * const end)549 static zend_always_inline unsigned int check_null(const char * const start, const char * const end)
550 {
551 ZEND_UNREACHABLE();
552 return 0;
553 }
554
mysqlnd_mbcharlen_dispatch(enum mysqlnd_encoding_charlen encoding,const unsigned int c)555 static unsigned int mysqlnd_mbcharlen_dispatch(enum mysqlnd_encoding_charlen encoding, const unsigned int c)
556 {
557 switch (encoding) {
558 #define ENUMERATOR_DISPATCH(x) case x##_id: return x(c);
559 ENUMERATE_ENCODINGS_CHARLEN(ENUMERATOR_DISPATCH)
560 #undef ENUMERATOR_DISPATCH
561 default: return mysqlnd_mbcharlen_null(c);
562 }
563 }
564
mysqlnd_mbvalid_dispatch(enum mysqlnd_encoding_valid encoding,const char * const start,const char * const end)565 static unsigned int mysqlnd_mbvalid_dispatch(enum mysqlnd_encoding_valid encoding, const char * const start, const char * const end)
566 {
567 switch (encoding) {
568 #define ENUMERATOR_DISPATCH(x) case x##_id: return x(start, end);
569 ENUMERATE_ENCODINGS_VALID(ENUMERATOR_DISPATCH)
570 #undef ENUMERATOR_DISPATCH
571 default: return check_null(start, end);
572 }
573 }
574
575 /* {{{ mysqlnd_charsets */
576 const MYSQLND_CHARSET mysqlnd_charsets[] =
577 {
578 { 1, "big5","big5_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_big5_id, check_mb_big5_id, LOWEST_MB_BIG5},
579 { 3, "dec8", "dec8_swedish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
580 { 4, "cp850", "cp850_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
581 { 6, "hp8", "hp8_english_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
582 { 7, "koi8r", "koi8r_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
583 { 8, "latin1", "latin1_swedish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
584 { 5, "latin1", "latin1_german1_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100}, /* should be after 0x8 because swedish_ci is the default collation */
585 { 9, "latin2", "latin2_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
586 { 2, "latin2", "latin2_czech_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100}, /* should be after 0x9 because general_ci is the default collation */
587 { 10, "swe7", "swe7_swedish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
588 { 11, "ascii", "ascii_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
589 { 12, "ujis", "ujis_japanese_ci", 1, 3, "", mysqlnd_mbcharlen_ujis_id, check_mb_ujis_id, LOWEST_MB_UJIS},
590 { 13, "sjis", "sjis_japanese_ci", 1, 2, "", mysqlnd_mbcharlen_sjis_id, check_mb_sjis_id, LOWEST_MB_SJIS},
591 { 16, "hebrew", "hebrew_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
592 { 17, "filename", "filename", 1, 5, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
593 { 18, "tis620", "tis620_thai_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
594 { 19, "euckr", "euckr_korean_ci", 1, 2, "", mysqlnd_mbcharlen_euckr_id, check_mb_euckr_id, LOWEST_MB_EUCKR},
595 { 21, "latin2", "latin2_hungarian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
596 { 27, "latin2", "latin2_croatian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
597 { 22, "koi8u", "koi8u_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
598 { 24, "gb2312", "gb2312_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_gb2312_id, check_mb_gb2312_id, LOWEST_MB_GB2312},
599 { 25, "greek", "greek_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
600 { 26, "cp1250", "cp1250_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
601 { 28, "gbk", "gbk_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_gbk_id, check_mb_gbk_id, LOWEST_MB_GBK},
602 { 30, "latin5", "latin5_turkish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
603 { 31, "latin1", "latin1_german2_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
604 { 15, "latin1", "latin1_danish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
605 { 32, "armscii8", "armscii8_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
606 { 33, UTF8_MB3, UTF8_MB3"_general_ci", 1, 3, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
607 { 35, "ucs2", "ucs2_general_ci", 2, 2, "UCS-2 Unicode", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
608 { 36, "cp866", "cp866_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
609 { 37, "keybcs2", "keybcs2_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
610 { 38, "macce", "macce_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
611 { 39, "macroman", "macroman_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
612 { 40, "cp852", "cp852_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
613 { 41, "latin7", "latin7_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
614 { 20, "latin7", "latin7_estonian_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
615 { 57, "cp1256", "cp1256_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
616 { 59, "cp1257", "cp1257_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
617 { 63, "binary", "binary", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
618 { 97, "eucjpms", "eucjpms_japanese_ci", 1, 3, "", mysqlnd_mbcharlen_eucjpms_id, check_mb_eucjpms_id, LOWEST_MB_EUCJPMS},
619 { 29, "cp1257", "cp1257_lithuanian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
620 { 31, "latin1", "latin1_german2_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
621 { 34, "cp1250", "cp1250_czech_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
622 { 42, "latin7", "latin7_general_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
623 { 43, "macce", "macce_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
624 { 44, "cp1250", "cp1250_croatian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
625 { 45, UTF8_MB4, UTF8_MB4"_general_ci", 1, 4, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
626 { 46, UTF8_MB4, UTF8_MB4"_bin", 1, 4, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
627 { 47, "latin1", "latin1_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
628 { 48, "latin1", "latin1_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
629 { 49, "latin1", "latin1_general_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
630 { 51, "cp1251", "cp1251_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
631 { 14, "cp1251", "cp1251_bulgarian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
632 { 23, "cp1251", "cp1251_ukrainian_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
633 { 50, "cp1251", "cp1251_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
634 { 52, "cp1251", "cp1251_general_cs", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
635 { 53, "macroman", "macroman_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
636 { 54, "utf16", "utf16_general_ci", 2, 4, "UTF-16 Unicode", mysqlnd_mbcharlen_utf16_id, check_mb_utf16_id, LOWEST_MB_UTF16},
637 { 55, "utf16", "utf16_bin", 2, 4, "UTF-16 Unicode", mysqlnd_mbcharlen_utf16_id, check_mb_utf16_id, LOWEST_MB_UTF16},
638 { 56, "utf16le", "utf16le_general_ci", 2, 4, "UTF-16LE Unicode", mysqlnd_mbcharlen_utf16_id, check_mb_utf16_id, LOWEST_MB_UTF16},
639 { 58, "cp1257", "cp1257_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
640 /*55*/{ 60, "utf32", "utf32_general_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
641 /*55*/{ 61, "utf32", "utf32_bin", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
642 { 62, "utf16le", "utf16le_bin", 2, 4, "UTF-16LE Unicode", mysqlnd_mbcharlen_utf16_id, check_mb_utf16_id, LOWEST_MB_UTF16},
643 { 64, "armscii8", "armscii8_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
644 { 65, "ascii", "ascii_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
645 { 66, "cp1250", "cp1250_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
646 { 67, "cp1256", "cp1256_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
647 { 68, "cp866", "cp866_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
648 { 69, "dec8", "dec8_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
649 { 70, "greek", "greek_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
650 { 71, "hebrew", "hebrew_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
651 { 72, "hp8", "hp8_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
652 { 73, "keybcs2", "keybcs2_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
653 { 74, "koi8r", "koi8r_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
654 { 75, "koi8u", "koi8u_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
655 { 77, "latin2", "latin2_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
656 { 78, "latin5", "latin5_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
657 { 79, "latin7", "latin7_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
658 { 80, "cp850", "cp850_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
659 { 81, "cp852", "cp852_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
660 { 82, "swe7", "swe7_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
661 { 83, UTF8_MB3, UTF8_MB3"_bin", 1, 3, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
662 { 84, "big5", "big5_bin", 1, 2, "", mysqlnd_mbcharlen_big5_id, check_mb_big5_id, LOWEST_MB_BIG5},
663 { 85, "euckr", "euckr_bin", 1, 2, "", mysqlnd_mbcharlen_euckr_id, check_mb_euckr_id, LOWEST_MB_EUCKR},
664 { 86, "gb2312", "gb2312_bin", 1, 2, "", mysqlnd_mbcharlen_gb2312_id, check_mb_gb2312_id, LOWEST_MB_GB2312},
665 { 87, "gbk", "gbk_bin", 1, 2, "", mysqlnd_mbcharlen_gbk_id, check_mb_gbk_id, LOWEST_MB_GBK},
666 { 88, "sjis", "sjis_bin", 1, 2, "", mysqlnd_mbcharlen_sjis_id, check_mb_sjis_id, LOWEST_MB_SJIS},
667 { 89, "tis620", "tis620_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
668 { 90, "ucs2", "ucs2_bin", 2, 2, "UCS-2 Unicode", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
669 { 91, "ujis", "ujis_bin", 1, 3, "", mysqlnd_mbcharlen_ujis_id, check_mb_ujis_id, LOWEST_MB_UJIS},
670 { 92, "geostd8", "geostd8_general_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
671 { 93, "geostd8", "geostd8_bin", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
672 { 94, "latin1", "latin1_spanish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
673 { 95, "cp932", "cp932_japanese_ci", 1, 2, "", mysqlnd_mbcharlen_cp932_id, check_mb_cp932_id, LOWEST_MB_CP932},
674 { 96, "cp932", "cp932_bin", 1, 2, "", mysqlnd_mbcharlen_cp932_id, check_mb_cp932_id, LOWEST_MB_CP932},
675 { 97, "eucjpms", "eucjpms_japanese_ci", 1, 3, "", mysqlnd_mbcharlen_eucjpms_id, check_mb_eucjpms_id, LOWEST_MB_EUCJPMS},
676 { 98, "eucjpms", "eucjpms_bin", 1, 3, "", mysqlnd_mbcharlen_eucjpms_id, check_mb_eucjpms_id, LOWEST_MB_EUCJPMS},
677 { 99, "cp1250", "cp1250_polish_ci", 1, 1, "", mysqlnd_mbcharlen_null_id, check_null_id, 0x100},
678 { 128, "ucs2", "ucs2_unicode_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
679 { 129, "ucs2", "ucs2_icelandic_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
680 { 130, "ucs2", "ucs2_latvian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
681 { 131, "ucs2", "ucs2_romanian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
682 { 132, "ucs2", "ucs2_slovenian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
683 { 133, "ucs2", "ucs2_polish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
684 { 134, "ucs2", "ucs2_estonian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
685 { 135, "ucs2", "ucs2_spanish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
686 { 136, "ucs2", "ucs2_swedish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
687 { 137, "ucs2", "ucs2_turkish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
688 { 138, "ucs2", "ucs2_czech_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
689 { 139, "ucs2", "ucs2_danish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
690 { 140, "ucs2", "ucs2_lithuanian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
691 { 141, "ucs2", "ucs2_slovak_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
692 { 142, "ucs2", "ucs2_spanish2_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
693 { 143, "ucs2", "ucs2_roman_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
694 { 144, "ucs2", "ucs2_persian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
695 { 145, "ucs2", "ucs2_esperanto_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
696 { 146, "ucs2", "ucs2_hungarian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
697 { 147, "ucs2", "ucs2_sinhala_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
698 { 148, "ucs2", "ucs2_german2_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
699 { 149, "ucs2", "ucs2_croatian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
700 { 150, "ucs2", "ucs2_unicode_520_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
701 { 151, "ucs2", "ucs2_vietnamese_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2_id, check_mb_ucs2_id, LOWEST_MB_UCS2},
702
703 /*56*/{160, "utf32", "utf32_unicode_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
704 /*56*/{161, "utf32", "utf32_icelandic_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
705 /*56*/{162, "utf32", "utf32_latvian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
706 /*56*/{163, "utf32", "utf32_romanian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
707 /*56*/{164, "utf32", "utf32_slovenian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
708 /*56*/{165, "utf32", "utf32_polish_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
709 /*56*/{166, "utf32", "utf32_estonian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
710 /*56*/{167, "utf32", "utf32_spanish_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
711 /*56*/{168, "utf32", "utf32_swedish_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
712 /*56*/{169, "utf32", "utf32_turkish_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
713 /*56*/{170, "utf32", "utf32_czech_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
714 /*56*/{171, "utf32", "utf32_danish_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
715 /*56*/{172, "utf32", "utf32_lithuanian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
716 /*56*/{173, "utf32", "utf32_slovak_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
717 /*56*/{174, "utf32", "utf32_spanish2_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
718 /*56*/{175, "utf32", "utf32_roman_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
719 /*56*/{176, "utf32", "utf32_persian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
720 /*56*/{177, "utf32", "utf32_esperanto_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
721 /*56*/{178, "utf32", "utf32_hungarian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
722 /*56*/{179, "utf32", "utf32_sinhala_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
723 /*56*/{180, "utf32", "utf32_german2_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
724 /*56*/{181, "utf32", "utf32_croatian_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
725 /*56*/{182, "utf32", "utf32_unicode_520_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
726 /*56*/{183, "utf32", "utf32_vietnamese_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32_id, check_mb_utf32_id, LOWEST_MB_UTF32},
727
728 { 192, UTF8_MB3, UTF8_MB3"_unicode_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
729 { 193, UTF8_MB3, UTF8_MB3"_icelandic_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
730 { 194, UTF8_MB3, UTF8_MB3"_latvian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
731 { 195, UTF8_MB3, UTF8_MB3"_romanian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
732 { 196, UTF8_MB3, UTF8_MB3"_slovenian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
733 { 197, UTF8_MB3, UTF8_MB3"_polish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
734 { 198, UTF8_MB3, UTF8_MB3"_estonian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
735 { 199, UTF8_MB3, UTF8_MB3"_spanish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
736 { 200, UTF8_MB3, UTF8_MB3"_swedish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
737 { 201, UTF8_MB3, UTF8_MB3"_turkish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
738 { 202, UTF8_MB3, UTF8_MB3"_czech_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
739 { 203, UTF8_MB3, UTF8_MB3"_danish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
740 { 204, UTF8_MB3, UTF8_MB3"_lithuanian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
741 { 205, UTF8_MB3, UTF8_MB3"_slovak_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
742 { 206, UTF8_MB3, UTF8_MB3"_spanish2_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
743 { 207, UTF8_MB3, UTF8_MB3"_roman_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
744 { 208, UTF8_MB3, UTF8_MB3"_persian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
745 { 209, UTF8_MB3, UTF8_MB3"_esperanto_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
746 { 210, UTF8_MB3, UTF8_MB3"_hungarian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
747 { 211, UTF8_MB3, UTF8_MB3"_sinhala_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
748 { 212, UTF8_MB3, UTF8_MB3"_german2_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
749 { 213, UTF8_MB3, UTF8_MB3"_croatian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
750 { 214, UTF8_MB3, UTF8_MB3"_unicode_520_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
751 { 215, UTF8_MB3, UTF8_MB3"_vietnamese_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3_id, check_mb_utf8mb3_valid_id, LOWEST_MB_UTF8MB3},
752
753 { 224, UTF8_MB4, UTF8_MB4"_unicode_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
754 { 225, UTF8_MB4, UTF8_MB4"_icelandic_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
755 { 226, UTF8_MB4, UTF8_MB4"_latvian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
756 { 227, UTF8_MB4, UTF8_MB4"_romanian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
757 { 228, UTF8_MB4, UTF8_MB4"_slovenian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
758 { 229, UTF8_MB4, UTF8_MB4"_polish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
759 { 230, UTF8_MB4, UTF8_MB4"_estonian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
760 { 231, UTF8_MB4, UTF8_MB4"_spanish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
761 { 232, UTF8_MB4, UTF8_MB4"_swedish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
762 { 233, UTF8_MB4, UTF8_MB4"_turkish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
763 { 234, UTF8_MB4, UTF8_MB4"_czech_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
764 { 235, UTF8_MB4, UTF8_MB4"_danish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
765 { 236, UTF8_MB4, UTF8_MB4"_lithuanian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
766 { 237, UTF8_MB4, UTF8_MB4"_slovak_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
767 { 238, UTF8_MB4, UTF8_MB4"_spanish2_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
768 { 239, UTF8_MB4, UTF8_MB4"_roman_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
769 { 240, UTF8_MB4, UTF8_MB4"_persian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
770 { 241, UTF8_MB4, UTF8_MB4"_esperanto_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
771 { 242, UTF8_MB4, UTF8_MB4"_hungarian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
772 { 243, UTF8_MB4, UTF8_MB4"_sinhala_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
773 { 244, UTF8_MB4, UTF8_MB4"_german2_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
774 { 245, UTF8_MB4, UTF8_MB4"_croatian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
775 { 246, UTF8_MB4, UTF8_MB4"_unicode_520_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
776 { 247, UTF8_MB4, UTF8_MB4"_vietnamese_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
777 { 248, "gb18030", "gb18030_chinese_ci", 1, 4, "", mysqlnd_mbcharlen_gb18030_id, my_ismbchar_gb18030_id, LOWEST_MB_GB18030},
778 { 249, "gb18030", "gb18030_bin", 1, 4, "", mysqlnd_mbcharlen_gb18030_id, my_ismbchar_gb18030_id, LOWEST_MB_GB18030},
779
780 { 254, UTF8_MB3, UTF8_MB3"_general_cs", 1, 3, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
781 { 255, UTF8_MB4, UTF8_MB4"_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
782 { 256, UTF8_MB4, UTF8_MB4"_de_pb_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
783 { 257, UTF8_MB4, UTF8_MB4"_is_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
784 { 258, UTF8_MB4, UTF8_MB4"_lv_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
785 { 259, UTF8_MB4, UTF8_MB4"_ro_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
786 { 260, UTF8_MB4, UTF8_MB4"_sl_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
787 { 261, UTF8_MB4, UTF8_MB4"_pl_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
788 { 262, UTF8_MB4, UTF8_MB4"_et_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
789 { 263, UTF8_MB4, UTF8_MB4"_es_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
790 { 264, UTF8_MB4, UTF8_MB4"_sv_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
791 { 265, UTF8_MB4, UTF8_MB4"_tr_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
792 { 266, UTF8_MB4, UTF8_MB4"_cs_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
793 { 267, UTF8_MB4, UTF8_MB4"_da_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
794 { 268, UTF8_MB4, UTF8_MB4"_lt_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
795 { 269, UTF8_MB4, UTF8_MB4"_sk_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
796 { 270, UTF8_MB4, UTF8_MB4"_es_trad_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
797 { 271, UTF8_MB4, UTF8_MB4"_la_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
798 { 272, UTF8_MB4, UTF8_MB4"_fa_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
799 { 273, UTF8_MB4, UTF8_MB4"_eo_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
800 { 274, UTF8_MB4, UTF8_MB4"_hu_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
801 { 275, UTF8_MB4, UTF8_MB4"_hr_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
802 { 276, UTF8_MB4, UTF8_MB4"_si_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
803 { 277, UTF8_MB4, UTF8_MB4"_vi_0900_ai_ci", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
804 { 278, UTF8_MB4, UTF8_MB4"_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
805 { 279, UTF8_MB4, UTF8_MB4"_de_pb_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
806 { 280, UTF8_MB4, UTF8_MB4"_is_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
807 { 281, UTF8_MB4, UTF8_MB4"_lv_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
808 { 282, UTF8_MB4, UTF8_MB4"_ro_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
809 { 283, UTF8_MB4, UTF8_MB4"_sl_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
810 { 284, UTF8_MB4, UTF8_MB4"_pl_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
811 { 285, UTF8_MB4, UTF8_MB4"_et_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
812 { 286, UTF8_MB4, UTF8_MB4"_es_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
813 { 287, UTF8_MB4, UTF8_MB4"_sv_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
814 { 288, UTF8_MB4, UTF8_MB4"_tr_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
815 { 289, UTF8_MB4, UTF8_MB4"_cs_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
816 { 290, UTF8_MB4, UTF8_MB4"_da_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
817 { 291, UTF8_MB4, UTF8_MB4"_lt_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
818 { 292, UTF8_MB4, UTF8_MB4"_sk_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
819 { 293, UTF8_MB4, UTF8_MB4"_es_trad_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
820 { 294, UTF8_MB4, UTF8_MB4"_la_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
821 { 295, UTF8_MB4, UTF8_MB4"_fa_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
822 { 296, UTF8_MB4, UTF8_MB4"_eo_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
823 { 297, UTF8_MB4, UTF8_MB4"_hu_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
824 { 298, UTF8_MB4, UTF8_MB4"_hr_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
825 { 299, UTF8_MB4, UTF8_MB4"_si_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
826 { 300, UTF8_MB4, UTF8_MB4"_vi_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
827 { 303, UTF8_MB4, UTF8_MB4"_ja_0900_as_cs", 1, 4, "", mysqlnd_mbcharlen_utf8_id, check_mb_utf8_valid_id, LOWEST_MB_UTF8},
828 { 0, NULL, NULL, 0, 0, NULL, mysqlnd_mbcharlen_null_id, check_null_id, 0x100}
829 };
830 /* }}} */
831
832
833 /* {{{ mysqlnd_find_charset_nr */
mysqlnd_find_charset_nr(const unsigned int charsetnr)834 PHPAPI const MYSQLND_CHARSET * mysqlnd_find_charset_nr(const unsigned int charsetnr)
835 {
836 const MYSQLND_CHARSET * c = mysqlnd_charsets;
837
838 do {
839 if (c->nr == charsetnr) {
840 return c;
841 }
842 ++c;
843 } while (c[0].nr != 0);
844 return NULL;
845 }
846 /* }}} */
847
848
849 /* {{{ mysqlnd_find_charset_name */
mysqlnd_find_charset_name(const char * const name)850 PHPAPI const MYSQLND_CHARSET * mysqlnd_find_charset_name(const char * const name)
851 {
852 if (name) {
853 const MYSQLND_CHARSET * c = mysqlnd_charsets;
854 do {
855 if (!strcasecmp(c->name, name)) {
856 return c;
857 }
858 ++c;
859 } while (c[0].nr != 0);
860 }
861 return NULL;
862 }
863 /* }}} */
864
865 /* {{{ mysqlnd_cset_escape_quotes */
mysqlnd_cset_escape_quotes(const MYSQLND_CHARSET * const cset,char * newstr,const char * escapestr,const size_t escapestr_len)866 PHPAPI zend_ulong mysqlnd_cset_escape_quotes(const MYSQLND_CHARSET * const cset, char * newstr,
867 const char * escapestr, const size_t escapestr_len)
868 {
869 const char *newstr_s = newstr;
870 const char *newstr_e = newstr + 2 * escapestr_len;
871 const char *end = escapestr + escapestr_len;
872
873 DBG_ENTER("mysqlnd_cset_escape_quotes");
874
875 for (;escapestr < end; escapestr++) {
876 unsigned int len = 0;
877 /* check unicode characters */
878
879 if (*((zend_uchar *) escapestr) >= cset->lowest_mb_byte && (len = mysqlnd_mbvalid_dispatch(cset->mb_valid, escapestr, end))) {
880 ZEND_ASSERT(newstr + len <= newstr_e);
881 /* copy mb char without escaping it */
882 while (len--) {
883 *newstr++ = *escapestr++;
884 }
885 escapestr--;
886 continue;
887 }
888 if (*escapestr == '\'') {
889 ZEND_ASSERT(newstr + 2 <= newstr_e);
890 *newstr++ = '\'';
891 *newstr++ = '\'';
892 } else {
893 ZEND_ASSERT(newstr + 1 <= newstr_e);
894 *newstr++ = *escapestr;
895 }
896 }
897 *newstr = '\0';
898
899 DBG_RETURN((zend_ulong)(newstr - newstr_s));
900 }
901 /* }}} */
902
903
904 /* {{{ mysqlnd_cset_escape_slashes */
mysqlnd_cset_escape_slashes(const MYSQLND_CHARSET * const cset,char * newstr,const char * escapestr,const size_t escapestr_len)905 PHPAPI zend_ulong mysqlnd_cset_escape_slashes(const MYSQLND_CHARSET * const cset, char *newstr,
906 const char * escapestr, const size_t escapestr_len)
907 {
908 const char *newstr_s = newstr;
909 const char *newstr_e = newstr + 2 * escapestr_len;
910 const char *end = escapestr + escapestr_len;
911
912 DBG_ENTER("mysqlnd_cset_escape_slashes");
913 DBG_INF_FMT("charset=%s", cset->name);
914
915 for (;escapestr < end; escapestr++) {
916 char esc = '\0';
917
918 /* check unicode characters */
919 if (*((zend_uchar *) escapestr) >= cset->lowest_mb_byte) {
920 unsigned int len = mysqlnd_mbvalid_dispatch(cset->mb_valid, escapestr, end);
921 if (len) {
922 ZEND_ASSERT(newstr + len <= newstr_e);
923 /* copy mb char without escaping it */
924 while (len--) {
925 *newstr++ = *escapestr++;
926 }
927 escapestr--;
928 continue;
929 } else if (mysqlnd_mbcharlen_dispatch(cset->mb_charlen, *escapestr) > 1) {
930 esc = *escapestr;
931 }
932 }
933 if (!esc) {
934 switch (*escapestr) {
935 case 0:
936 esc = '0';
937 break;
938 case '\n':
939 esc = 'n';
940 break;
941 case '\r':
942 esc = 'r';
943 break;
944 case '\\':
945 case '\'':
946 case '"':
947 esc = *escapestr;
948 break;
949 case '\032':
950 esc = 'Z';
951 break;
952 }
953 }
954 if (esc) {
955 ZEND_ASSERT(newstr + 2 <= newstr_e);
956 /* copy escaped character */
957 *newstr++ = '\\';
958 *newstr++ = esc;
959 } else {
960 ZEND_ASSERT(newstr + 1 <= newstr_e);
961 /* copy non escaped character */
962 *newstr++ = *escapestr;
963 }
964 }
965 *newstr = '\0';
966
967 DBG_RETURN((zend_ulong)(newstr - newstr_s));
968 }
969 /* }}} */
970
971
972 static struct st_mysqlnd_plugin_charsets mysqlnd_plugin_charsets_plugin =
973 {
974 {
975 MYSQLND_PLUGIN_API_VERSION,
976 "charsets",
977 MYSQLND_VERSION_ID,
978 PHP_MYSQLND_VERSION,
979 "PHP License 3.01",
980 "Andrey Hristov <andrey@php.net>, Ulf Wendel <uw@php.net>, Georg Richter <georg@php.net>",
981 {
982 NULL, /* no statistics , will be filled later if there are some */
983 NULL, /* no statistics */
984 },
985 {
986 NULL /* plugin shutdown */
987 }
988 },
989 {/* methods */
990 mysqlnd_find_charset_nr,
991 mysqlnd_find_charset_name,
992 mysqlnd_cset_escape_quotes,
993 mysqlnd_cset_escape_slashes
994 }
995 };
996
997
998 /* {{{ mysqlnd_charsets_plugin_register */
999 void
mysqlnd_charsets_plugin_register(void)1000 mysqlnd_charsets_plugin_register(void)
1001 {
1002 mysqlnd_plugin_register_ex((struct st_mysqlnd_plugin_header *) &mysqlnd_plugin_charsets_plugin);
1003 }
1004 /* }}} */
1005
1006 #if MYSQLND_CHARSETS_SANITY_CHECK
mysqlnd_charsets_sanity_check(void)1007 void mysqlnd_charsets_sanity_check(void)
1008 {
1009 /* Sanity check fast path checks for encoding handling. */
1010 for (size_t i = 0; i < sizeof(mysqlnd_charsets) / sizeof(*mysqlnd_charsets); i++) {
1011 const MYSQLND_CHARSET *c = &mysqlnd_charsets[i];
1012 printf("charset nr %u\n", c->nr);
1013
1014 /* Charset configuration consistency. */
1015 if (c->mb_valid != check_null_id) {
1016 ZEND_ASSERT(c->mb_valid != mysqlnd_mbcharlen_null_id);
1017 }
1018 if (c->mb_charlen != mysqlnd_mbcharlen_null_id) {
1019 ZEND_ASSERT(c->mb_charlen != check_null_id);
1020 }
1021 ZEND_ASSERT(c->char_minlen <= c->char_maxlen);
1022 if (c->char_maxlen == 1) {
1023 ZEND_ASSERT(c->lowest_mb_byte == 0x100);
1024 }
1025 if (c->char_minlen > 1) {
1026 ZEND_ASSERT(c->mb_valid != check_null_id);
1027 ZEND_ASSERT(c->mb_charlen != mysqlnd_mbcharlen_null_id);
1028 ZEND_ASSERT(c->lowest_mb_byte == 0x00);
1029 } else if (c->mb_valid != check_null_id) {
1030 /* Charset behavioural consistency w.r.t. charlen. */
1031 for (unsigned int ch = 0; ch < 255; ch++) {
1032 if (mysqlnd_mbcharlen_dispatch(c->mb_charlen, ch) > 1) {
1033 ZEND_ASSERT(ch >= c->lowest_mb_byte);
1034 }
1035 }
1036
1037 /* Charset behavioural consistency w.r.t. validity. */
1038 union {
1039 zend_uchar c[4];
1040 uint32_t l;
1041 } input;
1042 for (uint32_t i = 0; i < 0xFFFFFFFF; i++) {
1043 input.l = i;
1044 if (mysqlnd_mbvalid_dispatch(c->mb_valid, (char *)input.c, (char *)input.c + 4) > 1) {
1045 if (input.c[0] < c->lowest_mb_byte) {
1046 printf("Input %u (%u %u)\n", input.l, input.c[0], c->lowest_mb_byte);
1047 ZEND_ASSERT(false);
1048 }
1049 }
1050 }
1051 }
1052 }
1053 }
1054 #endif
1055
