Annotation of embedaddon/php/ext/mysqlnd/mysqlnd_charset.c, revision 1.1.1.2
1.1 misho 1: /*
2: +----------------------------------------------------------------------+
3: | PHP Version 5 |
4: +----------------------------------------------------------------------+
5: | Copyright (c) 2006-2012 The PHP Group |
6: +----------------------------------------------------------------------+
7: | This source file is subject to version 3.01 of the PHP license, |
8: | that is bundled with this package in the file LICENSE, and is |
9: | available through the world-wide-web at the following url: |
10: | http://www.php.net/license/3_01.txt |
11: | If you did not receive a copy of the PHP license and are unable to |
12: | obtain it through the world-wide-web, please send a note to |
13: | license@php.net so we can mail you a copy immediately. |
14: +----------------------------------------------------------------------+
15: | Authors: Georg Richter <georg@mysql.com> |
16: | Andrey Hristov <andrey@mysql.com> |
17: | Ulf Wendel <uwendel@mysql.com> |
18: +----------------------------------------------------------------------+
19: */
20: #include "php.h"
21: #include "php_globals.h"
22: #include "mysqlnd.h"
23: #include "mysqlnd_priv.h"
24: #include "mysqlnd_debug.h"
1.1.1.2 ! misho 25: #include "mysqlnd_charset.h"
1.1 misho 26:
27: /* {{{ utf8 functions */
28: static unsigned int check_mb_utf8mb3_sequence(const char *start, const char *end)
29: {
30: zend_uchar c;
31:
32: if (start >= end) {
33: return 0;
34: }
35:
36: c = (zend_uchar) start[0];
37:
38: if (c < 0x80) {
39: return 1; /* single byte character */
40: }
41: if (c < 0xC2) {
42: return 0; /* invalid mb character */
43: }
44: if (c < 0xE0) {
45: if (start + 2 > end) {
46: return 0; /* too small */
47: }
48: if (!(((zend_uchar)start[1] ^ 0x80) < 0x40)) {
49: return 0;
50: }
51: return 2;
52: }
53: if (c < 0xF0) {
54: if (start + 3 > end) {
55: return 0; /* too small */
56: }
57: if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 && ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
58: (c >= 0xE1 || (zend_uchar)start[1] >= 0xA0))) {
59: return 0; /* invalid utf8 character */
60: }
61: return 3;
62: }
63: return 0;
64: }
65:
66:
67: static unsigned int check_mb_utf8_sequence(const char *start, const char *end)
68: {
69: zend_uchar c;
70:
71: if (start >= end) {
72: return 0;
73: }
74:
75: c = (zend_uchar) start[0];
76:
77: if (c < 0x80) {
78: return 1; /* single byte character */
79: }
80: if (c < 0xC2) {
81: return 0; /* invalid mb character */
82: }
83: if (c < 0xE0) {
84: if (start + 2 > end) {
85: return 0; /* too small */
86: }
87: if (!(((zend_uchar)start[1] ^ 0x80) < 0x40)) {
88: return 0;
89: }
90: return 2;
91: }
92: if (c < 0xF0) {
93: if (start + 3 > end) {
94: return 0; /* too small */
95: }
96: if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 && ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
97: (c >= 0xE1 || (zend_uchar)start[1] >= 0xA0))) {
98: return 0; /* invalid utf8 character */
99: }
100: return 3;
101: }
102: if (c < 0xF5) {
103: if (start + 4 > end) { /* We need 4 characters */
104: return 0; /* too small */
105: }
106:
107: /*
108: UTF-8 quick four-byte mask:
109: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
110: Encoding allows to encode U+00010000..U+001FFFFF
111:
112: The maximum character defined in the Unicode standard is U+0010FFFF.
113: Higher characters U+00110000..U+001FFFFF are not used.
114:
115: 11110000.10010000.10xxxxxx.10xxxxxx == F0.90.80.80 == U+00010000 (min)
116: 11110100.10001111.10111111.10111111 == F4.8F.BF.BF == U+0010FFFF (max)
117:
118: Valid codes:
119: [F0][90..BF][80..BF][80..BF]
120: [F1][80..BF][80..BF][80..BF]
121: [F2][80..BF][80..BF][80..BF]
122: [F3][80..BF][80..BF][80..BF]
123: [F4][80..8F][80..BF][80..BF]
124: */
125:
126: if (!(((zend_uchar)start[1] ^ 0x80) < 0x40 &&
127: ((zend_uchar)start[2] ^ 0x80) < 0x40 &&
128: ((zend_uchar)start[3] ^ 0x80) < 0x40 &&
129: (c >= 0xf1 || (zend_uchar)start[1] >= 0x90) &&
130: (c <= 0xf3 || (zend_uchar)start[1] <= 0x8F)))
131: {
132: return 0; /* invalid utf8 character */
133: }
134: return 4;
135: }
136: return 0;
137: }
138:
139: static unsigned int check_mb_utf8mb3_valid(const char *start, const char *end)
140: {
141: unsigned int len = check_mb_utf8mb3_sequence(start, end);
142: return (len > 1)? len:0;
143: }
144:
145: static unsigned int check_mb_utf8_valid(const char *start, const char *end)
146: {
147: unsigned int len = check_mb_utf8_sequence(start, end);
148: return (len > 1)? len:0;
149: }
150:
151:
152: static unsigned int mysqlnd_mbcharlen_utf8mb3(unsigned int utf8)
153: {
154: if (utf8 < 0x80) {
155: return 1; /* single byte character */
156: }
157: if (utf8 < 0xC2) {
158: return 0; /* invalid multibyte header */
159: }
160: if (utf8 < 0xE0) {
161: return 2; /* double byte character */
162: }
163: if (utf8 < 0xF0) {
164: return 3; /* triple byte character */
165: }
166: return 0;
167: }
168:
169:
170: static unsigned int mysqlnd_mbcharlen_utf8(unsigned int utf8)
171: {
172: if (utf8 < 0x80) {
173: return 1; /* single byte character */
174: }
175: if (utf8 < 0xC2) {
176: return 0; /* invalid multibyte header */
177: }
178: if (utf8 < 0xE0) {
179: return 2; /* double byte character */
180: }
181: if (utf8 < 0xF0) {
182: return 3; /* triple byte character */
183: }
184: if (utf8 < 0xF8) {
185: return 4; /* four byte character */
186: }
187: return 0;
188: }
189: /* }}} */
190:
191:
192: /* {{{ big5 functions */
193: #define valid_big5head(c) (0xA1 <= (unsigned int)(c) && (unsigned int)(c) <= 0xF9)
194: #define valid_big5tail(c) ((0x40 <= (unsigned int)(c) && (unsigned int)(c) <= 0x7E) || \
195: (0xA1 <= (unsigned int)(c) && (unsigned int)(c) <= 0xFE))
196:
197: #define isbig5code(c,d) (isbig5head(c) && isbig5tail(d))
198:
199: static unsigned int check_mb_big5(const char *start, const char *end)
200: {
201: return (valid_big5head(*(start)) && (end - start) > 1 && valid_big5tail(*(start + 1)) ? 2 : 0);
202: }
203:
204:
205: static unsigned int mysqlnd_mbcharlen_big5(unsigned int big5)
206: {
207: return (valid_big5head(big5)) ? 2 : 1;
208: }
209: /* }}} */
210:
211:
212: /* {{{ cp932 functions */
213: #define valid_cp932head(c) ((0x81 <= (c) && (c) <= 0x9F) || (0xE0 <= (c) && c <= 0xFC))
214: #define valid_cp932tail(c) ((0x40 <= (c) && (c) <= 0x7E) || (0x80 <= (c) && c <= 0xFC))
215:
216:
217: static unsigned int check_mb_cp932(const char *start, const char *end)
218: {
219: return (valid_cp932head((zend_uchar)start[0]) && (end - start > 1) &&
220: valid_cp932tail((zend_uchar)start[1])) ? 2 : 0;
221: }
222:
223:
224: static unsigned int mysqlnd_mbcharlen_cp932(unsigned int cp932)
225: {
226: return (valid_cp932head((zend_uchar)cp932)) ? 2 : 1;
227: }
228: /* }}} */
229:
230:
231: /* {{{ euckr functions */
232: #define valid_euckr(c) ((0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xFE))
233:
234: static unsigned int check_mb_euckr(const char *start, const char *end)
235: {
236: if (end - start <= 1) {
237: return 0; /* invalid length */
238: }
239: if (*(zend_uchar *)start < 0x80) {
240: return 0; /* invalid euckr character */
241: }
242: if (valid_euckr(start[1])) {
243: return 2;
244: }
245: return 0;
246: }
247:
248:
249: static unsigned int mysqlnd_mbcharlen_euckr(unsigned int kr)
250: {
251: return (valid_euckr(kr)) ? 2 : 1;
252: }
253: /* }}} */
254:
255:
256: /* {{{ eucjpms functions */
257: #define valid_eucjpms(c) (((c) & 0xFF) >= 0xA1 && ((c) & 0xFF) <= 0xFE)
258: #define valid_eucjpms_kata(c) (((c) & 0xFF) >= 0xA1 && ((c) & 0xFF) <= 0xDF)
259: #define valid_eucjpms_ss2(c) (((c) & 0xFF) == 0x8E)
260: #define valid_eucjpms_ss3(c) (((c) & 0xFF) == 0x8F)
261:
262: static unsigned int check_mb_eucjpms(const char *start, const char *end)
263: {
264: if (*((zend_uchar *)start) < 0x80) {
265: return 0; /* invalid eucjpms character */
266: }
267: if (valid_eucjpms(start[0]) && (end - start) > 1 && valid_eucjpms(start[1])) {
268: return 2;
269: }
270: if (valid_eucjpms_ss2(start[0]) && (end - start) > 1 && valid_eucjpms_kata(start[1])) {
271: return 2;
272: }
273: if (valid_eucjpms_ss3(start[0]) && (end - start) > 2 && valid_eucjpms(start[1]) &&
274: valid_eucjpms(start[2])) {
275: return 2;
276: }
277: return 0;
278: }
279:
280:
281: static unsigned int mysqlnd_mbcharlen_eucjpms(unsigned int jpms)
282: {
283: if (valid_eucjpms(jpms) || valid_eucjpms_ss2(jpms)) {
284: return 2;
285: }
286: if (valid_eucjpms_ss3(jpms)) {
287: return 3;
288: }
289: return 1;
290: }
291: /* }}} */
292:
293:
294: /* {{{ gb2312 functions */
295: #define valid_gb2312_head(c) (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xF7)
296: #define valid_gb2312_tail(c) (0xA1 <= (zend_uchar)(c) && (zend_uchar)(c) <= 0xFE)
297:
298:
299: static unsigned int check_mb_gb2312(const char *start, const char *end)
300: {
301: return (valid_gb2312_head((unsigned int)start[0]) && end - start > 1 &&
302: valid_gb2312_tail((unsigned int)start[1])) ? 2 : 0;
303: }
304:
305:
306: static unsigned int mysqlnd_mbcharlen_gb2312(unsigned int gb)
307: {
308: return (valid_gb2312_head(gb)) ? 2 : 1;
309: }
310: /* }}} */
311:
312:
313: /* {{{ gbk functions */
314: #define valid_gbk_head(c) (0x81<=(zend_uchar)(c) && (zend_uchar)(c)<=0xFE)
315: #define valid_gbk_tail(c) ((0x40<=(zend_uchar)(c) && (zend_uchar)(c)<=0x7E) || (0x80<=(zend_uchar)(c) && (zend_uchar)(c)<=0xFE))
316:
317: static unsigned int check_mb_gbk(const char *start, const char *end)
318: {
319: return (valid_gbk_head(start[0]) && (end) - (start) > 1 && valid_gbk_tail(start[1])) ? 2 : 0;
320: }
321:
322: static unsigned int mysqlnd_mbcharlen_gbk(unsigned int gbk)
323: {
324: return (valid_gbk_head(gbk) ? 2 : 1);
325: }
326: /* }}} */
327:
328:
1.1.1.2 ! misho 329: /* {{{ sjis functions */
1.1 misho 330: #define valid_sjis_head(c) ((0x81 <= (c) && (c) <= 0x9F) || (0xE0 <= (c) && (c) <= 0xFC))
331: #define valid_sjis_tail(c) ((0x40 <= (c) && (c) <= 0x7E) || (0x80 <= (c) && (c) <= 0xFC))
332:
333:
334: static unsigned int check_mb_sjis(const char *start, const char *end)
335: {
336: return (valid_sjis_head((zend_uchar)start[0]) && (end - start) > 1 && valid_sjis_tail((zend_uchar)start[1])) ? 2 : 0;
337: }
338:
339:
340: static unsigned int mysqlnd_mbcharlen_sjis(unsigned int sjis)
341: {
342: return (valid_sjis_head((zend_uchar)sjis)) ? 2 : 1;
343: }
344: /* }}} */
345:
346:
347: /* {{{ ucs2 functions */
348: static unsigned int check_mb_ucs2(const char *start __attribute((unused)), const char *end __attribute((unused)))
349: {
350: return 2; /* always 2 */
351: }
352:
353: static unsigned int mysqlnd_mbcharlen_ucs2(unsigned int ucs2 __attribute((unused)))
354: {
355: return 2; /* always 2 */
356: }
357: /* }}} */
358:
359:
360: /* {{{ ujis functions */
361: #define valid_ujis(c) ((0xA1 <= ((c)&0xFF) && ((c)&0xFF) <= 0xFE))
362: #define valid_ujis_kata(c) ((0xA1 <= ((c)&0xFF) && ((c)&0xFF) <= 0xDF))
363: #define valid_ujis_ss2(c) (((c)&0xFF) == 0x8E)
364: #define valid_ujis_ss3(c) (((c)&0xFF) == 0x8F)
365:
366: static unsigned int check_mb_ujis(const char *start, const char *end)
367: {
368: if (*(zend_uchar*)start < 0x80) {
369: return 0; /* invalid ujis character */
370: }
371: if (valid_ujis(*(start)) && valid_ujis(*((start)+1))) {
372: return 2;
373: }
374: if (valid_ujis_ss2(*(start)) && valid_ujis_kata(*((start)+1))) {
375: return 2;
376: }
377: if (valid_ujis_ss3(*(start)) && (end-start) > 2 && valid_ujis(*((start)+1)) && valid_ujis(*((start)+2))) {
378: return 3;
379: }
380: return 0;
381: }
382:
383:
384: static unsigned int mysqlnd_mbcharlen_ujis(unsigned int ujis)
385: {
386: return (valid_ujis(ujis)? 2: valid_ujis_ss2(ujis)? 2: valid_ujis_ss3(ujis)? 3: 1);
387: }
388: /* }}} */
389:
390:
391:
392: /* {{{ utf16 functions */
393: #define UTF16_HIGH_HEAD(x) ((((zend_uchar) (x)) & 0xFC) == 0xD8)
394: #define UTF16_LOW_HEAD(x) ((((zend_uchar) (x)) & 0xFC) == 0xDC)
395:
396: static unsigned int check_mb_utf16(const char *start, const char *end)
397: {
398: if (start + 2 > end) {
399: return 0;
400: }
401:
402: if (UTF16_HIGH_HEAD(*start)) {
403: return (start + 4 <= end) && UTF16_LOW_HEAD(start[2]) ? 4 : 0;
404: }
405:
406: if (UTF16_LOW_HEAD(*start)) {
407: return 0;
408: }
409: return 2;
410: }
411:
412:
413: static uint mysqlnd_mbcharlen_utf16(unsigned int utf16)
414: {
415: return UTF16_HIGH_HEAD(utf16) ? 4 : 2;
416: }
417: /* }}} */
418:
419:
420: /* {{{ utf32 functions */
421: static uint
422: check_mb_utf32(const char *start __attribute((unused)), const char *end __attribute((unused)))
423: {
424: return 4;
425: }
426:
427:
428: static uint
429: mysqlnd_mbcharlen_utf32(unsigned int utf32 __attribute((unused)))
430: {
431: return 4;
432: }
433: /* }}} */
434:
435: /*
436: The server compiles sometimes the full utf-8 (the mb4) as utf8m4, and the old as utf8,
437: for BC reasons. Sometimes, utf8mb4 is just utf8 but the old charsets are utf8mb3.
438: Change easily now, with a macro, could be made compilastion dependable.
439: */
440:
441: #define UTF8_MB4 "utf8mb4"
442: #define UTF8_MB3 "utf8"
443:
444: /* {{{ mysqlnd_charsets */
445: const MYSQLND_CHARSET mysqlnd_charsets[] =
446: {
447: { 1, "big5","big5_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_big5, check_mb_big5},
448: { 3, "dec8", "dec8_swedisch_ci", 1, 1, "", NULL, NULL},
449: { 4, "cp850", "cp850_general_ci", 1, 1, "", NULL, NULL},
450: { 6, "hp8", "hp8_english_ci", 1, 1, "", NULL, NULL},
451: { 7, "koi8r", "koi8r_general_ci", 1, 1, "", NULL, NULL},
452: { 8, "latin1", "latin1_swedish_ci", 1, 1, "", NULL, NULL},
453: { 9, "latin2", "latin2_general_ci", 1, 1, "", NULL, NULL},
454: { 10, "swe7", "swe7_swedish_ci", 1, 1, "", NULL, NULL},
455: { 11, "ascii", "ascii_general_ci", 1, 1, "", NULL, NULL},
456: { 12, "ujis", "ujis_japanese_ci", 1, 3, "", mysqlnd_mbcharlen_ujis, check_mb_ujis},
457: { 13, "sjis", "sjis_japanese_ci", 1, 2, "", mysqlnd_mbcharlen_sjis, check_mb_sjis},
458: { 16, "hebrew", "hebrew_general_ci", 1, 1, "", NULL, NULL},
459: { 18, "tis620", "tis620_thai_ci", 1, 1, "", NULL, NULL},
460: { 19, "euckr", "euckr_korean_ci", 1, 2, "", mysqlnd_mbcharlen_euckr, check_mb_euckr},
461: { 22, "koi8u", "koi8u_general_ci", 1, 1, "", NULL, NULL},
462: { 24, "gb2312", "gb2312_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_gb2312, check_mb_gb2312},
463: { 25, "greek", "greek_general_ci", 1, 1, "", NULL, NULL},
464: { 26, "cp1250", "cp1250_general_ci", 1, 1, "", NULL, NULL},
465: { 28, "gbk", "gbk_chinese_ci", 1, 2, "", mysqlnd_mbcharlen_gbk, check_mb_gbk},
466: { 30, "latin5", "latin5_turkish_ci", 1, 1, "", NULL, NULL},
467: { 32, "armscii8", "armscii8_general_ci", 1, 1, "", NULL, NULL},
468: { 33, UTF8_MB3, UTF8_MB3"_general_ci", 1, 3, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
469: { 35, "ucs2", "ucs2_general_ci", 2, 2, "UCS-2 Unicode", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
470: { 36, "cp866", "cp866_general_ci", 1, 1, "", NULL, NULL},
471: { 37, "keybcs2", "keybcs2_general_ci", 1, 1, "", NULL, NULL},
472: { 38, "macce", "macce_general_ci", 1, 1, "", NULL, NULL},
473: { 39, "macroman", "macroman_general_ci", 1, 1, "", NULL, NULL},
474: { 40, "cp852", "cp852_general_ci", 1, 1, "", NULL, NULL},
475: { 41, "latin7", "latin7_general_ci", 1, 1, "", NULL, NULL},
476: { 51, "cp1251", "cp1251_general_ci", 1, 1, "", NULL, NULL},
477: { 57, "cp1256", "cp1256_general_ci", 1, 1, "", NULL, NULL},
478: { 59, "cp1257", "cp1257_general_ci", 1, 1, "", NULL, NULL},
479: { 63, "binary", "binary", 1, 1, "", NULL, NULL},
480: { 92, "geostd8", "geostd8_general_ci", 1, 1, "", NULL, NULL},
481: { 95, "cp932", "cp932_japanese_ci", 1, 2, "", mysqlnd_mbcharlen_cp932, check_mb_cp932},
482: { 97, "eucjpms", "eucjpms_japanese_ci", 1, 3, "", mysqlnd_mbcharlen_eucjpms, check_mb_eucjpms},
483: { 2, "latin2", "latin2_czech_cs", 1, 1, "", NULL, NULL},
484: { 5, "latin1", "latin1_german_ci", 1, 1, "", NULL, NULL},
485: { 14, "cp1251", "cp1251_bulgarian_ci", 1, 1, "", NULL, NULL},
486: { 15, "latin1", "latin1_danish_ci", 1, 1, "", NULL, NULL},
487: { 17, "filename", "filename", 1, 5, "", NULL, NULL},
488: { 20, "latin7", "latin7_estonian_cs", 1, 1, "", NULL, NULL},
489: { 21, "latin2", "latin2_hungarian_ci", 1, 1, "", NULL, NULL},
490: { 23, "cp1251", "cp1251_ukrainian_ci", 1, 1, "", NULL, NULL},
491: { 27, "latin2", "latin2_croatian_ci", 1, 1, "", NULL, NULL},
492: { 29, "cp1257", "cp1257_lithunian_ci", 1, 1, "", NULL, NULL},
493: { 31, "latin1", "latin1_german2_ci", 1, 1, "", NULL, NULL},
494: { 34, "cp1250", "cp1250_czech_cs", 1, 1, "", NULL, NULL},
495: { 42, "latin7", "latin7_general_cs", 1, 1, "", NULL, NULL},
496: { 43, "macce", "macce_bin", 1, 1, "", NULL, NULL},
497: { 44, "cp1250", "cp1250_croatian_ci", 1, 1, "", NULL, NULL},
1.1.1.2 ! misho 498: { 45, UTF8_MB4, UTF8_MB4"_general_ci", 1, 4, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
! 499: { 46, UTF8_MB4, UTF8_MB4"_bin", 1, 4, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
1.1 misho 500: { 47, "latin1", "latin1_bin", 1, 1, "", NULL, NULL},
501: { 48, "latin1", "latin1_general_ci", 1, 1, "", NULL, NULL},
502: { 49, "latin1", "latin1_general_cs", 1, 1, "", NULL, NULL},
503: { 50, "cp1251", "cp1251_bin", 1, 1, "", NULL, NULL},
504: { 52, "cp1251", "cp1251_general_cs", 1, 1, "", NULL, NULL},
505: { 53, "macroman", "macroman_bin", 1, 1, "", NULL, NULL},
506: { 54, "utf16", "utf16_general_ci", 2, 4, "UTF-16 Unicode", mysqlnd_mbcharlen_utf16, check_mb_utf16},
507: { 55, "utf16", "utf16_bin", 2, 4, "UTF-16 Unicode", mysqlnd_mbcharlen_utf16, check_mb_utf16},
508: { 58, "cp1257", "cp1257_bin", 1, 1, "", NULL, NULL},
509: #ifdef USED_TO_BE_SO_BEFORE_MYSQL_5_5
510: { 60, "armascii8", "armascii8_bin", 1, 1, "", NULL, NULL},
511: #endif
512: { 60, "utf32", "utf32_general_ci", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32, check_mb_utf32},
513: { 61, "utf32", "utf32_bin", 4, 4, "UTF-32 Unicode", mysqlnd_mbcharlen_utf32, check_mb_utf32},
514: { 65, "ascii", "ascii_bin", 1, 1, "", NULL, NULL},
515: { 66, "cp1250", "cp1250_bin", 1, 1, "", NULL, NULL},
516: { 67, "cp1256", "cp1256_bin", 1, 1, "", NULL, NULL},
517: { 68, "cp866", "cp866_bin", 1, 1, "", NULL, NULL},
518: { 69, "dec8", "dec8_bin", 1, 1, "", NULL, NULL},
519: { 70, "greek", "greek_bin", 1, 1, "", NULL, NULL},
520: { 71, "hebew", "hebrew_bin", 1, 1, "", NULL, NULL},
521: { 72, "hp8", "hp8_bin", 1, 1, "", NULL, NULL},
522: { 73, "keybcs2", "keybcs2_bin", 1, 1, "", NULL, NULL},
523: { 74, "koi8r", "koi8r_bin", 1, 1, "", NULL, NULL},
524: { 75, "koi8u", "koi8u_bin", 1, 1, "", NULL, NULL},
525: { 77, "latin2", "latin2_bin", 1, 1, "", NULL, NULL},
526: { 78, "latin5", "latin5_bin", 1, 1, "", NULL, NULL},
527: { 79, "latin7", "latin7_bin", 1, 1, "", NULL, NULL},
528: { 80, "cp850", "cp850_bin", 1, 1, "", NULL, NULL},
529: { 81, "cp852", "cp852_bin", 1, 1, "", NULL, NULL},
530: { 82, "swe7", "swe7_bin", 1, 1, "", NULL, NULL},
531: { 93, "geostd8", "geostd8_bin", 1, 1, "", NULL, NULL},
532: { 83, UTF8_MB3, UTF8_MB3"_bin", 1, 3, "UTF-8 Unicode", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
533: { 84, "big5", "big5_bin", 1, 2, "", mysqlnd_mbcharlen_big5, check_mb_big5},
534: { 85, "euckr", "euckr_bin", 1, 2, "", mysqlnd_mbcharlen_euckr, check_mb_euckr},
535: { 86, "gb2312", "gb2312_bin", 1, 2, "", mysqlnd_mbcharlen_gb2312, check_mb_gb2312},
536: { 87, "gbk", "gbk_bin", 1, 2, "", mysqlnd_mbcharlen_gbk, check_mb_gbk},
537: { 88, "sjis", "sjis_bin", 1, 2, "", mysqlnd_mbcharlen_sjis, check_mb_sjis},
538: { 89, "tis620", "tis620_bin", 1, 1, "", NULL, NULL},
539: { 90, "ucs2", "ucs2_bin", 2, 2, "UCS-2 Unicode", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
540: { 91, "ujis", "ujis_bin", 1, 3, "", mysqlnd_mbcharlen_ujis, check_mb_ujis},
541: { 94, "latin1", "latin1_spanish_ci", 1, 1, "", NULL, NULL},
542: { 96, "cp932", "cp932_bin", 1, 2, "", mysqlnd_mbcharlen_cp932, check_mb_cp932},
543: { 99, "cp1250", "cp1250_polish_ci", 1, 1, "", NULL, NULL},
544: { 98, "eucjpms", "eucjpms_bin", 1, 3, "", mysqlnd_mbcharlen_eucjpms, check_mb_eucjpms},
545: { 128, "ucs2", "ucs2_unicode_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
546: { 129, "ucs2", "ucs2_icelandic_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
547: { 130, "ucs2", "ucs2_latvian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
548: { 131, "ucs2", "ucs2_romanian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
549: { 132, "ucs2", "ucs2_slovenian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
550: { 133, "ucs2", "ucs2_polish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
551: { 134, "ucs2", "ucs2_estonian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
552: { 135, "ucs2", "ucs2_spanish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
553: { 136, "ucs2", "ucs2_swedish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
554: { 137, "ucs2", "ucs2_turkish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
555: { 138, "ucs2", "ucs2_czech_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
556: { 139, "ucs2", "ucs2_danish_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
557: { 140, "ucs2", "ucs2_lithunian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
558: { 141, "ucs2", "ucs2_slovak_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
559: { 142, "ucs2", "ucs2_spanish2_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
560: { 143, "ucs2", "ucs2_roman_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
561: { 144, "ucs2", "ucs2_persian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
562: { 145, "ucs2", "ucs2_esperanto_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
563: { 146, "ucs2", "ucs2_hungarian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
564: { 147, "ucs2", "ucs2_sinhala_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2},
565: { 149, "ucs2", "ucs2_croatian_ci", 2, 2, "", mysqlnd_mbcharlen_ucs2, check_mb_ucs2}, /* MDB */
566:
567: { 192, UTF8_MB3, UTF8_MB3"_general_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
568: { 193, UTF8_MB3, UTF8_MB3"_icelandic_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
569: { 194, UTF8_MB3, UTF8_MB3"_latvian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
570: { 195, UTF8_MB3, UTF8_MB3"_romanian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
571: { 196, UTF8_MB3, UTF8_MB3"_slovenian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
572: { 197, UTF8_MB3, UTF8_MB3"_polish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
573: { 198, UTF8_MB3, UTF8_MB3"_estonian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
574: { 119, UTF8_MB3, UTF8_MB3"_spanish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
575: { 200, UTF8_MB3, UTF8_MB3"_swedish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
576: { 201, UTF8_MB3, UTF8_MB3"_turkish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
577: { 202, UTF8_MB3, UTF8_MB3"_czech_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
578: { 203, UTF8_MB3, UTF8_MB3"_danish_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid },
579: { 204, UTF8_MB3, UTF8_MB3"_lithunian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid },
580: { 205, UTF8_MB3, UTF8_MB3"_slovak_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
581: { 206, UTF8_MB3, UTF8_MB3"_spanish2_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
582: { 207, UTF8_MB3, UTF8_MB3"_roman_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
583: { 208, UTF8_MB3, UTF8_MB3"_persian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
584: { 209, UTF8_MB3, UTF8_MB3"_esperanto_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
585: { 210, UTF8_MB3, UTF8_MB3"_hungarian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
586: { 211, UTF8_MB3, UTF8_MB3"_sinhala_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid},
587: { 213, UTF8_MB3, UTF8_MB3"_croatian_ci", 1, 3, "", mysqlnd_mbcharlen_utf8mb3, check_mb_utf8mb3_valid}, /*MDB*/
588:
589: { 224, UTF8_MB4, UTF8_MB4"_unicode_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
590: { 225, UTF8_MB4, UTF8_MB4"_icelandic_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
591: { 226, UTF8_MB4, UTF8_MB4"_latvian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
592: { 227, UTF8_MB4, UTF8_MB4"_romanian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
593: { 228, UTF8_MB4, UTF8_MB4"_slovenian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
594: { 229, UTF8_MB4, UTF8_MB4"_polish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
595: { 230, UTF8_MB4, UTF8_MB4"_estonian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
596: { 231, UTF8_MB4, UTF8_MB4"_spanish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
597: { 232, UTF8_MB4, UTF8_MB4"_swedish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
598: { 233, UTF8_MB4, UTF8_MB4"_turkish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
599: { 234, UTF8_MB4, UTF8_MB4"_czech_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
600: { 235, UTF8_MB4, UTF8_MB4"_danish_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
601: { 236, UTF8_MB4, UTF8_MB4"_lithuanian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
602: { 237, UTF8_MB4, UTF8_MB4"_slovak_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
603: { 238, UTF8_MB4, UTF8_MB4"_spanish2_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
604: { 239, UTF8_MB4, UTF8_MB4"_roman_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
605: { 240, UTF8_MB4, UTF8_MB4"_persian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
606: { 241, UTF8_MB4, UTF8_MB4"_esperanto_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
607: { 242, UTF8_MB4, UTF8_MB4"_hungarian_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
608: { 243, UTF8_MB4, UTF8_MB4"_sinhala_ci", 1, 4, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
609:
610: { 254, UTF8_MB3, UTF8_MB3"_general_cs", 1, 3, "", mysqlnd_mbcharlen_utf8, check_mb_utf8_valid},
611: { 0, NULL, NULL, 0, 0, NULL, NULL, NULL}
612: };
613: /* }}} */
614:
615:
616: /* {{{ mysqlnd_find_charset_nr */
617: PHPAPI const MYSQLND_CHARSET * mysqlnd_find_charset_nr(unsigned int charsetnr)
618: {
619: const MYSQLND_CHARSET * c = mysqlnd_charsets;
620:
621: do {
622: if (c->nr == charsetnr) {
623: return c;
624: }
625: ++c;
626: } while (c[0].nr != 0);
627: return NULL;
628: }
629: /* }}} */
630:
631:
632: /* {{{ mysqlnd_find_charset_name */
633: PHPAPI const MYSQLND_CHARSET * mysqlnd_find_charset_name(const char * const name)
634: {
1.1.1.2 ! misho 635: const MYSQLND_CHARSET * c = mysqlnd_charsets;
1.1 misho 636:
637: do {
638: if (!strcasecmp(c->name, name)) {
639: return c;
640: }
641: ++c;
642: } while (c[0].nr != 0);
643: return NULL;
644: }
645: /* }}} */
646:
647:
648: /* {{{ mysqlnd_cset_escape_quotes */
649: PHPAPI ulong mysqlnd_cset_escape_quotes(const MYSQLND_CHARSET * const cset, char *newstr,
650: const char * escapestr, size_t escapestr_len TSRMLS_DC)
651: {
652: const char *newstr_s = newstr;
653: const char *newstr_e = newstr + 2 * escapestr_len;
654: const char *end = escapestr + escapestr_len;
655: zend_bool escape_overflow = FALSE;
656:
657: DBG_ENTER("mysqlnd_cset_escape_quotes");
658:
659: for (;escapestr < end; escapestr++) {
660: unsigned int len = 0;
661: /* check unicode characters */
662:
663: if (cset->char_maxlen > 1 && (len = cset->mb_valid(escapestr, end))) {
664:
665: /* check possible overflow */
666: if ((newstr + len) > newstr_e) {
667: escape_overflow = TRUE;
668: break;
669: }
670: /* copy mb char without escaping it */
671: while (len--) {
672: *newstr++ = *escapestr++;
673: }
674: escapestr--;
675: continue;
676: }
677: if (*escapestr == '\'') {
678: if (newstr + 2 > newstr_e) {
679: escape_overflow = TRUE;
680: break;
681: }
682: *newstr++ = '\'';
683: *newstr++ = '\'';
684: } else {
685: if (newstr + 1 > newstr_e) {
686: escape_overflow = TRUE;
687: break;
688: }
689: *newstr++ = *escapestr;
690: }
691: }
692: *newstr = '\0';
693:
694: if (escape_overflow) {
695: DBG_RETURN((ulong)~0);
696: }
697: DBG_RETURN((ulong)(newstr - newstr_s));
698: }
699: /* }}} */
700:
701:
702: /* {{{ mysqlnd_cset_escape_slashes */
703: PHPAPI ulong mysqlnd_cset_escape_slashes(const MYSQLND_CHARSET * const cset, char *newstr,
704: const char * escapestr, size_t escapestr_len TSRMLS_DC)
705: {
706: const char *newstr_s = newstr;
707: const char *newstr_e = newstr + 2 * escapestr_len;
708: const char *end = escapestr + escapestr_len;
709: zend_bool escape_overflow = FALSE;
710:
711: DBG_ENTER("mysqlnd_cset_escape_slashes");
712: DBG_INF_FMT("charset=%s", cset->name);
713:
714: for (;escapestr < end; escapestr++) {
715: char esc = '\0';
716: unsigned int len = 0;
717:
718: /* check unicode characters */
719: if (cset->char_maxlen > 1 && (len = cset->mb_valid(escapestr, end))) {
720: /* check possible overflow */
721: if ((newstr + len) > newstr_e) {
722: escape_overflow = TRUE;
723: break;
724: }
725: /* copy mb char without escaping it */
726: while (len--) {
727: *newstr++ = *escapestr++;
728: }
729: escapestr--;
730: continue;
731: }
732: if (cset->char_maxlen > 1 && cset->mb_charlen(*escapestr) > 1) {
733: esc = *escapestr;
734: } else {
735: switch (*escapestr) {
736: case 0:
737: esc = '0';
738: break;
739: case '\n':
740: esc = 'n';
741: break;
742: case '\r':
743: esc = 'r';
744: break;
745: case '\\':
746: case '\'':
747: case '"':
748: esc = *escapestr;
749: break;
750: case '\032':
751: esc = 'Z';
752: break;
753: }
754: }
755: if (esc) {
756: if (newstr + 2 > newstr_e) {
757: escape_overflow = TRUE;
758: break;
759: }
760: /* copy escaped character */
761: *newstr++ = '\\';
762: *newstr++ = esc;
763: } else {
764: if (newstr + 1 > newstr_e) {
765: escape_overflow = TRUE;
766: break;
767: }
768: /* copy non escaped character */
769: *newstr++ = *escapestr;
770: }
771: }
772: *newstr = '\0';
773:
774: if (escape_overflow) {
775: DBG_RETURN((ulong)~0);
776: }
777: DBG_RETURN((ulong)(newstr - newstr_s));
778: }
779: /* }}} */
780:
1.1.1.2 ! misho 781:
! 782: static struct st_mysqlnd_plugin_charsets mysqlnd_plugin_charsets_plugin =
! 783: {
! 784: {
! 785: MYSQLND_PLUGIN_API_VERSION,
! 786: "charsets",
! 787: MYSQLND_VERSION_ID,
! 788: MYSQLND_VERSION,
! 789: "PHP License 3.01",
! 790: "Andrey Hristov <andrey@mysql.com>, Ulf Wendel <uwendel@mysql.com>, Georg Richter <georg@mysql.com>",
! 791: {
! 792: NULL, /* no statistics , will be filled later if there are some */
! 793: NULL, /* no statistics */
! 794: },
! 795: {
! 796: NULL /* plugin shutdown */
! 797: }
! 798: },
! 799: {/* methods */
! 800: mysqlnd_find_charset_nr,
! 801: mysqlnd_find_charset_name,
! 802: mysqlnd_cset_escape_quotes,
! 803: mysqlnd_cset_escape_slashes
! 804: }
! 805: };
! 806:
! 807:
! 808: /* {{{ mysqlnd_charsets_plugin_register */
! 809: void
! 810: mysqlnd_charsets_plugin_register(TSRMLS_D)
! 811: {
! 812: mysqlnd_plugin_register_ex((struct st_mysqlnd_plugin_header *) &mysqlnd_plugin_charsets_plugin TSRMLS_CC);
! 813: }
! 814: /* }}} */
! 815:
! 816:
1.1 misho 817: /*
818: * Local variables:
819: * tab-width: 4
820: * c-basic-offset: 4
821: * End:
822: * vim600: noet sw=4 ts=4 fdm=marker
823: * vim<600: noet sw=4 ts=4
824: */
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>