1
/* Copyright (C) 2000 MySQL AB
3
This program is free software; you can redistribute it and/or modify
4
it under the terms of the GNU General Public License as published by
5
the Free Software Foundation; version 2 of the License.
7
This program is distributed in the hope that it will be useful,
8
but WITHOUT ANY WARRANTY; without even the implied warranty of
9
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
10
GNU General Public License for more details.
12
You should have received a copy of the GNU General Public License
13
along with this program; if not, write to the Free Software
14
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
17
A better inplementation of the UNIX ctype(3) library.
18
Notes: my_global.h should be included before ctype.h
24
#include <my_attribute.h>
30
#define MY_CS_NAME_SIZE 32
31
#define MY_CS_CTYPE_TABLE_SIZE 257
32
#define MY_CS_TO_LOWER_TABLE_SIZE 256
33
#define MY_CS_TO_UPPER_TABLE_SIZE 256
34
#define MY_CS_SORT_ORDER_TABLE_SIZE 256
35
#define MY_CS_TO_UNI_TABLE_SIZE 256
37
#define CHARSET_DIR "charsets/"
41
typedef struct unicase_info_st
49
extern MY_UNICASE_INFO *my_unicase_default[256];
50
extern MY_UNICASE_INFO *my_unicase_turkish[256];
52
typedef struct uni_ctype_st
58
extern MY_UNI_CTYPE my_uni_ctype[256];
60
/* wm_wc and wc_mb return codes */
61
#define MY_CS_ILSEQ 0 /* Wrong by sequence: wb_wc */
62
#define MY_CS_ILUNI 0 /* Cannot encode Unicode to charset: wc_mb */
63
#define MY_CS_TOOSMALL -101 /* Need at least one byte: wc_mb and mb_wc */
64
#define MY_CS_TOOSMALL2 -102 /* Need at least two bytes: wc_mb and mb_wc */
65
#define MY_CS_TOOSMALL3 -103 /* Need at least three bytes: wc_mb and mb_wc */
66
/* These following three are currently not really used */
67
#define MY_CS_TOOSMALL4 -104 /* Need at least 4 bytes: wc_mb and mb_wc */
68
#define MY_CS_TOOSMALL5 -105 /* Need at least 5 bytes: wc_mb and mb_wc */
69
#define MY_CS_TOOSMALL6 -106 /* Need at least 6 bytes: wc_mb and mb_wc */
70
/* A helper macros for "need at least n bytes" */
71
#define MY_CS_TOOSMALLN(n) (-100-(n))
73
#define MY_SEQ_INTTAIL 1
74
#define MY_SEQ_SPACES 2
76
/* My charsets_list flags */
77
#define MY_CS_COMPILED 1 /* compiled-in sets */
78
#define MY_CS_CONFIG 2 /* sets that have a *.conf file */
79
#define MY_CS_INDEX 4 /* sets listed in the Index file */
80
#define MY_CS_LOADED 8 /* sets that are currently loaded */
81
#define MY_CS_BINSORT 16 /* if binary sort order */
82
#define MY_CS_PRIMARY 32 /* if primary collation */
83
#define MY_CS_STRNXFRM 64 /* if strnxfrm is used for sort */
84
#define MY_CS_UNICODE 128 /* is a charset is full unicode */
85
#define MY_CS_READY 256 /* if a charset is initialized */
86
#define MY_CS_AVAILABLE 512 /* If either compiled-in or loaded*/
87
#define MY_CS_CSSORT 1024 /* if case sensitive sort order */
88
#define MY_CS_HIDDEN 2048 /* don't display in SHOW */
89
#define MY_CS_PUREASCII 4096 /* if a charset is pure ascii */
90
#define MY_CS_NONASCII 8192 /* if not ASCII-compatible */
91
#define MY_CHARSET_UNDEFINED 0
93
/* Character repertoire flags */
94
#define MY_REPERTOIRE_ASCII 1 /* Pure ASCII U+0000..U+007F */
95
#define MY_REPERTOIRE_EXTENDED 2 /* Extended characters: U+0080..U+FFFF */
96
#define MY_REPERTOIRE_UNICODE30 3 /* ASCII | EXTENDED: U+0000..U+FFFF */
98
/* Flags for strxfrm */
99
#define MY_STRXFRM_LEVEL1 0x00000001 /* for primary weights */
100
#define MY_STRXFRM_LEVEL2 0x00000002 /* for secondary weights */
101
#define MY_STRXFRM_LEVEL3 0x00000004 /* for tertiary weights */
102
#define MY_STRXFRM_LEVEL4 0x00000008 /* fourth level weights */
103
#define MY_STRXFRM_LEVEL5 0x00000010 /* fifth level weights */
104
#define MY_STRXFRM_LEVEL6 0x00000020 /* sixth level weights */
105
#define MY_STRXFRM_LEVEL_ALL 0x0000003F /* Bit OR for the above six */
106
#define MY_STRXFRM_NLEVELS 6 /* Number of possible levels*/
108
#define MY_STRXFRM_PAD_WITH_SPACE 0x00000040 /* if pad result with spaces */
109
#define MY_STRXFRM_UNUSED_00000080 0x00000080 /* for future extensions */
111
#define MY_STRXFRM_DESC_LEVEL1 0x00000100 /* if desc order for level1 */
112
#define MY_STRXFRM_DESC_LEVEL2 0x00000200 /* if desc order for level2 */
113
#define MY_STRXFRM_DESC_LEVEL3 0x00000300 /* if desc order for level3 */
114
#define MY_STRXFRM_DESC_LEVEL4 0x00000800 /* if desc order for level4 */
115
#define MY_STRXFRM_DESC_LEVEL5 0x00001000 /* if desc order for level5 */
116
#define MY_STRXFRM_DESC_LEVEL6 0x00002000 /* if desc order for level6 */
117
#define MY_STRXFRM_DESC_SHIFT 8
119
#define MY_STRXFRM_UNUSED_00004000 0x00004000 /* for future extensions */
120
#define MY_STRXFRM_UNUSED_00008000 0x00008000 /* for future extensions */
122
#define MY_STRXFRM_REVERSE_LEVEL1 0x00010000 /* if reverse order for level1 */
123
#define MY_STRXFRM_REVERSE_LEVEL2 0x00020000 /* if reverse order for level2 */
124
#define MY_STRXFRM_REVERSE_LEVEL3 0x00040000 /* if reverse order for level3 */
125
#define MY_STRXFRM_REVERSE_LEVEL4 0x00080000 /* if reverse order for level4 */
126
#define MY_STRXFRM_REVERSE_LEVEL5 0x00100000 /* if reverse order for level5 */
127
#define MY_STRXFRM_REVERSE_LEVEL6 0x00200000 /* if reverse order for level6 */
128
#define MY_STRXFRM_REVERSE_SHIFT 16
131
typedef struct my_uni_idx_st
147
MY_LEX_START, MY_LEX_CHAR, MY_LEX_IDENT,
148
MY_LEX_IDENT_SEP, MY_LEX_IDENT_START,
149
MY_LEX_REAL, MY_LEX_HEX_NUMBER, MY_LEX_BIN_NUMBER,
150
MY_LEX_CMP_OP, MY_LEX_LONG_CMP_OP, MY_LEX_STRING, MY_LEX_COMMENT, MY_LEX_END,
151
MY_LEX_OPERATOR_OR_IDENT, MY_LEX_NUMBER_IDENT, MY_LEX_INT_OR_REAL,
152
MY_LEX_REAL_OR_POINT, MY_LEX_BOOL, MY_LEX_EOL, MY_LEX_ESCAPE,
153
MY_LEX_LONG_COMMENT, MY_LEX_END_LONG_COMMENT, MY_LEX_SEMICOLON,
154
MY_LEX_SET_VAR, MY_LEX_USER_END, MY_LEX_HOSTNAME, MY_LEX_SKIP,
155
MY_LEX_USER_VARIABLE_DELIMITER, MY_LEX_SYSTEM_VAR,
156
MY_LEX_IDENT_OR_KEYWORD,
157
MY_LEX_IDENT_OR_HEX, MY_LEX_IDENT_OR_BIN, MY_LEX_IDENT_OR_NCHAR,
158
MY_LEX_STRING_OR_DELIMITER
161
struct charset_info_st;
164
/* See strings/CHARSET_INFO.txt for information about this structure */
165
typedef struct my_collation_handler_st
167
my_bool (*init)(struct charset_info_st *, void *(*alloc)(size_t));
168
/* Collation routines */
169
int (*strnncoll)(struct charset_info_st *,
170
const uchar *, size_t, const uchar *, size_t, my_bool);
171
int (*strnncollsp)(struct charset_info_st *,
172
const uchar *, size_t, const uchar *, size_t,
173
my_bool diff_if_only_endspace_difference);
174
size_t (*strnxfrm)(struct charset_info_st *,
175
uchar *dst, size_t dstlen, uint nweights,
176
const uchar *src, size_t srclen, uint flags);
177
size_t (*strnxfrmlen)(struct charset_info_st *, size_t);
178
my_bool (*like_range)(struct charset_info_st *,
179
const char *s, size_t s_length,
180
pchar w_prefix, pchar w_one, pchar w_many,
182
char *min_str, char *max_str,
183
size_t *min_len, size_t *max_len);
184
int (*wildcmp)(struct charset_info_st *,
185
const char *str,const char *str_end,
186
const char *wildstr,const char *wildend,
187
int escape,int w_one, int w_many);
189
int (*strcasecmp)(struct charset_info_st *, const char *, const char *);
191
uint (*instr)(struct charset_info_st *,
192
const char *b, size_t b_length,
193
const char *s, size_t s_length,
194
my_match_t *match, uint nmatch);
196
/* Hash calculation */
197
void (*hash_sort)(struct charset_info_st *cs, const uchar *key, size_t len,
198
ulong *nr1, ulong *nr2);
199
my_bool (*propagate)(struct charset_info_st *cs, const uchar *str, size_t len);
200
} MY_COLLATION_HANDLER;
202
extern MY_COLLATION_HANDLER my_collation_mb_bin_handler;
203
extern MY_COLLATION_HANDLER my_collation_8bit_bin_handler;
204
extern MY_COLLATION_HANDLER my_collation_8bit_simple_ci_handler;
205
extern MY_COLLATION_HANDLER my_collation_ucs2_uca_handler;
207
/* Some typedef to make it easy for C++ to make function pointers */
208
typedef int (*my_charset_conv_mb_wc)(struct charset_info_st *, my_wc_t *,
209
const uchar *, const uchar *);
210
typedef int (*my_charset_conv_wc_mb)(struct charset_info_st *, my_wc_t,
212
typedef size_t (*my_charset_conv_case)(struct charset_info_st *,
213
char *, size_t, char *, size_t);
216
/* See strings/CHARSET_INFO.txt about information on this structure */
217
typedef struct my_charset_handler_st
219
my_bool (*init)(struct charset_info_st *, void *(*alloc)(size_t));
220
/* Multibyte routines */
221
uint (*ismbchar)(struct charset_info_st *, const char *, const char *);
222
uint (*mbcharlen)(struct charset_info_st *, uint c);
223
size_t (*numchars)(struct charset_info_st *, const char *b, const char *e);
224
size_t (*charpos)(struct charset_info_st *, const char *b, const char *e,
226
size_t (*well_formed_len)(struct charset_info_st *,
227
const char *b,const char *e,
228
size_t nchars, int *error);
229
size_t (*lengthsp)(struct charset_info_st *, const char *ptr, size_t length);
230
size_t (*numcells)(struct charset_info_st *, const char *b, const char *e);
232
/* Unicode conversion */
233
my_charset_conv_mb_wc mb_wc;
234
my_charset_conv_wc_mb wc_mb;
237
int (*ctype)(struct charset_info_st *cs, int *ctype,
238
const uchar *s, const uchar *e);
240
/* Functions for case and sort conversion */
241
size_t (*caseup_str)(struct charset_info_st *, char *);
242
size_t (*casedn_str)(struct charset_info_st *, char *);
244
my_charset_conv_case caseup;
245
my_charset_conv_case casedn;
247
/* Charset dependant snprintf() */
248
size_t (*snprintf)(struct charset_info_st *, char *to, size_t n,
250
...) ATTRIBUTE_FORMAT_FPTR(printf, 4, 5);
251
size_t (*long10_to_str)(struct charset_info_st *, char *to, size_t n,
252
int radix, long int val);
253
size_t (*longlong10_to_str)(struct charset_info_st *, char *to, size_t n,
254
int radix, longlong val);
256
void (*fill)(struct charset_info_st *, char *to, size_t len, int fill);
258
/* String-to-number conversion routines */
259
long (*strntol)(struct charset_info_st *, const char *s, size_t l,
260
int base, char **e, int *err);
261
ulong (*strntoul)(struct charset_info_st *, const char *s, size_t l,
262
int base, char **e, int *err);
263
longlong (*strntoll)(struct charset_info_st *, const char *s, size_t l,
264
int base, char **e, int *err);
265
ulonglong (*strntoull)(struct charset_info_st *, const char *s, size_t l,
266
int base, char **e, int *err);
267
double (*strntod)(struct charset_info_st *, char *s, size_t l, char **e,
269
longlong (*strtoll10)(struct charset_info_st *cs,
270
const char *nptr, char **endptr, int *error);
271
ulonglong (*strntoull10rnd)(struct charset_info_st *cs,
272
const char *str, size_t length,
274
char **endptr, int *error);
275
size_t (*scan)(struct charset_info_st *, const char *b, const char *e,
277
} MY_CHARSET_HANDLER;
279
extern MY_CHARSET_HANDLER my_charset_8bit_handler;
280
extern MY_CHARSET_HANDLER my_charset_ucs2_handler;
283
/* See strings/CHARSET_INFO.txt about information on this structure */
284
typedef struct charset_info_st
293
const char *tailoring;
298
uint16 *contractions;
299
uint16 **sort_order_big;
301
MY_UNI_IDX *tab_from_uni;
302
MY_UNICASE_INFO **caseinfo;
305
uint strxfrm_multiply;
306
uchar caseup_multiply;
307
uchar casedn_multiply;
310
uint16 min_sort_char;
311
uint16 max_sort_char; /* For LIKE optimization */
313
my_bool escape_with_backslash_is_dangerous;
314
uchar levels_for_compare;
315
uchar levels_for_order;
317
MY_CHARSET_HANDLER *cset;
318
MY_COLLATION_HANDLER *coll;
321
#define ILLEGAL_CHARSET_INFO_NUMBER (~0U)
324
extern CHARSET_INFO my_charset_bin;
325
extern CHARSET_INFO my_charset_big5_chinese_ci;
326
extern CHARSET_INFO my_charset_big5_bin;
327
extern CHARSET_INFO my_charset_cp932_japanese_ci;
328
extern CHARSET_INFO my_charset_cp932_bin;
329
extern CHARSET_INFO my_charset_cp1250_czech_ci;
330
extern CHARSET_INFO my_charset_eucjpms_japanese_ci;
331
extern CHARSET_INFO my_charset_eucjpms_bin;
332
extern CHARSET_INFO my_charset_euckr_korean_ci;
333
extern CHARSET_INFO my_charset_euckr_bin;
334
extern CHARSET_INFO my_charset_filename;
335
extern CHARSET_INFO my_charset_gb2312_chinese_ci;
336
extern CHARSET_INFO my_charset_gb2312_bin;
337
extern CHARSET_INFO my_charset_gbk_chinese_ci;
338
extern CHARSET_INFO my_charset_gbk_bin;
339
extern CHARSET_INFO my_charset_latin1;
340
extern CHARSET_INFO my_charset_latin1_german2_ci;
341
extern CHARSET_INFO my_charset_latin1_bin;
342
extern CHARSET_INFO my_charset_latin2_czech_ci;
343
extern CHARSET_INFO my_charset_sjis_japanese_ci;
344
extern CHARSET_INFO my_charset_sjis_bin;
345
extern CHARSET_INFO my_charset_tis620_thai_ci;
346
extern CHARSET_INFO my_charset_tis620_bin;
347
extern CHARSET_INFO my_charset_ucs2_general_ci;
348
extern CHARSET_INFO my_charset_ucs2_bin;
349
extern CHARSET_INFO my_charset_ucs2_unicode_ci;
350
extern CHARSET_INFO my_charset_ujis_japanese_ci;
351
extern CHARSET_INFO my_charset_ujis_bin;
352
extern CHARSET_INFO my_charset_utf16_bin;
353
extern CHARSET_INFO my_charset_utf16_general_ci;
354
extern CHARSET_INFO my_charset_utf16_unicode_ci;
355
extern CHARSET_INFO my_charset_utf32_bin;
356
extern CHARSET_INFO my_charset_utf32_general_ci;
357
extern CHARSET_INFO my_charset_utf32_unicode_ci;
358
extern CHARSET_INFO my_charset_utf8mb3_bin;
359
extern CHARSET_INFO my_charset_utf8mb3_general_ci;
360
extern CHARSET_INFO my_charset_utf8mb3_unicode_ci;
361
extern CHARSET_INFO my_charset_utf8mb4_bin;
362
extern CHARSET_INFO my_charset_utf8mb4_general_ci;
363
extern CHARSET_INFO my_charset_utf8mb4_unicode_ci;
365
#define MY_UTF8MB3 "utf8mb3"
366
#define MY_UTF8MB4 "utf8"
367
#define my_charset_utf8_general_ci my_charset_utf8mb4_general_ci
368
#define my_charset_utf8_bin my_charset_utf8mb4_bin
371
/* declarations for simple charsets */
372
extern size_t my_strnxfrm_simple(CHARSET_INFO *,
373
uchar *dst, size_t dstlen, uint nweights,
374
const uchar *src, size_t srclen, uint flags);
375
size_t my_strnxfrmlen_simple(CHARSET_INFO *, size_t);
376
extern int my_strnncoll_simple(CHARSET_INFO *, const uchar *, size_t,
377
const uchar *, size_t, my_bool);
379
extern int my_strnncollsp_simple(CHARSET_INFO *, const uchar *, size_t,
380
const uchar *, size_t,
381
my_bool diff_if_only_endspace_difference);
383
extern void my_hash_sort_simple(CHARSET_INFO *cs,
384
const uchar *key, size_t len,
385
ulong *nr1, ulong *nr2);
387
extern size_t my_lengthsp_8bit(CHARSET_INFO *cs, const char *ptr, size_t length);
389
extern uint my_instr_simple(struct charset_info_st *,
390
const char *b, size_t b_length,
391
const char *s, size_t s_length,
392
my_match_t *match, uint nmatch);
395
/* Functions for 8bit */
396
extern size_t my_caseup_str_8bit(CHARSET_INFO *, char *);
397
extern size_t my_casedn_str_8bit(CHARSET_INFO *, char *);
398
extern size_t my_caseup_8bit(CHARSET_INFO *, char *src, size_t srclen,
399
char *dst, size_t dstlen);
400
extern size_t my_casedn_8bit(CHARSET_INFO *, char *src, size_t srclen,
401
char *dst, size_t dstlen);
403
extern int my_strcasecmp_8bit(CHARSET_INFO * cs, const char *, const char *);
405
int my_mb_wc_8bit(CHARSET_INFO *cs,my_wc_t *wc, const uchar *s,const uchar *e);
406
int my_wc_mb_8bit(CHARSET_INFO *cs,my_wc_t wc, uchar *s, uchar *e);
408
int my_mb_ctype_8bit(CHARSET_INFO *,int *, const uchar *,const uchar *);
409
int my_mb_ctype_mb(CHARSET_INFO *,int *, const uchar *,const uchar *);
411
size_t my_scan_8bit(CHARSET_INFO *cs, const char *b, const char *e, int sq);
413
size_t my_snprintf_8bit(struct charset_info_st *, char *to, size_t n,
414
const char *fmt, ...)
415
ATTRIBUTE_FORMAT(printf, 4, 5);
417
long my_strntol_8bit(CHARSET_INFO *, const char *s, size_t l, int base,
419
ulong my_strntoul_8bit(CHARSET_INFO *, const char *s, size_t l, int base,
421
longlong my_strntoll_8bit(CHARSET_INFO *, const char *s, size_t l, int base,
423
ulonglong my_strntoull_8bit(CHARSET_INFO *, const char *s, size_t l, int base,
425
double my_strntod_8bit(CHARSET_INFO *, char *s, size_t l,char **e,
427
size_t my_long10_to_str_8bit(CHARSET_INFO *, char *to, size_t l, int radix,
429
size_t my_longlong10_to_str_8bit(CHARSET_INFO *, char *to, size_t l, int radix,
432
longlong my_strtoll10_8bit(CHARSET_INFO *cs,
433
const char *nptr, char **endptr, int *error);
434
longlong my_strtoll10_ucs2(CHARSET_INFO *cs,
435
const char *nptr, char **endptr, int *error);
437
ulonglong my_strntoull10rnd_8bit(CHARSET_INFO *cs,
438
const char *str, size_t length, int
439
unsigned_fl, char **endptr, int *error);
440
ulonglong my_strntoull10rnd_ucs2(CHARSET_INFO *cs,
441
const char *str, size_t length,
442
int unsigned_fl, char **endptr, int *error);
444
void my_fill_8bit(CHARSET_INFO *cs, char* to, size_t l, int fill);
446
my_bool my_like_range_simple(CHARSET_INFO *cs,
447
const char *ptr, size_t ptr_length,
448
pbool escape, pbool w_one, pbool w_many,
450
char *min_str, char *max_str,
451
size_t *min_length, size_t *max_length);
453
my_bool my_like_range_mb(CHARSET_INFO *cs,
454
const char *ptr, size_t ptr_length,
455
pbool escape, pbool w_one, pbool w_many,
457
char *min_str, char *max_str,
458
size_t *min_length, size_t *max_length);
460
my_bool my_like_range_ucs2(CHARSET_INFO *cs,
461
const char *ptr, size_t ptr_length,
462
pbool escape, pbool w_one, pbool w_many,
464
char *min_str, char *max_str,
465
size_t *min_length, size_t *max_length);
467
my_bool my_like_range_utf16(CHARSET_INFO *cs,
468
const char *ptr, size_t ptr_length,
469
pbool escape, pbool w_one, pbool w_many,
471
char *min_str, char *max_str,
472
size_t *min_length, size_t *max_length);
474
my_bool my_like_range_utf32(CHARSET_INFO *cs,
475
const char *ptr, size_t ptr_length,
476
pbool escape, pbool w_one, pbool w_many,
478
char *min_str, char *max_str,
479
size_t *min_length, size_t *max_length);
482
int my_wildcmp_8bit(CHARSET_INFO *,
483
const char *str,const char *str_end,
484
const char *wildstr,const char *wildend,
485
int escape, int w_one, int w_many);
487
int my_wildcmp_bin(CHARSET_INFO *,
488
const char *str,const char *str_end,
489
const char *wildstr,const char *wildend,
490
int escape, int w_one, int w_many);
492
size_t my_numchars_8bit(CHARSET_INFO *, const char *b, const char *e);
493
size_t my_numcells_8bit(CHARSET_INFO *, const char *b, const char *e);
494
size_t my_charpos_8bit(CHARSET_INFO *, const char *b, const char *e, size_t pos);
495
size_t my_well_formed_len_8bit(CHARSET_INFO *, const char *b, const char *e,
496
size_t pos, int *error);
497
uint my_mbcharlen_8bit(CHARSET_INFO *, uint c);
500
/* Functions for multibyte charsets */
501
extern size_t my_caseup_str_mb(CHARSET_INFO *, char *);
502
extern size_t my_casedn_str_mb(CHARSET_INFO *, char *);
503
extern size_t my_caseup_mb(CHARSET_INFO *, char *src, size_t srclen,
504
char *dst, size_t dstlen);
505
extern size_t my_casedn_mb(CHARSET_INFO *, char *src, size_t srclen,
506
char *dst, size_t dstlen);
507
extern int my_strcasecmp_mb(CHARSET_INFO * cs,const char *, const char *);
509
int my_wildcmp_mb(CHARSET_INFO *,
510
const char *str,const char *str_end,
511
const char *wildstr,const char *wildend,
512
int escape, int w_one, int w_many);
513
size_t my_numchars_mb(CHARSET_INFO *, const char *b, const char *e);
514
size_t my_numcells_mb(CHARSET_INFO *, const char *b, const char *e);
515
size_t my_charpos_mb(CHARSET_INFO *, const char *b, const char *e, size_t pos);
516
size_t my_well_formed_len_mb(CHARSET_INFO *, const char *b, const char *e,
517
size_t pos, int *error);
518
uint my_instr_mb(struct charset_info_st *,
519
const char *b, size_t b_length,
520
const char *s, size_t s_length,
521
my_match_t *match, uint nmatch);
523
int my_strnncoll_mb_bin(CHARSET_INFO * cs,
524
const uchar *s, size_t slen,
525
const uchar *t, size_t tlen,
526
my_bool t_is_prefix);
528
int my_strnncollsp_mb_bin(CHARSET_INFO *cs,
529
const uchar *a, size_t a_length,
530
const uchar *b, size_t b_length,
531
my_bool diff_if_only_endspace_difference);
533
int my_wildcmp_mb_bin(CHARSET_INFO *cs,
534
const char *str,const char *str_end,
535
const char *wildstr,const char *wildend,
536
int escape, int w_one, int w_many);
538
int my_strcasecmp_mb_bin(CHARSET_INFO * cs __attribute__((unused)),
539
const char *s, const char *t);
541
void my_hash_sort_mb_bin(CHARSET_INFO *cs __attribute__((unused)),
542
const uchar *key, size_t len,ulong *nr1, ulong *nr2);
544
size_t my_strnxfrm_mb(CHARSET_INFO *,
545
uchar *dst, size_t dstlen, uint nweights,
546
const uchar *src, size_t srclen, uint flags);
548
int my_wildcmp_unicode(CHARSET_INFO *cs,
549
const char *str, const char *str_end,
550
const char *wildstr, const char *wildend,
551
int escape, int w_one, int w_many,
552
MY_UNICASE_INFO **weights);
554
extern my_bool my_parse_charset_xml(const char *bug, size_t len,
555
int (*add)(CHARSET_INFO *cs));
556
extern char *my_strchr(CHARSET_INFO *cs, const char *str, const char *end,
559
my_bool my_propagate_simple(CHARSET_INFO *cs, const uchar *str, size_t len);
560
my_bool my_propagate_complex(CHARSET_INFO *cs, const uchar *str, size_t len);
563
uint my_string_repertoire(CHARSET_INFO *cs, const char *str, ulong len);
564
my_bool my_charset_is_ascii_based(CHARSET_INFO *cs);
565
my_bool my_charset_is_8bit_pure_ascii(CHARSET_INFO *cs);
568
uint my_strxfrm_flag_normalize(uint flags, uint nlevels);
569
void my_strxfrm_desc_and_reverse(uchar *str, uchar *strend,
570
uint flags, uint level);
571
size_t my_strxfrm_pad_desc_and_reverse(CHARSET_INFO *cs,
572
uchar *str, uchar *frmend, uchar *strend,
573
uint nweights, uint flags, uint level);
575
my_bool my_charset_is_ascii_compatible(CHARSET_INFO *cs);
577
#define _MY_U 01 /* Upper case */
578
#define _MY_L 02 /* Lower case */
579
#define _MY_NMR 04 /* Numeral (digit) */
580
#define _MY_SPC 010 /* Spacing character */
581
#define _MY_PNT 020 /* Punctuation */
582
#define _MY_CTR 040 /* Control character */
583
#define _MY_B 0100 /* Blank */
584
#define _MY_X 0200 /* heXadecimal digit */
587
#define my_isascii(c) (!((c) & ~0177))
588
#define my_toascii(c) ((c) & 0177)
589
#define my_tocntrl(c) ((c) & 31)
590
#define my_toprint(c) ((c) | 64)
591
#define my_toupper(s,c) (char) ((s)->to_upper[(uchar) (c)])
592
#define my_tolower(s,c) (char) ((s)->to_lower[(uchar) (c)])
593
#define my_isalpha(s, c) (((s)->ctype+1)[(uchar) (c)] & (_MY_U | _MY_L))
594
#define my_isupper(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_U)
595
#define my_islower(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_L)
596
#define my_isdigit(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_NMR)
597
#define my_isxdigit(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_X)
598
#define my_isalnum(s, c) (((s)->ctype+1)[(uchar) (c)] & (_MY_U | _MY_L | _MY_NMR))
599
#define my_isspace(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_SPC)
600
#define my_ispunct(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_PNT)
601
#define my_isprint(s, c) (((s)->ctype+1)[(uchar) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR | _MY_B))
602
#define my_isgraph(s, c) (((s)->ctype+1)[(uchar) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR))
603
#define my_iscntrl(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_CTR)
605
/* Some macros that should be cleaned up a little */
606
#define my_isvar(s,c) (my_isalnum(s,c) || (c) == '_')
607
#define my_isvar_start(s,c) (my_isalpha(s,c) || (c) == '_')
609
#define my_binary_compare(s) ((s)->state & MY_CS_BINSORT)
610
#define use_strnxfrm(s) ((s)->state & MY_CS_STRNXFRM)
611
#define my_strnxfrm(cs, d, dl, s, sl) \
612
((cs)->coll->strnxfrm((cs), (d), (dl), (dl), (s), (sl), MY_STRXFRM_PAD_WITH_SPACE))
613
#define my_strnncoll(s, a, b, c, d) ((s)->coll->strnncoll((s), (a), (b), (c), (d), 0))
614
#define my_like_range(s, a, b, c, d, e, f, g, h, i, j) \
615
((s)->coll->like_range((s), (a), (b), (c), (d), (e), (f), (g), (h), (i), (j)))
616
#define my_wildcmp(cs,s,se,w,we,e,o,m) ((cs)->coll->wildcmp((cs),(s),(se),(w),(we),(e),(o),(m)))
617
#define my_strcasecmp(s, a, b) ((s)->coll->strcasecmp((s), (a), (b)))
618
#define my_charpos(cs, b, e, num) (cs)->cset->charpos((cs), (const char*) (b), (const char *)(e), (num))
621
#define use_mb(s) ((s)->cset->ismbchar != NULL)
622
#define my_ismbchar(s, a, b) ((s)->cset->ismbchar((s), (a), (b)))
624
#define my_mbcharlen(s, a) ((s)->cset->mbcharlen((s),(a)))
626
#define my_mbcharlen(s, a) 1
629
#define my_caseup_str(s, a) ((s)->cset->caseup_str((s), (a)))
630
#define my_casedn_str(s, a) ((s)->cset->casedn_str((s), (a)))
631
#define my_strntol(s, a, b, c, d, e) ((s)->cset->strntol((s),(a),(b),(c),(d),(e)))
632
#define my_strntoul(s, a, b, c, d, e) ((s)->cset->strntoul((s),(a),(b),(c),(d),(e)))
633
#define my_strntoll(s, a, b, c, d, e) ((s)->cset->strntoll((s),(a),(b),(c),(d),(e)))
634
#define my_strntoull(s, a, b, c,d, e) ((s)->cset->strntoull((s),(a),(b),(c),(d),(e)))
635
#define my_strntod(s, a, b, c, d) ((s)->cset->strntod((s),(a),(b),(c),(d)))
638
/* XXX: still need to take care of this one */
639
#ifdef MY_CHARSET_TIS620
640
#error The TIS620 charset is broken at the moment. Tell tim to fix it.
649
#endif /* _m_ctype_h */