1
by brian
clean slate |
1 |
/* Copyright (C) 2000 MySQL AB
|
2 |
||
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.
|
|
6 |
||
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.
|
|
11 |
||
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 */
|
|
15 |
||
16 |
/*
|
|
17 |
A better inplementation of the UNIX ctype(3) library.
|
|
264.1.12
by Monty Taylor
Fixed ref to my_global.h in mystrings/longlong2str_asm.c. |
18 |
Notes: drizzled/global.h should be included before ctype.h
|
1
by brian
clean slate |
19 |
*/
|
20 |
||
21 |
#ifndef _m_ctype_h
|
|
22 |
#define _m_ctype_h
|
|
23 |
||
548
by Monty Taylor
Moved my_handler to myisam, which is where it actually belongs. |
24 |
#include <stdint.h> |
612.2.4
by Monty Taylor
Moved some defines to config.h. Stopped including config.h directly anywhere. |
25 |
#include <sys/types.h> |
629.1.1
by Monty Taylor
More solaris fixes. |
26 |
#if !defined(__cplusplus)
|
27 |
# include <stdbool.h>
|
|
28 |
#endif
|
|
548
by Monty Taylor
Moved my_handler to myisam, which is where it actually belongs. |
29 |
|
1
by brian
clean slate |
30 |
#ifdef __cplusplus
|
31 |
extern "C" { |
|
32 |
#endif
|
|
33 |
||
34 |
#define MY_CS_NAME_SIZE 32
|
|
35 |
#define MY_CS_CTYPE_TABLE_SIZE 257
|
|
36 |
#define MY_CS_TO_LOWER_TABLE_SIZE 256
|
|
37 |
#define MY_CS_TO_UPPER_TABLE_SIZE 256
|
|
38 |
#define MY_CS_SORT_ORDER_TABLE_SIZE 256
|
|
39 |
#define MY_CS_TO_UNI_TABLE_SIZE 256
|
|
40 |
||
41 |
#define CHARSET_DIR "charsets/"
|
|
42 |
||
612.2.4
by Monty Taylor
Moved some defines to config.h. Stopped including config.h directly anywhere. |
43 |
#define my_wc_t unsigned long
|
1
by brian
clean slate |
44 |
|
45 |
typedef struct unicase_info_st |
|
46 |
{
|
|
206
by Brian Aker
Removed final uint dead types. |
47 |
uint16_t toupper; |
48 |
uint16_t tolower; |
|
49 |
uint16_t sort; |
|
1
by brian
clean slate |
50 |
} MY_UNICASE_INFO; |
51 |
||
52 |
||
53 |
extern MY_UNICASE_INFO *my_unicase_default[256]; |
|
54 |
extern MY_UNICASE_INFO *my_unicase_turkish[256]; |
|
55 |
||
56 |
typedef struct uni_ctype_st |
|
57 |
{
|
|
481
by Brian Aker
Remove all of uchar. |
58 |
unsigned char pctype; |
59 |
unsigned char *ctype; |
|
1
by brian
clean slate |
60 |
} MY_UNI_CTYPE; |
61 |
||
62 |
extern MY_UNI_CTYPE my_uni_ctype[256]; |
|
63 |
||
64 |
/* wm_wc and wc_mb return codes */
|
|
65 |
#define MY_CS_ILSEQ 0 /* Wrong by sequence: wb_wc */ |
|
66 |
#define MY_CS_ILUNI 0 /* Cannot encode Unicode to charset: wc_mb */ |
|
67 |
#define MY_CS_TOOSMALL -101 /* Need at least one byte: wc_mb and mb_wc */ |
|
68 |
#define MY_CS_TOOSMALL2 -102 /* Need at least two bytes: wc_mb and mb_wc */ |
|
69 |
#define MY_CS_TOOSMALL3 -103 /* Need at least three bytes: wc_mb and mb_wc */ |
|
70 |
/* These following three are currently not really used */
|
|
71 |
#define MY_CS_TOOSMALL4 -104 /* Need at least 4 bytes: wc_mb and mb_wc */ |
|
72 |
#define MY_CS_TOOSMALL5 -105 /* Need at least 5 bytes: wc_mb and mb_wc */ |
|
73 |
#define MY_CS_TOOSMALL6 -106 /* Need at least 6 bytes: wc_mb and mb_wc */ |
|
74 |
/* A helper macros for "need at least n bytes" */
|
|
75 |
#define MY_CS_TOOSMALLN(n) (-100-(n))
|
|
76 |
||
77 |
#define MY_SEQ_INTTAIL 1
|
|
78 |
#define MY_SEQ_SPACES 2
|
|
79 |
||
80 |
/* My charsets_list flags */
|
|
81 |
#define MY_CS_COMPILED 1 /* compiled-in sets */ |
|
82 |
#define MY_CS_CONFIG 2 /* sets that have a *.conf file */ |
|
83 |
#define MY_CS_INDEX 4 /* sets listed in the Index file */ |
|
84 |
#define MY_CS_LOADED 8 /* sets that are currently loaded */ |
|
85 |
#define MY_CS_BINSORT 16 /* if binary sort order */ |
|
86 |
#define MY_CS_PRIMARY 32 /* if primary collation */ |
|
87 |
#define MY_CS_STRNXFRM 64 /* if strnxfrm is used for sort */ |
|
88 |
#define MY_CS_UNICODE 128 /* is a charset is full unicode */ |
|
89 |
#define MY_CS_READY 256 /* if a charset is initialized */ |
|
90 |
#define MY_CS_AVAILABLE 512 /* If either compiled-in or loaded*/ |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
91 |
#define MY_CS_CSSORT 1024 /* if case sensitive sort order */ |
92 |
#define MY_CS_HIDDEN 2048 /* don't display in SHOW */ |
|
1
by brian
clean slate |
93 |
#define MY_CS_PUREASCII 4096 /* if a charset is pure ascii */ |
94 |
#define MY_CS_NONASCII 8192 /* if not ASCII-compatible */ |
|
95 |
#define MY_CHARSET_UNDEFINED 0
|
|
96 |
||
97 |
/* Character repertoire flags */
|
|
98 |
#define MY_REPERTOIRE_ASCII 1 /* Pure ASCII U+0000..U+007F */ |
|
99 |
#define MY_REPERTOIRE_EXTENDED 2 /* Extended characters: U+0080..U+FFFF */ |
|
100 |
#define MY_REPERTOIRE_UNICODE30 3 /* ASCII | EXTENDED: U+0000..U+FFFF */ |
|
101 |
||
102 |
/* Flags for strxfrm */
|
|
103 |
#define MY_STRXFRM_LEVEL1 0x00000001 /* for primary weights */ |
|
104 |
#define MY_STRXFRM_LEVEL2 0x00000002 /* for secondary weights */ |
|
105 |
#define MY_STRXFRM_LEVEL3 0x00000004 /* for tertiary weights */ |
|
106 |
#define MY_STRXFRM_LEVEL4 0x00000008 /* fourth level weights */ |
|
107 |
#define MY_STRXFRM_LEVEL5 0x00000010 /* fifth level weights */ |
|
108 |
#define MY_STRXFRM_LEVEL6 0x00000020 /* sixth level weights */ |
|
109 |
#define MY_STRXFRM_LEVEL_ALL 0x0000003F /* Bit OR for the above six */ |
|
110 |
#define MY_STRXFRM_NLEVELS 6 /* Number of possible levels*/ |
|
111 |
||
112 |
#define MY_STRXFRM_PAD_WITH_SPACE 0x00000040 /* if pad result with spaces */ |
|
113 |
#define MY_STRXFRM_UNUSED_00000080 0x00000080 /* for future extensions */ |
|
114 |
||
115 |
#define MY_STRXFRM_DESC_LEVEL1 0x00000100 /* if desc order for level1 */ |
|
116 |
#define MY_STRXFRM_DESC_LEVEL2 0x00000200 /* if desc order for level2 */ |
|
117 |
#define MY_STRXFRM_DESC_LEVEL3 0x00000300 /* if desc order for level3 */ |
|
118 |
#define MY_STRXFRM_DESC_LEVEL4 0x00000800 /* if desc order for level4 */ |
|
119 |
#define MY_STRXFRM_DESC_LEVEL5 0x00001000 /* if desc order for level5 */ |
|
120 |
#define MY_STRXFRM_DESC_LEVEL6 0x00002000 /* if desc order for level6 */ |
|
121 |
#define MY_STRXFRM_DESC_SHIFT 8
|
|
122 |
||
123 |
#define MY_STRXFRM_UNUSED_00004000 0x00004000 /* for future extensions */ |
|
124 |
#define MY_STRXFRM_UNUSED_00008000 0x00008000 /* for future extensions */ |
|
125 |
||
126 |
#define MY_STRXFRM_REVERSE_LEVEL1 0x00010000 /* if reverse order for level1 */ |
|
127 |
#define MY_STRXFRM_REVERSE_LEVEL2 0x00020000 /* if reverse order for level2 */ |
|
128 |
#define MY_STRXFRM_REVERSE_LEVEL3 0x00040000 /* if reverse order for level3 */ |
|
129 |
#define MY_STRXFRM_REVERSE_LEVEL4 0x00080000 /* if reverse order for level4 */ |
|
130 |
#define MY_STRXFRM_REVERSE_LEVEL5 0x00100000 /* if reverse order for level5 */ |
|
131 |
#define MY_STRXFRM_REVERSE_LEVEL6 0x00200000 /* if reverse order for level6 */ |
|
132 |
#define MY_STRXFRM_REVERSE_SHIFT 16
|
|
133 |
||
134 |
||
135 |
typedef struct my_uni_idx_st |
|
136 |
{
|
|
206
by Brian Aker
Removed final uint dead types. |
137 |
uint16_t from; |
138 |
uint16_t to; |
|
481
by Brian Aker
Remove all of uchar. |
139 |
unsigned char *tab; |
1
by brian
clean slate |
140 |
} MY_UNI_IDX; |
141 |
||
142 |
typedef struct |
|
143 |
{
|
|
411
by Brian Aker
Removed legacy bits around enum. |
144 |
uint32_t beg; |
145 |
uint32_t end; |
|
146 |
uint32_t mb_len; |
|
1
by brian
clean slate |
147 |
} my_match_t; |
148 |
||
149 |
enum my_lex_states |
|
150 |
{
|
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
151 |
MY_LEX_START, MY_LEX_CHAR, MY_LEX_IDENT, |
1
by brian
clean slate |
152 |
MY_LEX_IDENT_SEP, MY_LEX_IDENT_START, |
153 |
MY_LEX_REAL, MY_LEX_HEX_NUMBER, MY_LEX_BIN_NUMBER, |
|
154 |
MY_LEX_CMP_OP, MY_LEX_LONG_CMP_OP, MY_LEX_STRING, MY_LEX_COMMENT, MY_LEX_END, |
|
155 |
MY_LEX_OPERATOR_OR_IDENT, MY_LEX_NUMBER_IDENT, MY_LEX_INT_OR_REAL, |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
156 |
MY_LEX_REAL_OR_POINT, MY_LEX_BOOL, MY_LEX_EOL, MY_LEX_ESCAPE, |
157 |
MY_LEX_LONG_COMMENT, MY_LEX_END_LONG_COMMENT, MY_LEX_SEMICOLON, |
|
158 |
MY_LEX_SET_VAR, MY_LEX_USER_END, MY_LEX_HOSTNAME, MY_LEX_SKIP, |
|
1
by brian
clean slate |
159 |
MY_LEX_USER_VARIABLE_DELIMITER, MY_LEX_SYSTEM_VAR, |
160 |
MY_LEX_IDENT_OR_KEYWORD, |
|
235
by Brian Aker
Final bit of NCHAR removed. |
161 |
MY_LEX_IDENT_OR_HEX, MY_LEX_IDENT_OR_BIN, |
1
by brian
clean slate |
162 |
MY_LEX_STRING_OR_DELIMITER
|
163 |
};
|
|
164 |
||
165 |
struct charset_info_st; |
|
166 |
||
167 |
||
168 |
/* See strings/CHARSET_INFO.txt for information about this structure */
|
|
169 |
typedef struct my_collation_handler_st |
|
170 |
{
|
|
276
by Brian Aker
Cleaned out my_bool from strings. |
171 |
bool (*init)(struct charset_info_st *, void *(*alloc)(size_t)); |
1
by brian
clean slate |
172 |
/* Collation routines */
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
173 |
int (*strnncoll)(const struct charset_info_st * const, |
481
by Brian Aker
Remove all of uchar. |
174 |
const unsigned char *, size_t, const unsigned char *, size_t, bool); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
175 |
int (*strnncollsp)(const struct charset_info_st * const, |
481
by Brian Aker
Remove all of uchar. |
176 |
const unsigned char *, size_t, const unsigned char *, size_t, |
276
by Brian Aker
Cleaned out my_bool from strings. |
177 |
bool diff_if_only_endspace_difference); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
178 |
size_t (*strnxfrm)(const struct charset_info_st * const, |
481
by Brian Aker
Remove all of uchar. |
179 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
180 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
181 |
size_t (*strnxfrmlen)(const struct charset_info_st * const, size_t); |
276
by Brian Aker
Cleaned out my_bool from strings. |
182 |
bool (*like_range)(const struct charset_info_st * const, |
77.1.95
by Monty Taylor
Fixed silly my_bool==char nonsense. |
183 |
const char *s, size_t s_length, |
184 |
char escape, char w_one, char w_many, |
|
185 |
size_t res_length, |
|
186 |
char *min_str, char *max_str, |
|
187 |
size_t *min_len, size_t *max_len); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
188 |
int (*wildcmp)(const struct charset_info_st * const, |
1
by brian
clean slate |
189 |
const char *str,const char *str_end, |
190 |
const char *wildstr,const char *wildend, |
|
191 |
int escape,int w_one, int w_many); |
|
192 |
||
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
193 |
int (*strcasecmp)(const struct charset_info_st * const, const char *, const char *); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
194 |
|
411
by Brian Aker
Removed legacy bits around enum. |
195 |
uint32_t (*instr)(const struct charset_info_st * const, |
1
by brian
clean slate |
196 |
const char *b, size_t b_length, |
197 |
const char *s, size_t s_length, |
|
411
by Brian Aker
Removed legacy bits around enum. |
198 |
my_match_t *match, uint32_t nmatch); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
199 |
|
1
by brian
clean slate |
200 |
/* Hash calculation */
|
481
by Brian Aker
Remove all of uchar. |
201 |
void (*hash_sort)(const struct charset_info_st *cs, const unsigned char *key, size_t len, |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
202 |
uint32_t *nr1, uint32_t *nr2); |
481
by Brian Aker
Remove all of uchar. |
203 |
bool (*propagate)(const struct charset_info_st *cs, const unsigned char *str, size_t len); |
1
by brian
clean slate |
204 |
} MY_COLLATION_HANDLER; |
205 |
||
206 |
extern MY_COLLATION_HANDLER my_collation_mb_bin_handler; |
|
207 |
extern MY_COLLATION_HANDLER my_collation_8bit_bin_handler; |
|
208 |
extern MY_COLLATION_HANDLER my_collation_8bit_simple_ci_handler; |
|
209 |
extern MY_COLLATION_HANDLER my_collation_ucs2_uca_handler; |
|
210 |
||
211 |
/* Some typedef to make it easy for C++ to make function pointers */
|
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
212 |
typedef int (*my_charset_conv_mb_wc)(const struct charset_info_st * const, my_wc_t *, |
481
by Brian Aker
Remove all of uchar. |
213 |
const unsigned char *, const unsigned char *); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
214 |
typedef int (*my_charset_conv_wc_mb)(const struct charset_info_st * const, my_wc_t, |
481
by Brian Aker
Remove all of uchar. |
215 |
unsigned char *, unsigned char *); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
216 |
typedef size_t (*my_charset_conv_case)(const struct charset_info_st * const, |
1
by brian
clean slate |
217 |
char *, size_t, char *, size_t); |
218 |
||
219 |
||
220 |
/* See strings/CHARSET_INFO.txt about information on this structure */
|
|
221 |
typedef struct my_charset_handler_st |
|
222 |
{
|
|
276
by Brian Aker
Cleaned out my_bool from strings. |
223 |
bool (*init)(struct charset_info_st *, void *(*alloc)(size_t)); |
1
by brian
clean slate |
224 |
/* Multibyte routines */
|
411
by Brian Aker
Removed legacy bits around enum. |
225 |
uint32_t (*ismbchar)(const struct charset_info_st * const, const char *, const char *); |
226 |
uint32_t (*mbcharlen)(const struct charset_info_st * const, uint32_t c); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
227 |
size_t (*numchars)(const struct charset_info_st * const, const char *b, const char *e); |
228 |
size_t (*charpos)(const struct charset_info_st * const, const char *b, const char *e, |
|
1
by brian
clean slate |
229 |
size_t pos); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
230 |
size_t (*well_formed_len)(const struct charset_info_st * const, |
1
by brian
clean slate |
231 |
const char *b,const char *e, |
232 |
size_t nchars, int *error); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
233 |
size_t (*lengthsp)(const struct charset_info_st * const, const char *ptr, size_t length); |
234 |
size_t (*numcells)(const struct charset_info_st * const, const char *b, const char *e); |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
235 |
|
1
by brian
clean slate |
236 |
/* Unicode conversion */
|
237 |
my_charset_conv_mb_wc mb_wc; |
|
238 |
my_charset_conv_wc_mb wc_mb; |
|
239 |
||
240 |
/* CTYPE scanner */
|
|
236.3.9
by Andrey Hristov
- Fix build of exotic, mostly non-western, charsets (--with-extra-charsets) |
241 |
int (*ctype)(const struct charset_info_st *cs, int *ctype, |
481
by Brian Aker
Remove all of uchar. |
242 |
const unsigned char *s, const unsigned char *e); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
243 |
|
1
by brian
clean slate |
244 |
/* Functions for case and sort conversion */
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
245 |
size_t (*caseup_str)(const struct charset_info_st * const, char *); |
246 |
size_t (*casedn_str)(const struct charset_info_st * const, char *); |
|
1
by brian
clean slate |
247 |
|
248 |
my_charset_conv_case caseup; |
|
249 |
my_charset_conv_case casedn; |
|
250 |
||
251 |
/* Charset dependant snprintf() */
|
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
252 |
size_t (*snprintf)(const struct charset_info_st * const, char *to, size_t n, |
1
by brian
clean slate |
253 |
const char *fmt, |
212.5.26
by Monty Taylor
Removed my_attribute. Renaming __attribute__((format(x,y,z))) to ATTRIBUTE_FORMAT(x,y,z) is retarded. So we don't do it anymore. |
254 |
...) __attribute__((format(printf, 4, 5))); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
255 |
size_t (*long10_to_str)(const struct charset_info_st * const, char *to, size_t n, |
1
by brian
clean slate |
256 |
int radix, long int val); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
257 |
size_t (*int64_t10_to_str)(const struct charset_info_st * const, char *to, size_t n, |
152
by Brian Aker
longlong replacement |
258 |
int radix, int64_t val); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
259 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
260 |
void (*fill)(const struct charset_info_st * const, char *to, size_t len, int fill); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
261 |
|
1
by brian
clean slate |
262 |
/* String-to-number conversion routines */
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
263 |
long (*strntol)(const struct charset_info_st * const, const char *s, size_t l, |
264 |
int base, char **e, int *err); |
|
612.2.4
by Monty Taylor
Moved some defines to config.h. Stopped including config.h directly anywhere. |
265 |
unsigned long (*strntoul)(const struct charset_info_st * const, const char *s, size_t l, |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
266 |
int base, char **e, int *err); |
267 |
int64_t (*strntoll)(const struct charset_info_st * const, const char *s, size_t l, |
|
268 |
int base, char **e, int *err); |
|
269 |
uint64_t (*strntoull)(const struct charset_info_st * const, const char *s, size_t l, |
|
270 |
int base, char **e, int *err); |
|
271 |
double (*strntod)(const struct charset_info_st * const, char *s, size_t l, char **e, |
|
1
by brian
clean slate |
272 |
int *err); |
236.3.9
by Andrey Hristov
- Fix build of exotic, mostly non-western, charsets (--with-extra-charsets) |
273 |
int64_t (*strtoll10)(const struct charset_info_st *cs, |
1
by brian
clean slate |
274 |
const char *nptr, char **endptr, int *error); |
236.3.9
by Andrey Hristov
- Fix build of exotic, mostly non-western, charsets (--with-extra-charsets) |
275 |
uint64_t (*strntoull10rnd)(const struct charset_info_st *cs, |
1
by brian
clean slate |
276 |
const char *str, size_t length, |
277 |
int unsigned_fl, |
|
278 |
char **endptr, int *error); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
279 |
size_t (*scan)(const struct charset_info_st * const, const char *b, const char *e, |
1
by brian
clean slate |
280 |
int sq); |
281 |
} MY_CHARSET_HANDLER; |
|
282 |
||
283 |
extern MY_CHARSET_HANDLER my_charset_8bit_handler; |
|
284 |
extern MY_CHARSET_HANDLER my_charset_ucs2_handler; |
|
285 |
||
286 |
||
287 |
/* See strings/CHARSET_INFO.txt about information on this structure */
|
|
288 |
typedef struct charset_info_st |
|
289 |
{
|
|
411
by Brian Aker
Removed legacy bits around enum. |
290 |
uint32_t number; |
291 |
uint32_t primary_number; |
|
292 |
uint32_t binary_number; |
|
293 |
uint32_t state; |
|
1
by brian
clean slate |
294 |
const char *csname; |
295 |
const char *name; |
|
296 |
const char *comment; |
|
297 |
const char *tailoring; |
|
481
by Brian Aker
Remove all of uchar. |
298 |
unsigned char *ctype; |
299 |
unsigned char *to_lower; |
|
300 |
unsigned char *to_upper; |
|
301 |
unsigned char *sort_order; |
|
206
by Brian Aker
Removed final uint dead types. |
302 |
uint16_t *contractions; |
303 |
uint16_t **sort_order_big; |
|
304 |
uint16_t *tab_to_uni; |
|
1
by brian
clean slate |
305 |
MY_UNI_IDX *tab_from_uni; |
306 |
MY_UNICASE_INFO **caseinfo; |
|
481
by Brian Aker
Remove all of uchar. |
307 |
unsigned char *state_map; |
308 |
unsigned char *ident_map; |
|
411
by Brian Aker
Removed legacy bits around enum. |
309 |
uint32_t strxfrm_multiply; |
481
by Brian Aker
Remove all of uchar. |
310 |
unsigned char caseup_multiply; |
311 |
unsigned char casedn_multiply; |
|
411
by Brian Aker
Removed legacy bits around enum. |
312 |
uint32_t mbminlen; |
313 |
uint32_t mbmaxlen; |
|
206
by Brian Aker
Removed final uint dead types. |
314 |
uint16_t min_sort_char; |
315 |
uint16_t max_sort_char; /* For LIKE optimization */ |
|
481
by Brian Aker
Remove all of uchar. |
316 |
unsigned char pad_char; |
276
by Brian Aker
Cleaned out my_bool from strings. |
317 |
bool escape_with_backslash_is_dangerous; |
481
by Brian Aker
Remove all of uchar. |
318 |
unsigned char levels_for_compare; |
319 |
unsigned char levels_for_order; |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
320 |
|
1
by brian
clean slate |
321 |
MY_CHARSET_HANDLER *cset; |
322 |
MY_COLLATION_HANDLER *coll; |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
323 |
|
1
by brian
clean slate |
324 |
} CHARSET_INFO; |
520.6.7
by Monty Taylor
Moved a bunch of crap out of common_includes. |
325 |
|
326 |
uint32_t strconvert(const CHARSET_INFO *from_cs, const char *from, |
|
327 |
const CHARSET_INFO *to_cs, char *to, uint32_t to_length, |
|
328 |
uint32_t *errors); |
|
329 |
||
365.2.9
by Monty Taylor
Got rid of all instances of ~0 |
330 |
#define ILLEGAL_CHARSET_INFO_NUMBER (UINT32_MAX)
|
1
by brian
clean slate |
331 |
|
332 |
||
333 |
extern CHARSET_INFO my_charset_bin; |
|
334 |
extern CHARSET_INFO my_charset_filename; |
|
335 |
extern CHARSET_INFO my_charset_latin1; |
|
336 |
extern CHARSET_INFO my_charset_latin1_german2_ci; |
|
337 |
extern CHARSET_INFO my_charset_latin1_bin; |
|
338 |
extern CHARSET_INFO my_charset_latin2_czech_ci; |
|
339 |
extern CHARSET_INFO my_charset_utf8mb3_bin; |
|
340 |
extern CHARSET_INFO my_charset_utf8mb3_general_ci; |
|
341 |
extern CHARSET_INFO my_charset_utf8mb3_unicode_ci; |
|
342 |
extern CHARSET_INFO my_charset_utf8mb4_bin; |
|
343 |
extern CHARSET_INFO my_charset_utf8mb4_general_ci; |
|
344 |
extern CHARSET_INFO my_charset_utf8mb4_unicode_ci; |
|
345 |
||
346 |
#define MY_UTF8MB3 "utf8mb3"
|
|
347 |
#define MY_UTF8MB4 "utf8"
|
|
348 |
#define my_charset_utf8_general_ci my_charset_utf8mb4_general_ci
|
|
349 |
#define my_charset_utf8_bin my_charset_utf8mb4_bin
|
|
350 |
||
351 |
||
352 |
/* declarations for simple charsets */
|
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
353 |
extern size_t my_strnxfrm_simple(const CHARSET_INFO * const, |
481
by Brian Aker
Remove all of uchar. |
354 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
355 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
356 |
size_t my_strnxfrmlen_simple(const CHARSET_INFO * const, size_t); |
481
by Brian Aker
Remove all of uchar. |
357 |
extern int my_strnncoll_simple(const CHARSET_INFO * const, const unsigned char *, size_t, |
358 |
const unsigned char *, size_t, bool); |
|
1
by brian
clean slate |
359 |
|
481
by Brian Aker
Remove all of uchar. |
360 |
extern int my_strnncollsp_simple(const CHARSET_INFO * const, const unsigned char *, size_t, |
361 |
const unsigned char *, size_t, |
|
276
by Brian Aker
Cleaned out my_bool from strings. |
362 |
bool diff_if_only_endspace_difference); |
1
by brian
clean slate |
363 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
364 |
extern void my_hash_sort_simple(const CHARSET_INFO * const cs, |
481
by Brian Aker
Remove all of uchar. |
365 |
const unsigned char *key, size_t len, |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
366 |
uint32_t *nr1, uint32_t *nr2); |
1
by brian
clean slate |
367 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
368 |
extern size_t my_lengthsp_8bit(const CHARSET_INFO * const cs, const char *ptr, size_t length); |
1
by brian
clean slate |
369 |
|
411
by Brian Aker
Removed legacy bits around enum. |
370 |
extern uint32_t my_instr_simple(const CHARSET_INFO * const, |
1
by brian
clean slate |
371 |
const char *b, size_t b_length, |
372 |
const char *s, size_t s_length, |
|
411
by Brian Aker
Removed legacy bits around enum. |
373 |
my_match_t *match, uint32_t nmatch); |
1
by brian
clean slate |
374 |
|
375 |
||
376 |
/* Functions for 8bit */
|
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
377 |
extern size_t my_caseup_str_8bit(const CHARSET_INFO * const, char *); |
378 |
extern size_t my_casedn_str_8bit(const CHARSET_INFO * const, char *); |
|
379 |
extern size_t my_caseup_8bit(const CHARSET_INFO * const, char *src, size_t srclen, |
|
380 |
char *dst, size_t dstlen); |
|
381 |
extern size_t my_casedn_8bit(const CHARSET_INFO * const, char *src, size_t srclen, |
|
382 |
char *dst, size_t dstlen); |
|
383 |
||
384 |
extern int my_strcasecmp_8bit(const CHARSET_INFO * const cs, const char *, const char *); |
|
385 |
||
481
by Brian Aker
Remove all of uchar. |
386 |
int my_mb_wc_8bit(const CHARSET_INFO * const cs,my_wc_t *wc, const unsigned char *s,const unsigned char *e); |
387 |
int my_wc_mb_8bit(const CHARSET_INFO * const cs,my_wc_t wc, unsigned char *s, unsigned char *e); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
388 |
|
481
by Brian Aker
Remove all of uchar. |
389 |
int my_mb_ctype_8bit(const CHARSET_INFO * const,int *, const unsigned char *,const unsigned char *); |
390 |
int my_mb_ctype_mb(const CHARSET_INFO * const,int *, const unsigned char *,const unsigned char *); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
391 |
|
392 |
size_t my_scan_8bit(const CHARSET_INFO * const cs, const char *b, const char *e, int sq); |
|
393 |
||
394 |
size_t my_snprintf_8bit(const CHARSET_INFO * const, char *to, size_t n, |
|
1
by brian
clean slate |
395 |
const char *fmt, ...) |
212.5.26
by Monty Taylor
Removed my_attribute. Renaming __attribute__((format(x,y,z))) to ATTRIBUTE_FORMAT(x,y,z) is retarded. So we don't do it anymore. |
396 |
__attribute__((format(printf, 4, 5))); |
1
by brian
clean slate |
397 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
398 |
long my_strntol_8bit(const CHARSET_INFO * const, const char *s, size_t l, int base, |
1
by brian
clean slate |
399 |
char **e, int *err); |
612.2.4
by Monty Taylor
Moved some defines to config.h. Stopped including config.h directly anywhere. |
400 |
unsigned long my_strntoul_8bit(const CHARSET_INFO * const, const char *s, size_t l, int base, |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
401 |
char **e, int *err); |
402 |
int64_t my_strntoll_8bit(const CHARSET_INFO * const, const char *s, size_t l, int base, |
|
403 |
char **e, int *err); |
|
404 |
uint64_t my_strntoull_8bit(const CHARSET_INFO * const, const char *s, size_t l, int base, |
|
405 |
char **e, int *err); |
|
406 |
double my_strntod_8bit(const CHARSET_INFO * const, char *s, size_t l,char **e, |
|
1
by brian
clean slate |
407 |
int *err); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
408 |
size_t my_long10_to_str_8bit(const CHARSET_INFO * const, char *to, size_t l, int radix, |
1
by brian
clean slate |
409 |
long int val); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
410 |
size_t my_int64_t10_to_str_8bit(const CHARSET_INFO * const, char *to, size_t l, int radix, |
152
by Brian Aker
longlong replacement |
411 |
int64_t val); |
1
by brian
clean slate |
412 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
413 |
int64_t my_strtoll10_8bit(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
414 |
const char *nptr, char **endptr, int *error); |
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
415 |
int64_t my_strtoll10_ucs2(CHARSET_INFO *cs, |
1
by brian
clean slate |
416 |
const char *nptr, char **endptr, int *error); |
417 |
||
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
418 |
uint64_t my_strntoull10rnd_8bit(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
419 |
const char *str, size_t length, int |
420 |
unsigned_fl, char **endptr, int *error); |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
421 |
uint64_t my_strntoull10rnd_ucs2(CHARSET_INFO *cs, |
1
by brian
clean slate |
422 |
const char *str, size_t length, |
423 |
int unsigned_fl, char **endptr, int *error); |
|
424 |
||
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
425 |
void my_fill_8bit(const CHARSET_INFO * const cs, char* to, size_t l, int fill); |
1
by brian
clean slate |
426 |
|
276
by Brian Aker
Cleaned out my_bool from strings. |
427 |
bool my_like_range_simple(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
428 |
const char *ptr, size_t ptr_length, |
77.1.95
by Monty Taylor
Fixed silly my_bool==char nonsense. |
429 |
char escape, char w_one, char w_many, |
1
by brian
clean slate |
430 |
size_t res_length, |
431 |
char *min_str, char *max_str, |
|
432 |
size_t *min_length, size_t *max_length); |
|
433 |
||
276
by Brian Aker
Cleaned out my_bool from strings. |
434 |
bool my_like_range_mb(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
435 |
const char *ptr, size_t ptr_length, |
77.1.95
by Monty Taylor
Fixed silly my_bool==char nonsense. |
436 |
char escape, char w_one, char w_many, |
1
by brian
clean slate |
437 |
size_t res_length, |
438 |
char *min_str, char *max_str, |
|
439 |
size_t *min_length, size_t *max_length); |
|
440 |
||
276
by Brian Aker
Cleaned out my_bool from strings. |
441 |
bool my_like_range_ucs2(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
442 |
const char *ptr, size_t ptr_length, |
77.1.95
by Monty Taylor
Fixed silly my_bool==char nonsense. |
443 |
char escape, char w_one, char w_many, |
1
by brian
clean slate |
444 |
size_t res_length, |
445 |
char *min_str, char *max_str, |
|
446 |
size_t *min_length, size_t *max_length); |
|
447 |
||
276
by Brian Aker
Cleaned out my_bool from strings. |
448 |
bool my_like_range_utf16(const CHARSET_INFO * const cs, |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
449 |
const char *ptr, size_t ptr_length, |
450 |
char escape, char w_one, char w_many, |
|
451 |
size_t res_length, |
|
452 |
char *min_str, char *max_str, |
|
453 |
size_t *min_length, size_t *max_length); |
|
454 |
||
276
by Brian Aker
Cleaned out my_bool from strings. |
455 |
bool my_like_range_utf32(const CHARSET_INFO * const cs, |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
456 |
const char *ptr, size_t ptr_length, |
457 |
char escape, char w_one, char w_many, |
|
458 |
size_t res_length, |
|
459 |
char *min_str, char *max_str, |
|
460 |
size_t *min_length, size_t *max_length); |
|
461 |
||
462 |
||
463 |
int my_wildcmp_8bit(const CHARSET_INFO * const, |
|
1
by brian
clean slate |
464 |
const char *str,const char *str_end, |
465 |
const char *wildstr,const char *wildend, |
|
466 |
int escape, int w_one, int w_many); |
|
467 |
||
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
468 |
int my_wildcmp_bin(const CHARSET_INFO * const, |
1
by brian
clean slate |
469 |
const char *str,const char *str_end, |
470 |
const char *wildstr,const char *wildend, |
|
471 |
int escape, int w_one, int w_many); |
|
472 |
||
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
473 |
size_t my_numchars_8bit(const CHARSET_INFO * const, const char *b, const char *e); |
474 |
size_t my_numcells_8bit(const CHARSET_INFO * const, const char *b, const char *e); |
|
475 |
size_t my_charpos_8bit(const CHARSET_INFO * const, const char *b, const char *e, size_t pos); |
|
476 |
size_t my_well_formed_len_8bit(const CHARSET_INFO * const, const char *b, const char *e, |
|
1
by brian
clean slate |
477 |
size_t pos, int *error); |
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
478 |
typedef void *(*cs_alloc_func)(size_t); |
479 |
bool my_coll_init_simple(CHARSET_INFO *cs, cs_alloc_func alloc); |
|
480 |
bool my_cset_init_8bit(CHARSET_INFO *cs, cs_alloc_func alloc); |
|
411
by Brian Aker
Removed legacy bits around enum. |
481 |
uint32_t my_mbcharlen_8bit(const CHARSET_INFO * const, uint32_t c); |
1
by brian
clean slate |
482 |
|
483 |
||
484 |
/* Functions for multibyte charsets */
|
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
485 |
extern size_t my_caseup_str_mb(const CHARSET_INFO * const, char *); |
486 |
extern size_t my_casedn_str_mb(const CHARSET_INFO * const, char *); |
|
487 |
extern size_t my_caseup_mb(const CHARSET_INFO * const, char *src, size_t srclen, |
|
488 |
char *dst, size_t dstlen); |
|
489 |
extern size_t my_casedn_mb(const CHARSET_INFO * const, char *src, size_t srclen, |
|
490 |
char *dst, size_t dstlen); |
|
491 |
extern int my_strcasecmp_mb(const CHARSET_INFO * const cs, const char *s, const char *t); |
|
1
by brian
clean slate |
492 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
493 |
int my_wildcmp_mb(const CHARSET_INFO * const, |
1
by brian
clean slate |
494 |
const char *str,const char *str_end, |
495 |
const char *wildstr,const char *wildend, |
|
496 |
int escape, int w_one, int w_many); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
497 |
size_t my_numchars_mb(const CHARSET_INFO * const, const char *b, const char *e); |
498 |
size_t my_numcells_mb(const CHARSET_INFO * const, const char *b, const char *e); |
|
499 |
size_t my_charpos_mb(const CHARSET_INFO * const, const char *b, const char *e, size_t pos); |
|
500 |
size_t my_well_formed_len_mb(const CHARSET_INFO * const, const char *b, const char *e, |
|
1
by brian
clean slate |
501 |
size_t pos, int *error); |
411
by Brian Aker
Removed legacy bits around enum. |
502 |
uint32_t my_instr_mb(const CHARSET_INFO * const, |
1
by brian
clean slate |
503 |
const char *b, size_t b_length, |
504 |
const char *s, size_t s_length, |
|
411
by Brian Aker
Removed legacy bits around enum. |
505 |
my_match_t *match, uint32_t nmatch); |
1
by brian
clean slate |
506 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
507 |
int my_strnncoll_mb_bin(const CHARSET_INFO * const cs, |
481
by Brian Aker
Remove all of uchar. |
508 |
const unsigned char *s, size_t slen, |
509 |
const unsigned char *t, size_t tlen, |
|
276
by Brian Aker
Cleaned out my_bool from strings. |
510 |
bool t_is_prefix); |
1
by brian
clean slate |
511 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
512 |
int my_strnncollsp_mb_bin(const CHARSET_INFO * const cs, |
481
by Brian Aker
Remove all of uchar. |
513 |
const unsigned char *a, size_t a_length, |
514 |
const unsigned char *b, size_t b_length, |
|
276
by Brian Aker
Cleaned out my_bool from strings. |
515 |
bool diff_if_only_endspace_difference); |
1
by brian
clean slate |
516 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
517 |
int my_wildcmp_mb_bin(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
518 |
const char *str,const char *str_end, |
519 |
const char *wildstr,const char *wildend, |
|
520 |
int escape, int w_one, int w_many); |
|
521 |
||
647
by Brian Aker
Clean up of code for Solaris (removed dead lock call). |
522 |
int my_strcasecmp_mb_bin(const CHARSET_INFO * const, const char *s, const char *t); |
1
by brian
clean slate |
523 |
|
647
by Brian Aker
Clean up of code for Solaris (removed dead lock call). |
524 |
void my_hash_sort_mb_bin(const CHARSET_INFO *, |
481
by Brian Aker
Remove all of uchar. |
525 |
const unsigned char *key, size_t len, uint32_t *nr1, uint32_t *nr2); |
1
by brian
clean slate |
526 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
527 |
size_t my_strnxfrm_mb(const CHARSET_INFO * const, |
481
by Brian Aker
Remove all of uchar. |
528 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
529 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
1
by brian
clean slate |
530 |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
531 |
int my_wildcmp_unicode(const CHARSET_INFO * const cs, |
1
by brian
clean slate |
532 |
const char *str, const char *str_end, |
533 |
const char *wildstr, const char *wildend, |
|
534 |
int escape, int w_one, int w_many, |
|
535 |
MY_UNICASE_INFO **weights); |
|
536 |
||
276
by Brian Aker
Cleaned out my_bool from strings. |
537 |
extern bool my_parse_charset_xml(const char *bug, size_t len, |
1
by brian
clean slate |
538 |
int (*add)(CHARSET_INFO *cs)); |
539 |
||
481
by Brian Aker
Remove all of uchar. |
540 |
bool my_propagate_simple(const CHARSET_INFO * const cs, const unsigned char *str, size_t len); |
541 |
bool my_propagate_complex(const CHARSET_INFO * const cs, const unsigned char *str, size_t len); |
|
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
542 |
|
543 |
||
612.2.4
by Monty Taylor
Moved some defines to config.h. Stopped including config.h directly anywhere. |
544 |
uint32_t my_string_repertoire(const CHARSET_INFO * const cs, const char *str, unsigned long len); |
276
by Brian Aker
Cleaned out my_bool from strings. |
545 |
bool my_charset_is_ascii_based(const CHARSET_INFO * const cs); |
546 |
bool my_charset_is_8bit_pure_ascii(const CHARSET_INFO * const cs); |
|
1
by brian
clean slate |
547 |
|
548 |
||
411
by Brian Aker
Removed legacy bits around enum. |
549 |
uint32_t my_strxfrm_flag_normalize(uint32_t flags, uint32_t nlevels); |
481
by Brian Aker
Remove all of uchar. |
550 |
void my_strxfrm_desc_and_reverse(unsigned char *str, unsigned char *strend, |
411
by Brian Aker
Removed legacy bits around enum. |
551 |
uint32_t flags, uint32_t level); |
264.2.6
by Andrey Hristov
Constify the usage of CHARSET_INFO almost to the last place in the code. |
552 |
size_t my_strxfrm_pad_desc_and_reverse(const CHARSET_INFO * const cs, |
481
by Brian Aker
Remove all of uchar. |
553 |
unsigned char *str, unsigned char *frmend, unsigned char *strend, |
411
by Brian Aker
Removed legacy bits around enum. |
554 |
uint32_t nweights, uint32_t flags, uint32_t level); |
1
by brian
clean slate |
555 |
|
276
by Brian Aker
Cleaned out my_bool from strings. |
556 |
bool my_charset_is_ascii_compatible(const CHARSET_INFO * const cs); |
1
by brian
clean slate |
557 |
|
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
558 |
int
|
559 |
my_strnncoll_utf8mb3(const CHARSET_INFO * const cs, |
|
560 |
const unsigned char *s, size_t slen, |
|
561 |
const unsigned char *t, size_t tlen, |
|
562 |
bool t_is_prefix); |
|
563 |
||
564 |
int
|
|
565 |
my_strnncollsp_utf8mb3(const CHARSET_INFO * const cs, |
|
566 |
const unsigned char *s, size_t slen, |
|
567 |
const unsigned char *t, size_t tlen, |
|
568 |
bool diff_if_only_endspace_difference); |
|
569 |
||
570 |
size_t
|
|
571 |
my_strnxfrm_utf8mb3(const CHARSET_INFO * const cs, |
|
572 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
|
573 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
574 |
||
575 |
size_t
|
|
576 |
my_strnxfrmlen_utf8mb3(const CHARSET_INFO * const cs, size_t len); |
|
577 |
||
578 |
size_t my_strnxfrm_utf8mb3_bin(const CHARSET_INFO * const cs, |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
579 |
unsigned char *dst, size_t dstlen, |
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
580 |
uint32_t nweights, const unsigned char *src, |
581 |
size_t srclen, uint32_t flags); |
|
582 |
/*
|
|
583 |
Compare 0-terminated UTF8 strings.
|
|
584 |
||
585 |
SYNOPSIS
|
|
586 |
my_strcasecmp_utf8mb3()
|
|
587 |
cs character set handler
|
|
588 |
s First 0-terminated string to compare
|
|
589 |
t Second 0-terminated string to compare
|
|
590 |
||
591 |
IMPLEMENTATION
|
|
592 |
||
593 |
RETURN
|
|
594 |
- negative number if s < t
|
|
595 |
- positive number if s > t
|
|
596 |
- 0 is the strings are equal
|
|
597 |
*/
|
|
598 |
int
|
|
599 |
my_strcasecmp_utf8mb3(const CHARSET_INFO * const cs, |
|
600 |
const char *s, const char *t); |
|
601 |
||
602 |
int
|
|
603 |
my_strnncoll_utf8mb3_cs(CHARSET_INFO *cs, |
|
604 |
const unsigned char *s, size_t slen, |
|
605 |
const unsigned char *t, size_t tlen, |
|
606 |
bool t_is_prefix); |
|
607 |
||
608 |
int
|
|
609 |
my_strnncollsp_utf8mb3_cs(CHARSET_INFO *cs, |
|
610 |
const unsigned char *s, size_t slen, |
|
611 |
const unsigned char *t, size_t tlen, |
|
612 |
bool diff_if_only_endspace_difference); |
|
613 |
||
614 |
int
|
|
615 |
my_wildcmp_utf8mb3(const CHARSET_INFO * const cs, |
|
616 |
const char *str,const char *str_end, |
|
617 |
const char *wildstr,const char *wildend, |
|
618 |
int escape, int w_one, int w_many); |
|
619 |
||
620 |
unsigned int |
|
621 |
my_ismbchar_utf8mb3(const CHARSET_INFO * const cs, |
|
622 |
const char *b, const char *e); |
|
623 |
||
624 |
int
|
|
625 |
my_wc_mb_filename(const CHARSET_INFO * const, |
|
626 |
my_wc_t wc, unsigned char *s, unsigned char *e); |
|
627 |
||
628 |
int
|
|
629 |
my_mb_wc_filename(const CHARSET_INFO * const, |
|
630 |
my_wc_t *pwc, const unsigned char *s, const unsigned char *e); |
|
631 |
||
632 |
size_t
|
|
633 |
my_caseup_str_utf8mb3(const CHARSET_INFO * const cs, char *src); |
|
634 |
size_t
|
|
635 |
my_casedn_str_utf8mb3(const CHARSET_INFO * const cs, char *src); |
|
636 |
size_t
|
|
637 |
my_caseup_utf8mb3(const CHARSET_INFO * const cs, char *src, size_t srclen, |
|
638 |
char *dst, size_t dstlen); |
|
639 |
size_t
|
|
640 |
my_casedn_utf8mb3(const CHARSET_INFO * const cs, char *src, size_t srclen, |
|
641 |
char *dst, size_t dstlen); |
|
642 |
||
643 |
int my_mb_wc_utf8mb3(const CHARSET_INFO * const, my_wc_t * pwc, |
|
644 |
const unsigned char *s, const unsigned char *e); |
|
645 |
int my_wc_mb_utf8mb3(const CHARSET_INFO * const, |
|
646 |
my_wc_t wc, unsigned char *r, unsigned char *e); |
|
647 |
||
648 |
unsigned int |
|
649 |
my_mbcharlen_utf8mb3(const CHARSET_INFO * const, uint32_t c); |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
650 |
|
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
651 |
unsigned int my_ismbchar_utf8mb4(const CHARSET_INFO * const cs, const char *b, const char *e); |
652 |
unsigned int my_mbcharlen_utf8mb4(const CHARSET_INFO * const, uint32_t c); |
|
653 |
||
654 |
size_t my_strnxfrmlen_utf8mb4(const CHARSET_INFO * const, size_t len); |
|
655 |
size_t
|
|
656 |
my_strnxfrm_utf8mb4(const CHARSET_INFO * const cs, |
|
657 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
|
658 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
659 |
||
660 |
int
|
|
661 |
my_wildcmp_utf8mb4(const CHARSET_INFO * const cs, |
|
662 |
const char *str, const char *strend, |
|
663 |
const char *wildstr, const char *wildend, |
|
664 |
int escape, int w_one, int w_many); |
|
665 |
int
|
|
666 |
my_strnncollsp_utf8mb4(const CHARSET_INFO * const cs, |
|
667 |
const unsigned char *s, size_t slen, |
|
668 |
const unsigned char *t, size_t tlen, |
|
669 |
bool diff_if_only_endspace_difference); |
|
670 |
int my_strcasecmp_utf8mb4(const CHARSET_INFO * const cs, |
|
671 |
const char *s, const char *t); |
|
672 |
||
673 |
int
|
|
674 |
my_strnncoll_utf8mb4(const CHARSET_INFO * const cs, |
|
675 |
const unsigned char *s, size_t slen, |
|
676 |
const unsigned char *t, size_t tlen, |
|
677 |
bool t_is_prefix); |
|
678 |
||
679 |
int
|
|
680 |
my_mb_wc_utf8mb4(const CHARSET_INFO * const cs, |
|
681 |
my_wc_t * pwc, const unsigned char *s, const unsigned char *e); |
|
682 |
||
683 |
int
|
|
684 |
my_wc_mb_utf8mb4(const CHARSET_INFO * const cs, |
|
685 |
my_wc_t wc, unsigned char *r, unsigned char *e); |
|
686 |
||
687 |
size_t my_caseup_str_utf8mb4(const CHARSET_INFO * const cs, char *src); |
|
688 |
size_t my_casedn_str_utf8mb4(const CHARSET_INFO * const cs, char *src); |
|
689 |
||
690 |
size_t
|
|
691 |
my_caseup_utf8mb4(const CHARSET_INFO * const cs, char *src, size_t srclen, |
|
692 |
char *dst, size_t dstlen); |
|
693 |
size_t
|
|
694 |
my_casedn_utf8mb4(const CHARSET_INFO * const cs, |
|
695 |
char *src, size_t srclen, |
|
696 |
char *dst, size_t dstlen); |
|
697 |
||
698 |
||
699 |
bool my_coll_init_uca(CHARSET_INFO *cs, cs_alloc_func alloc); |
|
700 |
||
701 |
int my_strnncoll_any_uca(const CHARSET_INFO * const cs, |
|
702 |
const unsigned char *s, size_t slen, |
|
703 |
const unsigned char *t, size_t tlen, |
|
704 |
bool t_is_prefix); |
|
705 |
||
706 |
int my_strnncollsp_any_uca(const CHARSET_INFO * const cs, |
|
707 |
const unsigned char *s, size_t slen, |
|
708 |
const unsigned char *t, size_t tlen, |
|
709 |
bool diff_if_only_endspace_difference); |
|
710 |
||
711 |
void my_hash_sort_any_uca(const CHARSET_INFO * const cs, |
|
712 |
const unsigned char *s, size_t slen, |
|
713 |
uint32_t *n1, uint32_t *n2); |
|
714 |
||
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
715 |
size_t my_strnxfrm_any_uca(const CHARSET_INFO * const cs, |
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
716 |
unsigned char *dst, size_t dstlen, uint32_t nweights, |
717 |
const unsigned char *src, size_t srclen, |
|
718 |
uint32_t flags); |
|
719 |
||
720 |
int my_wildcmp_uca(const CHARSET_INFO * const cs, |
|
721 |
const char *str,const char *str_end, |
|
722 |
const char *wildstr,const char *wildend, |
|
723 |
int escape, int w_one, int w_many); |
|
724 |
||
725 |
int my_strnncoll_8bit_bin(const CHARSET_INFO * const, |
|
726 |
const unsigned char *s, size_t slen, |
|
727 |
const unsigned char *t, size_t tlen, |
|
728 |
bool t_is_prefix); |
|
729 |
int my_strnncollsp_8bit_bin(const CHARSET_INFO * const, |
|
660.1.3
by Eric Herman
removed trailing whitespace with simple script: |
730 |
const unsigned char *a, size_t a_length, |
632.1.10
by Monty Taylor
Got rid of Sun Studio warnings. |
731 |
const unsigned char *b, size_t b_length, |
732 |
bool diff_if_only_endspace_difference); |
|
733 |
size_t my_case_str_bin(const CHARSET_INFO * const, char *); |
|
734 |
size_t my_case_bin(const CHARSET_INFO * const, char *, |
|
735 |
size_t srclen, char *, size_t); |
|
736 |
int my_strcasecmp_bin(const CHARSET_INFO * const, |
|
737 |
const char *s, const char *t); |
|
738 |
size_t
|
|
739 |
my_strnxfrm_8bit_bin(const CHARSET_INFO * const cs, |
|
740 |
unsigned char * dst, size_t dstlen, uint32_t nweights, |
|
741 |
const unsigned char *src, size_t srclen, uint32_t flags); |
|
742 |
uint32_t my_instr_bin(const CHARSET_INFO * const, |
|
743 |
const char *b, size_t b_length, |
|
744 |
const char *s, size_t s_length, |
|
745 |
my_match_t *match, uint32_t nmatch); |
|
746 |
size_t my_lengthsp_binary(const CHARSET_INFO * const, |
|
747 |
const char *, size_t length); |
|
748 |
int my_mb_wc_bin(const CHARSET_INFO * const, |
|
749 |
my_wc_t *wc, const unsigned char *str, |
|
750 |
const unsigned char *end); |
|
751 |
int my_wc_mb_bin(const CHARSET_INFO * const, my_wc_t wc, |
|
752 |
unsigned char *str, unsigned char *end); |
|
753 |
void my_hash_sort_8bit_bin(const CHARSET_INFO * const, |
|
754 |
const unsigned char *key, size_t len, |
|
755 |
uint32_t *nr1, uint32_t *nr2); |
|
756 |
bool my_coll_init_8bit_bin(CHARSET_INFO *cs, |
|
757 |
cs_alloc_func); |
|
758 |
int my_strnncoll_binary(const CHARSET_INFO * const, |
|
759 |
const unsigned char *s, size_t slen, |
|
760 |
const unsigned char *t, size_t tlen, |
|
761 |
bool t_is_prefix); |
|
762 |
int my_strnncollsp_binary(const CHARSET_INFO * const cs, |
|
763 |
const unsigned char *s, size_t slen, |
|
764 |
const unsigned char *t, size_t tlen, |
|
765 |
bool); |
|
766 |
||
767 |
||
1
by brian
clean slate |
768 |
#define _MY_U 01 /* Upper case */ |
769 |
#define _MY_L 02 /* Lower case */ |
|
770 |
#define _MY_NMR 04 /* Numeral (digit) */ |
|
771 |
#define _MY_SPC 010 /* Spacing character */ |
|
772 |
#define _MY_PNT 020 /* Punctuation */ |
|
773 |
#define _MY_CTR 040 /* Control character */ |
|
774 |
#define _MY_B 0100 /* Blank */ |
|
775 |
#define _MY_X 0200 /* heXadecimal digit */ |
|
776 |
||
777 |
||
778 |
#define my_isascii(c) (!((c) & ~0177))
|
|
779 |
#define my_toascii(c) ((c) & 0177)
|
|
780 |
#define my_tocntrl(c) ((c) & 31)
|
|
781 |
#define my_toprint(c) ((c) | 64)
|
|
481
by Brian Aker
Remove all of uchar. |
782 |
#define my_toupper(s,c) (char) ((s)->to_upper[(unsigned char) (c)])
|
783 |
#define my_tolower(s,c) (char) ((s)->to_lower[(unsigned char) (c)])
|
|
784 |
#define my_isalpha(s, c) (((s)->ctype+1)[(unsigned char) (c)] & (_MY_U | _MY_L))
|
|
785 |
#define my_isupper(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_U)
|
|
786 |
#define my_islower(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_L)
|
|
787 |
#define my_isdigit(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_NMR)
|
|
788 |
#define my_isxdigit(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_X)
|
|
789 |
#define my_isalnum(s, c) (((s)->ctype+1)[(unsigned char) (c)] & (_MY_U | _MY_L | _MY_NMR))
|
|
790 |
#define my_isspace(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_SPC)
|
|
791 |
#define my_ispunct(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_PNT)
|
|
792 |
#define my_isprint(s, c) (((s)->ctype+1)[(unsigned char) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR | _MY_B))
|
|
793 |
#define my_isgraph(s, c) (((s)->ctype+1)[(unsigned char) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR))
|
|
794 |
#define my_iscntrl(s, c) (((s)->ctype+1)[(unsigned char) (c)] & _MY_CTR)
|
|
1
by brian
clean slate |
795 |
|
796 |
/* Some macros that should be cleaned up a little */
|
|
797 |
#define my_isvar(s,c) (my_isalnum(s,c) || (c) == '_')
|
|
798 |
#define my_isvar_start(s,c) (my_isalpha(s,c) || (c) == '_')
|
|
799 |
||
800 |
#define my_binary_compare(s) ((s)->state & MY_CS_BINSORT)
|
|
801 |
#define use_strnxfrm(s) ((s)->state & MY_CS_STRNXFRM)
|
|
802 |
#define my_strnxfrm(cs, d, dl, s, sl) \
|
|
803 |
((cs)->coll->strnxfrm((cs), (d), (dl), (dl), (s), (sl), MY_STRXFRM_PAD_WITH_SPACE))
|
|
804 |
#define my_strnncoll(s, a, b, c, d) ((s)->coll->strnncoll((s), (a), (b), (c), (d), 0))
|
|
805 |
#define my_like_range(s, a, b, c, d, e, f, g, h, i, j) \
|
|
806 |
((s)->coll->like_range((s), (a), (b), (c), (d), (e), (f), (g), (h), (i), (j)))
|
|
807 |
#define my_wildcmp(cs,s,se,w,we,e,o,m) ((cs)->coll->wildcmp((cs),(s),(se),(w),(we),(e),(o),(m)))
|
|
808 |
#define my_strcasecmp(s, a, b) ((s)->coll->strcasecmp((s), (a), (b)))
|
|
809 |
#define my_charpos(cs, b, e, num) (cs)->cset->charpos((cs), (const char*) (b), (const char *)(e), (num))
|
|
810 |
||
811 |
||
812 |
#define use_mb(s) ((s)->cset->ismbchar != NULL)
|
|
813 |
#define my_ismbchar(s, a, b) ((s)->cset->ismbchar((s), (a), (b)))
|
|
814 |
#ifdef USE_MB
|
|
815 |
#define my_mbcharlen(s, a) ((s)->cset->mbcharlen((s),(a)))
|
|
816 |
#else
|
|
817 |
#define my_mbcharlen(s, a) 1
|
|
818 |
#endif
|
|
819 |
||
820 |
#define my_caseup_str(s, a) ((s)->cset->caseup_str((s), (a)))
|
|
821 |
#define my_casedn_str(s, a) ((s)->cset->casedn_str((s), (a)))
|
|
822 |
#define my_strntol(s, a, b, c, d, e) ((s)->cset->strntol((s),(a),(b),(c),(d),(e)))
|
|
823 |
#define my_strntoul(s, a, b, c, d, e) ((s)->cset->strntoul((s),(a),(b),(c),(d),(e)))
|
|
824 |
#define my_strntoll(s, a, b, c, d, e) ((s)->cset->strntoll((s),(a),(b),(c),(d),(e)))
|
|
825 |
#define my_strntoull(s, a, b, c,d, e) ((s)->cset->strntoull((s),(a),(b),(c),(d),(e)))
|
|
826 |
#define my_strntod(s, a, b, c, d) ((s)->cset->strntod((s),(a),(b),(c),(d)))
|
|
827 |
||
828 |
||
829 |
/* XXX: still need to take care of this one */
|
|
830 |
#ifdef MY_CHARSET_TIS620
|
|
831 |
#error The TIS620 charset is broken at the moment. Tell tim to fix it.
|
|
832 |
#define USE_TIS620
|
|
212.5.19
by Monty Taylor
Coupla small include cleanups. |
833 |
#include <mystrings/t_ctype.h> |
1
by brian
clean slate |
834 |
#endif
|
835 |
||
836 |
#ifdef __cplusplus
|
|
837 |
}
|
|
838 |
#endif
|
|
839 |
||
840 |
#endif /* _m_ctype_h */ |