MySQL 8.4.2
Source Code Documentation
Macros | Functions | Variables
ctype-ujis.cc File Reference
#include <cassert>
#include <cstddef>
#include <cstdint>
#include "my_compiler.h"
#include "mysql/strings/m_ctype.h"
#include "strings/m_ctype_internals.h"
#include "template_utils.h"

Macros

#define isujis(c)   ((0xa1 <= ((c)&0xff) && ((c)&0xff) <= 0xfe))
 
#define iskata(c)   ((0xa1 <= ((c)&0xff) && ((c)&0xff) <= 0xdf))
 
#define isujis_ss2(c)   (((c)&0xff) == 0x8e)
 
#define isujis_ss3(c)   (((c)&0xff) == 0x8f)
 

Functions

static unsigned ismbchar_ujis (const CHARSET_INFO *cs, const char *p, const char *e)
 
static unsigned mbcharlen_ujis (const CHARSET_INFO *cs, unsigned c)
 
static size_t my_well_formed_len_ujis (const CHARSET_INFO *cs, const char *beg, const char *end, size_t pos, int *error)
 
static size_t my_numcells_eucjp (const CHARSET_INFO *cs, const char *str, const char *str_end)
 
static int my_mb_wc_euc_jp (const CHARSET_INFO *cs, my_wc_t *pwc, const uint8_t *s, const uint8_t *e)
 Scans an UJIS character from the input string and converts to Unicode code point. More...
 
static int my_wc_mb_euc_jp (const CHARSET_INFO *cs, my_wc_t wc, uint8_t *s, uint8_t *e)
 Puts the given Unicode character into an UJIS string. More...
 
static const MY_UNICASE_CHARACTER * get_case_info_for_ch (const CHARSET_INFO *cs, unsigned plane, unsigned page, unsigned offs)
 
static size_t my_casefold_ujis (const CHARSET_INFO *cs, char *src, size_t srclen, char *dst, size_t dstlen, const uint8_t *map, size_t is_upper)
 
size_t my_casedn_ujis (const CHARSET_INFO *cs, char *src, size_t srclen, char *dst, size_t dstlen)
 
size_t my_caseup_ujis (const CHARSET_INFO *cs, char *src, size_t srclen, char *dst, size_t dstlen)
 

Variables

static const uint8_t ctype_ujis [257]
 
static const uint8_t to_lower_ujis []
 
static const uint8_t to_upper_ujis []
 
static const uint8_t sort_order_ujis []
 
static const uint16_t jisx0208_eucjp_to_unicode [65536]
 
static const uint16_t unicode_to_jisx0208_eucjp [65536]
 
static const uint16_t jisx0212_eucjp_to_unicode [65536]
 
static const uint16_t unicode_to_jisx0212_eucjp [65536]
 
static const MY_UNICASE_CHARACTER cA2 [256]
 
static const MY_UNICASE_CHARACTER cA3 [256]
 
static const MY_UNICASE_CHARACTER cA6 [256]
 
static const MY_UNICASE_CHARACTER cA7 [256]
 
static const MY_UNICASE_CHARACTER c8FA6 []
 
static const MY_UNICASE_CHARACTER c8FA7 []
 
static const MY_UNICASE_CHARACTER c8FA9 []
 
static const MY_UNICASE_CHARACTER c8FAA []
 
static const MY_UNICASE_CHARACTER c8FAB []
 
static const MY_UNICASE_CHARACTER * my_caseinfo_pages_ujis [512]
 
static MY_UNICASE_INFO my_caseinfo_ujis = {0x0FFFF, my_caseinfo_pages_ujis}
 
static MY_COLLATION_HANDLER my_collation_ci_handler
 
static MY_CHARSET_HANDLER my_charset_handler
 
CHARSET_INFO my_charset_ujis_japanese_ci
 
CHARSET_INFO my_charset_ujis_bin
 

Macro Definition Documentation

◆ iskata

#define iskata (   c)    ((0xa1 <= ((c)&0xff) && ((c)&0xff) <= 0xdf))

◆ isujis

#define isujis (   c)    ((0xa1 <= ((c)&0xff) && ((c)&0xff) <= 0xfe))

◆ isujis_ss2

#define isujis_ss2 (   c)    (((c)&0xff) == 0x8e)

◆ isujis_ss3

#define isujis_ss3 (   c)    (((c)&0xff) == 0x8f)

Function Documentation

◆ get_case_info_for_ch()

static const MY_UNICASE_CHARACTER * get_case_info_for_ch ( const CHARSET_INFO *  cs,
unsigned  plane,
unsigned  page,
unsigned  offs 
)
static

◆ ismbchar_ujis()

static unsigned ismbchar_ujis ( const CHARSET_INFO *  cs,
const char *  p,
const char *  e 
)
static

◆ mbcharlen_ujis()

static unsigned mbcharlen_ujis ( const CHARSET_INFO *  cs,
unsigned  c 
)
static

◆ my_casedn_ujis()

size_t my_casedn_ujis ( const CHARSET_INFO *  cs,
char *  src,
size_t  srclen,
char *  dst,
size_t  dstlen 
)

◆ my_casefold_ujis()

static size_t my_casefold_ujis ( const CHARSET_INFO *  cs,
char *  src,
size_t  srclen,
char *  dst,
size_t  dstlen,
const uint8_t *  map,
size_t  is_upper 
)
static

◆ my_caseup_ujis()

size_t my_caseup_ujis ( const CHARSET_INFO *  cs,
char *  src,
size_t  srclen,
char *  dst,
size_t  dstlen 
)

◆ my_mb_wc_euc_jp()

static int my_mb_wc_euc_jp ( const CHARSET_INFO *  cs,
my_wc_t *  pwc,
const uint8_t *  s,
const uint8_t *  e 
)
static

Scans an UJIS character from the input string and converts to Unicode code point.

Parameters
[in]csCharacter set info, unused
[out]pwcUnicode code point
[in]sBeginning of the input string
[in]eEnd of the input string
Return values
MY_CS_TOOSMALLIf the string was too short to scan a character
1If a 1-byte character was scanned
2If a 2-byte character was scanned
3If a 3-byte character was scanned
-2If a 2-byte unassigned character was scanned
-3If a 3-byte unassigned character was scanned
MY_CS_ILSEQIf a wrong byte sequence was found

◆ my_numcells_eucjp()

static size_t my_numcells_eucjp ( const CHARSET_INFO *  cs,
const char *  str,
const char *  str_end 
)
static

◆ my_wc_mb_euc_jp()

static int my_wc_mb_euc_jp ( const CHARSET_INFO *  cs,
my_wc_t  wc,
uint8_t *  s,
uint8_t *  e 
)
static

Puts the given Unicode character into an UJIS string.

Parameters
[in]csCharacter set info, unused
[in]wcUnicode code point
[in]sBeginning of the out string
[in]eEnd of the out string
Return values
MY_CS_TOOSMALLIf the string was too short to put a character
1If a 1-byte character was put
2If a 2-byte character was put
MY_CS_ILUNIIf the Unicode character does not exist in UJIS

◆ my_well_formed_len_ujis()

static size_t my_well_formed_len_ujis ( const CHARSET_INFO *  cs,
const char *  beg,
const char *  end,
size_t  pos,
int *  error 
)
static

Variable Documentation

◆ c8FA6

const MY_UNICASE_CHARACTER c8FA6[]
static

◆ c8FA7

const MY_UNICASE_CHARACTER c8FA7[]
static

◆ c8FA9

const MY_UNICASE_CHARACTER c8FA9[]
static

◆ c8FAA

const MY_UNICASE_CHARACTER c8FAA[]
static

◆ c8FAB

const MY_UNICASE_CHARACTER c8FAB[]
static

◆ cA2

const MY_UNICASE_CHARACTER cA2[256]
static

◆ cA3

const MY_UNICASE_CHARACTER cA3[256]
static

◆ cA6

const MY_UNICASE_CHARACTER cA6[256]
static

◆ cA7

const MY_UNICASE_CHARACTER cA7[256]
static

◆ ctype_ujis

const uint8_t ctype_ujis[257]
static
Initial value:
= {
0,
0040, 0040, 0040, 0040, 0040, 0040, 0040, 0040,
0040, 0050, 0050, 0050, 0050, 0050, 0040, 0040,
0040, 0040, 0040, 0040, 0040, 0040, 0040, 0040,
0040, 0040, 0040, 0040, 0040, 0040, 0040, 0040,
0110, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0204, 0204, 0204, 0204, 0204, 0204, 0204, 0204,
0204, 0204, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0201, 0201, 0201, 0201, 0201, 0201, 0001,
0001, 0001, 0001, 0001, 0001, 0001, 0001, 0001,
0001, 0001, 0001, 0001, 0001, 0001, 0001, 0001,
0001, 0001, 0001, 0020, 0020, 0020, 0020, 0020,
0020, 0202, 0202, 0202, 0202, 0202, 0202, 0002,
0002, 0002, 0002, 0002, 0002, 0002, 0002, 0002,
0002, 0002, 0002, 0002, 0002, 0002, 0002, 0002,
0002, 0002, 0002, 0020, 0020, 0020, 0020, 0040,
0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000,
0000, 0000, 0020, 0020, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000,
0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0000, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020, 0020,
0020, 0020, 0020, 0020, 0020, 0020, 0020, 0000,
}

◆ jisx0208_eucjp_to_unicode

const uint16_t jisx0208_eucjp_to_unicode[65536]
static

◆ jisx0212_eucjp_to_unicode

const uint16_t jisx0212_eucjp_to_unicode[65536]
static

◆ my_caseinfo_pages_ujis

const MY_UNICASE_CHARACTER* my_caseinfo_pages_ujis[512]
static

◆ my_caseinfo_ujis

MY_UNICASE_INFO my_caseinfo_ujis = {0x0FFFF, my_caseinfo_pages_ujis}
static

◆ my_charset_handler

MY_CHARSET_HANDLER my_charset_handler
static
Initial value:
= {nullptr,
ismbchar_ujis,
mbcharlen_ujis,
my_numchars_mb,
my_charpos_mb3,
my_well_formed_len_ujis,
my_lengthsp_8bit,
my_numcells_eucjp,
my_mb_wc_euc_jp,
my_wc_mb_euc_jp,
my_mb_ctype_mb,
my_caseup_str_mb,
my_casedn_str_mb,
my_caseup_ujis,
my_casedn_ujis,
my_snprintf_8bit,
my_long10_to_str_8bit,
my_longlong10_to_str_8bit,
my_fill_8bit,
my_strntol_8bit,
my_strntoul_8bit,
my_strntoll_8bit,
my_strntoull_8bit,
my_strntod_8bit,
my_strtoll10_8bit,
my_strntoull10rnd_8bit,
my_scan_8bit}
my_mb_ctype_mb
int my_mb_ctype_mb(const CHARSET_INFO *cs, int *ctype, const uint8_t *s, const uint8_t *e)
Definition: ctype-mb.cc:1322
my_numchars_mb
size_t my_numchars_mb(const CHARSET_INFO *cs, const char *pos, const char *end)
Definition: ctype-mb.cc:315
my_casedn_str_mb
size_t my_casedn_str_mb(const CHARSET_INFO *cs, char *str)
Definition: ctype-mb.cc:58
my_charpos_mb3
size_t my_charpos_mb3(const CHARSET_INFO *cs, const char *pos, const char *end, size_t length)
Definition: ctype-mb.cc:326
my_caseup_str_mb
size_t my_caseup_str_mb(const CHARSET_INFO *cs, char *str)
Definition: ctype-mb.cc:41
my_strtoll10_8bit
long long my_strtoll10_8bit(const CHARSET_INFO *cs, const char *nptr, const char **endptr, int *error)
Definition: ctype-simple.cc:1125
my_strntoull10rnd_8bit
unsigned long long my_strntoull10rnd_8bit(const CHARSET_INFO *cs, const char *str, size_t length, int unsigned_flag, const char **endptr, int *error)
Definition: ctype-simple.cc:1221
my_strntod_8bit
double my_strntod_8bit(const CHARSET_INFO *cs, const char *str, size_t length, const char **end, int *err)
Definition: ctype-simple.cc:636
my_strntoll_8bit
long long my_strntoll_8bit(const CHARSET_INFO *cs, const char *nptr, size_t l, int base, const char **endptr, int *err)
Definition: ctype-simple.cc:461
my_longlong10_to_str_8bit
size_t my_longlong10_to_str_8bit(const CHARSET_INFO *cs, char *dst, size_t len, int radix, long long val)
Definition: ctype-simple.cc:686
my_long10_to_str_8bit
size_t my_long10_to_str_8bit(const CHARSET_INFO *cs, char *dst, size_t len, int radix, long int val)
Definition: ctype-simple.cc:650
my_lengthsp_8bit
size_t my_lengthsp_8bit(const CHARSET_INFO *cs, const char *ptr, size_t length)
Definition: ctype-simple.cc:927
my_strntoull_8bit
unsigned long long my_strntoull_8bit(const CHARSET_INFO *cs, const char *nptr, size_t l, int base, const char **endptr, int *err)
Definition: ctype-simple.cc:539
my_snprintf_8bit
size_t my_snprintf_8bit(const CHARSET_INFO *cs, char *to, size_t n, const char *fmt,...)
Definition: ctype-simple.cc:278
my_fill_8bit
void my_fill_8bit(const CHARSET_INFO *cs, char *s, size_t l, int fill)
Definition: ctype-simple.cc:898
my_strntoul_8bit
unsigned long my_strntoul_8bit(const CHARSET_INFO *cs, const char *nptr, size_t l, int base, const char **endptr, int *err)
Definition: ctype-simple.cc:389
my_strntol_8bit
long my_strntol_8bit(const CHARSET_INFO *cs, const char *nptr, size_t l, int base, const char **endptr, int *err)
Definition: ctype-simple.cc:311
my_scan_8bit
size_t my_scan_8bit(const CHARSET_INFO *cs, const char *str, const char *end, int sq)
Definition: ctype-simple.cc:876
mbcharlen_ujis
static unsigned mbcharlen_ujis(const CHARSET_INFO *cs, unsigned c)
Definition: ctype-ujis.cc:303
my_mb_wc_euc_jp
static int my_mb_wc_euc_jp(const CHARSET_INFO *cs, my_wc_t *pwc, const uint8_t *s, const uint8_t *e)
Scans an UJIS character from the input string and converts to Unicode code point.
Definition: ctype-ujis.cc:33198
my_casedn_ujis
size_t my_casedn_ujis(const CHARSET_INFO *cs, char *src, size_t srclen, char *dst, size_t dstlen)
Definition: ctype-ujis.cc:35769
my_caseup_ujis
size_t my_caseup_ujis(const CHARSET_INFO *cs, char *src, size_t srclen, char *dst, size_t dstlen)
Definition: ctype-ujis.cc:35779
my_numcells_eucjp
static size_t my_numcells_eucjp(const CHARSET_INFO *cs, const char *str, const char *str_end)
Definition: ctype-ujis.cc:360
ismbchar_ujis
static unsigned ismbchar_ujis(const CHARSET_INFO *cs, const char *p, const char *e)
Definition: ctype-ujis.cc:289
my_well_formed_len_ujis
static size_t my_well_formed_len_ujis(const CHARSET_INFO *cs, const char *beg, const char *end, size_t pos, int *error)
Definition: ctype-ujis.cc:316
my_wc_mb_euc_jp
static int my_wc_mb_euc_jp(const CHARSET_INFO *cs, my_wc_t wc, uint8_t *s, uint8_t *e)
Puts the given Unicode character into an UJIS string.
Definition: ctype-ujis.cc:33252

◆ my_charset_ujis_bin

CHARSET_INFO my_charset_ujis_bin

◆ my_charset_ujis_japanese_ci

CHARSET_INFO my_charset_ujis_japanese_ci

◆ my_collation_ci_handler

MY_COLLATION_HANDLER my_collation_ci_handler
static
Initial value:
= {
nullptr,
nullptr,
my_strnncoll_simple,
my_strnncollsp_simple,
my_strnxfrm_mb,
my_strnxfrmlen_simple,
my_like_range_mb,
my_wildcmp_mb,
my_strcasecmp_mb,
my_instr_mb,
my_hash_sort_simple,
my_propagate_simple}
my_instr_mb
unsigned my_instr_mb(const CHARSET_INFO *cs, const char *b, size_t b_length, const char *s, size_t s_length, my_match_t *match, unsigned nmatch)
Definition: ctype-mb.cc:357
my_wildcmp_mb
int my_wildcmp_mb(const CHARSET_INFO *cs, const char *str, const char *str_end, const char *wildstr, const char *wildend, int escape, int w_one, int w_many)
Definition: ctype-mb.cc:308
my_strcasecmp_mb
int my_strcasecmp_mb(const CHARSET_INFO *cs, const char *s, const char *t)
Definition: ctype-mb.cc:194
my_strnxfrm_mb
size_t my_strnxfrm_mb(const CHARSET_INFO *cs, uint8_t *dst, size_t dstlen, unsigned nweights, const uint8_t *src, size_t srclen, unsigned flags)
Definition: ctype-mb.cc:496
my_like_range_mb
bool my_like_range_mb(const CHARSET_INFO *cs, const char *ptr, size_t ptr_length, char escape, char w_one, char w_many, size_t res_length, char *min_str, char *max_str, size_t *min_length, size_t *max_length)
Definition: ctype-mb.cc:657
my_strnncoll_simple
int my_strnncoll_simple(const CHARSET_INFO *cs, const uint8_t *s, size_t slen, const uint8_t *t, size_t tlen, bool t_is_prefix)
Definition: ctype-simple.cc:133
my_propagate_simple
bool my_propagate_simple(const CHARSET_INFO *cs, const uint8_t *str, size_t length)
Definition: ctype-simple.cc:1474
my_strnxfrmlen_simple
size_t my_strnxfrmlen_simple(const CHARSET_INFO *cs, size_t len)
Definition: ctype-simple.cc:62
my_hash_sort_simple
void my_hash_sort_simple(const CHARSET_INFO *cs, const uint8_t *key, size_t len, uint64_t *nr1, uint64_t *nr2)
Definition: ctype-simple.cc:288
my_strnncollsp_simple
int my_strnncollsp_simple(const CHARSET_INFO *cs, const uint8_t *a, size_t a_length, const uint8_t *b, size_t b_length)
Definition: ctype-simple.cc:175

◆ sort_order_ujis

const uint8_t sort_order_ujis[]
static

◆ to_lower_ujis

const uint8_t to_lower_ujis[]
static

◆ to_upper_ujis

const uint8_t to_upper_ujis[]
static

◆ unicode_to_jisx0208_eucjp

const uint16_t unicode_to_jisx0208_eucjp[65536]
static

◆ unicode_to_jisx0212_eucjp

const uint16_t unicode_to_jisx0212_eucjp[65536]
static