ctype.cc File Reference

#include <assert.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>
#include <algorithm>
#include "m_ctype.h"
#include "m_string.h"
#include "my_byteorder.h"
#include "my_inttypes.h"
#include "my_loglevel.h"
#include "my_macros.h"
#include "my_xml.h"
#include "mysys_err.h"

Classes
struct	my_cs_file_section_st
struct	my_cs_file_info

Macros
#define	_CS_MISC   1
#define	_CS_ID   2
#define	_CS_CSNAME   3
#define	_CS_FAMILY   4
#define	_CS_ORDER   5
#define	_CS_COLNAME   6
#define	_CS_FLAG   7
#define	_CS_CHARSET   8
#define	_CS_COLLATION   9
#define	_CS_UPPERMAP   10
#define	_CS_LOWERMAP   11
#define	_CS_UNIMAP   12
#define	_CS_COLLMAP   13
#define	_CS_CTYPEMAP   14
#define	_CS_PRIMARY_ID   15
#define	_CS_BINARY_ID   16
#define	_CS_CSDESCRIPT   17
#define	_CS_UCA_VERSION   100
#define	_CS_CL_SUPPRESS_CONTRACTIONS   101
#define	_CS_CL_OPTIMIZE   102
#define	_CS_CL_SHIFT_AFTER_METHOD   103
#define	_CS_ST_SETTINGS   200
#define	_CS_ST_STRENGTH   201
#define	_CS_ST_ALTERNATE   202
#define	_CS_ST_BACKWARDS   203
#define	_CS_ST_NORMALIZATION   204
#define	_CS_ST_CASE_LEVEL   205
#define	_CS_ST_CASE_FIRST   206
#define	_CS_ST_HIRAGANA_QUATERNARY   207
#define	_CS_ST_NUMERIC   208
#define	_CS_ST_VARIABLE_TOP   209
#define	_CS_ST_MATCH_BOUNDARIES   210
#define	_CS_ST_MATCH_STYLE   211
#define	_CS_RULES   300
#define	_CS_RESET   301
#define	_CS_DIFF1   302
#define	_CS_DIFF2   303
#define	_CS_DIFF3   304
#define	_CS_DIFF4   305
#define	_CS_IDENTICAL   306
#define	_CS_EXP_X   320
#define	_CS_EXP_EXTEND   321
#define	_CS_EXP_DIFF1   322
#define	_CS_EXP_DIFF2   323
#define	_CS_EXP_DIFF3   324
#define	_CS_EXP_DIFF4   325
#define	_CS_EXP_IDENTICAL   326
#define	_CS_A_DIFF1   351
#define	_CS_A_DIFF2   352
#define	_CS_A_DIFF3   353
#define	_CS_A_DIFF4   354
#define	_CS_A_IDENTICAL   355
#define	_CS_CONTEXT   370
#define	_CS_RESET_BEFORE   380
#define	_CS_RESET_FIRST_PRIMARY_IGNORABLE   401
#define	_CS_RESET_LAST_PRIMARY_IGNORABLE   402
#define	_CS_RESET_FIRST_SECONDARY_IGNORABLE   403
#define	_CS_RESET_LAST_SECONDARY_IGNORABLE   404
#define	_CS_RESET_FIRST_TERTIARY_IGNORABLE   405
#define	_CS_RESET_LAST_TERTIARY_IGNORABLE   406
#define	_CS_RESET_FIRST_TRAILING   407
#define	_CS_RESET_LAST_TRAILING   408
#define	_CS_RESET_FIRST_VARIABLE   409
#define	_CS_RESET_LAST_VARIABLE   410
#define	_CS_RESET_FIRST_NON_IGNORABLE   411
#define	_CS_RESET_LAST_NON_IGNORABLE   412
#define	MY_CS_CSDESCR_SIZE   64
#define	MY_CS_CONTEXT_SIZE   64

Typedefs
typedef struct my_cs_file_info	MY_CHARSET_FILE

Functions
static char *	mstr (char *str, const char *src, size_t l1, size_t l2)
static struct my_cs_file_section_st *	cs_file_sec (const char *attr, size_t len)
static void	my_charset_file_reset_charset (MY_CHARSET_FILE *i)
static void	my_charset_file_reset_collation (MY_CHARSET_FILE *i)
static void	my_charset_file_init (MY_CHARSET_FILE *i)
static void	my_charset_file_free (MY_CHARSET_FILE *i)
static int	my_charset_file_tailoring_realloc (MY_CHARSET_FILE *i, size_t newlen)
static int	fill_uchar (uchar *a, uint size, const char *str, size_t len)
static int	fill_uint16 (uint16 *a, uint size, const char *str, size_t len)
static int	tailoring_append (MY_XML_PARSER *st, const char *fmt, size_t len, const char *attr)
static int	tailoring_append2 (MY_XML_PARSER *st, const char *fmt, size_t len1, const char *attr1, size_t len2, const char *attr2)
static size_t	scan_one_character (const char *s, const char *e, my_wc_t *wc)
static int	tailoring_append_abbreviation (MY_XML_PARSER *st, const char *fmt, size_t len, const char *attr)
static int	cs_enter (MY_XML_PARSER *st, const char *attr, size_t len)
static int	cs_leave (MY_XML_PARSER *st, const char *attr, size_t len)
static int	cs_value (MY_XML_PARSER *st, const char *attr, size_t len)
bool	my_parse_charset_xml (MY_CHARSET_LOADER *loader, const char *buf, size_t len)
uint	my_string_repertoire (const CHARSET_INFO *cs, const char *str, size_t length)
uint	my_charset_repertoire (const CHARSET_INFO *cs)
bool	my_charset_is_8bit_pure_ascii (const CHARSET_INFO *cs)
bool	my_charset_is_ascii_compatible (const CHARSET_INFO *cs)
static size_t	my_convert_internal (char *to, size_t to_length, const CHARSET_INFO *to_cs, const char *from, size_t from_length, const CHARSET_INFO *from_cs, uint *errors)
	Convert a string between two character sets. More...
size_t	my_convert (char *to, size_t to_length, const CHARSET_INFO *to_cs, const char *from, size_t from_length, const CHARSET_INFO *from_cs, uint *errors)
	Convert a string between two character sets. More...
uint	my_mbcharlen_ptr (const CHARSET_INFO *cs, const char *s, const char *e)
	Get the length of the first code in given sequence of chars. More...
bool	my_is_prefixidx_cand (const CHARSET_INFO *cs, const char *wildstr, const char *wildend, int escape, int w_many, size_t *prefix_len)
	Identify whether given like pattern looks like a prefix pattern, which can become candidate for index only scan on prefix indexes. More...

Variables
int(*	my_string_stack_guard )(int) = nullptr
static struct my_cs_file_section_st	sec []
static const char *	diff_fmt [5]
static const char *	context_diff_fmt [5]

Macro Definition Documentation

_CS_A_DIFF1

#define _CS_A_DIFF1   351

_CS_A_DIFF2

#define _CS_A_DIFF2   352

_CS_A_DIFF3

#define _CS_A_DIFF3   353

_CS_A_DIFF4

#define _CS_A_DIFF4   354

_CS_A_IDENTICAL

#define _CS_A_IDENTICAL   355

_CS_BINARY_ID

#define _CS_BINARY_ID   16

_CS_CHARSET

#define _CS_CHARSET   8

_CS_CL_OPTIMIZE

#define _CS_CL_OPTIMIZE   102

_CS_CL_SHIFT_AFTER_METHOD

#define _CS_CL_SHIFT_AFTER_METHOD   103

_CS_CL_SUPPRESS_CONTRACTIONS

#define _CS_CL_SUPPRESS_CONTRACTIONS   101

_CS_COLLATION

#define _CS_COLLATION   9

_CS_COLLMAP

#define _CS_COLLMAP   13

_CS_COLNAME

#define _CS_COLNAME   6

_CS_CONTEXT

#define _CS_CONTEXT   370

_CS_CSDESCRIPT

#define _CS_CSDESCRIPT   17

_CS_CSNAME

#define _CS_CSNAME   3

_CS_CTYPEMAP

#define _CS_CTYPEMAP   14

_CS_DIFF1

#define _CS_DIFF1   302

_CS_DIFF2

#define _CS_DIFF2   303

_CS_DIFF3

#define _CS_DIFF3   304

_CS_DIFF4

#define _CS_DIFF4   305

_CS_EXP_DIFF1

#define _CS_EXP_DIFF1   322

_CS_EXP_DIFF2

#define _CS_EXP_DIFF2   323

_CS_EXP_DIFF3

#define _CS_EXP_DIFF3   324

_CS_EXP_DIFF4

#define _CS_EXP_DIFF4   325

_CS_EXP_EXTEND

#define _CS_EXP_EXTEND   321

_CS_EXP_IDENTICAL

#define _CS_EXP_IDENTICAL   326

_CS_EXP_X

#define _CS_EXP_X   320

_CS_FAMILY

#define _CS_FAMILY   4

_CS_FLAG

#define _CS_FLAG   7

_CS_ID

#define _CS_ID   2

_CS_IDENTICAL

#define _CS_IDENTICAL   306

_CS_LOWERMAP

#define _CS_LOWERMAP   11

_CS_MISC

#define _CS_MISC   1

_CS_ORDER

#define _CS_ORDER   5

_CS_PRIMARY_ID

#define _CS_PRIMARY_ID   15

_CS_RESET

#define _CS_RESET   301

_CS_RESET_BEFORE

#define _CS_RESET_BEFORE   380

_CS_RESET_FIRST_NON_IGNORABLE

#define _CS_RESET_FIRST_NON_IGNORABLE   411

_CS_RESET_FIRST_PRIMARY_IGNORABLE

#define _CS_RESET_FIRST_PRIMARY_IGNORABLE   401

_CS_RESET_FIRST_SECONDARY_IGNORABLE

#define _CS_RESET_FIRST_SECONDARY_IGNORABLE   403

_CS_RESET_FIRST_TERTIARY_IGNORABLE

#define _CS_RESET_FIRST_TERTIARY_IGNORABLE   405

_CS_RESET_FIRST_TRAILING

#define _CS_RESET_FIRST_TRAILING   407

_CS_RESET_FIRST_VARIABLE

#define _CS_RESET_FIRST_VARIABLE   409

_CS_RESET_LAST_NON_IGNORABLE

#define _CS_RESET_LAST_NON_IGNORABLE   412

_CS_RESET_LAST_PRIMARY_IGNORABLE

#define _CS_RESET_LAST_PRIMARY_IGNORABLE   402

_CS_RESET_LAST_SECONDARY_IGNORABLE

#define _CS_RESET_LAST_SECONDARY_IGNORABLE   404

_CS_RESET_LAST_TERTIARY_IGNORABLE

#define _CS_RESET_LAST_TERTIARY_IGNORABLE   406

_CS_RESET_LAST_TRAILING

#define _CS_RESET_LAST_TRAILING   408

_CS_RESET_LAST_VARIABLE

#define _CS_RESET_LAST_VARIABLE   410

_CS_RULES

#define _CS_RULES   300

_CS_ST_ALTERNATE

#define _CS_ST_ALTERNATE   202

_CS_ST_BACKWARDS

#define _CS_ST_BACKWARDS   203

_CS_ST_CASE_FIRST

#define _CS_ST_CASE_FIRST   206

_CS_ST_CASE_LEVEL

#define _CS_ST_CASE_LEVEL   205

_CS_ST_HIRAGANA_QUATERNARY

#define _CS_ST_HIRAGANA_QUATERNARY   207

_CS_ST_MATCH_BOUNDARIES

#define _CS_ST_MATCH_BOUNDARIES   210

_CS_ST_MATCH_STYLE

#define _CS_ST_MATCH_STYLE   211

_CS_ST_NORMALIZATION

#define _CS_ST_NORMALIZATION   204

_CS_ST_NUMERIC

#define _CS_ST_NUMERIC   208

_CS_ST_SETTINGS

#define _CS_ST_SETTINGS   200

_CS_ST_STRENGTH

#define _CS_ST_STRENGTH   201

_CS_ST_VARIABLE_TOP

#define _CS_ST_VARIABLE_TOP   209

_CS_UCA_VERSION

#define _CS_UCA_VERSION   100

_CS_UNIMAP

#define _CS_UNIMAP   12

_CS_UPPERMAP

#define _CS_UPPERMAP   10

MY_CS_CONTEXT_SIZE

#define MY_CS_CONTEXT_SIZE   64

MY_CS_CSDESCR_SIZE

#define MY_CS_CSDESCR_SIZE   64

Typedef Documentation

MY_CHARSET_FILE

typedef struct my_cs_file_info MY_CHARSET_FILE

Function Documentation

cs_enter()

static int cs_enter ( MY_XML_PARSER *  st, const char *  attr, size_t  len  )

static

cs_file_sec()

static struct my_cs_file_section_st * cs_file_sec ( const char *  attr, size_t  len  )

static

cs_leave()

static int cs_leave ( MY_XML_PARSER *  st, const char *  attr, size_t  len  )

static

cs_value()

static int cs_value ( MY_XML_PARSER *  st, const char *  attr, size_t  len  )

static

fill_uchar()

static int fill_uchar ( uchar *  a, uint  size, const char *  str, size_t  len  )

static

fill_uint16()

static int fill_uint16 ( uint16 *  a, uint  size, const char *  str, size_t  len  )

static

mstr()

static char * mstr ( char *  str, const char *  src, size_t  l1, size_t  l2  )

static

my_charset_file_free()

static void my_charset_file_free ( MY_CHARSET_FILE *  i )

static

my_charset_file_init()

static void my_charset_file_init ( MY_CHARSET_FILE *  i )

static

my_charset_file_reset_charset()

static void my_charset_file_reset_charset ( MY_CHARSET_FILE *  i )

static

my_charset_file_reset_collation()

static void my_charset_file_reset_collation ( MY_CHARSET_FILE *  i )

static

my_charset_file_tailoring_realloc()

static int my_charset_file_tailoring_realloc ( MY_CHARSET_FILE *  i, size_t  newlen  )

static

my_charset_is_8bit_pure_ascii()

bool my_charset_is_8bit_pure_ascii

(

const CHARSET_INFO *

cs

)

my_charset_is_ascii_compatible()

bool my_charset_is_ascii_compatible

(

const CHARSET_INFO *

cs

)

my_charset_repertoire()

uint my_charset_repertoire

(

const CHARSET_INFO *

cs

)

my_convert()

size_t my_convert	(	char *	to,
		size_t	to_length,
		const CHARSET_INFO *	to_cs,
		const char *	from,
		size_t	from_length,
		const CHARSET_INFO *	from_cs,
		uint *	errors
	)

Convert a string between two character sets.

Optimized for quick copying of ASCII characters in the range 0x00..0x7F. 'to' must be large enough to store (form_length * to_cs->mbmaxlen) bytes.

Parameters

[out]	to	Store result here
	to_length	Size of "to" buffer
	to_cs	Character set of result string
	from	Copy from here
	from_length	Length of the "from" string
	from_cs	Character set of the "from" string
[out]	errors	Number of conversion errors

Returns

Number of bytes copied to 'to' string

my_convert_internal()

static size_t my_convert_internal ( char *  to, size_t  to_length, const CHARSET_INFO *  to_cs, const char *  from, size_t  from_length, const CHARSET_INFO *  from_cs, uint *  errors  )

static

Convert a string between two character sets.

'to' must be large enough to store (form_length * to_cs->mbmaxlen) bytes.

Parameters

[out]	to	Store result here
	to_length	Size of "to" buffer
	to_cs	Character set of result string
	from	Copy from here
	from_length	Length of the "from" string
	from_cs	Character set of the "from" string
[out]	errors	Number of conversion errors

Returns

Number of bytes copied to 'to' string

my_is_prefixidx_cand()

bool my_is_prefixidx_cand	(	const CHARSET_INFO *	cs,
		const char *	wildstr,
		const char *	wildend,
		int	escape,
		int	w_many,
		size_t *	prefix_len
	)

Identify whether given like pattern looks like a prefix pattern, which can become candidate for index only scan on prefix indexes.

Parameters

	cs	Character set and collation pointer
	wildstr	Pointer to LIKE pattern.
	wildend	Pointer to end of LIKE pattern.
	escape	Escape character pattern, typically '\'.
	w_many	'Many characters' pattern, typically ''.
[out]	prefix_len	Length of LIKE pattern.

Returns

Optimization status.

Return values

TRUE	if LIKE pattern can be used for prefix index only scan.
FALSE	else.

my_mbcharlen_ptr()

uint my_mbcharlen_ptr	(	const CHARSET_INFO *	cs,
		const char *	s,
		const char *	e
	)

Get the length of the first code in given sequence of chars.

This func is introduced because we can't determine the length by checking the first byte only for gb18030, so we first try my_mbcharlen, and then my_mbcharlen_2 if necessary to get the length

Parameters

[in]	cs	charset_info
[in]	s	start of the char sequence
[in]	e	end of the char sequence

Returns

The length of the first code, or 0 for invalid code

my_parse_charset_xml()

bool my_parse_charset_xml	(	MY_CHARSET_LOADER *	loader,
		const char *	buf,
		size_t	len
	)

my_string_repertoire()

uint my_string_repertoire	(	const CHARSET_INFO *	cs,
		const char *	str,
		size_t	length
	)

scan_one_character()

static size_t scan_one_character ( const char *  s, const char *  e, my_wc_t *  wc  )

static

tailoring_append()

static int tailoring_append ( MY_XML_PARSER *  st, const char *  fmt, size_t  len, const char *  attr  )

static

tailoring_append2()

static int tailoring_append2 ( MY_XML_PARSER *  st, const char *  fmt, size_t  len1, const char *  attr1, size_t  len2, const char *  attr2  )

static

tailoring_append_abbreviation()

static int tailoring_append_abbreviation ( MY_XML_PARSER *  st, const char *  fmt, size_t  len, const char *  attr  )

static

Variable Documentation

context_diff_fmt

const char* context_diff_fmt[5]

static

Initial value:

= {
    "<%.*s|%.*s", "<<%.*s|%.*s", "<<<%.*s|%.*s", "<<<<%.*s|%.*s", "=%.*s|%.*s"}

diff_fmt

const char* diff_fmt[5]

static

Initial value:

= {"<%.*s", "<<%.*s", "<<<%.*s", "<<<<%.*s",
                                  "=%.*s"}

my_string_stack_guard

int(* my_string_stack_guard) (int)

(

int

)

= nullptr

sec

struct my_cs_file_section_st sec[]

static