m_ctype.h

网页抓取程序

extern CHARSET_INFO my_charset_bin;
extern CHARSET_INFO my_charset_big5_chinese_ci;
extern CHARSET_INFO my_charset_big5_bin;
extern CHARSET_INFO my_charset_cp932_japanese_ci;
extern CHARSET_INFO my_charset_cp932_bin;
extern CHARSET_INFO my_charset_eucjpms_japanese_ci;
extern CHARSET_INFO my_charset_eucjpms_bin;
extern CHARSET_INFO my_charset_euckr_korean_ci;
extern CHARSET_INFO my_charset_euckr_bin;
extern CHARSET_INFO my_charset_gb2312_chinese_ci;
extern CHARSET_INFO my_charset_gb2312_bin;
extern CHARSET_INFO my_charset_gbk_chinese_ci;
extern CHARSET_INFO my_charset_gbk_bin;
extern CHARSET_INFO my_charset_latin1;
extern CHARSET_INFO my_charset_latin1_german2_ci;
extern CHARSET_INFO my_charset_latin1_bin;
extern CHARSET_INFO my_charset_latin2_czech_ci;
extern CHARSET_INFO my_charset_sjis_japanese_ci;
extern CHARSET_INFO my_charset_sjis_bin;
extern CHARSET_INFO my_charset_tis620_thai_ci;
extern CHARSET_INFO my_charset_tis620_bin;
extern CHARSET_INFO my_charset_ucs2_general_ci;
extern CHARSET_INFO my_charset_ucs2_bin;
extern CHARSET_INFO my_charset_ucs2_general_uca;
extern CHARSET_INFO my_charset_ujis_japanese_ci;
extern CHARSET_INFO my_charset_ujis_bin;
extern CHARSET_INFO my_charset_utf8_general_ci;
extern CHARSET_INFO my_charset_utf8_bin;
extern CHARSET_INFO my_charset_cp1250_czech_ci;

/* declarations for simple charsets */
extern int  my_strnxfrm_simple(CHARSET_INFO *, uchar *, uint, const uchar *,
                               uint); 
uint  my_strnxfrmlen_simple(CHARSET_INFO *, uint); 
extern int  my_strnncoll_simple(CHARSET_INFO *, const uchar *, uint,
				const uchar *, uint, my_bool);

extern int  my_strnncollsp_simple(CHARSET_INFO *, const uchar *, uint,
                                  const uchar *, uint,
                                  my_bool diff_if_only_endspace_difference);

extern void my_hash_sort_simple(CHARSET_INFO *cs,
				const uchar *key, uint len,
				ulong *nr1, ulong *nr2); 

extern uint my_lengthsp_8bit(CHARSET_INFO *cs, const char *ptr, uint length);

extern uint my_instr_simple(struct charset_info_st *,
                            const char *b, uint b_length,
                            const char *s, uint s_length,
                            my_match_t *match, uint nmatch);


/* Functions for 8bit */
extern void my_caseup_str_8bit(CHARSET_INFO *, char *);
extern void my_casedn_str_8bit(CHARSET_INFO *, char *);
extern uint my_caseup_8bit(CHARSET_INFO *, char *src, uint srclen,
                                           char *dst, uint dstlen);
extern uint my_casedn_8bit(CHARSET_INFO *, char *src, uint srclen,
                                           char *dst, uint dstlen);

extern int my_strcasecmp_8bit(CHARSET_INFO * cs, const char *, const char *);

int my_mb_wc_8bit(CHARSET_INFO *cs,my_wc_t *wc, const uchar *s,const uchar *e);
int my_wc_mb_8bit(CHARSET_INFO *cs,my_wc_t wc, uchar *s, uchar *e);

ulong my_scan_8bit(CHARSET_INFO *cs, const char *b, const char *e, int sq);

int my_snprintf_8bit(struct charset_info_st *, char *to, uint n,
		     const char *fmt, ...);

long        my_strntol_8bit(CHARSET_INFO *, const char *s, uint l, int base,
			    char **e, int *err);
ulong      my_strntoul_8bit(CHARSET_INFO *, const char *s, uint l, int base,
			    char **e, int *err);
longlong   my_strntoll_8bit(CHARSET_INFO *, const char *s, uint l, int base,
			    char **e, int *err);
ulonglong my_strntoull_8bit(CHARSET_INFO *, const char *s, uint l, int base,
			    char **e, int *err);
double      my_strntod_8bit(CHARSET_INFO *, char *s, uint l,char **e,
			    int *err);
int  my_long10_to_str_8bit(CHARSET_INFO *, char *to, uint l, int radix,
			   long int val);
int my_longlong10_to_str_8bit(CHARSET_INFO *, char *to, uint l, int radix,
			      longlong val);

longlong my_strtoll10_8bit(CHARSET_INFO *cs,
                           const char *nptr, char **endptr, int *error);
longlong my_strtoll10_ucs2(CHARSET_INFO *cs, 
                           const char *nptr, char **endptr, int *error);

void my_fill_8bit(CHARSET_INFO *cs, char* to, uint l, int fill);

my_bool  my_like_range_simple(CHARSET_INFO *cs,
			      const char *ptr, uint ptr_length,
			      pbool escape, pbool w_one, pbool w_many,
			      uint res_length,
			      char *min_str, char *max_str,
			      uint *min_length, uint *max_length);

my_bool  my_like_range_mb(CHARSET_INFO *cs,
			  const char *ptr, uint ptr_length,
			  pbool escape, pbool w_one, pbool w_many,
			  uint res_length,
			  char *min_str, char *max_str,
			  uint *min_length, uint *max_length);

my_bool  my_like_range_ucs2(CHARSET_INFO *cs,
			    const char *ptr, uint ptr_length,
			    pbool escape, pbool w_one, pbool w_many,
			    uint res_length,
			    char *min_str, char *max_str,
			    uint *min_length, uint *max_length);


int my_wildcmp_8bit(CHARSET_INFO *,
		    const char *str,const char *str_end,
		    const char *wildstr,const char *wildend,
		    int escape, int w_one, int w_many);

uint my_numchars_8bit(CHARSET_INFO *, const char *b, const char *e);
uint my_numcells_8bit(CHARSET_INFO *, const char *b, const char *e);
uint my_charpos_8bit(CHARSET_INFO *, const char *b, const char *e, uint pos);
uint my_well_formed_len_8bit(CHARSET_INFO *, const char *b, const char *e,
                             uint pos, int *error);
int my_mbcharlen_8bit(CHARSET_INFO *, uint c);


/* Functions for multibyte charsets */
extern void my_caseup_str_mb(CHARSET_INFO *, char *);
extern void my_casedn_str_mb(CHARSET_INFO *, char *);
extern uint my_caseup_mb(CHARSET_INFO *, char *src, uint srclen,
                                         char *dst, uint dstlen);
extern uint my_casedn_mb(CHARSET_INFO *, char *src, uint srclen,
                                         char *dst, uint dstlen);
extern int my_strcasecmp_mb(CHARSET_INFO * cs,const char *, const char *);

int my_wildcmp_mb(CHARSET_INFO *,
		  const char *str,const char *str_end,
		  const char *wildstr,const char *wildend,
		  int escape, int w_one, int w_many);
uint my_numchars_mb(CHARSET_INFO *, const char *b, const char *e);
uint my_numcells_mb(CHARSET_INFO *, const char *b, const char *e);
uint my_charpos_mb(CHARSET_INFO *, const char *b, const char *e, uint pos);
uint my_well_formed_len_mb(CHARSET_INFO *, const char *b, const char *e,
                           uint pos, int *error);
uint my_instr_mb(struct charset_info_st *,
                 const char *b, uint b_length,
                 const char *s, uint s_length,
                 my_match_t *match, uint nmatch);

int my_wildcmp_unicode(CHARSET_INFO *cs,
                       const char *str, const char *str_end,
                       const char *wildstr, const char *wildend,
                       int escape, int w_one, int w_many,
                       MY_UNICASE_INFO **weights);

extern my_bool my_parse_charset_xml(const char *bug, uint len,
				    int (*add)(CHARSET_INFO *cs));

my_bool my_propagate_simple(CHARSET_INFO *cs, const uchar *str, uint len);
my_bool my_propagate_complex(CHARSET_INFO *cs, const uchar *str, uint len);


#define	_MY_U	01	/* Upper case */
#define	_MY_L	02	/* Lower case */
#define	_MY_NMR	04	/* Numeral (digit) */
#define	_MY_SPC	010	/* Spacing character */
#define	_MY_PNT	020	/* Punctuation */
#define	_MY_CTR	040	/* Control character */
#define	_MY_B	0100	/* Blank */
#define	_MY_X	0200	/* heXadecimal digit */


#define	my_isascii(c)	(!((c) & ~0177))
#define	my_toascii(c)	((c) & 0177)
#define my_tocntrl(c)	((c) & 31)
#define my_toprint(c)	((c) | 64)
#define my_toupper(s,c)	(char) ((s)->to_upper[(uchar) (c)])
#define my_tolower(s,c)	(char) ((s)->to_lower[(uchar) (c)])
#define	my_isalpha(s, c)  (((s)->ctype+1)[(uchar) (c)] & (_MY_U | _MY_L))
#define	my_isupper(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_U)
#define	my_islower(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_L)
#define	my_isdigit(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_NMR)
#define	my_isxdigit(s, c) (((s)->ctype+1)[(uchar) (c)] & _MY_X)
#define	my_isalnum(s, c)  (((s)->ctype+1)[(uchar) (c)] & (_MY_U | _MY_L | _MY_NMR))
#define	my_isspace(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_SPC)
#define	my_ispunct(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_PNT)
#define	my_isprint(s, c)  (((s)->ctype+1)[(uchar) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR | _MY_B))
#define	my_isgraph(s, c)  (((s)->ctype+1)[(uchar) (c)] & (_MY_PNT | _MY_U | _MY_L | _MY_NMR))
#define	my_iscntrl(s, c)  (((s)->ctype+1)[(uchar) (c)] & _MY_CTR)

/* Some macros that should be cleaned up a little */
#define my_isvar(s,c)                 (my_isalnum(s,c) || (c) == '_')
#define my_isvar_start(s,c)           (my_isalpha(s,c) || (c) == '_')

#define my_binary_compare(s)	      ((s)->state  & MY_CS_BINSORT)
#define use_strnxfrm(s)               ((s)->state  & MY_CS_STRNXFRM)
#define my_strnxfrm(s, a, b, c, d)    ((s)->coll->strnxfrm((s), (a), (b), (c), (d)))
#define my_strnncoll(s, a, b, c, d) ((s)->coll->strnncoll((s), (a), (b), (c), (d), 0))
#define my_like_range(s, a, b, c, d, e, f, g, h, i, j) \
   ((s)->coll->like_range((s), (a), (b), (c), (d), (e), (f), (g), (h), (i), (j)))
#define my_wildcmp(cs,s,se,w,we,e,o,m) ((cs)->coll->wildcmp((cs),(s),(se),(w),(we),(e),(o),(m)))
#define my_strcasecmp(s, a, b)        ((s)->coll->strcasecmp((s), (a), (b)))
#define my_charpos(cs, b, e, num)     (cs)->cset->charpos((cs), (const char*) (b), (const char *)(e), (num))


#define use_mb(s)                     ((s)->cset->ismbchar != NULL)
#define my_ismbchar(s, a, b)          ((s)->cset->ismbchar((s), (a), (b)))
#ifdef USE_MB
#define my_mbcharlen(s, a)            ((s)->cset->mbcharlen((s),(a)))
#else
#define my_mbcharlen(s, a)            1
#endif

#define my_caseup_str(s, a)           ((s)->cset->caseup_str((s), (a)))
#define my_casedn_str(s, a)           ((s)->cset->casedn_str((s), (a)))
#define my_strntol(s, a, b, c, d, e)  ((s)->cset->strntol((s),(a),(b),(c),(d),(e)))
#define my_strntoul(s, a, b, c, d, e) ((s)->cset->strntoul((s),(a),(b),(c),(d),(e)))
#define my_strntoll(s, a, b, c, d, e) ((s)->cset->strntoll((s),(a),(b),(c),(d),(e)))
#define my_strntoull(s, a, b, c,d, e) ((s)->cset->strntoull((s),(a),(b),(c),(d),(e)))
#define my_strntod(s, a, b, c, d)     ((s)->cset->strntod((s),(a),(b),(c),(d)))


/* XXX: still need to take care of this one */
#ifdef MY_CHARSET_TIS620
#error The TIS620 charset is broken at the moment.  Tell tim to fix it.
#define USE_TIS620
#include "t_ctype.h"
#endif

#ifdef	__cplusplus
}
#endif

#endif /* _m_ctype_h */