oniguruma.h

Mac OS X 平台一文本编辑器的源代码

#define ONIGENC_MBC_MAXLEN_DIST(enc) \
    (ONIGENC_MBC_MAXLEN(enc) > 0 ? ONIGENC_MBC_MAXLEN(enc) \
                                 : ONIG_INFINITE_DISTANCE)
#define ONIGENC_MBC_MINLEN(enc)            1
#define ONIGENC_MBC_TO_CODE(enc,p,e)       m17n_codepoint((enc),(p),(e))
#define ONIGENC_CODE_TO_MBCLEN(enc,code)   m17n_codelen((enc),(code))
#define ONIGENC_CODE_TO_MBC(enc,code,buf)  onigenc_code_to_mbc(enc, code, buf)

#if 0     /* !! not supported !! */
#define ONIGENC_IS_MBC_NEWLINE(enc,p,end)
#define ONIGENC_STEP_BACK(enc,start,s,n)
#endif

#define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) \
        onigenc_is_code_ctype(enc,code,ctype)

#ifdef isblank
# define ONIGENC_IS_CODE_BLANK(enc,code) isblank((int )code)
#else
# define ONIGENC_IS_CODE_BLANK(enc,code) ((code) == ' ' || (code) == '\t')
#endif
#ifdef isgraph
# define ONIGENC_IS_CODE_GRAPH(enc,code) isgraph((int )code)
#else
# define ONIGENC_IS_CODE_GRAPH(enc,code) \
  (isprint((int )code) && !isspace((int )code))
#endif

#define ONIGENC_IS_CODE_PRINT(enc,code)     m17n_isprint(enc,code)
#define ONIGENC_IS_CODE_ALNUM(enc,code)     m17n_isalnum(enc,code)
#define ONIGENC_IS_CODE_ALPHA(enc,code)     m17n_isalpha(enc,code)
#define ONIGENC_IS_CODE_LOWER(enc,code)     m17n_islower(enc,code)
#define ONIGENC_IS_CODE_UPPER(enc,code)     m17n_isupper(enc,code)
#define ONIGENC_IS_CODE_CNTRL(enc,code)     m17n_iscntrl(enc,code)
#define ONIGENC_IS_CODE_PUNCT(enc,code)     m17n_ispunct(enc,code)
#define ONIGENC_IS_CODE_SPACE(enc,code)     m17n_isspace(enc,code)
#define ONIGENC_IS_CODE_DIGIT(enc,code)     m17n_isdigit(enc,code)
#define ONIGENC_IS_CODE_XDIGIT(enc,code)    m17n_isxdigit(enc,code)
#define ONIGENC_IS_CODE_WORD(enc,code)      m17n_iswchar(enc,code)

ONIG_EXTERN
int onigenc_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, int ctype));
ONIG_EXTERN
int onigenc_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, OnigUChar *buf));
ONIG_EXTERN
int onigenc_mbc_to_normalize P_((OnigEncoding enc, OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* buf));
ONIG_EXTERN
int onigenc_is_mbc_ambiguous P_((OnigEncoding enc, OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end));
ONIG_EXTERN
int onigenc_is_allowed_reverse_match P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end));

#else  /* ONIG_RUBY_M17N */

#define ONIGENC_NAME(enc)                      ((enc)->name)

#define ONIGENC_MBC_TO_NORMALIZE(enc,flag,pp,end,buf) \
  (enc)->mbc_to_normalize(flag,(const OnigUChar** )pp,end,buf)
#define ONIGENC_IS_MBC_AMBIGUOUS(enc,flag,pp,end) \
  (enc)->is_mbc_ambiguous(flag,(const OnigUChar** )pp,end)
#define ONIGENC_SUPPORT_AMBIG_FLAG(enc)        ((enc)->support_ambig_flag)
#define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
        (enc)->is_allowed_reverse_match(s,end)
#define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \
        (enc)->left_adjust_char_head(start, s)
#define ONIGENC_GET_ALL_PAIR_AMBIG_CODES(enc,ambig_flag,acs) \
        (enc)->get_all_pair_ambig_codes(ambig_flag,acs)
#define ONIGENC_GET_ALL_COMP_AMBIG_CODES(enc,ambig_flag,acs) \
        (enc)->get_all_comp_ambig_codes(ambig_flag,acs)
#define ONIGENC_STEP_BACK(enc,start,s,n) \
        onigenc_step_back((enc),(start),(s),(n))

#define ONIGENC_MBC_ENC_LEN(enc,p)             (enc)->mbc_enc_len(p)
#define ONIGENC_MBC_MAXLEN(enc)               ((enc)->max_enc_len)
#define ONIGENC_MBC_MAXLEN_DIST(enc)           ONIGENC_MBC_MAXLEN(enc)
#define ONIGENC_MBC_MINLEN(enc)               ((enc)->min_enc_len)
#define ONIGENC_IS_MBC_NEWLINE(enc,p,end)      (enc)->is_mbc_newline((p),(end))
#define ONIGENC_MBC_TO_CODE(enc,p,end)         (enc)->mbc_to_code((p),(end))
#define ONIGENC_CODE_TO_MBCLEN(enc,code)       (enc)->code_to_mbclen(code)
#define ONIGENC_CODE_TO_MBC(enc,code,buf)      (enc)->code_to_mbc(code,buf)

#define ONIGENC_IS_CODE_CTYPE(enc,code,ctype)  (enc)->is_code_ctype(code,ctype)

#define ONIGENC_IS_CODE_NEWLINE(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE)
#define ONIGENC_IS_CODE_GRAPH(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH)
#define ONIGENC_IS_CODE_PRINT(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT)
#define ONIGENC_IS_CODE_ALNUM(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM)
#define ONIGENC_IS_CODE_ALPHA(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA)
#define ONIGENC_IS_CODE_LOWER(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER)
#define ONIGENC_IS_CODE_UPPER(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER)
#define ONIGENC_IS_CODE_CNTRL(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL)
#define ONIGENC_IS_CODE_PUNCT(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT)
#define ONIGENC_IS_CODE_SPACE(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE)
#define ONIGENC_IS_CODE_BLANK(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK)
#define ONIGENC_IS_CODE_DIGIT(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT)
#define ONIGENC_IS_CODE_XDIGIT(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT)
#define ONIGENC_IS_CODE_WORD(enc,code) \
        ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD)

#define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbr,mbr) \
        (enc)->get_ctype_code_range(ctype,sbr,mbr)

ONIG_EXTERN
OnigUChar* onigenc_step_back P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, int n));

#endif /* is not ONIG_RUBY_M17N */


/* encoding API */
ONIG_EXTERN
int onigenc_init P_(());
ONIG_EXTERN
int onigenc_set_default_encoding P_((OnigEncoding enc));
ONIG_EXTERN
OnigEncoding onigenc_get_default_encoding P_(());
ONIG_EXTERN
void  onigenc_set_default_caseconv_table P_((const OnigUChar* table));
ONIG_EXTERN
OnigUChar* onigenc_get_right_adjust_char_head_with_prev P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, const OnigUChar** prev));
ONIG_EXTERN
OnigUChar* onigenc_get_prev_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
ONIG_EXTERN
OnigUChar* onigenc_get_left_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
ONIG_EXTERN
OnigUChar* onigenc_get_right_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
ONIG_EXTERN
int onigenc_strlen P_((OnigEncoding enc, const OnigUChar* p, const OnigUChar* end));
ONIG_EXTERN
int onigenc_strlen_null P_((OnigEncoding enc, const OnigUChar* p));
ONIG_EXTERN
int onigenc_str_bytelen_null P_((OnigEncoding enc, const OnigUChar* p));



/* PART: regular expression */

/* config parameters */
#define ONIG_NREGION                          10
#define ONIG_MAX_BACKREF_NUM                1000
#define ONIG_MAX_REPEAT_NUM               100000
#define ONIG_MAX_MULTI_BYTE_RANGES_NUM      1000
/* constants */
#define ONIG_MAX_ERROR_MESSAGE_LEN            90

typedef unsigned int        OnigOptionType;

#define ONIG_OPTION_DEFAULT            ONIG_OPTION_NONE

/* options */
#define ONIG_OPTION_NONE                 0
#define ONIG_OPTION_IGNORECASE           1L
#define ONIG_OPTION_EXTEND               (ONIG_OPTION_IGNORECASE         << 1)
#define ONIG_OPTION_MULTILINE            (ONIG_OPTION_EXTEND             << 1)
#define ONIG_OPTION_SINGLELINE           (ONIG_OPTION_MULTILINE          << 1)
#define ONIG_OPTION_FIND_LONGEST         (ONIG_OPTION_SINGLELINE         << 1)
#define ONIG_OPTION_FIND_NOT_EMPTY       (ONIG_OPTION_FIND_LONGEST       << 1)
#define ONIG_OPTION_NEGATE_SINGLELINE    (ONIG_OPTION_FIND_NOT_EMPTY     << 1)
#define ONIG_OPTION_DONT_CAPTURE_GROUP   (ONIG_OPTION_NEGATE_SINGLELINE  << 1)
#define ONIG_OPTION_CAPTURE_GROUP        (ONIG_OPTION_DONT_CAPTURE_GROUP << 1)
/* options (search time) */
#define ONIG_OPTION_NOTBOL               (ONIG_OPTION_CAPTURE_GROUP << 1)
#define ONIG_OPTION_NOTEOL               (ONIG_OPTION_NOTBOL << 1)
#define ONIG_OPTION_POSIX_REGION         (ONIG_OPTION_NOTEOL << 1)
#define ONIG_OPTION_MAXBIT               ONIG_OPTION_POSIX_REGION  /* limit */

#define ONIG_OPTION_ON(options,regopt)      ((options) |= (regopt))
#define ONIG_OPTION_OFF(options,regopt)     ((options) &= ~(regopt))
#define ONIG_IS_OPTION_ON(options,option)   ((options) & (option))

/* syntax */
typedef struct {
  unsigned int  op;
  unsigned int  op2;
  unsigned int  behavior;
  OnigOptionType options;    /* default option */
} OnigSyntaxType;

ONIG_EXTERN OnigSyntaxType OnigSyntaxASIS;
ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixBasic;
ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixExtended;
ONIG_EXTERN OnigSyntaxType OnigSyntaxEmacs;
ONIG_EXTERN OnigSyntaxType OnigSyntaxGrep;
ONIG_EXTERN OnigSyntaxType OnigSyntaxGnuRegex;
ONIG_EXTERN OnigSyntaxType OnigSyntaxJava;
ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl;
ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl_NG;
ONIG_EXTERN OnigSyntaxType OnigSyntaxRuby;

/* predefined syntaxes (see regsyntax.c) */
#define ONIG_SYNTAX_ASIS               (&OnigSyntaxASIS)
#define ONIG_SYNTAX_POSIX_BASIC        (&OnigSyntaxPosixBasic)
#define ONIG_SYNTAX_POSIX_EXTENDED     (&OnigSyntaxPosixExtended)
#define ONIG_SYNTAX_EMACS              (&OnigSyntaxEmacs)
#define ONIG_SYNTAX_GREP               (&OnigSyntaxGrep)
#define ONIG_SYNTAX_GNU_REGEX          (&OnigSyntaxGnuRegex)
#define ONIG_SYNTAX_JAVA               (&OnigSyntaxJava)
#define ONIG_SYNTAX_PERL               (&OnigSyntaxPerl)
#define ONIG_SYNTAX_PERL_NG            (&OnigSyntaxPerl_NG)
#define ONIG_SYNTAX_RUBY               (&OnigSyntaxRuby)

/* default syntax */
ONIG_EXTERN OnigSyntaxType*   OnigDefaultSyntax;
#define ONIG_SYNTAX_DEFAULT   OnigDefaultSyntax

/* syntax (operators) */
#define ONIG_SYN_OP_VARIABLE_META_CHARACTERS    (1<<0)
#define ONIG_SYN_OP_DOT_ANYCHAR                 (1<<1)   /* . */
#define ONIG_SYN_OP_ASTERISK_ZERO_INF           (1<<2)   /* * */
#define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF       (1<<3)
#define ONIG_SYN_OP_PLUS_ONE_INF                (1<<4)   /* + */
#define ONIG_SYN_OP_ESC_PLUS_ONE_INF            (1<<5)
#define ONIG_SYN_OP_QMARK_ZERO_ONE              (1<<6)   /* ? */
#define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE          (1<<7)
#define ONIG_SYN_OP_BRACE_INTERVAL              (1<<8)   /* {lower,upper} */
#define ONIG_SYN_OP_ESC_BRACE_INTERVAL          (1<<9)   /* \{lower,upper\} */
#define ONIG_SYN_OP_VBAR_ALT                    (1<<10)   /* | */
#define ONIG_SYN_OP_ESC_VBAR_ALT                (1<<11)  /* \| */
#define ONIG_SYN_OP_LPAREN_SUBEXP               (1<<12)  /* (...)   */
#define ONIG_SYN_OP_ESC_LPAREN_SUBEXP           (1<<13)  /* \(...\) */
#define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR           (1<<14)  /* \A, \Z, \z */
#define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR  (1<<15)  /* \G     */
#define ONIG_SYN_OP_DECIMAL_BACKREF             (1<<16)  /* \num   */
#define ONIG_SYN_OP_BRACKET_CC                  (1<<17)  /* [...]  */
#define ONIG_SYN_OP_ESC_W_WORD                  (1<<18)  /* \w, \W */
#define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END     (1<<19)  /* \<. \> */
#define ONIG_SYN_OP_ESC_B_WORD_BOUND            (1<<20)  /* \b, \B */
#define ONIG_SYN_OP_ESC_S_WHITE_SPACE           (1<<21)  /* \s, \S */
#define ONIG_SYN_OP_ESC_D_DIGIT                 (1<<22)  /* \d, \D */
#define ONIG_SYN_OP_LINE_ANCHOR                 (1<<23)  /* ^, $   */
#define ONIG_SYN_OP_POSIX_BRACKET               (1<<24)  /* [:xxxx:] */
#define ONIG_SYN_OP_QMARK_NON_GREEDY            (1<<25)  /* ??,*?,+?,{n,m}? */
#define ONIG_SYN_OP_ESC_CONTROL_CHARS           (1<<26)  /* \n,\r,\t,\a ... */
#define ONIG_SYN_OP_ESC_C_CONTROL               (1<<27)  /* \cx  */
#define ONIG_SYN_OP_ESC_OCTAL3                  (1<<28)  /* \OOO */
#define ONIG_SYN_OP_ESC_X_HEX2                  (1<<29)  /* \xHH */
#define ONIG_SYN_OP_ESC_X_BRACE_HEX8            (1<<30)  /* \x{7HHHHHHH} */

#define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE        (1<<0)   /* \Q...\E */
#define ONIG_SYN_OP2_QMARK_GROUP_EFFECT         (1<<1)   /* (?...) */
#define ONIG_SYN_OP2_OPTION_PERL                (1<<2)   /* (?imsx),(?-imsx) */
#define ONIG_SYN_OP2_OPTION_RUBY                (1<<3)   /* (?imx), (?-imx)  */
#define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT     (1<<4)   /* ?+,*+,++ */
#define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL   (1<<5)   /* {n,m}+   */
#define ONIG_SYN_OP2_CCLASS_SET_OP              (1<<6)   /* [...&&..[..]..] */
#define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP       (1<<7)   /* (?<name>...) */
#define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF        (1<<8)   /* \k<name> */
#define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL          (1<<9)   /* \g<name>, \g<n> */
#define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY     (1<<10)  /* (?@..),(?@<x>..) */
#define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL  (1<<11)  /* \C-x */
#define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META     (1<<12)  /* \M-x */
#define ONIG_SYN_OP2_ESC_V_VTAB                 (1<<13)  /* \v as VTAB */
#define ONIG_SYN_OP2_ESC_U_HEX4                 (1<<14)  /* \uHHHH */
#define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR         (1<<15)  /* \`, \' */
#define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY  (1<<16)  /* \p{...}, \P{...} */
#define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1<<17)  /* \p{^..}, \P{^..} */
#define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS    (1<<18)  /* \p{IsXDigit} */
#define ONIG_SYN_OP2_ESC_H_XDIGIT               (1<<19)  /* \h, \H */
#define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE         (1<<20)  /* \ */

/* syntax (behavior) */
#define ONIG_SYN_CONTEXT_INDEP_ANCHORS           (1<<31) /* not implemented */
#define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS        (1<<0)  /* ?, *, +, {n,m} */
#define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS      (1<<1)  /* error or ignore */
#define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP    (1<<2)  /* ...)... */
#define ONIG_SYN_ALLOW_INVALID_INTERVAL          (1<<3)  /* {??? */
#define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV       (1<<4)  /* {,n} => {0,n} */
#define ONIG_SYN_STRICT_CHECK_BACKREF            (1<<5)  /* /(\1)/,/\1()/ ..*/
#define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND   (1<<6)  /* (?<=a|bc) */
#define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP        (1<<7)  /* see doc/RE */
#define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1<<8)  /* (?<x>)(?<x>) */
#define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY   (1<<9)  /* a{n}?=(?:a{n})? */

/* syntax (behavior) in char class [...] */
#define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC      (1<<20) /* [^...] */
#define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC          (1<<21) /* [..\w..] etc.. */
#define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC         (1<<22)
#define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC     (1<<23) /* [0-9-a]=[0-9\-a] */
/* syntax (behavior) warning */
#define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED          (1<<24) /* [,-,] */
#define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT    (1<<25) /* (?:a*)+ */

/* meta character specifiers (onig_set_meta_char()) */
#define ONIG_META_CHAR_ESCAPE               0
#define ONIG_META_CHAR_ANYCHAR              1
#define ONIG_META_CHAR_ANYTIME              2
#define ONIG_META_CHAR_ZERO_OR_ONE_TIME     3