2
* First, the stuff that ends up in the outside-world include file
4
= #define API_EXPORT(type) __declspec(dllexport) type __stdcall
6
= #define API_EXPORT(type) type
9
= typedef off_t regoff_t;
12
= size_t re_nsub; // number of parenthesized subexpressions
13
= const char *re_endp; // end pointer for REG_PEND
14
= struct re_guts *re_g; // none of your business :-)
17
= regoff_t rm_so; // start of match
18
= regoff_t rm_eo; // end of match
22
* internals of regex_t
24
#define MAGIC1 ((('r'^0200)<<8) | 'e')
27
* The internal representation is a *strip*, a sequence of
28
* operators ending with an endmarker. (Some terminology etc. is a
29
* historical relic of earlier versions which used multiple strips.)
30
* Certain oddities in the representation are there to permit running
31
* the machinery backwards; in particular, any deviation from sequential
32
* flow must be marked at both its source and its destination. Some
35
* - OPLUS_ and O_PLUS are *inside* the loop they create.
36
* - OQUEST_ and O_QUEST are *outside* the bypass they create.
37
* - OCH_ and O_CH are *outside* the multi-way branch they create, while
38
* OOR1 and OOR2 are respectively the end and the beginning of one of
39
* the branches. Note that there is an implicit OOR2 following OCH_
40
* and an implicit OOR1 preceding O_CH.
42
* In state representations, an operator's bit is on to signify a state
43
* immediately *preceding* "execution" of that operator.
45
typedef unsigned long sop; /* strip operator */
47
#define OPRMASK 0xf8000000
48
#define OPDMASK 0x07ffffff
49
#define OPSHIFT ((unsigned)27)
50
#define OP(n) ((n)&OPRMASK)
51
#define OPND(n) ((n)&OPDMASK)
52
#define SOP(op, opnd) ((op)|(opnd))
53
/* operators meaning operand */
54
/* (back, fwd are offsets) */
55
#define OEND (1<<OPSHIFT) /* endmarker - */
56
#define OCHAR (2<<OPSHIFT) /* character unsigned char */
57
#define OBOL (3<<OPSHIFT) /* left anchor - */
58
#define OEOL (4<<OPSHIFT) /* right anchor - */
59
#define OANY (5<<OPSHIFT) /* . - */
60
#define OANYOF (6<<OPSHIFT) /* [...] set number */
61
#define OBACK_ (7<<OPSHIFT) /* begin \d paren number */
62
#define O_BACK (8<<OPSHIFT) /* end \d paren number */
63
#define OPLUS_ (9<<OPSHIFT) /* + prefix fwd to suffix */
64
#define O_PLUS (10<<OPSHIFT) /* + suffix back to prefix */
65
#define OQUEST_ (11<<OPSHIFT) /* ? prefix fwd to suffix */
66
#define O_QUEST (12<<OPSHIFT) /* ? suffix back to prefix */
67
#define OLPAREN (13<<OPSHIFT) /* ( fwd to ) */
68
#define ORPAREN (14<<OPSHIFT) /* ) back to ( */
69
#define OCH_ (15<<OPSHIFT) /* begin choice fwd to OOR2 */
70
#define OOR1 (16u<<OPSHIFT) /* | pt. 1 back to OOR1 or OCH_ */
71
#define OOR2 (17u<<OPSHIFT) /* | pt. 2 fwd to OOR2 or O_CH */
72
#define O_CH (18u<<OPSHIFT) /* end choice back to OOR1 */
73
#define OBOW (19u<<OPSHIFT) /* begin word - */
74
#define OEOW (20u<<OPSHIFT) /* end word - */
77
* Structure for [] character-set representation. Character sets are
78
* done as bit vectors, grouped 8 to a byte vector for compactness.
79
* The individual set therefore has both a pointer to the byte vector
80
* and a mask to pick out the relevant bit of each byte. A hash code
81
* simplifies testing whether two sets could be identical.
83
* This will get trickier for multicharacter collating elements. As
84
* preliminary hooks for dealing with such things, we also carry along
85
* a string of multi-character elements, and decide the size of the
86
* vectors at run time.
89
uch *ptr; /* -> uch [csetsize] */
90
uch mask; /* bit within array */
91
uch hash; /* hash code */
93
char *multis; /* -> char[smulti] ab\0cd\0ef\0\0 */
95
/* note that CHadd and CHsub are unsafe, and CHIN doesn't yield 0/1 */
96
#define CHadd(cs, c) ((cs)->ptr[(uch)(c)] |= (cs)->mask, (cs)->hash += (c))
97
#define CHsub(cs, c) ((cs)->ptr[(uch)(c)] &= ~(cs)->mask, (cs)->hash -= (c))
98
#define CHIN(cs, c) ((cs)->ptr[(uch)(c)] & (cs)->mask)
99
#define MCadd(p, cs, cp) mcadd(p, cs, cp) /* regcomp() internal fns */
101
/* stuff for character categories */
102
typedef unsigned char cat_t;
105
* main compiled-expression structure
109
# define MAGIC2 ((('R'^0200)<<8)|'E')
110
sop *strip; /* malloced area for strip */
111
int csetsize; /* number of bits in a cset vector */
112
int ncsets; /* number of csets in use */
113
cset *sets; /* -> cset [ncsets] */
114
uch *setbits; /* -> uch[csetsize][ncsets/CHAR_BIT] */
115
int cflags; /* copy of regcomp() cflags argument */
116
sopno nstates; /* = number of sops */
117
sopno firststate; /* the initial OEND (normally 0) */
118
sopno laststate; /* the final OEND */
119
int iflags; /* internal flags */
120
# define USEBOL 01 /* used ^ */
121
# define USEEOL 02 /* used $ */
122
# define BAD 04 /* something wrong */
123
int nbol; /* number of ^ used */
124
int neol; /* number of $ used */
125
int ncategories; /* how many character categories */
126
cat_t *categories; /* ->catspace[-CHAR_MIN] */
127
char *must; /* match must contain this string */
128
int mlen; /* length of must */
129
size_t nsub; /* copy of re_nsub */
130
int backrefs; /* does it use back references? */
131
sopno nplus; /* how deep does it nest +s? */
132
/* catspace must be last */
133
cat_t catspace[1]; /* actually [NC] */
137
#define OUT (CHAR_MAX+1) /* a non-character value */
138
#define ISWORD(c) (isalnum(c) || (c) == '_')
added
removed
plug-ins/script-fu/re/regex2.h
