[perl5.git] / regexp.h

/*    regexp.h
 */

/*
 * Definitions etc. for regexp(3) routines.
 *
 * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
 * not the System V one.
 */


struct regnode {
    U8	flags;
    U8  type;
    U16 next_off;
};

typedef struct regnode regnode;

struct reg_data {
    U32 count;
    U8 *what;
    void* data[1];
};

struct reg_substr_datum {
    I32 min_offset;
    I32 max_offset;
    SV *substr;
};

struct reg_substr_data {
    struct reg_substr_datum data[3];	/* Actual array */
};

typedef struct regexp {
	I32 *startp;
	I32 *endp;
	regnode *regstclass;
#if 0
        SV *anchored_substr;	/* Substring at fixed position wrt start. */
	I32 anchored_offset;	/* Position of it. */
        SV *float_substr;	/* Substring at variable position wrt start. */
	I32 float_min_offset;	/* Minimal position of it. */
	I32 float_max_offset;	/* Maximal position of it. */
        SV *check_substr;	/* Substring to check before matching. */
        I32 check_offset_min;	/* Offset of the above. */
        I32 check_offset_max;	/* Offset of the above. */
#else
        struct reg_substr_data *substrs;
#endif
	char *precomp;		/* pre-compilation regular expression */
        struct reg_data *data;	/* Additional data. */
	char *subbeg;		/* saved or original string 
				   so \digit works forever. */
	I32 sublen;		/* Length of string pointed by subbeg */
	I32 refcnt;
	I32 minlen;		/* mininum possible length of $& */
	I32 prelen;		/* length of precomp */
	U32 nparens;		/* number of parentheses */
	U32 lastparen;		/* last paren matched */
	U32 reganch;		/* Internal use only +
				   Tainted information used by regexec? */
	regnode program[1];	/* Unwarranted chumminess with compiler. */
} regexp;

#define anchored_substr substrs->data[0].substr
#define anchored_offset substrs->data[0].min_offset
#define float_substr substrs->data[1].substr
#define float_min_offset substrs->data[1].min_offset
#define float_max_offset substrs->data[1].max_offset
#define check_substr substrs->data[2].substr
#define check_offset_min substrs->data[2].min_offset
#define check_offset_max substrs->data[2].max_offset

#define ROPT_ANCH		(ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS)
#define ROPT_ANCH_SINGLE	(ROPT_ANCH_BOL|ROPT_ANCH_GPOS)
#define ROPT_ANCH_BOL	 	0x00001
#define ROPT_ANCH_MBOL	 	0x00002
#define ROPT_ANCH_GPOS	 	0x00004
#define ROPT_SKIP		0x00008
#define ROPT_IMPLICIT		0x00010	/* Converted .* to ^.* */
#define ROPT_NOSCAN		0x00020	/* Check-string always at start. */
#define ROPT_GPOS_SEEN		0x00040
#define ROPT_CHECK_ALL		0x00080
#define ROPT_LOOKBEHIND_SEEN	0x00100
#define ROPT_EVAL_SEEN		0x00200
#define ROPT_TAINTED_SEEN	0x00400
#define ROPT_ANCH_SBOL	 	0x00800

/* 0xf800 of reganch is used by PMf_COMPILETIME */

#define ROPT_UTF8		0x10000
#define ROPT_NAUGHTY		0x20000 /* how exponential is this pattern? */
#define ROPT_COPY_DONE		0x40000	/* subbeg is a copy of the string */

#define RX_MATCH_TAINTED(prog)	((prog)->reganch & ROPT_TAINTED_SEEN)
#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
#define RX_MATCH_TAINTED_set(prog, t) ((t) \
				       ? RX_MATCH_TAINTED_on(prog) \
				       : RX_MATCH_TAINTED_off(prog))

#define RX_MATCH_COPIED(prog)		((prog)->reganch & ROPT_COPY_DONE)
#define RX_MATCH_COPIED_on(prog)	((prog)->reganch |= ROPT_COPY_DONE)
#define RX_MATCH_COPIED_off(prog)	((prog)->reganch &= ~ROPT_COPY_DONE)
#define RX_MATCH_COPIED_set(prog,t)	((t) \
					 ? RX_MATCH_COPIED_on(prog) \
					 : RX_MATCH_COPIED_off(prog))

#define REXEC_COPY_STR	1		/* Need to copy the string. */
#define REXEC_CHECKED	2		/* check_substr already checked. */
#define REXEC_SCREAM	4		/* use scream table. */
#define REXEC_IGNOREPOS	8		/* \G matches at start. */
#define REXEC_NOT_FIRST	0x10		/* This is another iteration of //g. */

#define ReREFCNT_inc(re) ((re && re->refcnt++), re)
#define ReREFCNT_dec(re) pregfree(re)

#define FBMcf_TAIL_DOLLAR	1
#define FBMcf_TAIL_Z		2
#define FBMcf_TAIL_z		4
#define FBMcf_TAIL		(FBMcf_TAIL_DOLLAR|FBMcf_TAIL_Z|FBMcf_TAIL_z)

#define FBMrf_MULTILINE	1
Commit	Line	Data
a0d0e21e LW	1	/* regexp.h
	2	*/
	3
378cc40b LW	4	/*
	5	* Definitions etc. for regexp(3) routines.
	6	*
	7	* Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
	8	* not the System V one.
	9	*/
	10
378cc40b	11
c277df42 IZ	12	struct regnode {
	13	U8 flags;
	14	U8 type;
	15	U16 next_off;
	16	};
	17
	18	typedef struct regnode regnode;
	19
c31fac66 GS	20	struct reg_data {
	21	U32 count;
	22	U8 *what;
	23	void* data[1];
	24	};
	25
2779dcf1 IZ	26	struct reg_substr_datum {
	27	I32 min_offset;
	28	I32 max_offset;
	29	SV *substr;
	30	};
	31
	32	struct reg_substr_data {
	33	struct reg_substr_datum data[3]; /* Actual array */
	34	};
	35
378cc40b	36	typedef struct regexp {
cf93c79d IZ	37	I32 *startp;
cf93c79d IZ	38	I32 *endp;
c277df42	39	regnode *regstclass;
2779dcf1	40	#if 0
c277df42 IZ	41	SV anchored_substr; / Substring at fixed position wrt start. */
	42	I32 anchored_offset; /* Position of it. */
	43	SV float_substr; / Substring at variable position wrt start. */
	44	I32 float_min_offset; /* Minimal position of it. */
	45	I32 float_max_offset; /* Maximal position of it. */
	46	SV check_substr; / Substring to check before matching. */
	47	I32 check_offset_min; /* Offset of the above. */
	48	I32 check_offset_max; /* Offset of the above. */
2779dcf1 IZ	49	#else
	50	struct reg_substr_data *substrs;
	51	#endif
cf93c79d	52	char precomp; / pre-compilation regular expression */
c277df42	53	struct reg_data data; / Additional data. */
cf93c79d IZ	54	char subbeg; / saved or original string
	55	so \digit works forever. */
	56	I32 sublen; /* Length of string pointed by subbeg */
	57	I32 refcnt;
	58	I32 minlen; /* mininum possible length of $& */
	59	I32 prelen; /* length of precomp */
	60	U32 nparens; /* number of parentheses */
	61	U32 lastparen; /* last paren matched */
	62	U32 reganch; /* Internal use only +
	63	Tainted information used by regexec? */
c277df42	64	regnode program[1]; /* Unwarranted chumminess with compiler. */
378cc40b LW	65	} regexp;
378cc40b LW	66
2779dcf1 IZ	67	#define anchored_substr substrs->data[0].substr
	68	#define anchored_offset substrs->data[0].min_offset
	69	#define float_substr substrs->data[1].substr
	70	#define float_min_offset substrs->data[1].min_offset
	71	#define float_max_offset substrs->data[1].max_offset
	72	#define check_substr substrs->data[2].substr
	73	#define check_offset_min substrs->data[2].min_offset
	74	#define check_offset_max substrs->data[2].max_offset
	75
c277df42 IZ	76	#define ROPT_ANCH (ROPT_ANCH_BOL\|ROPT_ANCH_MBOL\|ROPT_ANCH_GPOS)
c277df42 IZ	77	#define ROPT_ANCH_SINGLE (ROPT_ANCH_BOL\|ROPT_ANCH_GPOS)
a0ed51b3 LW	78	#define ROPT_ANCH_BOL 0x00001
	79	#define ROPT_ANCH_MBOL 0x00002
	80	#define ROPT_ANCH_GPOS 0x00004
	81	#define ROPT_SKIP 0x00008
	82	#define ROPT_IMPLICIT 0x00010 /* Converted .* to ^.* */
	83	#define ROPT_NOSCAN 0x00020 /* Check-string always at start. */
	84	#define ROPT_GPOS_SEEN 0x00040
	85	#define ROPT_CHECK_ALL 0x00080
	86	#define ROPT_LOOKBEHIND_SEEN 0x00100
	87	#define ROPT_EVAL_SEEN 0x00200
	88	#define ROPT_TAINTED_SEEN 0x00400
22e551b9	89	#define ROPT_ANCH_SBOL 0x00800
a0ed51b3	90
8782bef2	91	/* 0xf800 of reganch is used by PMf_COMPILETIME */
c277df42	92
a0ed51b3 LW	93	#define ROPT_UTF8 0x10000
a0ed51b3 LW	94	#define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */
cf93c79d	95	#define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */
a0ed51b3	96
c277df42	97	#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
72311751 GS	98	#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch \|= ROPT_TAINTED_SEEN)
	99	#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
	100	#define RX_MATCH_TAINTED_set(prog, t) ((t) \
	101	? RX_MATCH_TAINTED_on(prog) \
	102	: RX_MATCH_TAINTED_off(prog))
c277df42	103
cf93c79d IZ	104	#define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
	105	#define RX_MATCH_COPIED_on(prog) ((prog)->reganch \|= ROPT_COPY_DONE)
	106	#define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
	107	#define RX_MATCH_COPIED_set(prog,t) ((t) \
	108	? RX_MATCH_COPIED_on(prog) \
	109	: RX_MATCH_COPIED_off(prog))
	110
c277df42 IZ	111	#define REXEC_COPY_STR 1 /* Need to copy the string. */
c277df42 IZ	112	#define REXEC_CHECKED 2 /* check_substr already checked. */
22e551b9	113	#define REXEC_SCREAM 4 /* use scream table. */
ad94a511	114	#define REXEC_IGNOREPOS 8 /* \G matches at start. */
cf93c79d	115	#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 IZ	116
	117	#define ReREFCNT_inc(re) ((re && re->refcnt++), re)
	118	#define ReREFCNT_dec(re) pregfree(re)
cf93c79d IZ	119
	120	#define FBMcf_TAIL_DOLLAR 1
	121	#define FBMcf_TAIL_Z 2
	122	#define FBMcf_TAIL_z 4
	123	#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR\|FBMcf_TAIL_Z\|FBMcf_TAIL_z)
	124
	125	#define FBMrf_MULTILINE 1