regcomp.h

视频监控网络部分的协议ddns,的模块的实现代码,请大家大胆指正.

/*    regcomp.h
 *
 *    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
 *    2000, 2001, 2002, 2003, 2005, 2006, 2007, by Larry Wall and others
 *
 *    You may distribute under the terms of either the GNU General Public
 *    License or the Artistic License, as specified in the README file.
 *
 */
#include "regcharclass.h"

typedef OP OP_4tree;			/* Will be redefined later. */


/* Convert branch sequences to more efficient trie ops? */
#define PERL_ENABLE_TRIE_OPTIMISATION 1

/* Be really agressive about optimising patterns with trie sequences? */
#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1

/* Should the optimiser take positive assertions into account? */
#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0

/* Not for production use: */
#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0

/* Activate offsets code - set to if 1 to enable */
#ifdef DEBUGGING
#define RE_TRACK_PATTERN_OFFSETS
#endif

/* Unless the next line is uncommented it is illegal to combine lazy 
   matching with possessive matching. Frankly it doesn't make much sense 
   to allow it as X*?+ matches nothing, X+?+ matches a single char only, 
   and X{min,max}?+ matches min times only.
 */
/* #define REG_ALLOW_MINMOD_SUSPEND */

/*
 * The "internal use only" fields in regexp.h are present to pass info from
 * compile to execute that permits the execute phase to run lots faster on
 * simple cases.  They are:
 *
 * regstart	sv that must begin a match; NULL if none obvious
 * reganch	is the match anchored (at beginning-of-line only)?
 * regmust	string (pointer into program) that match must include, or NULL
 *  [regmust changed to SV* for bminstr()--law]
 * regmlen	length of regmust string
 *  [regmlen not used currently]
 *
 * Regstart and reganch permit very fast decisions on suitable starting points
 * for a match, cutting down the work a lot.  Regmust permits fast rejection
 * of lines that cannot possibly match.  The regmust tests are costly enough
 * that pregcomp() supplies a regmust only if the r.e. contains something
 * potentially expensive (at present, the only such thing detected is * or +
 * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 * supplied because the test in pregexec() needs it and pregcomp() is computing
 * it anyway.
 * [regmust is now supplied always.  The tests that use regmust have a
 * heuristic that disables the test if it usually matches.]
 *
 * [In fact, we now use regmust in many cases to locate where the search
 * starts in the string, so if regback is >= 0, the regmust search is never
 * wasted effort.  The regback variable says how many characters back from
 * where regmust matched is the earliest possible start of the match.
 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
 */

/*
 * Structure for regexp "program".  This is essentially a linear encoding
 * of a nondeterministic finite-state machine (aka syntax charts or
 * "railroad normal form" in parsing technology).  Each node is an opcode
 * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 * all nodes except BRANCH implement concatenation; a "next" pointer with
 * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 * have one of the subtle syntax dependencies:  an individual BRANCH (as
 * opposed to a collection of them) is never concatenated with anything
 * because of operator precedence.)  The operand of some types of node is
 * a literal string; for others, it is a node leading into a sub-FSM.  In
 * particular, the operand of a BRANCH node is the first node of the branch.
 * (NB this is *not* a tree structure:  the tail of the branch connects
 * to the thing following the set of BRANCHes.)  The opcodes are defined
 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
 */

/*
 * A node is one char of opcode followed by two chars of "next" pointer.
 * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 * value is a positive offset from the opcode of the node containing it.
 * An operand, if any, simply follows the node.  (Note that much of the
 * code generation knows about this implicit relationship.)
 *
 * Using two bytes for the "next" pointer is vast overkill for most things,
 * but allows patterns to get big without disasters.
 *
 * [The "next" pointer is always aligned on an even
 * boundary, and reads the offset directly as a short.  Also, there is no
 * special test to reverse the sign of BACK pointers since the offset is
 * stored negative.]
 */

/* This is the stuff that used to live in regexp.h that was truly
   private to the engine itself. It now lives here. */



 typedef struct regexp_internal {
        int name_list_idx;	/* Optional data index of an array of paren names */
        union {
	    U32 *offsets;           /* offset annotations 20001228 MJD
                                       data about mapping the program to the
                                       string -
                                       offsets[0] is proglen when this is used
                                       */
            U32 proglen;
        } u;

        regnode *regstclass;    /* Optional startclass as identified or constructed
                                   by the optimiser */
        struct reg_data *data;	/* Additional miscellaneous data used by the program.
                                   Used to make it easier to clone and free arbitrary
                                   data that the regops need. Often the ARG field of
                                   a regop is an index into this structure */
	regnode program[1];	/* Unwarranted chumminess with compiler. */
} regexp_internal;

#define RXi_SET(x,y) (x)->pprivate = (void*)(y)   
#define RXi_GET(x)   ((regexp_internal *)((x)->pprivate))
#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
/*
 * Flags stored in regexp->intflags
 * These are used only internally to the regexp engine
 *
 * See regexp.h for flags used externally to the regexp engine
 */
#define PREGf_SKIP		0x00000001
#define PREGf_IMPLICIT		0x00000002 /* Converted .* to ^.* */
#define PREGf_NAUGHTY		0x00000004 /* how exponential is this pattern? */
#define PREGf_VERBARG_SEEN	0x00000008
#define PREGf_CUTGROUP_SEEN	0x00000010


/* this is where the old regcomp.h started */

struct regnode_string {
    U8	str_len;
    U8  type;
    U16 next_off;
    char string[1];
};

/* Argument bearing node - workhorse, 
   arg1 is often for the data field */
struct regnode_1 {
    U8	flags;
    U8  type;
    U16 next_off;
    U32 arg1;
};

/* Similar to a regnode_1 but with an extra signed argument */
struct regnode_2L {
    U8	flags;
    U8  type;
    U16 next_off;
    U32 arg1;
    I32 arg2;
};

/* 'Two field' -- Two 16 bit unsigned args */
struct regnode_2 {
    U8	flags;
    U8  type;
    U16 next_off;
    U16 arg1;
    U16 arg2;
};


#define ANYOF_BITMAP_SIZE	32	/* 256 b/(8 b/B) */
#define ANYOF_CLASSBITMAP_SIZE	 4	/* up to 40 (8*5) named classes */

/* also used by trie */
struct regnode_charclass {
    U8	flags;
    U8  type;
    U16 next_off;
    U32 arg1;
    char bitmap[ANYOF_BITMAP_SIZE];	/* only compile-time */
};

struct regnode_charclass_class {	/* has [[:blah:]] classes */
    U8	flags;				/* should have ANYOF_CLASS here */
    U8  type;
    U16 next_off;
    U32 arg1;
    char bitmap[ANYOF_BITMAP_SIZE];		/* both compile-time */
    char classflags[ANYOF_CLASSBITMAP_SIZE];	/* and run-time */
};

/* XXX fix this description.
   Impose a limit of REG_INFTY on various pattern matching operations
   to limit stack growth and to avoid "infinite" recursions.
*/
/* The default size for REG_INFTY is I16_MAX, which is the same as
   SHORT_MAX (see perl.h).  Unfortunately I16 isn't necessarily 16 bits
   (see handy.h).  On the Cray C90, sizeof(short)==4 and hence I16_MAX is
   ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
   ((1<<63)-1).  To limit stack growth to reasonable sizes, supply a
   smaller default.
	--Andy Dougherty  11 June 1998
*/
#if SHORTSIZE > 2
#  ifndef REG_INFTY
#    define REG_INFTY ((1<<15)-1)
#  endif
#endif

#ifndef REG_INFTY
#  define REG_INFTY I16_MAX
#endif

#define ARG_VALUE(arg) (arg)
#define ARG__SET(arg,val) ((arg) = (val))

#undef ARG
#undef ARG1
#undef ARG2

#define ARG(p) ARG_VALUE(ARG_LOC(p))
#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))

#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))

#undef NEXT_OFF
#undef NODE_ALIGN

#define NEXT_OFF(p) ((p)->next_off)
#define NODE_ALIGN(node)
#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */

#define SIZE_ALIGN NODE_ALIGN

#undef OP
#undef OPERAND
#undef MASK
#undef STRING

#define	OP(p)		((p)->type)
#define	OPERAND(p)	(((struct regnode_string *)p)->string)
#define MASK(p)		((char*)OPERAND(p))
#define	STR_LEN(p)	(((struct regnode_string *)p)->str_len)
#define	STRING(p)	(((struct regnode_string *)p)->string)
#define STR_SZ(l)	((l + sizeof(regnode) - 1) / sizeof(regnode))
#define NODE_SZ_STR(p)	(STR_SZ(STR_LEN(p))+1)

#undef NODE_ALIGN
#undef ARG_LOC
#undef NEXTOPER
#undef PREVOPER

#define	NODE_ALIGN(node)
#define	ARG_LOC(p)	(((struct regnode_1 *)p)->arg1)
#define	ARG1_LOC(p)	(((struct regnode_2 *)p)->arg1)
#define	ARG2_LOC(p)	(((struct regnode_2 *)p)->arg2)
#define ARG2L_LOC(p)	(((struct regnode_2L *)p)->arg2)

#define NODE_STEP_REGNODE	1	/* sizeof(regnode)/sizeof(regnode) */
#define EXTRA_STEP_2ARGS	EXTRA_SIZE(struct regnode_2)

#define NODE_STEP_B	4

#define	NEXTOPER(p)	((p) + NODE_STEP_REGNODE)
#define	PREVOPER(p)	((p) - NODE_STEP_REGNODE)

#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
    (ptr)->type = op;    (ptr)->next_off = 0;   (ptr)++; } STMT_END
#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
    ARG_SET(ptr, arg);  FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END

#define REG_MAGIC 0234

#define SIZE_ONLY (RExC_emit == &PL_regdummy)

/* Flags for node->flags of ANYOF */

#define ANYOF_CLASS		0x08	/* has [[:blah:]] classes */
#define ANYOF_INVERT		0x04
#define ANYOF_FOLD		0x02
#define ANYOF_LOCALE		0x01

/* Used for regstclass only */
#define ANYOF_EOS		0x10		/* Can match an empty string too */

/* There is a character or a range past 0xff */
#define ANYOF_UNICODE		0x20
#define ANYOF_UNICODE_ALL	0x40	/* Can match any char past 0xff */

/* size of node is large (includes class pointer) */
#define ANYOF_LARGE 		0x80

/* Are there any runtime flags on in this node? */
#define ANYOF_RUNTIME(s)	(ANYOF_FLAGS(s) & 0x0f)

#define ANYOF_FLAGS_ALL		0xff

/* Character classes for node->classflags of ANYOF */
/* Should be synchronized with a table in regprop() */
/* 2n should pair with 2n+1 */

#define ANYOF_ALNUM	 0	/* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
#define ANYOF_NALNUM	 1
#define ANYOF_SPACE	 2	/* \s */
#define ANYOF_NSPACE	 3
#define ANYOF_DIGIT	 4
#define ANYOF_NDIGIT	 5
#define ANYOF_ALNUMC	 6	/* isalnum(3), utf8::IsAlnum, ALNUMC */
#define ANYOF_NALNUMC	 7
#define ANYOF_ALPHA	 8
#define ANYOF_NALPHA	 9
#define ANYOF_ASCII	10
#define ANYOF_NASCII	11
#define ANYOF_CNTRL	12
#define ANYOF_NCNTRL	13
#define ANYOF_GRAPH	14
#define ANYOF_NGRAPH	15
#define ANYOF_LOWER	16
#define ANYOF_NLOWER	17
#define ANYOF_PRINT	18
#define ANYOF_NPRINT	19
#define ANYOF_PUNCT	20
#define ANYOF_NPUNCT	21
#define ANYOF_UPPER	22
#define ANYOF_NUPPER	23
#define ANYOF_XDIGIT	24
#define ANYOF_NXDIGIT	25
#define ANYOF_PSXSPC	26	/* POSIX space: \s plus the vertical tab */
#define ANYOF_NPSXSPC	27
#define ANYOF_BLANK	28	/* GNU extension: space and tab: non-vertical space */
#define ANYOF_NBLANK	29

#define ANYOF_MAX	32

/* pseudo classes, not stored in the class bitmap, but used as flags
   during compilation of char classes */

#define ANYOF_VERTWS	(ANYOF_MAX+1)
#define ANYOF_NVERTWS	(ANYOF_MAX+2)
#define ANYOF_HORIZWS	(ANYOF_MAX+3)
#define ANYOF_NHORIZWS	(ANYOF_MAX+4)

/* Backward source code compatibility. */

#define ANYOF_ALNUML	 ANYOF_ALNUM
#define ANYOF_NALNUML	 ANYOF_NALNUM
#define ANYOF_SPACEL	 ANYOF_SPACE
#define ANYOF_NSPACEL	 ANYOF_NSPACE

/* Utility macros for the bitmap and classes of ANYOF */

#define ANYOF_SIZE		(sizeof(struct regnode_charclass))
#define ANYOF_CLASS_SIZE	(sizeof(struct regnode_charclass_class))

#define ANYOF_FLAGS(p)		((p)->flags)

#define ANYOF_BIT(c)		(1 << ((c) & 7))

#define ANYOF_CLASS_BYTE(p, c)	(((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
#define ANYOF_CLASS_SET(p, c)	(ANYOF_CLASS_BYTE(p, c) |=  ANYOF_BIT(c))
#define ANYOF_CLASS_CLEAR(p, c)	(ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
#define ANYOF_CLASS_TEST(p, c)	(ANYOF_CLASS_BYTE(p, c) &   ANYOF_BIT(c))

#define ANYOF_CLASS_ZERO(ret)	Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
#define ANYOF_BITMAP_ZERO(ret)	Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)

#define ANYOF_BITMAP(p)		(((struct regnode_charclass*)(p))->bitmap)
#define ANYOF_BITMAP_BYTE(p, c)	(ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
#define ANYOF_BITMAP_SET(p, c)	(ANYOF_BITMAP_BYTE(p, c) |=  ANYOF_BIT(c))
#define ANYOF_BITMAP_CLEAR(p,c)	(ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
#define ANYOF_BITMAP_TEST(p, c)	(ANYOF_BITMAP_BYTE(p, c) &   ANYOF_BIT(c))

#define ANYOF_BITMAP_SETALL(p)		\
	memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
#define ANYOF_BITMAP_CLEARALL(p)	\
	Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
/* Check that all 256 bits are all set.  Used in S_cl_is_anything()  */
#define ANYOF_BITMAP_TESTALLSET(p)	\
	memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE)

#define ANYOF_SKIP		((ANYOF_SIZE - 1)/sizeof(regnode))
#define ANYOF_CLASS_SKIP	((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
#define ANYOF_CLASS_ADD_SKIP	(ANYOF_CLASS_SKIP - ANYOF_SKIP)


/*
 * Utility definitions.
 */
#ifndef CHARMASK
#  define UCHARAT(p)	((int)*(const U8*)(p))
#else
#  define UCHARAT(p)	((int)*(p)&CHARMASK)
#endif

#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))