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.

#define REG_SEEN_ZERO_LEN	0x00000001
#define REG_SEEN_LOOKBEHIND	0x00000002
#define REG_SEEN_GPOS		0x00000004
#define REG_SEEN_EVAL		0x00000008
#define REG_SEEN_CANY		0x00000010
#define REG_SEEN_SANY		REG_SEEN_CANY /* src bckwrd cmpt */
#define REG_SEEN_RECURSE        0x00000020
#define REG_TOP_LEVEL_BRANCHES  0x00000040
#define REG_SEEN_VERBARG        0x00000080
#define REG_SEEN_CUTGROUP       0x00000100
#define REG_SEEN_RUN_ON_COMMENT 0x00000200

START_EXTERN_C

#ifdef PLUGGABLE_RE_EXTENSION
#include "re_nodes.h"
#else
#include "regnodes.h"
#endif

/* The following have no fixed length. U8 so we can do strchr() on it. */
#ifndef DOINIT
EXTCONST U8 PL_varies[];
#else
EXTCONST U8 PL_varies[] = {
    BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
    WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
    NREF, NREFF, NREFFL,
    0
};
#endif

/* The following always have a length of 1. U8 we can do strchr() on it. */
/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
#ifndef DOINIT
EXTCONST U8 PL_simple[];
#else
EXTCONST U8 PL_simple[] = {
    REG_ANY,	SANY,	CANY,
    ANYOF,
    ALNUM,	ALNUML,
    NALNUM,	NALNUML,
    SPACE,	SPACEL,
    NSPACE,	NSPACEL,
    DIGIT,	NDIGIT,
    VERTWS,     NVERTWS,
    HORIZWS,    NHORIZWS,
    0
};
#endif

#ifndef PLUGGABLE_RE_EXTENSION
#ifndef DOINIT
EXTCONST regexp_engine PL_core_reg_engine;
#else /* DOINIT */
EXTCONST regexp_engine PL_core_reg_engine = { 
        Perl_re_compile,
        Perl_regexec_flags,
        Perl_re_intuit_start,
        Perl_re_intuit_string, 
        Perl_regfree_internal,
        Perl_reg_numbered_buff_fetch,
        Perl_reg_numbered_buff_store,
        Perl_reg_numbered_buff_length,
        Perl_reg_named_buff,
        Perl_reg_named_buff_iter,
        Perl_reg_qr_package,
#if defined(USE_ITHREADS)        
        Perl_regdupe_internal
#endif        
};
#endif /* DOINIT */
#endif /* PLUGGABLE_RE_EXTENSION */


END_EXTERN_C


/* .what is a character array with one character for each member of .data
 * The character describes the function of the corresponding .data item:
 *   f - start-class data for regstclass optimization
 *   n - Root of op tree for (?{EVAL}) item
 *   o - Start op for (?{EVAL}) item
 *   p - Pad for (?{EVAL}) item
 *   s - swash for Unicode-style character class, and the multicharacter
 *       strings resulting from casefolding the single-character entries
 *       in the character class
 *   t - trie struct
 *   u - trie struct's widecharmap (a HV, so can't share, must dup)
 *       also used for revcharmap and words under DEBUGGING
 *   T - aho-trie struct
 *   S - sv for named capture lookup
 * 20010712 [e-mail]
 * (Remember to update re_dup() and pregfree() if you add any items.)
 */
struct reg_data {
    U32 count;
    U8 *what;
    void* data[1];
};

/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
   anchored* and float* via array indexes 0 and 1.  */
#define anchored_substr substrs->data[0].substr
#define anchored_utf8 substrs->data[0].utf8_substr
#define anchored_offset substrs->data[0].min_offset
#define anchored_end_shift substrs->data[0].end_shift

#define float_substr substrs->data[1].substr
#define float_utf8 substrs->data[1].utf8_substr
#define float_min_offset substrs->data[1].min_offset
#define float_max_offset substrs->data[1].max_offset
#define float_end_shift substrs->data[1].end_shift

#define check_substr substrs->data[2].substr
#define check_utf8 substrs->data[2].utf8_substr
#define check_offset_min substrs->data[2].min_offset
#define check_offset_max substrs->data[2].max_offset
#define check_end_shift substrs->data[2].end_shift



/* trie related stuff */

/* a transition record for the state machine. the
   check field determines which state "owns" the
   transition. the char the transition is for is
   determined by offset from the owning states base
   field.  the next field determines which state
   is to be transitioned to if any.
*/
struct _reg_trie_trans {
  U32 next;
  U32 check;
};

/* a transition list element for the list based representation */
struct _reg_trie_trans_list_elem {
    U16 forid;
    U32 newstate;
};
typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;

/* a state for compressed nodes. base is an offset
  into an array of reg_trie_trans array. If wordnum is
  nonzero the state is accepting. if base is zero then
  the state has no children (and will be accepting)
*/
struct _reg_trie_state {
  U16 wordnum;
  union {
    U32                base;
    reg_trie_trans_le* list;
  } trans;
};



typedef struct _reg_trie_state    reg_trie_state;
typedef struct _reg_trie_trans    reg_trie_trans;


/* anything in here that needs to be freed later
   should be dealt with in pregfree.
   refcount is first in both this and _reg_ac_data to allow a space
   optimisation in Perl_regdupe.  */
struct _reg_trie_data {
    U32             refcount;        /* number of times this trie is referenced */
    U32             lasttrans;       /* last valid transition element */
    U16             *charmap;        /* byte to charid lookup array */
    reg_trie_state  *states;         /* state data */
    reg_trie_trans  *trans;          /* array of transition elements */
    char            *bitmap;         /* stclass bitmap */
    U32             *wordlen;        /* array of lengths of words */
    U16 	    *jump;           /* optional 1 indexed array of offsets before tail 
                                        for the node following a given word. */
    U16	            *nextword;       /* optional 1 indexed array to support linked list
                                        of duplicate wordnums */
    U16             uniquecharcount; /* unique chars in trie (width of trans table) */
    U32             startstate;      /* initial state - used for common prefix optimisation */
    STRLEN          minlen;          /* minimum length of words in trie - build/opt only? */
    STRLEN          maxlen;          /* maximum length of words in trie - build/opt only? */
    U32             statecount;      /* Build only - number of states in the states array 
                                        (including the unused zero state) */
    U32             wordcount;       /* Build only */
#ifdef DEBUGGING
    STRLEN          charcount;       /* Build only */
#endif
};
/* There is one (3 under DEBUGGING) pointers that logically belong in this
   structure, but are held outside as they need duplication on thread cloning,
   whereas the rest of the structure can be read only:
    HV              *widecharmap;    code points > 255 to charid
#ifdef DEBUGGING
    AV              *words;          Array of words contained in trie, for dumping
    AV              *revcharmap;     Map of each charid back to its character representation
#endif
*/

#define TRIE_WORDS_OFFSET 2

typedef struct _reg_trie_data reg_trie_data;

/* refcount is first in both this and _reg_trie_data to allow a space
   optimisation in Perl_regdupe.  */
struct _reg_ac_data {
    U32              refcount;
    U32              trie;
    U32              *fail;
    reg_trie_state   *states;
};
typedef struct _reg_ac_data reg_ac_data;

/* ANY_BIT doesnt use the structure, so we can borrow it here.
   This is simpler than refactoring all of it as wed end up with
   three different sets... */

#define TRIE_BITMAP(p)		(((reg_trie_data *)(p))->bitmap)
#define TRIE_BITMAP_BYTE(p, c)	(TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
#define TRIE_BITMAP_SET(p, c)	(TRIE_BITMAP_BYTE(p, c) |=  ANYOF_BIT((U8)c))
#define TRIE_BITMAP_CLEAR(p,c)	(TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
#define TRIE_BITMAP_TEST(p, c)	(TRIE_BITMAP_BYTE(p, c) &   ANYOF_BIT((U8)c))

#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)


#define BITMAP_BYTE(p, c)	(((U8*)p)[(((U8)(c)) >> 3) & 31])
#define BITMAP_TEST(p, c)	(BITMAP_BYTE(p, c) &   ANYOF_BIT((U8)c))

/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))

#ifdef DEBUGGING
#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
#else
#define TRIE_CHARCOUNT(trie) (trie_charcount)
#endif

#define RE_TRIE_MAXBUF_INIT 65536
#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"

/*

RE_DEBUG_FLAGS is used to control what debug output is emitted
its divided into three groups of options, some of which interact.
The three groups are: Compile, Execute, Extra. There is room for a
further group, as currently only the low three bytes are used.

    Compile Options:
    
    PARSE
    PEEP
    TRIE
    PROGRAM
    OFFSETS

    Execute Options:

    INTUIT
    MATCH
    TRIE

    Extra Options

    TRIE
    OFFSETS

If you modify any of these make sure you make corresponding changes to
re.pm, especially to the documentation.

*/


/* Compile */
#define RE_DEBUG_COMPILE_MASK      0x0000FF
#define RE_DEBUG_COMPILE_PARSE     0x000001
#define RE_DEBUG_COMPILE_OPTIMISE  0x000002
#define RE_DEBUG_COMPILE_TRIE      0x000004
#define RE_DEBUG_COMPILE_DUMP      0x000008
#define RE_DEBUG_COMPILE_FLAGS     0x000010

/* Execute */
#define RE_DEBUG_EXECUTE_MASK      0x00FF00
#define RE_DEBUG_EXECUTE_INTUIT    0x000100
#define RE_DEBUG_EXECUTE_MATCH     0x000200
#define RE_DEBUG_EXECUTE_TRIE      0x000400

/* Extra */
#define RE_DEBUG_EXTRA_MASK        0xFF0000
#define RE_DEBUG_EXTRA_TRIE        0x010000
#define RE_DEBUG_EXTRA_OFFSETS     0x020000
#define RE_DEBUG_EXTRA_OFFDEBUG    0x040000
#define RE_DEBUG_EXTRA_STATE       0x080000
#define RE_DEBUG_EXTRA_OPTIMISE    0x100000
#define RE_DEBUG_EXTRA_BUFFERS     0x400000
/* combined */
#define RE_DEBUG_EXTRA_STACK       0x280000

#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
/* Compile */
#define DEBUG_COMPILE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x  )
#define DEBUG_PARSE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x  )
#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x  )
#define DEBUG_PARSE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x  )
#define DEBUG_DUMP_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x  )
#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
#define DEBUG_FLAGS_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
/* Execute */
#define DEBUG_EXECUTE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x  )
#define DEBUG_INTUIT_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x  )
#define DEBUG_MATCH_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x  )
#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )

/* Extra */
#define DEBUG_EXTRA_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x  )
#define DEBUG_OFFSETS_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x  )
#define DEBUG_STATE_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
#define DEBUG_STACK_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
#define DEBUG_BUFFERS_r(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )

#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
    if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
         (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
        Perl_warn_nocontext x )
#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
    if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )

#define DEBUG_TRIE_r(x) DEBUG_r( \
    if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
        | RE_DEBUG_EXECUTE_TRIE )) x )

/* initialization */
/* get_sv() can return NULL during global destruction. */
#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
        SV * re_debug_flags_sv = NULL; \
        re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
        if (re_debug_flags_sv) { \
            if (!SvIOK(re_debug_flags_sv)) \
                sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
            re_debug_flags=SvIV(re_debug_flags_sv); \
        }\
})

#ifdef DEBUGGING

#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;

#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
    const char * const rpv =                          \
        pv_pretty((dsv), (pv), (l), (m), \
            PL_colors[(c1)],PL_colors[(c2)], \
            PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) );         \
    const int rlen = SvCUR(dsv)

#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
    const char * const rpv =                          \
        pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
            PL_colors[(c1)],PL_colors[(c2)], \
            PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )

#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)                    \
    const char * const rpv =                                       \
        pv_pretty((dsv), (pv), (l), (m), \
            PL_colors[0], PL_colors[1], \
            ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELLIPSES | \
              ((isuni) ? PERL_PV_ESCAPE_UNI : 0))                  \
        )

#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
    
#else /* if not DEBUGGING */

#define GET_RE_DEBUG_FLAGS_DECL
#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
#define RE_SV_DUMPLEN(ItEm)
#define RE_SV_TAIL(ItEm)

#endif /* DEBUG RELATED DEFINES */