regex.c

这是一个开放源代码的与WINNT/WIN2K/WIN2003兼容的操作系统

/* Make sure we have at least N more bytes of space in buffer.  */
#define GET_BUFFER_SPACE(n)						\
    while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated)	\
      EXTEND_BUFFER ()

/* Make sure we have one more byte of buffer space and then add C to it.  */
#define BUF_PUSH(c)							\
  do {									\
    GET_BUFFER_SPACE (1);						\
    *b++ = (unsigned char) (c);						\
  } while (0)


/* Ensure we have two more bytes of buffer space and then append C1 and C2.  */
#define BUF_PUSH_2(c1, c2)						\
  do {									\
    GET_BUFFER_SPACE (2);						\
    *b++ = (unsigned char) (c1);					\
    *b++ = (unsigned char) (c2);					\
  } while (0)


/* As with BUF_PUSH_2, except for three bytes.  */
#define BUF_PUSH_3(c1, c2, c3)						\
  do {									\
    GET_BUFFER_SPACE (3);						\
    *b++ = (unsigned char) (c1);					\
    *b++ = (unsigned char) (c2);					\
    *b++ = (unsigned char) (c3);					\
  } while (0)


/* Store a jump with opcode OP at LOC to location TO.  We store a
   relative address offset by the three bytes the jump itself occupies.  */
#define STORE_JUMP(op, loc, to) \
  store_op1 (op, loc, (int) ((to) - (loc) - 3))

/* Likewise, for a two-argument jump.  */
#define STORE_JUMP2(op, loc, to, arg) \
  store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)

/* Like `STORE_JUMP', but for inserting.  Assume `b' is the buffer end.  */
#define INSERT_JUMP(op, loc, to) \
  insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)

/* Like `STORE_JUMP2', but for inserting.  Assume `b' is the buffer end.  */
#define INSERT_JUMP2(op, loc, to, arg) \
  insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)


/* This is not an arbitrary limit: the arguments which represent offsets
   into the pattern are two bytes long.  So if 2^16 bytes turns out to
   be too small, many things would have to change.  */
/* Any other compiler which, like MSC, has allocation limit below 2^16
   bytes will have to use approach similar to what was done below for
   MSC and drop MAX_BUF_SIZE a bit.  Otherwise you may end up
   reallocating to 0 bytes.  Such thing is not going to work too well.
   You have been warned!!  */
#if defined _MSC_VER  && !defined WIN32
/* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
   The REALLOC define eliminates a flurry of conversion warnings,
   but is not required. */
# define MAX_BUF_SIZE  65500L
# define REALLOC(p,s) realloc ((p), (size_t) (s))
#else
# define MAX_BUF_SIZE (1L << 16)
# define REALLOC(p,s) realloc ((p), (s))
#endif

/* Extend the buffer by twice its current size via realloc and
   reset the pointers that pointed into the old block to point to the
   correct places in the new one.  If extending the buffer results in it
   being larger than MAX_BUF_SIZE, then flag memory exhausted.  */
#define EXTEND_BUFFER()							\
  do { 									\
    unsigned char *old_buffer = bufp->buffer;				\
    if (bufp->allocated == MAX_BUF_SIZE) 				\
      return REG_ESIZE;							\
    bufp->allocated <<= 1;						\
    if (bufp->allocated > MAX_BUF_SIZE)					\
      bufp->allocated = MAX_BUF_SIZE; 					\
    bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
    if (bufp->buffer == NULL)						\
      return REG_ESPACE;						\
    /* If the buffer moved, move all the pointers into it.  */		\
    if (old_buffer != bufp->buffer)					\
      {									\
        b = (b - old_buffer) + bufp->buffer;				\
        begalt = (begalt - old_buffer) + bufp->buffer;			\
        if (fixup_alt_jump)						\
          fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
        if (laststart)							\
          laststart = (laststart - old_buffer) + bufp->buffer;		\
        if (pending_exact)						\
          pending_exact = (pending_exact - old_buffer) + bufp->buffer;	\
      }									\
  } while (0)


/* Since we have one byte reserved for the register number argument to
   {start,stop}_memory, the maximum number of groups we can report
   things about is what fits in that byte.  */
#define MAX_REGNUM 255

/* But patterns can have more than `MAX_REGNUM' registers.  We just
   ignore the excess.  */
typedef unsigned regnum_t;


/* Macros for the compile stack.  */

/* Since offsets can go either forwards or backwards, this type needs to
   be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1.  */
/* int may be not enough when sizeof(int) == 2.  */
typedef long pattern_offset_t;

typedef struct
{
  pattern_offset_t begalt_offset;
  pattern_offset_t fixup_alt_jump;
  pattern_offset_t inner_group_offset;
  pattern_offset_t laststart_offset;
  regnum_t regnum;
} compile_stack_elt_t;


typedef struct
{
  compile_stack_elt_t *stack;
  unsigned size;
  unsigned avail;			/* Offset of next open position.  */
} compile_stack_type;


#define INIT_COMPILE_STACK_SIZE 32

#define COMPILE_STACK_EMPTY  (compile_stack.avail == 0)
#define COMPILE_STACK_FULL  (compile_stack.avail == compile_stack.size)

/* The next available element.  */
#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])


/* Set the bit for character C in a list.  */
#define SET_LIST_BIT(c)                               \
  (b[((unsigned char) (c)) / BYTEWIDTH]               \
   |= 1 << (((unsigned char) c) % BYTEWIDTH))


/* Get the next unsigned number in the uncompiled pattern.  */
#define GET_UNSIGNED_NUMBER(num) 					\
  { if (p != pend)							\
     {									\
       PATFETCH (c); 							\
       while (ISDIGIT (c)) 						\
         { 								\
           if (num < 0)							\
              num = 0;							\
           num = num * 10 + c - '0'; 					\
           if (p == pend) 						\
              break; 							\
           PATFETCH (c);						\
         } 								\
       } 								\
    }

#if defined _LIBC || WIDE_CHAR_SUPPORT
/* The GNU C library provides support for user-defined character classes
   and the functions from ISO C amendement 1.  */
# ifdef CHARCLASS_NAME_MAX
#  define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
# else
/* This shouldn't happen but some implementation might still have this
   problem.  Use a reasonable default value.  */
#  define CHAR_CLASS_MAX_LENGTH 256
# endif

# ifdef _LIBC
#  define IS_CHAR_CLASS(string) __wctype (string)
# else
#  define IS_CHAR_CLASS(string) wctype (string)
# endif
#else
# define CHAR_CLASS_MAX_LENGTH  6 /* Namely, `xdigit'.  */

# define IS_CHAR_CLASS(string)						\
   (STREQ (string, "alpha") || STREQ (string, "upper")			\
    || STREQ (string, "lower") || STREQ (string, "digit")		\
    || STREQ (string, "alnum") || STREQ (string, "xdigit")		\
    || STREQ (string, "space") || STREQ (string, "print")		\
    || STREQ (string, "punct") || STREQ (string, "graph")		\
    || STREQ (string, "cntrl") || STREQ (string, "blank"))
#endif

#ifndef MATCH_MAY_ALLOCATE

/* If we cannot allocate large objects within re_match_2_internal,
   we make the fail stack and register vectors global.
   The fail stack, we grow to the maximum size when a regexp
   is compiled.
   The register vectors, we adjust in size each time we
   compile a regexp, according to the number of registers it needs.  */

static fail_stack_type fail_stack;

/* Size with which the following vectors are currently allocated.
   That is so we can make them bigger as needed,
   but never make them smaller.  */
static int regs_allocated_size;

static const char **     regstart, **     regend;
static const char ** old_regstart, ** old_regend;
static const char **best_regstart, **best_regend;
static register_info_type *reg_info;
static const char **reg_dummy;
static register_info_type *reg_info_dummy;

/* Make the register vectors big enough for NUM_REGS registers,
   but don't make them smaller.  */

static
regex_grow_registers (num_regs)
     int num_regs;
{
  if (num_regs > regs_allocated_size)
    {
      RETALLOC_IF (regstart,	 num_regs, const char *);
      RETALLOC_IF (regend,	 num_regs, const char *);
      RETALLOC_IF (old_regstart, num_regs, const char *);
      RETALLOC_IF (old_regend,	 num_regs, const char *);
      RETALLOC_IF (best_regstart, num_regs, const char *);
      RETALLOC_IF (best_regend,	 num_regs, const char *);
      RETALLOC_IF (reg_info,	 num_regs, register_info_type);
      RETALLOC_IF (reg_dummy,	 num_regs, const char *);
      RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);

      regs_allocated_size = num_regs;
    }
}

#endif /* not MATCH_MAY_ALLOCATE */

static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
						 compile_stack,
						 regnum_t regnum));

/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
   Returns one of error codes defined in `regex.h', or zero for success.

   Assumes the `allocated' (and perhaps `buffer') and `translate'
   fields are set in BUFP on entry.

   If it succeeds, results are put in BUFP (if it returns an error, the
   contents of BUFP are undefined):
     `buffer' is the compiled pattern;
     `syntax' is set to SYNTAX;
     `used' is set to the length of the compiled pattern;
     `fastmap_accurate' is zero;
     `re_nsub' is the number of subexpressions in PATTERN;
     `not_bol' and `not_eol' are zero;

   The `fastmap' and `newline_anchor' fields are neither
   examined nor set.  */

/* Return, freeing storage we allocated.  */
#define FREE_STACK_RETURN(value)		\
  return (free (compile_stack.stack), value)

static reg_errcode_t
regex_compile (pattern, size, syntax, bufp)
     const char *pattern;
     size_t size;
     reg_syntax_t syntax;
     struct re_pattern_buffer *bufp;
{
  /* We fetch characters from PATTERN here.  Even though PATTERN is
     `char *' (i.e., signed), we declare these variables as unsigned, so
     they can be reliably used as array indices.  */
  register unsigned char c, c1;

  /* A random temporary spot in PATTERN.  */
  const char *p1;

  /* Points to the end of the buffer, where we should append.  */
  register unsigned char *b;

  /* Keeps track of unclosed groups.  */
  compile_stack_type compile_stack;

  /* Points to the current (ending) position in the pattern.  */
  const char *p = pattern;
  const char *pend = pattern + size;

  /* How to translate the characters in the pattern.  */
  RE_TRANSLATE_TYPE translate = bufp->translate;

  /* Address of the count-byte of the most recently inserted `exactn'
     command.  This makes it possible to tell if a new exact-match
     character can be added to that command or if the character requires
     a new `exactn' command.  */
  unsigned char *pending_exact = 0;

  /* Address of start of the most recently finished expression.
     This tells, e.g., postfix * where to find the start of its
     operand.  Reset at the beginning of groups and alternatives.  */
  unsigned char *laststart = 0;

  /* Address of beginning of regexp, or inside of last group.  */
  unsigned char *begalt;

  /* Place in the uncompiled pattern (i.e., the {) to
     which to go back if the interval is invalid.  */
  const char *beg_interval;

  /* Address of the place where a forward jump should go to the end of
     the containing expression.  Each alternative of an `or' -- except the
     last -- ends with a forward jump of this sort.  */
  unsigned char *fixup_alt_jump = 0;

  /* Counts open-groups as they are encountered.  Remembered for the
     matching close-group on the compile stack, so the same register
     number is put in the stop_memory as the start_memory.  */
  regnum_t regnum = 0;

#ifdef DEBUG
  DEBUG_PRINT1 ("\nCompiling pattern: ");
  if (debug)
    {
      unsigned debug_count;

      for (debug_count = 0; debug_count < size; debug_count++)
        putchar (pattern[debug_count]);
      putchar ('\n');
    }
#endif /* DEBUG */

  /* Initialize the compile stack.  */
  compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
  if (compile_stack.stack == NULL)
    return REG_ESPACE;

  compile_stack.size = INIT_COMPILE_STACK_SIZE;
  compile_stack.avail = 0;

  /* Initialize the pattern buffer.  */
  bufp->syntax = syntax;
  bufp->fastmap_accurate = 0;
  bufp->not_bol = bufp->not_eol = 0;

  /* Set `used' to zero, so that if we return an error, the pattern
     printer (for debugging) will think there's no pattern.  We reset it
     at the end.  */
  bufp->used = 0;

  /* Always count groups, whether or not bufp->no_sub is set.  */
  bufp->re_nsub = 0;

#if !defined emacs && !defined SYNTAX_TABLE
  /* Initialize the syntax table.  */
   init_syntax_once ();
#endif

  if (bufp->allocated == 0)
    {
      if (bufp->buffer)
	{ /* If zero allocated, but buffer is non-null, try to realloc
             enough space.  This loses if buffer's address is bogus, but
             that is the user's responsibility.  */
          RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
        }
      else
        { /* Caller did not allocate a buffer.  Do it for them.  */
          bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
        }
      if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);

      bufp->allocated = INIT_BUF_SIZE;
    }

  begalt = b = bufp->buffer;

  /* Loop through the uncompiled pattern until we're at the end.  */
  while (p != pend)
    {
      PATFETCH (c);

      switch (c)
        {
        case '^':
          {
            if (   /* If at start of pattern, it's an operator.  */
                   p == pattern + 1
                   /* If context independent, it's an operator.  */
                || syntax & RE_CONTEXT_INDEP_ANCHORS
                   /* Otherwise, depen