regex.c

来自「WinMerge可以显示两个文件的不同之处」· C语言 代码 · 共 1,951 行 · 第 1/5 页

  printf ("re_nsub: %d\t", bufp->re_nsub);
  printf ("regs_alloc: %d\t", bufp->regs_allocated);
  printf ("can_be_null: %d\t", bufp->can_be_null);
  printf ("newline_anchor: %d\n", bufp->newline_anchor);
  printf ("no_sub: %d\t", bufp->no_sub);
  printf ("not_bol: %d\t", bufp->not_bol);
  printf ("not_eol: %d\t", bufp->not_eol);
  printf ("syntax: %d\n", bufp->syntax);
  /* Perhaps we should print the translate table?  */
}


void
print_double_string (const char *where, const char *string1, int size1, const char *string2, int size2)
{
  unsigned this_char;
  
  if (where == NULL)
    printf ("(null)");
  else
    {
      if (FIRST_STRING_P (where))
        {
          for (this_char = where - string1; this_char < size1; this_char++)
            printchar (string1[this_char]);

          where = string2;    
        }

      for (this_char = where - string2; this_char < size2; this_char++)
        printchar (string2[this_char]);
    }
}

#else /* not DEBUG */

#undef assert
#define assert(e)

#define DEBUG_STATEMENT(e)
#define DEBUG_PRINT1(x)
#define DEBUG_PRINT2(x1, x2)
#define DEBUG_PRINT3(x1, x2, x3)
#define DEBUG_PRINT4(x1, x2, x3, x4)
#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)

#endif /* not DEBUG */

/* Set by `re_set_syntax' to the current regexp syntax to recognize.  Can
   also be assigned to arbitrarily: each pattern buffer stores its own
   syntax, so it can be changed between regex compilations.  */
reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS;


/* Specify the precise syntax of regexps for compilation.  This provides
   for compatibility for various utilities which historically have
   different, incompatible syntaxes.

   The argument SYNTAX is a bit mask comprised of the various bits
   defined in regex.h.  We return the old syntax.  */

reg_syntax_t
re_set_syntax (reg_syntax_t syntax)
{
  reg_syntax_t ret = re_syntax_options;
  
  re_syntax_options = syntax;
  return ret;
}

/* This table gives an error message for each of the error codes listed
   in regex.h.  Obviously the order here has to be same as there.  */

static const char *re_error_msg[] =
  { NULL,					/* REG_NOERROR */
    "No match",					/* REG_NOMATCH */
    "Invalid regular expression",		/* REG_BADPAT */
    "Invalid collation character",		/* REG_ECOLLATE */
    "Invalid character class name",		/* REG_ECTYPE */
    "Trailing backslash",			/* REG_EESCAPE */
    "Invalid back reference",			/* REG_ESUBREG */
    "Unmatched [ or [^",			/* REG_EBRACK */
    "Unmatched ( or \\(",			/* REG_EPAREN */
    "Unmatched \\{",				/* REG_EBRACE */
    "Invalid content of \\{\\}",		/* REG_BADBR */
    "Invalid range end",			/* REG_ERANGE */
    "Memory exhausted",				/* REG_ESPACE */
    "Invalid preceding regular expression",	/* REG_BADRPT */
    "Premature end of regular expression",	/* REG_EEND */
    "Regular expression too big",		/* REG_ESIZE */
    "Unmatched ) or \\)",			/* REG_ERPAREN */
  };

/* Avoiding alloca during matching, to placate r_alloc.  */

/* Define MATCH_MAY_ALLOCATE if we need to make sure that the
   searching and matching functions should not call alloca.  On some
   systems, alloca is implemented in terms of malloc, and if we're
   using the relocating allocator routines, then malloc could cause a
   relocation, which might (if the strings being searched are in the
   ralloc heap) shift the data out from underneath the regexp
   routines.

   Here's another reason to avoid allocation: Emacs insists on
   processing input from X in a signal handler; processing X input may
   call malloc; if input arrives while a matching routine is calling
   malloc, then we're scrod.  But Emacs can't just block input while
   calling matching routines; then we don't notice interrupts when
   they come in.  So, Emacs blocks input around all regexp calls
   except the matching calls, which it leaves unprotected, in the
   faith that they will not malloc.  */

/* Normally, this is fine.  */
#define MATCH_MAY_ALLOCATE

/* But under some circumstances, it's not.  */
#if defined (emacs) || (defined (REL_ALLOC) && defined (C_ALLOCA))
#undef MATCH_MAY_ALLOCATE
#endif


/* Failure stack declarations and macros; both re_compile_fastmap and
   re_match_2 use a failure stack.  These have to be macros because of
   REGEX_ALLOCATE.  */
   

/* Number of failure points for which to initially allocate space
   when matching.  If this number is exceeded, we allocate more
   space, so it is not a hard limit.  */
#ifndef INIT_FAILURE_ALLOC
#define INIT_FAILURE_ALLOC 5
#endif

/* Roughly the maximum number of failure points on the stack.  Would be
   exactly that if always used MAX_FAILURE_SPACE each time we failed.
   This is a variable only so users of regex can assign to it; we never
   change it ourselves.  */
int re_max_failures = 2000;

typedef unsigned char *fail_stack_elt_t;

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

#define FAIL_STACK_EMPTY()     (fail_stack.avail == 0)
#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
#define FAIL_STACK_FULL()      (fail_stack.avail == fail_stack.size)
#define FAIL_STACK_TOP()       (fail_stack.stack[fail_stack.avail])


/* Initialize `fail_stack'.  Do `return -2' if the alloc fails.  */

#ifdef MATCH_MAY_ALLOCATE
#define INIT_FAIL_STACK()						\
  do {									\
    fail_stack.stack = (fail_stack_elt_t *)				\
      REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t));	\
									\
    if (fail_stack.stack == NULL)					\
      return -2;							\
									\
    fail_stack.size = INIT_FAILURE_ALLOC;				\
    fail_stack.avail = 0;						\
  } while (0)
#else
#define INIT_FAIL_STACK()						\
  do {									\
    fail_stack.avail = 0;						\
  } while (0)
#endif


/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.

   Return 1 if succeeds, and 0 if either ran out of memory
   allocating space for it or it was already too large.  
   
   REGEX_REALLOCATE requires `destination' be declared.   */

#define DOUBLE_FAIL_STACK(fail_stack)					\
  ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS		\
   ? 0									\
   : ((fail_stack).stack = (fail_stack_elt_t *)				\
        REGEX_REALLOCATE ((fail_stack).stack, 				\
          (fail_stack).size * sizeof (fail_stack_elt_t),		\
          ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)),	\
									\
      (fail_stack).stack == NULL					\
      ? 0								\
      : ((fail_stack).size <<= 1, 					\
         1)))


/* Push PATTERN_OP on FAIL_STACK. 

   Return 1 if was able to do so and 0 if ran out of memory allocating
   space to do so.  */
#define PUSH_PATTERN_OP(pattern_op, fail_stack)				\
  ((FAIL_STACK_FULL ()							\
    && !DOUBLE_FAIL_STACK (fail_stack))					\
    ? 0									\
    : ((fail_stack).stack[(fail_stack).avail++] = pattern_op,		\
       1))

/* This pushes an item onto the failure stack.  Must be a four-byte
   value.  Assumes the variable `fail_stack'.  Probably should only
   be called from within `PUSH_FAILURE_POINT'.  */
#define PUSH_FAILURE_ITEM(item)						\
  fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item

/* The complement operation.  Assumes `fail_stack' is nonempty.  */
#define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail]

/* Used to omit pushing failure point id's when we're not debugging.  */
#ifdef DEBUG
#define DEBUG_PUSH PUSH_FAILURE_ITEM
#define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM ()
#else
#define DEBUG_PUSH(item)
#define DEBUG_POP(item_addr)
#endif


/* Push the information about the state we will need
   if we ever fail back to it.  
   
   Requires variables fail_stack, regstart, regend, reg_info, and
   num_regs be declared.  DOUBLE_FAIL_STACK requires `destination' be
   declared.
   
   Does `return FAILURE_CODE' if runs out of memory.  */

#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code)	\
  do {									\
    /*char *destination;*/							\
    /* Must be int, so when we don't save any registers, the arithmetic	\
       of 0 + -1 isn't done as unsigned.  */				\
    int this_reg;							\
    									\
    DEBUG_STATEMENT (failure_id++);					\
    DEBUG_STATEMENT (nfailure_points_pushed++);				\
    DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id);		\
    DEBUG_PRINT2 ("  Before push, next avail: %d\n", (fail_stack).avail);\
    DEBUG_PRINT2 ("                     size: %d\n", (fail_stack).size);\
									\
    DEBUG_PRINT2 ("  slots needed: %d\n", NUM_FAILURE_ITEMS);		\
    DEBUG_PRINT2 ("     available: %d\n", REMAINING_AVAIL_SLOTS);	\
									\
    /* Ensure we have enough space allocated for what we will push.  */	\
    while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS)			\
      {									\
        if (!DOUBLE_FAIL_STACK (fail_stack))			\
          return failure_code;						\
									\
        DEBUG_PRINT2 ("\n  Doubled stack; size now: %d\n",		\
		       (fail_stack).size);				\
        DEBUG_PRINT2 ("  slots available: %d\n", REMAINING_AVAIL_SLOTS);\
      }									\
									\
    /* Push the info, starting with the registers.  */			\
    DEBUG_PRINT1 ("\n");						\
									\
    for (this_reg = (int)lowest_active_reg; this_reg <= (int)highest_active_reg;	\
         this_reg++)							\
      {									\
	DEBUG_PRINT2 ("  Pushing reg: %d\n", this_reg);			\
        DEBUG_STATEMENT (num_regs_pushed++);				\
									\
	DEBUG_PRINT2 ("    start: 0x%x\n", regstart[this_reg]);		\
        PUSH_FAILURE_ITEM (regstart[this_reg]);				\
                                                                        \
	DEBUG_PRINT2 ("    end: 0x%x\n", regend[this_reg]);		\
        PUSH_FAILURE_ITEM (regend[this_reg]);				\
									\
	DEBUG_PRINT2 ("    info: 0x%x\n      ", reg_info[this_reg]);	\
        DEBUG_PRINT2 (" match_null=%d",					\
                      REG_MATCH_NULL_STRING_P (reg_info[this_reg]));	\
        DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg]));	\
        DEBUG_PRINT2 (" matched_something=%d",				\
                      MATCHED_SOMETHING (reg_info[this_reg]));		\
        DEBUG_PRINT2 (" ever_matched=%d",				\
                      EVER_MATCHED_SOMETHING (reg_info[this_reg]));	\
	DEBUG_PRINT1 ("\n");						\
        PUSH_FAILURE_ITEM (reg_info[this_reg].word);			\
      }									\
									\
    DEBUG_PRINT2 ("  Pushing  low active reg: %d\n", lowest_active_reg);\
    PUSH_FAILURE_ITEM (lowest_active_reg);				\
									\
    DEBUG_PRINT2 ("  Pushing high active reg: %d\n", highest_active_reg);\
    PUSH_FAILURE_ITEM (highest_active_reg);				\
									\
    DEBUG_PRINT2 ("  Pushing pattern 0x%x: ", pattern_place);		\
    DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend);		\
    PUSH_FAILURE_ITEM (pattern_place);					\
									\
    DEBUG_PRINT2 ("  Pushing string 0x%x: `", string_place);		\
    DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2,   \
				 size2);				\
    DEBUG_PRINT1 ("'\n");						\
    PUSH_FAILURE_ITEM (string_place);					\
									\
    DEBUG_PRINT2 ("  Pushing failure id: %u\n", failure_id);		\
    DEBUG_PUSH (failure_id);						\
  } while (0)

/* This is the number of items that are pushed and popped on the stack
   for each register.  */
#define NUM_REG_ITEMS  3

/* Individual items aside from the registers.  */
#ifdef DEBUG
#define NUM_NONREG_ITEMS 5 /* Includes failure point id.  */
#else
#define NUM_NONREG_ITEMS 4
#endif

/* We push at most this many items on the stack.  */
#define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS)

/* We actually push this many items.  */
#define NUM_FAILURE_ITEMS						\
  ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS 	\
    + NUM_NONREG_ITEMS)

/* How many items can still be added to the stack without overflowing it.  */
#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)


/* Pops what PUSH_FAIL_STACK pushes.

   We restore into the parameters, all of which should be lvalues:
     STR -- the saved data position.
     PAT -- the saved pattern position.
     LOW_REG, HIGH_REG -- the highest and lowest active registers.
     REGSTART, REGEND -- arrays of string positions.
     REG_INFO -- array of information about each subexpression.
   
   Also assumes the variables `fail_stack' and (if debugging), `bufp',
   `pend', `string1', `size1', `string2', and `size2'.  */

#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
{									\
  DEBUG_STATEMENT (fail_stack_elt_t failure_id;)			\
  int this_reg;								\
  unsigned char *string_temp;					\
									\
  assert (!FAIL_STACK_EMPTY ());					\
									\
  /* Remove failure points and point to how many regs pushed.  */	\
  DEBUG_PRINT1 ("POP_FAILURE_POINT:\n");				\
  DEBUG_PRINT2 ("  Before pop, next avail: %d\n", fail_stack.avail);	\
  DEBUG_PRINT2 ("                    size: %d\n", fail_stack.size);	\
									\
  assert (fail_stack.avail >= NUM_NONREG_ITEMS);			\
									\
  DEBUG_POP (&failure_id);						\
  DEBUG_PRINT2 ("  Popping failure id: %u\n", failure_id);		\
									\
  /* If the saved string location is NULL, it came from an		\
     on_failure_keep_string_jump opcode, and we want to throw away the	\
     saved NULL, thus retaining our current position in the string.  */	\
  string_temp = POP_FAILURE_ITEM ();					\
  if (string_temp != NULL)						\
    str = (char *) string_temp;					\
									\
  DEBUG_PRINT2 ("  Popping string 0x%x: `", str);			\
  DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2);	\
  DEBUG_PRINT1 ("'\n");							\
									\
  pat = (unsigned char *) POP_FAILURE_ITEM ();				\
  DEBUG_PRINT2 ("  Popping pattern 0x%x: ", pat);			\
  DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend);			\
									\
  /* Restore register info.  */						\
  high_reg = (unsigned) POP_FAILURE_ITEM ();				\
  DEBUG_PRINT2 ("  Popping high active reg: %d\n", high_reg);		\
									\
  low_reg = (unsigned) POP_FAILURE_ITEM ();				\
  DEBUG_PRINT2 ("  Popping  low active reg: %d\n", low_reg);		\
									\
  for (this_reg = (int)high_reg; this_reg >= (int)low_reg; this_reg--)		\
    {									\
      DEBUG_PRINT2 ("    Popping reg: %d\n", this_reg);			\
									\
      reg_info[this_reg].word = POP_FAILURE_ITEM ();			\