pcre-module.c

一个C格式的脚本处理函数库源代码,可让你的C程序具有执行C格式的脚本文件

#include <stdio.h>
#include <slang.h>
#include <pcre.h>

SLANG_MODULE(pcre);

static int PCRE_Type_Id;

typedef struct
{
   pcre *p;
   pcre_extra *extra;
   int *ovector;
   unsigned int ovector_len;	       /* must be a multiple of 3 */
   unsigned int num_matches;	       /* return value of pcre_exec (>= 1)*/
}
PCRE_Type;

static void free_pcre_type (PCRE_Type *pt)
{
   if (pt->ovector != NULL)
     SLfree ((char *) pt->ovector);

   SLfree ((char *) pt);
}

static SLang_MMT_Type *allocate_pcre_type (pcre *p, pcre_extra *extra)
{
   PCRE_Type *pt;
   SLang_MMT_Type *mmt;
   int ovector_len;

   pt = (PCRE_Type *) SLmalloc (sizeof (PCRE_Type));
   if (pt == NULL)
     return NULL;
   memset ((char *) pt, 0, sizeof (PCRE_Type));

   pt->p = p;
   pt->extra = extra;

   if (0 != pcre_fullinfo (p, extra, PCRE_INFO_CAPTURECOUNT, &ovector_len))
     {
	free_pcre_type (pt);
	SLang_verror (SL_INTRINSIC_ERROR, "pcre_fullinfo failed");
	return NULL;
     }

   ovector_len += 1;		       /* allow for pattern matched */
   ovector_len *= 3;		       /* required to be multiple of 3 */
   if (NULL == (pt->ovector = (int *)SLmalloc (ovector_len * sizeof (int))))
     {
	free_pcre_type (pt);
	return NULL;
     }
   pt->ovector_len = ovector_len;

   if (NULL == (mmt = SLang_create_mmt (PCRE_Type_Id, (VOID_STAR) pt)))
     {
	free_pcre_type (pt);
	return NULL;
     }   
   return mmt;
}

static int _pcre_compile_1 (char *pattern, int options)
{
   pcre *p;
   pcre_extra *extra;
   SLCONST char *err;
   int erroffset;
   unsigned char *table;
   SLang_MMT_Type *mmt;

   table = NULL;
   p = pcre_compile (pattern, options, &err, &erroffset, table);
   if (NULL == p)
     {
	SLang_verror (SL_INTRINSIC_ERROR, "Error compiling pattern '%s' at offset %d: %s", 
		      pattern, erroffset, err);
	return -1;
     }

   extra = pcre_study (p, 0, &err);
   /* apparantly, a NULL return is ok */
   if (err != NULL)
     {
	SLang_verror (SL_INTRINSIC_ERROR, "pcre_study failed: %s", err);
	pcre_free (p);
	return -1;
     }
   
   if (NULL == (mmt = allocate_pcre_type (p, extra)))
     {
	pcre_free ((char *) p);
	pcre_free ((char *) extra);
	return -1;
     }

   if (-1 == SLang_push_mmt (mmt))
     {
	SLang_free_mmt (mmt);
	return -1;
     }
   return 0;
}

static void _pcre_compile (void)
{
   char *pattern;
   int options = 0;
   
   switch (SLang_Num_Function_Args)
     {
      case 2:
	if (-1 == SLang_pop_integer (&options))
	  return;
	/* drop */
      case 1:
      default:
	if (-1 == SLang_pop_slstring (&pattern))
	  return;
     }
   (void) _pcre_compile_1 (pattern, options);
   SLang_free_slstring (pattern);
}



/* returns number of matches */
static int _pcre_exec_1 (PCRE_Type *pt, char *str, int pos, int options)
{
   int rc;
   unsigned int len;
   
   pt->num_matches = 0;
   len = strlen (str);
   if ((unsigned int) pos > len)
     return 0;

   rc = pcre_exec (pt->p, pt->extra, str, len, pos,
		   options, pt->ovector, pt->ovector_len);

   if (rc == PCRE_ERROR_NOMATCH)
     return 0;

   if (rc <= 0)
     {
	SLang_verror (SL_INTRINSIC_ERROR, "pcre_exec returned %d", rc);
	return -1;
     }
   pt->num_matches = (unsigned int) rc;
   return rc;
}

static int _pcre_exec (void)
{
   PCRE_Type *p;
   SLang_MMT_Type *mmt;
   char *str;
   int pos = 0;
   int options = 0;
   int ret = -1;

   switch (SLang_Num_Function_Args)
     {
      case 4:
	if (-1 == SLang_pop_integer (&options))
	  return -1;
      case 3:
	if (-1 == SLang_pop_integer (&pos))
	  return -1;
      default:
	if (-1 == SLang_pop_slstring (&str))
	  return -1;

	if (NULL == (mmt = SLang_pop_mmt (PCRE_Type_Id)))
	  goto free_and_return;
	p = SLang_object_from_mmt (mmt);
     }
   ret = _pcre_exec_1 (p, str, pos, options);
   
   free_and_return:
   SLang_free_slstring (str);
   SLang_free_mmt (mmt);
   return ret;
}


static int get_nth_start_stop (PCRE_Type *pt, unsigned int n, 
			       unsigned int *a, unsigned int *b)
{
   int start, stop;

   if (n >= pt->num_matches) 
     return -1;
   
   start = pt->ovector[2*n];
   stop = pt->ovector[2*n+1];
   if ((start < 0) || (stop < start))
     return -1;
   
   *a = (unsigned int) start;
   *b = (unsigned int) stop;
   return 0;
}

static void _pcre_nth_match (PCRE_Type *pt, int *np)
{
   unsigned int start, stop;
   SLang_Array_Type *at;
   int two = 2;
   int *data;

   if (-1 == get_nth_start_stop (pt, (unsigned int) *np, &start, &stop))
     {
	SLang_push_null ();
	return;
     }
   
   if (NULL == (at = SLang_create_array (SLANG_INT_TYPE, 0, NULL, &two, 1)))
     return;
   
   data = (int *)at->data;
   data[0] = (int)start;
   data[1] = (int)stop;
   (void) SLang_push_array (at, 1);
}

static void _pcre_nth_substr (PCRE_Type *pt, char *str, int *np)
{
   unsigned int start, stop;
   unsigned int len;
   
   len = strlen (str);

   if ((-1 == get_nth_start_stop (pt, (unsigned int) *np, &start, &stop))
       || (start > len) || (stop > len))
     {
	SLang_push_null ();
	return;
     }
   
   str = SLang_create_nslstring (str + start, stop - start);
   (void) SLang_push_string (str);
   SLang_free_slstring (str);
}

/* This function converts a slang RE to a pcre expression.  It performs the
 * following transformations:
 *    (     -->   \(
 *    )     -->   \)
 *    #     -->   \#
 *    |     -->   \|
 *    {     -->   \{
 *    }     -->   \}
 *   \<     -->   \b
 *   \>     -->   \b
 *   \C     -->   (?i)
 *   \c     -->   (?-i)
 *   \(     -->   (
 *   \)     -->   )
 *   \{     -->   {
 *   \}     -->   }
 * Anything else?
 */
static char *_slang_to_pcre (char *slpattern)
{
   char *pattern, *p, *s;
   unsigned int len;
   int in_bracket;
   char ch;

   len = strlen (slpattern);
   pattern = SLmalloc (3*len + 1);
   if (pattern == NULL)
     return NULL;
   
   p = pattern;
   s = slpattern;
   in_bracket = 0;
   while ((ch = *s++) != 0)
     {
	switch (ch)
	  {
	   case '{':
	   case '}':
	   case '(':
	   case ')':
	   case '#':
	   case '|':
	     if (0 == in_bracket) *p++ = '\\';
	     *p++ = ch;
	     break;

	   case '[':
	     in_bracket = 1;
	     *p++ = ch;
	     break;
	     
	   case ']':
	     in_bracket = 0;
	     *p++ = ch;
	     break;

	   case '\\':
	     ch = *s++;
	     switch (ch)
	       {
		case 0:
		  s--;