my_strntod_ucs2 and my_strntol_ucs2 fixes

strings/ctype-utf8.c

@@ -22,7 +22,7 @@
 #include "m_string.h"
 #include "m_ctype.h"
 #include "my_sys.h"			/* defines errno */
-
+#include <errno.h>
 
 #ifdef HAVE_CHARSET_utf8
 #define HAVE_UNIDATA
@@ -2118,7 +2118,7 @@ static int my_ucs2_uni (CHARSET_INFO *cs __attribute__((unused)) ,
                  my_wc_t * pwc, const uchar *s, const uchar *e)
 {
   if (s+2 > e) /* Need 2 characters */
-    return MY_CS_ILSEQ;
+    return MY_CS_TOOFEW(0);
   
   *pwc= ((unsigned char)s[0]) * 256  + ((unsigned char)s[1]);
   return 2;
@@ -2441,118 +2441,114 @@ static int my_snprintf_ucs2(CHARSET_INFO *cs __attribute__((unused))
   return my_vsnprintf_ucs2(to, n, fmt, args);
 }
 
-#define MY_ERRNO(x)
 
 long        my_strntol_ucs2(CHARSET_INFO *cs,
 			   const char *nptr, uint l, char **endptr, int base)
 {
-  int negative;
-  register ulong cutoff;
-  register unsigned int cutlim;
-  register ulong i;
-  register const char *s;
-  register unsigned char c;
-  const char *save, *e;
+  int      negative=0;
   int      overflow;
+  int      cnv;
+  my_wc_t  wc;
+  register unsigned int cutlim;
+  register ulong cutoff;
+  register ulong res;
+  register const char *s=nptr;
+  register const char *e=nptr+l;
+  const char *save;
   
-  if (base < 0 || base == 1 || base > 36)
-    base = 10;
-  
-  s = nptr;
-  e = nptr+l;
-  
-  for ( ; s<e && my_isspace(cs, *s) ; s++);
-  
-  if (s == e)
+  do {
+    if ((cnv=cs->mb_wc(cs,&wc,s,e))>0)
     {
-    goto noconv;
-  }
-  
-  /* Check for a sign.	*/
-  if (*s == '-')
+      switch (wc)
       {
-    negative = 1;
-    ++s;
+        case ' ' : break;
+        case '\t': break;
+        case '-' : negative= !negative; break;
+        case '+' : break;
+        default  : goto bs;
       }
-  else if (*s == '+')
+    } 
+    else /* No more characters or bad multibyte sequence */
     {
-    negative = 0;
-    ++s;
+      if (endptr !=NULL )
+        *endptr = (char*)s;
+      my_errno = (cnv==MY_CS_ILSEQ) ? EILSEQ : EDOM;
+      return 0;
     } 
-  else
-    negative = 0;
+    s+=cnv;
+  } while (1);
   
-  if (base == 16 && s[0] == '0' && my_toupper(cs,s[1]) == 'X')
-    s += 2;
+bs:
   
-  if (base == 0)
-  {
-    if (*s == '0')
-    {
-      if (my_toupper(cs,s[1]) == 'X')
-      {
-	s += 2;
-	base = 16;
-      }
-      else
-	base = 8;
-    }
-    else
+  if (base <= 0 || base == 1 || base > 36)
     base = 10;
-  }
   
+  overflow = 0;
+  res = 0;
   save = s;
   cutoff = ((ulong)~0L) / (unsigned long int) base;
   cutlim = (uint) (((ulong)~0L) % (unsigned long int) base);
   
-  overflow = 0;
-  i = 0;
-  for (c = *s; s != e; c = *++s)
+  do {
+    if ((cnv=cs->mb_wc(cs,&wc,s,e))>0)
     {
-    if (my_isdigit(cs,c))
-      c -= '0';
-    else if (my_isalpha(cs,c))
-      c = my_toupper(cs,c) - 'A' + 10;
+      s+=cnv;
+      if ( wc>='0' && wc<='9')
+        wc -= '0';
+      else if ( wc>='A' && wc<='Z')
+        wc = wc - 'A' + 10;
+      else if ( wc>='a' && wc<='z')
+        wc = wc - 'a' + 10;
       else
         break;
-    if (c >= base)
+      if ((int)wc >= base)
         break;
-    if (i > cutoff || (i == cutoff && c > cutlim))
+      if (res > cutoff || (res == cutoff && wc > cutlim))
         overflow = 1;
       else
       {
-      i *= (ulong) base;
-      i += c;
+        res *= (ulong) base;
+        res += wc;
       }
     }
-  
-  if (s == save)
-    goto noconv;
+    else if (cnv==MY_CS_ILSEQ)
+    {
+      if (endptr !=NULL )
+        *endptr = (char*)s;
+      my_errno=EILSEQ;
+      return 0;
+    } 
+    else
+    {
+      /* No more characters */
+      break;
+    }
+  } while(1);
   
   if (endptr != NULL)
     *endptr = (char *) s;
   
+  if (s == save)
+  {
+    my_errno=EDOM;
+    return 0L;
+  }
+  
   if (negative)
   {
-    if (i  > (ulong) LONG_MIN)
+    if (res > (ulong) LONG_MIN)
       overflow = 1;
   }
-  else if (i > (ulong) LONG_MAX)
+  else if (res > (ulong) LONG_MAX)
     overflow = 1;
   
   if (overflow)
   {
-    MY_ERRNO(ERANGE);
+    my_errno=(ERANGE);
     return negative ? LONG_MIN : LONG_MAX;
   }
   
-  return (negative ? -((long) i) : (long) i);
-
-noconv:
-  MY_ERRNO(EDOM);
-  if (endptr != NULL)
-    *endptr = (char *) nptr;
-  return 0L;
+  return (negative ? -((long) res) : (long) res);
 }
 
 
@@ -2646,14 +2642,14 @@ ulong      my_strntoul_ucs2(CHARSET_INFO *cs,
 
   if (overflow)
   {
-    MY_ERRNO(ERANGE);
+    my_errno=(ERANGE);
     return ((ulong)~0L);
   }
   
   return (negative ? -((long) i) : (long) i);
   
 noconv:
-  MY_ERRNO(EDOM);
+  my_errno=(EDOM);
   if (endptr != NULL)
     *endptr = (char *) nptr;
   return 0L;
@@ -2759,14 +2755,14 @@ longlong   my_strntoll_ucs2(CHARSET_INFO *cs __attribute__((unused)),
 
   if (overflow)
   {
-    MY_ERRNO(ERANGE);
+    my_errno=(ERANGE);
     return negative ? LONGLONG_MIN : LONGLONG_MAX;
   }
 
   return (negative ? -((longlong) i) : (longlong) i);
 
 noconv:
-  MY_ERRNO(EDOM);
+  my_errno=(EDOM);
   if (endptr != NULL)
     *endptr = (char *) nptr;
   return 0L;
@@ -2864,35 +2860,53 @@ ulonglong my_strntoull_ucs2(CHARSET_INFO *cs,
 
   if (overflow)
   {
-    MY_ERRNO(ERANGE);
+    my_errno=(ERANGE);
     return (~(ulonglong) 0);
   }
 
   return (negative ? -((longlong) i) : (longlong) i);
 
 noconv:
-  MY_ERRNO(EDOM);
+  my_errno=(EDOM);
   if (endptr != NULL)
     *endptr = (char *) nptr;
   return 0L;
 }
 
 double      my_strntod_ucs2(CHARSET_INFO *cs __attribute__((unused)),
-			   const char *s, uint l, char **e)
+			   const char *nptr, uint l, char **endptr)
 {
   char     buf[256];
   double   res;
+  register char *b=buf;
+  register const char *s=nptr;
+  register const char *e=nptr+l;
+  my_wc_t  wc;
+  int      cnv;
+  
   if((l+1)>sizeof(buf))
   {
-    if (e)
-      memcpy(*e,s,sizeof(s));
+    if (endptr)
+      *endptr=(char*)nptr;
+    my_errno=ERANGE;
     return 0;
   }
-  strncpy(buf,s,l);
-  buf[l]='\0';
-  res=strtod(buf,e);
-  if (e)
-    memcpy(*e,*e-buf+s,sizeof(s));
+  
+  while ((cnv=cs->mb_wc(cs,&wc,s,e))>0)
+  {
+    s+=cnv;
+    if (wc < 128)
+    {
+      *b++=wc;
+    }
+    else
+      break;
+  }
+  *b='\0';
+  
+  res=strtod(buf,endptr);
+  if (endptr)
+    *endptr=(char*) (*endptr-buf+nptr);
   return res;
 }