Fixed BUILD script to use --with-berkeley-db instead of --with-bdb

Lots of small fixes to multi-precision-math path
Give Note for '123.4e'
Added helper functions type 'val_string_from_real()
Don't give warnings for end space for string2decimal()
Changed storage of values for SP so that we can detect length of argument without strlen()
Changed interface for str2dec() so that we must supple the pointer to the last character in the buffer



BUILD/SETUP.sh:
  with-bdb ->with-berkeley-db
include/decimal.h:
  Make string2decimal and string2decimal_fixed inline
mysql-test/r/func_group.result:
  More tests (to find bugs in precision math fixes)
mysql-test/r/func_set.result:
  Test to cover more Item_func_field::val_xxx() code
mysql-test/r/ps_6bdb.result:
  update results
mysql-test/r/type_decimal.result:
  New tests
  Give note for '123.4e'
mysql-test/r/type_newdecimal.result:
  Number of decimal changes (probably right, but hard to verify)
mysql-test/t/func_group.test:
  More tests (to find bugs in precision math fixes)
mysql-test/t/func_set.test:
  Test to cover more Item_func_field::val_xxx() code
mysql-test/t/type_decimal.test:
  New tests to cover more cases in decimal.c
sql/item.cc:
  Added helper functions type 'val_string_from_real()'
  Use new interfase to str2my_decimal()
  Moved nr_of_decimals() here (and made it static)
sql/item.h:
  Added helper functions type 'val_string_from_real()'
sql/item_func.cc:
  Style fixes
  Trivial optimizations
  Ensure null_value is set if my_decimal_add/sub/mul/div returns error
  Remove not needed Item_func_int_div::val_str()
  Join Item_func_signproc and Item_func_neg
  Fix that FIELD() works when first argument is NULL or one if it's arguments are NULL
  new str2my_decimal interface
sql/item_func.h:
  Make Item_func_int_div inherit from Item_int_func (allows us to remove some virtual functions)
  Join Item_func_signproc & Item_func_neg
sql/item_strfunc.cc:
  Move nr_of_decmails() to Item.cc (as it was only used here)
sql/item_sum.cc:
  Style fixes
  Change a lot of code to use new helper converter functions in item.cc
  Moved Item_sum::val_decimal() to Item_sum_int::val_decimal()
  Fixed calls to wrong functions (Item_sum_num::val_int())
  Ensure that all hybrid functions checks hybrid_type in val_xxx() (As there is no gurantee that they are called in the right context)
  Simplify key_length allocation in Item_sum_sum_distinct()
  Simplified create_tmp_field() and reset_field()
  Fixed potential error in Item_sum_hybrid::reset_field()
  Optimize Item_sum_avg::update_field()
  Item_std_field() functions musted be fully coded becasue Item_variance_field::val_xxx functions called helper functions
  Coded missing Item_sum_ufd_xxx::val_decimal() functions
sql/item_sum.h:
  Moved Item_sum::val_decimal() to Item_sum_int::val_decimal()
  Added missing Item_sum_ufd_xxx::val_decimal() functions
  Removed not used scale() function.
  Fixed that Item_std_field() works with decimal arguments
  Fixed that CREATE ... STD() will create a REAL field
sql/log_event.cc:
  Ensure that we use same format for all types
sql/my_decimal.cc:
  Don't give warnings for end space for string2decimal()
  Added dbug_print_decimal()
sql/my_decimal.h:
  Style fixes
  Prototypes for new functions
  New interface for str2my_decimal()
sql/mysql_priv.h:
  Made nr_of_decimals() static
sql/protocol.cc:
  Simplify code (by assume that decimal can't be bigger than DECIMAL_MAX_STR_LENGTH]
sql/protocol_cursor.cc:
  Changed storage of values for SP so that we can detect length of argument without strlen()
sql/sp_head.cc:
  Simplify code for decimal handling by letting item handling conversion to decimal
sql/sp_rcontext.cc:
  Use new method to get length of arguments
sql/sql_analyse.cc:
  if -> switch
  Increase 'empty' if decimal value=0
  Remove usage of strcat()
sql/sql_base.cc:
  Remove unnecessary checks
sql/sql_class.cc:
  Remove not needed 'else'
  Removed not used variables
sql/sql_select.cc:
  remove test for impossible condtion
strings/decimal.c:
  Made two trivial functions macros
  Changed interface for str2dec() so that we must supple the pointer to the last character in the buffer
  This safer than before as we don't require an end \0 anymore (old code gave wrong answers in MySQL for some internals strings that where not \0 terminated)
  Detect error numbers of type '12.55e'
  str2dec() will now set 'to' to zero in case of errors

1 files changed, 137 insertions, 96 deletions

diff --git a/strings/decimal.c b/strings/decimal.c
index 16c80158722..212bd7204f6 100644
--- a/strings/decimal.c
+++ b/strings/decimal.c
@@ -743,12 +743,12 @@ int decimal_shift(decimal *dec, int shift)
   Convert string to decimal
 
   SYNOPSIS
-    str2decl()
-      from    - value to convert
+    internal_str2decl()
+      from    - value to convert. Doesn't have to be \0 terminated!
       to      - decimal where where the result will be stored
                 to->buf and to->len must be set.
-      end     - if not NULL, *end will be set to the char where
-                conversion ended
+      end     - Pointer to pointer to end of string. Will on return be
+		set to the char after the last used character
       fixed   - use to->intg, to->frac as limits for input number
 
   NOTE
@@ -757,31 +757,36 @@ int decimal_shift(decimal *dec, int shift)
 
   RETURN VALUE
     E_DEC_OK/E_DEC_TRUNCATED/E_DEC_OVERFLOW/E_DEC_BAD_NUM/E_DEC_OOM
+    In case of E_DEC_FATAL_ERROR *to is set to decimal zero
+    (to make error handling easier)
 */
 
-static int str2dec(char *from, decimal *to, char **end, my_bool fixed)
+int internal_str2dec(const char *from, decimal *to, char **end, my_bool fixed)
 {
-  char *s=from, *s1, *endp;
+  const char *s= from, *s1, *endp, *end_of_string= *end;
   int i, intg, frac, error, intg1, frac1;
   dec1 x,*buf;
-  LINT_INIT(error);
   sanity(to);
 
-  while (my_isspace(&my_charset_latin1, *s))
+  error= E_DEC_BAD_NUM;                         /* In case of bad number */
+  while (s < end_of_string && my_isspace(&my_charset_latin1, *s))
     s++;
+  if (s == end_of_string)
+    goto fatal_error;
+
   if ((to->sign= (*s == '-')))
     s++;
   else if (*s == '+')
     s++;
 
   s1=s;
-  while (my_isdigit(&my_charset_latin1, *s))
+  while (s < end_of_string && my_isdigit(&my_charset_latin1, *s))
     s++;
   intg=s-s1;
-  if (*s=='.')
+  if (s < end_of_string && *s=='.')
   {
     endp= s+1;
-    while (my_isdigit(&my_charset_latin1, *endp))
+    while (s < end_of_string && my_isdigit(&my_charset_latin1, *endp))
       endp++;
     frac= endp - s - 1;
   }
@@ -791,12 +796,12 @@ static int str2dec(char *from, decimal *to, char **end, my_bool fixed)
     endp= s;
   }
 
-  if (end)
-    *end= endp;
+  *end= (char*) endp;
 
   if (frac+intg == 0)
-    return E_DEC_BAD_NUM;
+    goto fatal_error;
 
+  error= 0;
   if (fixed)
   {
     if (frac > to->frac)
@@ -812,7 +817,10 @@ static int str2dec(char *from, decimal *to, char **end, my_bool fixed)
     intg1=ROUND_UP(intg);
     frac1=ROUND_UP(frac);
     if (intg1+frac1 > to->len)
-      return E_DEC_OOM;
+    {
+      error= E_DEC_OOM;
+      goto fatal_error;
+    }
   }
   else
   {
@@ -861,36 +869,43 @@ static int str2dec(char *from, decimal *to, char **end, my_bool fixed)
   }
   if (i)
     *buf=x*powers10[DIG_PER_DEC1-i];
-  if (*endp == 'e' || *endp == 'E')
-  {
-    long exp= strtol(endp + 1, &endp, 10);
-
-    if (end)
-      *end= endp;
 
-    if (exp > INT_MAX/2)
-      return E_DEC_OVERFLOW;
-    if (exp < INT_MIN/2 && error != E_DEC_OVERFLOW)
-      return E_DEC_TRUNCATED;
+  /* Handle exponent */
+  if (endp+1 < end_of_string && (*endp == 'e' || *endp == 'E'))
+  {
+    int str_error;
+    longlong exp= my_strtoll10(endp+1, (char**) &end_of_string, &str_error);
 
-    if(error != E_DEC_OVERFLOW)
-      error= decimal_shift(to, exp);
+    if (end_of_string != endp +1)               /* If at least one digit */
+    {
+      *end= (char*) end_of_string;
+      if (str_error > 0)
+      {
+        error= E_DEC_BAD_NUM;
+        goto fatal_error;
+      }
+      if (exp > INT_MAX/2 || (str_error == 0 && exp < 0))
+      {
+        error= E_DEC_OVERFLOW;
+        goto fatal_error;
+      }
+      if (exp < INT_MIN/2 && error != E_DEC_OVERFLOW)
+      {
+        error= E_DEC_TRUNCATED;
+        goto fatal_error;
+      }
+      if (error != E_DEC_OVERFLOW)
+        error= decimal_shift(to, (int) exp);
+    }
   }
+  return error;
 
+fatal_error:
+  decimal_make_zero(to);
   return error;
 }
 
 
-int string2decimal(char *from, decimal *to, char **end)
-{
-  return str2dec(from, to, end, 0);
-}
-
-int string2decimal_fixed(char *from, decimal *to, char **end)
-{
-  return str2dec(from, to, end, 1);
-}
-
 /*
   Convert decimal to double
 
@@ -932,9 +947,10 @@ int decimal2double(decimal *from, double *to)
 int double2decimal(double from, decimal *to)
 {
   /* TODO: fix it, when we'll have dtoa */
-  char s[400];
+  char s[400], *end;
   sprintf(s, "%f", from);
-  return string2decimal(s, to, 0);
+  end= strend(s);
+  return string2decimal(s, to, &end);
 }
 
 static int ull2dec(ulonglong from, decimal *to)
@@ -2160,13 +2176,13 @@ void check_result_code(int actual, int want)
 {
   if (actual != want)
   {
-    printf("\n^^^^^^^^^^^^^ mast return %d\n", want);
+    printf("\n^^^^^^^^^^^^^ must return %d\n", want);
     exit(1);
   }
 }
 
 
-void print_decimal(decimal *d, char *orig, int actual, int want)
+void print_decimal(decimal *d, const char *orig, int actual, int want)
 {
   char s[100];
   int slen=sizeof(s);
@@ -2218,38 +2234,41 @@ void test_d2s()
   dump_decimal(&a); printf("  -->  res=%d str='%s' len=%d\n", res, s, slen);
 }
 
-void test_s2d(char *s, char *orig, int ex)
+void test_s2d(const char *s, const char *orig, int ex)
 {
-  char s1[100];
+  char s1[100], *end;
   int res;
   sprintf(s1, "'%s'", s);
+  end= strend(s);
   printf("len=%2d %-30s => res=%d    ", a.len, s1,
-         (res= string2decimal(s, &a, 0)));
+         (res= string2decimal(s, &a, &end)));
   print_decimal(&a, orig, res, ex);
   printf("\n");
 }
 
-void test_d2f(char *s, int ex)
+void test_d2f(const char *s, int ex)
 {
-  char s1[100];
+  char s1[100], *end;
   double x;
   int res;
 
   sprintf(s1, "'%s'", s);
-  string2decimal(s, &a, 0);
+  end= strend(s);
+  string2decimal(s, &a, &end);
   res=decimal2double(&a, &x);
   if (full) dump_decimal(&a);
   printf("%-40s => res=%d    %.*g\n", s1, res, a.intg+a.frac, x);
   check_result_code(res, ex);
 }
 
-void test_d2b2d(char *str, int p, int s, char *orig, int ex)
+void test_d2b2d(const char *str, int p, int s, const char *orig, int ex)
 {
-  char s1[100], buf[100];
+  char s1[100], buf[100], *end;
   int res, i, size=decimal_bin_size(p, s);
 
   sprintf(s1, "'%s'", str);
-  string2decimal(str, &a, 0);
+  end= strend(str);
+  string2decimal(str, &a, &end);
   res=decimal2bin(&a, buf, p, s);
   printf("%-31s {%2d, %2d} => res=%d size=%-2d ", s1, p, s, res, size);
   if (full)
@@ -2263,6 +2282,7 @@ void test_d2b2d(char *str, int p, int s, char *orig, int ex)
   print_decimal(&a, orig, res, ex);
   printf("\n");
 }
+
 void test_f2d(double from, int ex)
 {
   int res;
@@ -2273,7 +2293,7 @@ void test_f2d(double from, int ex)
   printf("\n");
 }
 
-void test_ull2d(ulonglong from, char *orig, int ex)
+void test_ull2d(ulonglong from, const char *orig, int ex)
 {
   char s[100];
   int res;
@@ -2285,7 +2305,7 @@ void test_ull2d(ulonglong from, char *orig, int ex)
   printf("\n");
 }
 
-void test_ll2d(longlong from, char *orig, int ex)
+void test_ll2d(longlong from, const char *orig, int ex)
 {
   char s[100];
   int res;
@@ -2297,13 +2317,14 @@ void test_ll2d(longlong from, char *orig, int ex)
   printf("\n");
 }
 
-void test_d2ull(char *s, char *orig, int ex)
+void test_d2ull(const char *s, const char *orig, int ex)
 {
-  char s1[100];
+  char s1[100], *end;
   ulonglong x;
   int res;
 
-  string2decimal(s, &a, 0);
+  end= strend(s);
+  string2decimal(s, &a, &end);
   res=decimal2ulonglong(&a, &x);
   if (full) dump_decimal(&a);
   longlong10_to_str(x,s1,10);
@@ -2316,13 +2337,14 @@ void test_d2ull(char *s, char *orig, int ex)
   }
 }
 
-void test_d2ll(char *s, char *orig, int ex)
+void test_d2ll(const char *s, const char *orig, int ex)
 {
-  char s1[100];
+  char s1[100], *end;
   longlong x;
   int res;
 
-  string2decimal(s, &a, 0);
+  end= strend(s);
+  string2decimal(s, &a, &end);
   res=decimal2longlong(&a, &x);
   if (full) dump_decimal(&a);
   longlong10_to_str(x,s1,-10);
@@ -2335,39 +2357,45 @@ void test_d2ll(char *s, char *orig, int ex)
   }
 }
 
-void test_da(char *s1, char *s2, char *orig, int ex)
+void test_da(const char *s1, const char *s2, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' + '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_add(&a, &b, &c);
   printf("%-40s => res=%d    ", s, res);
   print_decimal(&c, orig, res, ex);
   printf("\n");
 }
 
-void test_ds(char *s1, char *s2, char *orig, int ex)
+void test_ds(const char *s1, const char *s2, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' - '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_sub(&a, &b, &c);
   printf("%-40s => res=%d    ", s, res);
   print_decimal(&c, orig, res, ex);
   printf("\n");
 }
 
-void test_dc(char *s1, char *s2, int orig)
+void test_dc(const char *s1, const char *s2, int orig)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' <=> '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_cmp(&a, &b);
   printf("%-40s => res=%d\n", s, res);
   if (orig != res)
@@ -2377,26 +2405,30 @@ void test_dc(char *s1, char *s2, int orig)
   }
 }
 
-void test_dm(char *s1, char *s2, char *orig, int ex)
+void test_dm(const char *s1, const char *s2, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' * '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_mul(&a, &b, &c);
   printf("%-40s => res=%d    ", s, res);
   print_decimal(&c, orig, res, ex);
   printf("\n");
 }
 
-void test_dv(char *s1, char *s2, char *orig, int ex)
+void test_dv(const char *s1, const char *s2, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' / '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_div(&a, &b, &c, 5);
   printf("%-40s => res=%d    ", s, res);
   check_result_code(res, ex);
@@ -2407,13 +2439,15 @@ void test_dv(char *s1, char *s2, char *orig, int ex)
   printf("\n");
 }
 
-void test_md(char *s1, char *s2, char *orig, int ex)
+void test_md(const char *s1, const char *s2, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' %% '%s'", s1, s2);
-  string2decimal(s1, &a, 0);
-  string2decimal(s2, &b, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
+  end= strend(s2);
+  string2decimal(s2, &b, &end);
   res=decimal_mod(&a, &b, &c);
   printf("%-40s => res=%d    ", s, res);
   check_result_code(res, ex);
@@ -2424,14 +2458,17 @@ void test_md(char *s1, char *s2, char *orig, int ex)
   printf("\n");
 }
 
-char *round_mode[]={"TRUNCATE", "HALF_EVEN", "HALF_UP", "CEILING", "FLOOR"};
+const char *round_mode[]=
+{"TRUNCATE", "HALF_EVEN", "HALF_UP", "CEILING", "FLOOR"};
 
-void test_ro(char *s1, int n, decimal_round_mode mode, char *orig, int ex)
+void test_ro(const char *s1, int n, decimal_round_mode mode, const char *orig,
+             int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s', %d, %s", s1, n, round_mode[mode]);
-  string2decimal(s1, &a, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
   res=decimal_round(&a, &b, n, mode);
   printf("%-40s => res=%d    ", s, res);
   print_decimal(&b, orig, res, ex);
@@ -2439,7 +2476,7 @@ void test_ro(char *s1, int n, decimal_round_mode mode, char *orig, int ex)
 }
 
 
-void test_mx(int precision, int frac, char *orig)
+void test_mx(int precision, int frac, const char *orig)
 {
   char s[100];
   sprintf(s, "%d, %d", precision, frac);
@@ -2450,15 +2487,17 @@ void test_mx(int precision, int frac, char *orig)
 }
 
 
-void test_pr(char *s1, int prec, int dec, char filler, char *orig, int ex)
+void test_pr(const char *s1, int prec, int dec, char filler, const char *orig,
+             int ex)
 {
-  char s[100];
+  char s[100], *end;
   char s2[100];
   int slen= sizeof(s2);
   int res;
 
   sprintf(s, "'%s', %d, %d, '%c'", s1, prec, dec, filler);
-  string2decimal(s1, &a, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
   res= decimal2string(&a, s2, &slen, prec, dec, filler);
   printf("%-40s => res=%d    '%s'", s, res, s2);
   check_result_code(res, ex);
@@ -2471,12 +2510,13 @@ void test_pr(char *s1, int prec, int dec, char filler, char *orig, int ex)
 }
 
 
-void test_sh(char *s1, int shift, char *orig, int ex)
+void test_sh(const char *s1, int shift, const char *orig, int ex)
 {
-  char s[100];
+  char s[100], *end;
   int res;
   sprintf(s, "'%s' %s %d", s1, ((shift < 0) ? ">>" : "<<"), abs(shift));
-  string2decimal(s1, &a, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
   res= decimal_shift(&a, shift);
   printf("%-40s => res=%d    ", s, res);
   print_decimal(&a, orig, res, ex);
@@ -2484,12 +2524,13 @@ void test_sh(char *s1, int shift, char *orig, int ex)
 }
 
 
-void test_fr(char *s1, char *orig)
+void test_fr(const char *s1, const char *orig)
 {
-  char s[100];
+  char s[100], *end;
   sprintf(s, "'%s'", s1);
   printf("%-40s =>          ", s);
-  string2decimal(s1, &a, 0);
+  end= strend(s1);
+  string2decimal(s1, &a, &end);
   decimal_optimize_fraction(&a);
   print_decimal(&a, orig, 0, 0);
   printf("\n");