open-dsi
/
postgres
mirror of https://github.com/postgres/postgres
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add non-decimal integer support to type numeric.

Browse Source 
This enhances the numeric type input function, adding support for
hexadecimal, octal, and binary integers of any size, up to the limits
of the numeric type.

Since 6fcda9aba8, such non-decimal integers have been accepted by the
parser as integer literals and passed through to numeric_in(). This
commit gives numeric_in() the ability to handle them.

While at it, simplify the handling of NaN and infinities, reducing the
number of calls to pg_strncasecmp(), and arrange for pg_strncasecmp()
to not be called at all for regular numbers. This gives a significant
performance improvement for decimal inputs, more than offsetting the
small performance hit of checking for non-decimal input.

Discussion: https://postgr.es/m/CAEZATCV8XShnmT9HZy25C%2Bo78CVOFmUN5EM9FRAZ5xvYTggPMg%40mail.gmail.com
pull/117/head
Dean Rasheed 3 years ago
parent 62e1e28bf7
commit 6dfacbf72b
4 changed files with 405 additions and 95 deletions
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Show Stats Download Patch File Download Diff File
  1. 360
      src/backend/utils/adt/numeric.c
  2. 78
      src/test/regress/expected/numeric.out
  3. 48
      src/test/regress/expected/numerology.out
  4. 14
      src/test/regress/sql/numeric.sql

360
src/backend/utils/adt/numeric.c
Unescape View File

@ -500,6 +500,11 @@ static void zero_var(NumericVar *var);
static bool set_var_from_str(const char *str, const char *cp,
NumericVar *dest, const char **endptr,
Node *escontext);
static bool set_var_from_non_decimal_integer_str(const char *str,
const char *cp, int sign,
int base, NumericVar *dest,
const char **endptr,
Node *escontext);
static void set_var_from_num(Numeric num, NumericVar *dest);
static void init_var_from_num(Numeric num, NumericVar *dest);
static void set_var_from_var(const NumericVar *value, NumericVar *dest);
@ -629,6 +634,8 @@ numeric_in(PG_FUNCTION_ARGS)
Node *escontext = fcinfo->context;
Numeric res;
const char *cp;
const char *numstart;
int sign;
/* Skip leading spaces */
cp = str;
@ -640,70 +647,130 @@ numeric_in(PG_FUNCTION_ARGS)
}
/*
* Check for NaN and infinities. We recognize the same strings allowed by
* float8in().
* Process the number's sign. This duplicates logic in set_var_from_str(),
* but it's worth doing here, since it simplifies the handling of
* infinities and non-decimal integers.
*/
if (pg_strncasecmp(cp, "NaN", 3) == 0)
{
res = make_result(&const_nan);
cp += 3;
}
else if (pg_strncasecmp(cp, "Infinity", 8) == 0)
{
res = make_result(&const_pinf);
cp += 8;
}
else if (pg_strncasecmp(cp, "+Infinity", 9) == 0)
{
res = make_result(&const_pinf);
cp += 9;
}
else if (pg_strncasecmp(cp, "-Infinity", 9) == 0)
{
res = make_result(&const_ninf);
cp += 9;
}
else if (pg_strncasecmp(cp, "inf", 3) == 0)
{
res = make_result(&const_pinf);
cp += 3;
}
else if (pg_strncasecmp(cp, "+inf", 4) == 0)
numstart = cp;
sign = NUMERIC_POS;
if (*cp == '+')
cp++;
else if (*cp == '-')
{
res = make_result(&const_pinf);
cp += 4;
sign = NUMERIC_NEG;
cp++;
}
else if (pg_strncasecmp(cp, "-inf", 4) == 0)
/*
* Check for NaN and infinities. We recognize the same strings allowed by
* float8in().
*
* Since all other legal inputs have a digit or a decimal point after the
* sign, we need only check for NaN/infinity if that's not the case.
*/
if (!isdigit((unsigned char) *cp) && *cp != '.')
{
res = make_result(&const_ninf);
cp += 4;
/*
* The number must be NaN or infinity; anything else can only be a
* syntax error. Note that NaN mustn't have a sign.
*/
if (pg_strncasecmp(numstart, "NaN", 3) == 0)
{
res = make_result(&const_nan);
cp = numstart + 3;
}
else if (pg_strncasecmp(cp, "Infinity", 8) == 0)
{
res = make_result(sign == NUMERIC_POS ? &const_pinf : &const_ninf);
cp += 8;
}
else if (pg_strncasecmp(cp, "inf", 3) == 0)
{
res = make_result(sign == NUMERIC_POS ? &const_pinf : &const_ninf);
cp += 3;
}
else
goto invalid_syntax;
/*
* Check for trailing junk; there should be nothing left but spaces.
*
* We intentionally do this check before applying the typmod because
* we would like to throw any trailing-junk syntax error before any
* semantic error resulting from apply_typmod_special().
*/
while (*cp)
{
if (!isspace((unsigned char) *cp))
goto invalid_syntax;
cp++;
}
if (!apply_typmod_special(res, typmod, escontext))
PG_RETURN_NULL();
}
else
{
/*
* Use set_var_from_str() to parse a normal numeric value
* We have a normal numeric value, which may be a non-decimal integer
* or a regular decimal number.
*/
NumericVar value;
int base;
bool have_error;
init_var(&value);
if (!set_var_from_str(str, cp, &value, &cp, escontext))
PG_RETURN_NULL();
/*
* Determine the number's base by looking for a non-decimal prefix
* indicator ("0x", "0o", or "0b").
*/
if (cp[0] == '0')
{
switch (cp[1])
{
case 'x':
case 'X':
base = 16;
break;
case 'o':
case 'O':
base = 8;
break;
case 'b':
case 'B':
base = 2;
break;
default:
base = 10;
}
}
else
base = 10;
/* Parse the rest of the number and apply the sign */
if (base == 10)
{
if (!set_var_from_str(str, cp, &value, &cp, escontext))
PG_RETURN_NULL();
value.sign = sign;
}
else
{
if (!set_var_from_non_decimal_integer_str(str, cp + 2, sign, base,
&value, &cp, escontext))
PG_RETURN_NULL();
}
/*
* We duplicate a few lines of code here because we would like to
* throw any trailing-junk syntax error before any semantic error
* resulting from apply_typmod. We can't easily fold the two cases
* together because we mustn't apply apply_typmod to a NaN/Inf.
* Should be nothing left but spaces. As above, throw any typmod error
* after finishing syntax check.
*/
while (*cp)
{
if (!isspace((unsigned char) *cp))
ereturn(escontext, (Datum) 0,
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("invalid input syntax for type %s: \"%s\"",
"numeric", str)));
goto invalid_syntax;
cp++;
}
@ -718,26 +785,15 @@ numeric_in(PG_FUNCTION_ARGS)
errmsg("value overflows numeric format")));
free_var(&value);
PG_RETURN_NUMERIC(res);
}
/* Should be nothing left but spaces */
while (*cp)
{
if (!isspace((unsigned char) *cp))
ereturn(escontext, (Datum) 0,
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("invalid input syntax for type %s: \"%s\"",
"numeric", str)));
cp++;
}
/* As above, throw any typmod error after finishing syntax check */
if (!apply_typmod_special(res, typmod, escontext))
PG_RETURN_NULL();
PG_RETURN_NUMERIC(res);
invalid_syntax:
ereturn(escontext, (Datum) 0,
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("invalid input syntax for type %s: \"%s\"",
"numeric", str)));
}
@ -6992,6 +7048,188 @@ set_var_from_str(const char *str, const char *cp,
}
/*
* Return the numeric value of a single hex digit.
*/
static inline int
xdigit_value(char dig)
{
return dig >= '0' && dig <= '9' ? dig - '0' :
dig >= 'a' && dig <= 'f' ? dig - 'a' + 10 :
dig >= 'A' && dig <= 'F' ? dig - 'A' + 10 : -1;
}
/*
* set_var_from_non_decimal_integer_str()
*
* Parse a string containing a non-decimal integer
*
* This function does not handle leading or trailing spaces. It returns
* the end+1 position parsed into *endptr, so that caller can check for
* trailing spaces/garbage if deemed necessary.
*
* cp is the place to actually start parsing; str is what to use in error
* reports. The number's sign and base prefix indicator (e.g., "0x") are
* assumed to have already been parsed, so cp should point to the number's
* first digit in the base specified.
*
* base is expected to be 2, 8 or 16.
*
* Returns true on success, false on failure (if escontext points to an
* ErrorSaveContext; otherwise errors are thrown).
*/
static bool
set_var_from_non_decimal_integer_str(const char *str, const char *cp, int sign,
int base, NumericVar *dest,
const char **endptr, Node *escontext)
{
const char *firstdigit = cp;
int64 tmp;
int64 mul;
NumericVar tmp_var;
init_var(&tmp_var);
zero_var(dest);
/*
* Process input digits in groups that fit in int64. Here "tmp" is the
* value of the digits in the group, and "mul" is base^n, where n is the
* number of digits in the group. Thus tmp < mul, and we must start a new
* group when mul * base threatens to overflow PG_INT64_MAX.
*/
tmp = 0;
mul = 1;
if (base == 16)
{
while (*cp)
{
if (isxdigit((unsigned char) *cp))
{
if (mul > PG_INT64_MAX / 16)
{
/* Add the contribution from this group of digits */
int64_to_numericvar(mul, &tmp_var);
mul_var(dest, &tmp_var, dest, 0);
int64_to_numericvar(tmp, &tmp_var);
add_var(dest, &tmp_var, dest);
/* Result will overflow if weight overflows int16 */
if (dest->weight > SHRT_MAX)
goto out_of_range;
/* Begin a new group */
tmp = 0;
mul = 1;
}
tmp = tmp * 16 + xdigit_value(*cp++);
mul = mul * 16;
}
else
break;
}
}
else if (base == 8)
{
while (*cp)
{
if (*cp >= '0' && *cp <= '7')
{
if (mul > PG_INT64_MAX / 8)
{
/* Add the contribution from this group of digits */
int64_to_numericvar(mul, &tmp_var);
mul_var(dest, &tmp_var, dest, 0);
int64_to_numericvar(tmp, &tmp_var);
add_var(dest, &tmp_var, dest);
/* Result will overflow if weight overflows int16 */
if (dest->weight > SHRT_MAX)
goto out_of_range;
/* Begin a new group */
tmp = 0;
mul = 1;
}
tmp = tmp * 8 + (*cp++ - '0');
mul = mul * 8;
}
else
break;
}
}
else if (base == 2)
{
while (*cp)
{
if (*cp >= '0' && *cp <= '1')
{
if (mul > PG_INT64_MAX / 2)
{
/* Add the contribution from this group of digits */
int64_to_numericvar(mul, &tmp_var);
mul_var(dest, &tmp_var, dest, 0);
int64_to_numericvar(tmp, &tmp_var);
add_var(dest, &tmp_var, dest);
/* Result will overflow if weight overflows int16 */
if (dest->weight > SHRT_MAX)
goto out_of_range;
/* Begin a new group */
tmp = 0;
mul = 1;
}
tmp = tmp * 2 + (*cp++ - '0');
mul = mul * 2;
}
else
break;
}
}
else
/* Should never happen; treat as invalid input */
goto invalid_syntax;
/* Check that we got at least one digit */
if (unlikely(cp == firstdigit))
goto invalid_syntax;
/* Add the contribution from the final group of digits */
int64_to_numericvar(mul, &tmp_var);
mul_var(dest, &tmp_var, dest, 0);
int64_to_numericvar(tmp, &tmp_var);
add_var(dest, &tmp_var, dest);
if (dest->weight > SHRT_MAX)
goto out_of_range;
dest->sign = sign;
free_var(&tmp_var);
/* Return end+1 position for caller */
*endptr = cp;
return true;
out_of_range:
ereturn(escontext, false,
(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
errmsg("value overflows numeric format")));
invalid_syntax:
ereturn(escontext, false,
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("invalid input syntax for type %s: \"%s\"",
"numeric", str)));
}
/*
* set_var_from_num() -
*

78
src/test/regress/expected/numeric.out
Unescape View File

@ -2144,6 +2144,12 @@ INSERT INTO num_input_test(n1) VALUES (' -inf ');
INSERT INTO num_input_test(n1) VALUES (' Infinity ');
INSERT INTO num_input_test(n1) VALUES (' +inFinity ');
INSERT INTO num_input_test(n1) VALUES (' -INFINITY ');
INSERT INTO num_input_test(n1) VALUES ('0b10001110111100111100001001010');
INSERT INTO num_input_test(n1) VALUES (' -0B1010101101010100101010011000110011101011000111110000101011010010 ');
INSERT INTO num_input_test(n1) VALUES (' +0o112402761777 ');
INSERT INTO num_input_test(n1) VALUES ('-0O001255245230633431670261');
INSERT INTO num_input_test(n1) VALUES ('-0x0000000000000000000000000deadbeef');
INSERT INTO num_input_test(n1) VALUES (' 0X30b1F33a6DF0bD4E64DF9BdA7D15 ');
-- bad inputs
INSERT INTO num_input_test(n1) VALUES (' ');
ERROR: invalid input syntax for type numeric: " "
@ -2177,27 +2183,61 @@ INSERT INTO num_input_test(n1) VALUES (' N aN ');
ERROR: invalid input syntax for type numeric: " N aN "
LINE 1: INSERT INTO num_input_test(n1) VALUES (' N aN ');
^
INSERT INTO num_input_test(n1) VALUES ('+NaN');
ERROR: invalid input syntax for type numeric: "+NaN"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('+NaN');
^
INSERT INTO num_input_test(n1) VALUES ('-NaN');
ERROR: invalid input syntax for type numeric: "-NaN"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('-NaN');
^
INSERT INTO num_input_test(n1) VALUES ('+ infinity');
ERROR: invalid input syntax for type numeric: "+ infinity"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('+ infinity');
^
INSERT INTO num_input_test(n1) VALUES ('0b1112');
ERROR: invalid input syntax for type numeric: "0b1112"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('0b1112');
^
INSERT INTO num_input_test(n1) VALUES ('0c1112');
ERROR: invalid input syntax for type numeric: "0c1112"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('0c1112');
^
INSERT INTO num_input_test(n1) VALUES ('0o12345678');
ERROR: invalid input syntax for type numeric: "0o12345678"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('0o12345678');
^
INSERT INTO num_input_test(n1) VALUES ('0x1eg');
ERROR: invalid input syntax for type numeric: "0x1eg"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('0x1eg');
^
INSERT INTO num_input_test(n1) VALUES ('0x12.34');
ERROR: invalid input syntax for type numeric: "0x12.34"
LINE 1: INSERT INTO num_input_test(n1) VALUES ('0x12.34');
^
SELECT * FROM num_input_test;
n1
-----------
123
3245874
-93853
555.50
-555.50
NaN
NaN
Infinity
Infinity
-Infinity
Infinity
Infinity
-Infinity
(13 rows)
n1
-----------------------------------
123
3245874
-93853
555.50
-555.50
NaN
NaN
Infinity
Infinity
-Infinity
Infinity
Infinity
-Infinity
299792458
-12345678901234567890
9999999999
-12345678900987654321
-3735928559
987654321234567898765432123456789
(19 rows)
-- Also try it with non-error-throwing API
SELECT pg_input_is_valid('34.5', 'numeric');
@ -2242,6 +2282,12 @@ SELECT pg_input_error_message('1234.567', 'numeric(7,4)');
numeric field overflow
(1 row)
SELECT pg_input_error_message('0x1234.567', 'numeric');
pg_input_error_message
-----------------------------------------------------
invalid input syntax for type numeric: "0x1234.567"
(1 row)
--
-- Test precision and scale typemods
--

48
src/test/regress/expected/numerology.out
Unescape View File

@ -104,9 +104,11 @@ SELECT 0b111111111111111111111111111111111111111111111111111111111111111;
(1 row)
SELECT 0b1000000000000000000000000000000000000000000000000000000000000000;
ERROR: invalid input syntax for type numeric: "0b1000000000000000000000000000000000000000000000000000000000000000"
LINE 1: SELECT 0b100000000000000000000000000000000000000000000000000...
^
?column?
---------------------
9223372036854775808
(1 row)
SELECT 0o777777777777777777777;
?column?
---------------------
@ -114,9 +116,11 @@ SELECT 0o777777777777777777777;
(1 row)
SELECT 0o1000000000000000000000;
ERROR: invalid input syntax for type numeric: "0o1000000000000000000000"
LINE 1: SELECT 0o1000000000000000000000;
^
?column?
---------------------
9223372036854775808
(1 row)
SELECT 0x7FFFFFFFFFFFFFFF;
?column?
---------------------
@ -124,9 +128,11 @@ SELECT 0x7FFFFFFFFFFFFFFF;
(1 row)
SELECT 0x8000000000000000;
ERROR: invalid input syntax for type numeric: "0x8000000000000000"
LINE 1: SELECT 0x8000000000000000;
^
?column?
---------------------
9223372036854775808
(1 row)
SELECT -0b1000000000000000000000000000000000000000000000000000000000000000;
?column?
----------------------
@ -134,9 +140,11 @@ SELECT -0b1000000000000000000000000000000000000000000000000000000000000000;
(1 row)
SELECT -0b1000000000000000000000000000000000000000000000000000000000000001;
ERROR: invalid input syntax for type numeric: "-0b1000000000000000000000000000000000000000000000000000000000000001"
LINE 1: SELECT -0b10000000000000000000000000000000000000000000000000...
^
?column?
----------------------
-9223372036854775809
(1 row)
SELECT -0o1000000000000000000000;
?column?
----------------------
@ -144,9 +152,11 @@ SELECT -0o1000000000000000000000;
(1 row)
SELECT -0o1000000000000000000001;
ERROR: invalid input syntax for type numeric: "-0o1000000000000000000001"
LINE 1: SELECT -0o1000000000000000000001;
^
?column?
----------------------
-9223372036854775809
(1 row)
SELECT -0x8000000000000000;
?column?
----------------------
@ -154,9 +164,11 @@ SELECT -0x8000000000000000;
(1 row)
SELECT -0x8000000000000001;
ERROR: invalid input syntax for type numeric: "-0x8000000000000001"
LINE 1: SELECT -0x8000000000000001;
^
?column?
----------------------
-9223372036854775809
(1 row)
-- error cases
SELECT 123abc;
ERROR: trailing junk after numeric literal at or near "123a"

14
src/test/regress/sql/numeric.sql
Unescape View File

@ -1039,6 +1039,12 @@ INSERT INTO num_input_test(n1) VALUES (' -inf ');
INSERT INTO num_input_test(n1) VALUES (' Infinity ');
INSERT INTO num_input_test(n1) VALUES (' +inFinity ');
INSERT INTO num_input_test(n1) VALUES (' -INFINITY ');
INSERT INTO num_input_test(n1) VALUES ('0b10001110111100111100001001010');
INSERT INTO num_input_test(n1) VALUES (' -0B1010101101010100101010011000110011101011000111110000101011010010 ');
INSERT INTO num_input_test(n1) VALUES (' +0o112402761777 ');
INSERT INTO num_input_test(n1) VALUES ('-0O001255245230633431670261');
INSERT INTO num_input_test(n1) VALUES ('-0x0000000000000000000000000deadbeef');
INSERT INTO num_input_test(n1) VALUES (' 0X30b1F33a6DF0bD4E64DF9BdA7D15 ');
-- bad inputs
INSERT INTO num_input_test(n1) VALUES (' ');
@ -1049,7 +1055,14 @@ INSERT INTO num_input_test(n1) VALUES ('5 . 0');
INSERT INTO num_input_test(n1) VALUES ('5. 0 ');
INSERT INTO num_input_test(n1) VALUES ('');
INSERT INTO num_input_test(n1) VALUES (' N aN ');
INSERT INTO num_input_test(n1) VALUES ('+NaN');
INSERT INTO num_input_test(n1) VALUES ('-NaN');
INSERT INTO num_input_test(n1) VALUES ('+ infinity');
INSERT INTO num_input_test(n1) VALUES ('0b1112');
INSERT INTO num_input_test(n1) VALUES ('0c1112');
INSERT INTO num_input_test(n1) VALUES ('0o12345678');
INSERT INTO num_input_test(n1) VALUES ('0x1eg');
INSERT INTO num_input_test(n1) VALUES ('0x12.34');
SELECT * FROM num_input_test;
@ -1061,6 +1074,7 @@ SELECT pg_input_error_message('1e400000', 'numeric');
SELECT pg_input_is_valid('1234.567', 'numeric(8,4)');
SELECT pg_input_is_valid('1234.567', 'numeric(7,4)');
SELECT pg_input_error_message('1234.567', 'numeric(7,4)');
SELECT pg_input_error_message('0x1234.567', 'numeric');
--
-- Test precision and scale typemods

Write Preview
Loading…
Cancel
Save