fixed some number conversion bugs
diff --git a/inc/qcbor/qcbor_decode.h b/inc/qcbor/qcbor_decode.h
index 82cfaec..1d6a37a 100644
--- a/inc/qcbor/qcbor_decode.h
+++ b/inc/qcbor/qcbor_decode.h
@@ -1052,6 +1052,10 @@
Not that all these types can support numbers much larger that can be represented by
in a 64-bit integer, so \ref QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW will
often be encountered.
+
+ When converting bignums and decimal fractions \ref QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW
+ will be set if the result is below 1, unless the mantissa is zero, in which
+ case the coversion is successful and the value of 0 is returned.
*/
void QCBORDecode_GetInt64ConvertAll(QCBORDecodeContext *pCtx, uint32_t uOptions, int64_t *pnValue);
@@ -1091,7 +1095,7 @@
The sames as QCBORDecode_GetInt64ConvertAll(), but returns an unsigned integer and thus
sets \ref QCBOR_ERR_NUMBER_SIGN_CONVERSION
- is set if the value to be decoded is negatve.
+ if the value to be decoded is negatve.
See also QCBORDecode_GetUint64Convert() and QCBORDecode_GetUint64ConvertAll().
*/
diff --git a/src/qcbor_decode.c b/src/qcbor_decode.c
index c418677..d9117e5 100644
--- a/src/qcbor_decode.c
+++ b/src/qcbor_decode.c
@@ -2596,31 +2596,30 @@
// The main exponentiator that works on only positive numbers
-static QCBORError Exponentitate10UU(uint64_t uMantissa, int64_t nExponent, uint64_t *puResult)
+static QCBORError Exponentitate10(uint64_t uMantissa, int64_t nExponent, uint64_t *puResult)
{
- uint64_t uResult;
+ uint64_t uResult = uMantissa;
- uResult = uMantissa;
-
- /* This loop will run a maximum of 19 times because
- * UINT64_MAX < 10 ^^ 19. More than that will cause
- * exit with the overflow error
- */
- while(nExponent > 1) {
- if(uResult > UINT64_MAX / 10) {
- return QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW; // Error overflow
+ if(uResult != 0) {
+ /* This loop will run a maximum of 19 times because
+ * UINT64_MAX < 10 ^^ 19. More than that will cause
+ * exit with the overflow error
+ */
+ for(; nExponent > 0; nExponent--) {
+ if(uResult > UINT64_MAX / 10) {
+ return QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW; // Error overflow
+ }
+ uResult = uResult * 10;
}
- uResult = uResult * 10;
- nExponent--;
- }
- while(nExponent < 0 ) {
- if(uResult == 0) {
- return QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW; // Underflow error
+ for(; nExponent < 0; nExponent++) {
+ uResult = uResult / 10;
+ if(uResult == 0) {
+ return QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW; // Underflow error
+ }
}
- uResult = uResult / 10;
- nExponent--;
}
+ /* else, mantissa is zero so this returns zero */
*puResult = uResult;
@@ -2633,7 +2632,7 @@
will not fit in an int64_t and this will error out with
under or overflow
*/
-static QCBORError Exponentitate2UU(uint64_t uMantissa, int64_t nExponent, uint64_t *puResult)
+static QCBORError Exponentitate2(uint64_t uMantissa, int64_t nExponent, uint64_t *puResult)
{
uint64_t uResult;
@@ -2656,7 +2655,7 @@
return QCBOR_ERR_CONVERSION_UNDER_OVER_FLOW; // Underflow error
}
uResult = uResult >> 1;
- nExponent--;
+ nExponent++;
}
*puResult = uResult;
@@ -2664,12 +2663,15 @@
return QCBOR_SUCCESS;
}
-
+/*
+ Compute value with signed mantissa and signed result. Works with exponent of 2 or 10 based on exponentiator.
+ */
static inline QCBORError ExponentiateNN(int64_t nMantissa, int64_t nExponent, int64_t *pnResult, fExponentiator pfExp)
{
uint64_t uResult;
- // Take the absolute value of the mantissa
+ // Take the absolute value of the mantissa and convert to unsigned.
+ // TODO: this should be possible in one intruction
uint64_t uMantissa = nMantissa > 0 ? (uint64_t)nMantissa : (uint64_t)-nMantissa;
// Do the exponentiation of the positive mantissa
@@ -2699,7 +2701,9 @@
return QCBOR_SUCCESS;
}
-
+/*
+ Compute value with signed mantissa and unsigned result. Works with exponent of 2 or 10 based on exponentiator.
+ */
static inline QCBORError ExponentitateNU(int64_t nMantissa, int64_t nExponent, uint64_t *puResult, fExponentiator pfExp)
{
if(nMantissa < 0) {
@@ -2714,27 +2718,9 @@
#include <math.h>
-/*
-static inline uint8_t Exponentitate10F(uint64_t uMantissa, int64_t nExponent, double *pfResult)
-{
- // TODO: checkout exceptions; what is HUGE_VAL?
- *pfResult = pow((double)10, (double)nExponent) * (double)uMantissa;
-
- //if(*pfResult == HUGE_VAL)
- return 0;
-}
-*/
-/*
- A) bignum is positive
- A1) output is signed INT64_MAX
- A2) output is unsigned UINT64_MAX
- B) bignum is negative
- B1) output is signed INT64_MAX
- B2) output is unsigned error
- */
-static inline QCBORError ConvertBigNumToUnsigned(const UsefulBufC BigNum, uint64_t uMax, uint64_t *pResult)
+static QCBORError ConvertBigNumToUnsigned(const UsefulBufC BigNum, uint64_t uMax, uint64_t *pResult)
{
uint64_t uResult;
@@ -2752,27 +2738,6 @@
return QCBOR_SUCCESS;
}
-
-
-static double ConvertBigNumToDouble(const UsefulBufC BigNum)
-{
- double dResult;
-
- dResult = 0.0;
- const uint8_t *pByte = BigNum.ptr;
- size_t uLen = BigNum.len;
- /* This will overflow and become the float value INFINITY if the number
- is too large to fit. No error will be logged.
- TODO: should an error be logged? */
- while(uLen--) {
- dResult = (dResult * 256.0) + (double)*pByte++;
- }
-
- return dResult;
-}
-
-
-
static inline QCBORError ConvertPositiveBigNumToUnsigned(const UsefulBufC BigNum, uint64_t *pResult)
{
return ConvertBigNumToUnsigned(BigNum, UINT64_MAX, pResult);
@@ -2914,9 +2879,9 @@
case QCBOR_TYPE_DECIMAL_FRACTION:
if(uOptions & QCBOR_CONVERT_TYPE_DECIMAL_FRACTION) {
pMe->uLastError = (uint8_t)ExponentiateNN(Item.val.expAndMantissa.Mantissa.nInt,
- Item.val.expAndMantissa.nExponent,
- pValue,
- &Exponentitate10UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ &Exponentitate10);
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
}
@@ -2925,9 +2890,9 @@
case QCBOR_TYPE_BIGFLOAT:
if(uOptions & QCBOR_CONVERT_TYPE_BIGFLOAT) {
pMe->uLastError = (uint8_t)ExponentiateNN(Item.val.expAndMantissa.Mantissa.nInt,
- Item.val.expAndMantissa.nExponent,
- pValue,
- &Exponentitate2UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate2);
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
}
@@ -2939,16 +2904,11 @@
int64_t nMantissa;
pMe->uLastError = (uint8_t)ConvertPositiveBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
- int64_t nMultiplier;
- pMe->uLastError = (uint8_t)ExponentiateNN(10,
+ pMe->uLastError = (uint8_t)ExponentiateNN(nMantissa,
Item.val.expAndMantissa.nExponent,
- &nMultiplier,
- &Exponentitate10UU);
- if(!pMe->uLastError) {
- // TODO: overflow
- *pValue = nMantissa * nMultiplier;
- }
- }
+ pValue,
+ Exponentitate10);
+ }
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
}
@@ -2960,9 +2920,9 @@
pMe->uLastError = (uint8_t)ConvertNegativeBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
pMe->uLastError = (uint8_t)ExponentiateNN(nMantissa,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate10UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate10);
}
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
@@ -2975,9 +2935,9 @@
pMe->uLastError = (uint8_t)ConvertPositiveBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
pMe->uLastError = (uint8_t)ExponentiateNN(nMantissa,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate2UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate2);
}
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
@@ -2990,9 +2950,9 @@
pMe->uLastError = (uint8_t)ConvertNegativeBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
pMe->uLastError = (uint8_t)ExponentiateNN(nMantissa,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate2UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate2);
}
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
@@ -3119,9 +3079,9 @@
case QCBOR_TYPE_DECIMAL_FRACTION:
if(uOptions & QCBOR_CONVERT_TYPE_DECIMAL_FRACTION) {
pMe->uLastError = (uint8_t)ExponentitateNU(Item.val.expAndMantissa.Mantissa.nInt,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate10UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate10);
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
}
@@ -3130,9 +3090,9 @@
case QCBOR_TYPE_BIGFLOAT:
if(uOptions & QCBOR_CONVERT_TYPE_BIGFLOAT) {
pMe->uLastError = (uint8_t)ExponentitateNU(Item.val.expAndMantissa.Mantissa.nInt,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate2UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate2);
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
}
@@ -3142,17 +3102,15 @@
case QCBOR_TYPE_DECIMAL_FRACTION_POS_BIGNUM:
if(uOptions & QCBOR_CONVERT_TYPE_DECIMAL_FRACTION) {
- uint64_t uMantissa;
- pMe->uLastError = (uint8_t)ConvertPositiveBigNumToUnsigned(Item.val.expAndMantissa.Mantissa.bigNum, &uMantissa);
+ // TODO: Would be better to convert to unsigned
+ int64_t nMantissa;
+ pMe->uLastError = (uint8_t)ConvertPositiveBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
- uint64_t uMultiplier;
- pMe->uLastError = (uint8_t)ExponentitateNU(10,
- Item.val.expAndMantissa.nExponent,
- &uMultiplier,
- Exponentitate10UU);
- if(!pMe->uLastError) {
- *pValue = uMantissa * uMultiplier;
- }
+ pMe->uLastError = (uint8_t)ExponentitateNU(nMantissa,
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate10);
+
}
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
@@ -3169,13 +3127,14 @@
case QCBOR_TYPE_BIGFLOAT_POS_BIGNUM:
if(uOptions & QCBOR_CONVERT_TYPE_DECIMAL_FRACTION) {
+ // TODO: Would be better to convert to unsigned
int64_t nMantissa;
pMe->uLastError = (uint8_t)ConvertPositiveBigNumToSigned(Item.val.expAndMantissa.Mantissa.bigNum, &nMantissa);
if(!pMe->uLastError) {
pMe->uLastError = (uint8_t)ExponentitateNU(nMantissa,
- Item.val.expAndMantissa.nExponent,
- pValue,
- Exponentitate2UU);
+ Item.val.expAndMantissa.nExponent,
+ pValue,
+ Exponentitate2);
}
} else {
pMe->uLastError = QCBOR_ERR_CONVERSION_NOT_REQUESTED;
@@ -3252,6 +3211,23 @@
}
+static double ConvertBigNumToDouble(const UsefulBufC BigNum)
+{
+ double dResult;
+
+ dResult = 0.0;
+ const uint8_t *pByte = BigNum.ptr;
+ size_t uLen = BigNum.len;
+ /* This will overflow and become the float value INFINITY if the number
+ is too large to fit. No error will be logged.
+ TODO: should an error be logged? */
+ while(uLen--) {
+ dResult = (dResult * 256.0) + (double)*pByte++;
+ }
+
+ return dResult;
+}
+
/*
Public function, see header qcbor/qcbor_decode.h file
*/
diff --git a/test/qcbor_decode_tests.c b/test/qcbor_decode_tests.c
index 86e75e7..cd56325 100644
--- a/test/qcbor_decode_tests.c
+++ b/test/qcbor_decode_tests.c
@@ -4002,7 +4002,7 @@
static const struct NumberConversion NumberConversions[] = {
{
- "Decimal Fraction with negative bignum [3, 257]",
+ "Decimal Fraction with positive bignum 257 * 10e3",
{(uint8_t[]){0xC4, 0x82, 0x1B, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03,
0xC2, 0x42, 0x01, 0x01}, 15},
257000,
@@ -4013,7 +4013,7 @@
QCBOR_SUCCESS
},
{
- "bigfloat with negative bignum [3, 257]",
+ "bigfloat with negative bignum -257 * 2e3",
{(uint8_t[]){0xC5, 0x82, 0x1B, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03,
0xC3, 0x42, 0x01, 0x01}, 15},
-2056,
@@ -4024,7 +4024,7 @@
QCBOR_SUCCESS
},
{
- "bigfloat with positive bignum [3, 257]",
+ "bigfloat with positive bignum 257 * 2e3",
{(uint8_t[]){0xC5, 0x82, 0x1B, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03,
0xC2, 0x42, 0x01, 0x01}, 15},
2056,
@@ -4172,10 +4172,10 @@
int32_t IntegerConvertTest()
{
- const size_t nNumTests = sizeof(NumberConversions)/sizeof(struct NumberConversion);
+ const int nNumTests = sizeof(NumberConversions)/sizeof(struct NumberConversion);
- for(const struct NumberConversion *pF = NumberConversions; pF < NumberConversions + nNumTests; pF++) {
- const size_t nIndex = (size_t)(pF - NumberConversions)/sizeof(struct NumberConversion);
+ for(int nIndex = 0; nIndex < nNumTests; nIndex++) {
+ const struct NumberConversion *pF = &NumberConversions[nIndex];
// Set up the decoding context including a memory pool so that
// indefinite length items can be checked