py: Fix handling of "0" mpz in some functions.

This commit is contained in:
Damien George 2015-01-09 20:58:58 +00:00
parent f5465b9eb0
commit 58056b0f43

View File

@ -653,6 +653,11 @@ void mpz_set(mpz_t *dest, const mpz_t *src) {
}
void mpz_set_from_int(mpz_t *z, mp_int_t val) {
if (val == 0) {
z->len = 0;
return;
}
mpz_need_dig(z, MPZ_NUM_DIG_FOR_INT);
mp_uint_t uval;
@ -709,19 +714,19 @@ typedef uint32_t mp_float_int_t;
z->neg = u.p.sgn;
if (u.p.exp == 0) {
// value == 0 || value < 1
mpz_init_zero(z);
mpz_set_from_int(z, 0);
} else if (u.p.exp == ((1 << EXP_SZ) - 1)) {
// u.p.frc == 0 indicates inf, else NaN
// should be handled by caller
mpz_init_zero(z);
mpz_set_from_int(z, 0);
} else {
const int adj_exp = (int)u.p.exp - ((1 << (EXP_SZ - 1)) - 1);
if (adj_exp < 0) {
// value < 1 , truncates to 0
mpz_init_zero(z);
mpz_set_from_int(z, 0);
} else if (adj_exp == 0) {
// 1 <= value < 2 , so truncates to 1
mpz_init_from_int(z, 1);
mpz_set_from_int(z, 1);
} else {
// 2 <= value
const int dig_cnt = (adj_exp + 1 + (DIG_SIZE - 1)) / DIG_SIZE;
@ -1353,7 +1358,7 @@ mp_int_t mpz_hash(const mpz_t *z) {
mp_int_t val = 0;
mpz_dig_t *d = z->dig + z->len;
while (--d >= z->dig) {
while (d-- > z->dig) {
val = (val << DIG_SIZE) | *d;
}
@ -1368,7 +1373,7 @@ bool mpz_as_int_checked(const mpz_t *i, mp_int_t *value) {
mp_int_t val = 0;
mpz_dig_t *d = i->dig + i->len;
while (--d >= i->dig) {
while (d-- > i->dig) {
if (val > (~(WORD_MSBIT_HIGH) >> DIG_SIZE)) {
// will overflow
return false;
@ -1393,7 +1398,7 @@ bool mpz_as_uint_checked(const mpz_t *i, mp_uint_t *value) {
mp_uint_t val = 0;
mpz_dig_t *d = i->dig + i->len;
while (--d >= i->dig) {
while (d-- > i->dig) {
if (val > (~(WORD_MSBIT_HIGH) >> (DIG_SIZE - 1))) {
// will overflow
return false;
@ -1410,7 +1415,7 @@ mp_float_t mpz_as_float(const mpz_t *i) {
mp_float_t val = 0;
mpz_dig_t *d = i->dig + i->len;
while (--d >= i->dig) {
while (d-- > i->dig) {
val = val * DIG_BASE + *d;
}