Minor grisu improvements

This commit is contained in:
Victor Zverovich 2019-02-02 07:40:43 -08:00
parent 7fbbfed8c6
commit e61cac687d

View File

@ -332,14 +332,17 @@ const char basic_data<T>::BACKGROUND_COLOR[] = "\x1b[48;2;";
template <typename T> const char basic_data<T>::RESET_COLOR[] = "\x1b[0m";
template <typename T> const wchar_t basic_data<T>::WRESET_COLOR[] = L"\x1b[0m";
template <typename T> struct bits {
static FMT_CONSTEXPR_DECL const int value =
sizeof(T) * std::numeric_limits<unsigned char>::digits;
};
// A handmade floating-point number f * pow(2, e).
class fp {
private:
typedef uint64_t significand_type;
// All sizes are in bits.
static FMT_CONSTEXPR_DECL const int char_size =
std::numeric_limits<unsigned char>::digits;
// Subtract 1 to account for an implicit most significant bit in the
// normalized form.
static FMT_CONSTEXPR_DECL const int double_significand_size =
@ -352,7 +355,7 @@ class fp {
int e;
static FMT_CONSTEXPR_DECL const int significand_size =
sizeof(significand_type) * char_size;
bits<significand_type>::value;
fp() : f(0), e(0) {}
fp(uint64_t f_val, int e_val) : f(f_val), e(e_val) {}
@ -362,9 +365,8 @@ class fp {
template <typename Double> explicit fp(Double d) {
// Assume double is in the format [sign][exponent][significand].
typedef std::numeric_limits<Double> limits;
const int double_size = static_cast<int>(sizeof(Double) * char_size);
const int exponent_size =
double_size - double_significand_size - 1; // -1 for sign
bits<Double>::value - double_significand_size - 1; // -1 for sign
const uint64_t significand_mask = implicit_bit - 1;
const uint64_t exponent_mask = (~0ull >> 1) & ~significand_mask;
const int exponent_bias = (1 << exponent_size) - limits::max_exponent - 1;
@ -463,9 +465,10 @@ FMT_FUNC bool grisu2_round(char* buf, int& size, int max_digits, uint64_t delta,
}
// Generates output using Grisu2 digit-gen algorithm.
FMT_FUNC bool grisu2_gen_digits(char* buf, int& size, uint32_t hi, uint64_t lo,
int& exp, uint64_t delta, const fp& one,
const fp& diff, int max_digits) {
FMT_FUNC int grisu2_gen_digits(char* buf, uint32_t hi, uint64_t lo, int& exp,
uint64_t delta, const fp& one, const fp& diff,
int max_digits) {
int size = 0;
// Generate digits for the most significant part (hi).
while (exp > 0) {
uint32_t digit = 0;
@ -519,10 +522,12 @@ FMT_FUNC bool grisu2_gen_digits(char* buf, int& size, uint32_t hi, uint64_t lo,
--exp;
uint64_t remainder = (static_cast<uint64_t>(hi) << -one.e) + lo;
if (remainder <= delta || size > max_digits) {
return grisu2_round(
buf, size, max_digits, delta, remainder,
static_cast<uint64_t>(data::POWERS_OF_10_32[exp]) << -one.e, diff.f,
exp);
return grisu2_round(buf, size, max_digits, delta, remainder,
static_cast<uint64_t>(data::POWERS_OF_10_32[exp])
<< -one.e,
diff.f, exp)
? size
: -1;
}
}
// Generate digits for the least significant part (lo).
@ -535,7 +540,9 @@ FMT_FUNC bool grisu2_gen_digits(char* buf, int& size, uint32_t hi, uint64_t lo,
--exp;
if (lo < delta || size > max_digits) {
return grisu2_round(buf, size, max_digits, delta, lo, one.f,
diff.f * data::POWERS_OF_10_32[-exp], exp);
diff.f * data::POWERS_OF_10_32[-exp], exp)
? size
: -1;
}
}
}
@ -742,11 +749,10 @@ grisu2_format(Double value, buffer& buf, core_format_specs specs) {
upper = upper * cached_pow; // \tilde{M}^+ in Grisu.
--upper.f; // \tilde{M}^+ - 1 ulp -> M^+_{\downarrow}.
fp one(1ull << -upper.e, upper.e);
// hi (p1 in Grisu) contains the most significant digits of scaled_upper.
// hi (p1 in Grisu) contains the most significant digits of scaled upper.
// hi = floor(upper / one).
uint32_t hi = static_cast<uint32_t>(upper.f >> -one.e);
int exp = count_digits(hi); // kappa in Grisu.
gen_digits_params params = process_specs(specs, cached_exp + exp, buf);
fp_value.normalize();
fp scaled_value = fp_value * cached_pow;
lower = lower * cached_pow; // \tilde{M}^- in Grisu.
@ -754,11 +760,12 @@ grisu2_format(Double value, buffer& buf, core_format_specs specs) {
uint64_t delta = upper.f - lower.f;
fp diff = upper - scaled_value; // wp_w in Grisu.
// lo (p2 in Grisu) contains the least significants digits of scaled_upper.
// lo = supper % one.
// lo = upper % one.
uint64_t lo = upper.f & (one.f - 1);
int size = 0;
if (!grisu2_gen_digits(buf.data(), size, hi, lo, exp, delta, one, diff,
params.num_digits)) {
gen_digits_params params = process_specs(specs, cached_exp + exp, buf);
int size = grisu2_gen_digits(buf.data(), hi, lo, exp, delta, one, diff,
params.num_digits);
if (size < 0) {
buf.clear();
return false;
}