Skip to content

J
json
Unverified Commit 909b439b authored by Niels Lohmann's avatar Niels Lohmann
Browse files
Options

⚡ micro-optimizations for dump() 

numtostr now directly writes to a stream. Return value of snprintf is
reused to avoid finding end of string. Cachegrind suggests a 1%
performance increase.
parent 0f04e42d
Hide whitespace changes
Inline Side-by-side
Showing with 52 additions and 92 deletions
src/json.hpp
View file @ 909b439b
......@@ -8252,14 +8252,9 @@ class basic_json
{
public:
template<typename NumberType>
numtostr(NumberType value)
numtostr(NumberType value, std::ostream& o)
{
x_write(value, std::is_integral<NumberType>());
}
const char* c_str() const
{
return m_buf.data();
x_write(value, std::is_integral<NumberType>(), o);
}
private:
......@@ -8267,12 +8262,12 @@ class basic_json
std::array < char, 64 > m_buf{{}};
template<typename NumberType>
void x_write(NumberType x, /*is_integral=*/std::true_type)
void x_write(NumberType x, /*is_integral=*/std::true_type, std::ostream& o)
{
// special case for "0"
if (x == 0)
{
m_buf[0] = '0';
o.put('0');
return;
}
......@@ -8298,30 +8293,31 @@ class basic_json
}
std::reverse(m_buf.begin(), m_buf.begin() + i);
o.write(m_buf.data(), static_cast<std::streamsize>(i));
}
template<typename NumberType>
void x_write(NumberType x, /*is_integral=*/std::false_type)
void x_write(NumberType x, /*is_integral=*/std::false_type, std::ostream& o)
{
// special case for 0.0 and -0.0
if (x == 0)
{
size_t i = 0;
if (std::signbit(x))
{
m_buf[i++] = '-';
o.write("-0.0", 4);
}
else
{
o.write("0.0", 3);
}
m_buf[i++] = '0';
m_buf[i++] = '.';
m_buf[i] = '0';
return;
}
// get number of digits for a text -> float -> text round-trip
static constexpr auto d = std::numeric_limits<NumberType>::digits10;
// the actual conversion
const auto written_bytes = snprintf(m_buf.data(), m_buf.size(), "%.*g", d, x);
// the actual conversion; store how many bytes have been written
auto written_bytes = snprintf(m_buf.data(), m_buf.size(), "%.*g", d, x);
// negative value indicates an error
assert(written_bytes > 0);
......@@ -8342,6 +8338,7 @@ class basic_json
{
const auto end = std::remove(m_buf.begin(), m_buf.begin() + written_bytes, thousands_sep);
std::fill(end, m_buf.end(), '\0');
written_bytes -= m_buf.end() - end;
}
// convert decimal point to '.'
......@@ -8357,36 +8354,19 @@ class basic_json
}
}
o.write(m_buf.data(), static_cast<std::streamsize>(written_bytes));
// determine if need to append ".0"
size_t i = 0;
bool value_is_int_like = true;
for (i = 0; i < m_buf.size(); ++i)
const bool value_is_int_like = std::all_of(m_buf.begin(),
m_buf.begin() + written_bytes + 1,
[](char c)
{
// break when end of number is reached
if (m_buf[i] == '\0')
{
break;
}
// check if we find non-int character
value_is_int_like = value_is_int_like and m_buf[i] != '.' and
m_buf[i] != 'e' and m_buf[i] != 'E';
}
// we use %g above, so there cannot be an 'E' character
return c != '.' and c != 'e';
});
if (value_is_int_like)
{
// there must be 2 bytes left for ".0"
assert((i + 2) < m_buf.size());
// we write to the end of the number
assert(m_buf[i] == '\0');
assert(m_buf[i - 1] != '\0');
// add ".0"
m_buf[i] = '.';
m_buf[i + 1] = '0';
// the resulting string is properly terminated
assert(m_buf[i + 2] == '\0');
o.write(".0", 2);
}
}
};
......@@ -8566,19 +8546,19 @@ class basic_json
case value_t::number_integer:
{
o << numtostr(m_value.number_integer).c_str();
numtostr(m_value.number_integer, o);
return;
}
case value_t::number_unsigned:
{
o << numtostr(m_value.number_unsigned).c_str();
numtostr(m_value.number_unsigned, o);
return;
}
case value_t::number_float:
{
o << numtostr(m_value.number_float).c_str();
numtostr(m_value.number_float, o);
return;
}
......
src/json.hpp.re2c
View file @ 909b439b
......@@ -8252,14 +8252,9 @@ class basic_json
{
public:
template<typename NumberType>
numtostr(NumberType value)
numtostr(NumberType value, std::ostream& o)
{
x_write(value, std::is_integral<NumberType>());
}
const char* c_str() const
{
return m_buf.data();
x_write(value, std::is_integral<NumberType>(), o);
}
private:
......@@ -8267,12 +8262,12 @@ class basic_json
std::array < char, 64 > m_buf{{}};
template<typename NumberType>
void x_write(NumberType x, /*is_integral=*/std::true_type)
void x_write(NumberType x, /*is_integral=*/std::true_type, std::ostream& o)
{
// special case for "0"
if (x == 0)
{
m_buf[0] = '0';
o.put('0');
return;
}
......@@ -8298,30 +8293,31 @@ class basic_json
}
std::reverse(m_buf.begin(), m_buf.begin() + i);
o.write(m_buf.data(), static_cast<std::streamsize>(i));
}
template<typename NumberType>
void x_write(NumberType x, /*is_integral=*/std::false_type)
void x_write(NumberType x, /*is_integral=*/std::false_type, std::ostream& o)
{
// special case for 0.0 and -0.0
if (x == 0)
{
size_t i = 0;
if (std::signbit(x))
{
m_buf[i++] = '-';
o.write("-0.0", 4);
}
else
{
o.write("0.0", 3);
}
m_buf[i++] = '0';
m_buf[i++] = '.';
m_buf[i] = '0';
return;
}
// get number of digits for a text -> float -> text round-trip
static constexpr auto d = std::numeric_limits<NumberType>::digits10;
// the actual conversion
const auto written_bytes = snprintf(m_buf.data(), m_buf.size(), "%.*g", d, x);
// the actual conversion; store how many bytes have been written
auto written_bytes = snprintf(m_buf.data(), m_buf.size(), "%.*g", d, x);
// negative value indicates an error
assert(written_bytes > 0);
......@@ -8342,6 +8338,7 @@ class basic_json
{
const auto end = std::remove(m_buf.begin(), m_buf.begin() + written_bytes, thousands_sep);
std::fill(end, m_buf.end(), '\0');
written_bytes -= m_buf.end() - end;
}
// convert decimal point to '.'
......@@ -8357,36 +8354,19 @@ class basic_json
}
}
o.write(m_buf.data(), static_cast<std::streamsize>(written_bytes));
// determine if need to append ".0"
size_t i = 0;
bool value_is_int_like = true;
for (i = 0; i < m_buf.size(); ++i)
const bool value_is_int_like = std::all_of(m_buf.begin(),
m_buf.begin() + written_bytes + 1,
[](char c)
{
// break when end of number is reached
if (m_buf[i] == '\0')
{
break;
}
// check if we find non-int character
value_is_int_like = value_is_int_like and m_buf[i] != '.' and
m_buf[i] != 'e' and m_buf[i] != 'E';
}
// we use %g above, so there cannot be an 'E' character
return c != '.' and c != 'e';
});
if (value_is_int_like)
{
// there must be 2 bytes left for ".0"
assert((i + 2) < m_buf.size());
// we write to the end of the number
assert(m_buf[i] == '\0');
assert(m_buf[i - 1] != '\0');
// add ".0"
m_buf[i] = '.';
m_buf[i + 1] = '0';
// the resulting string is properly terminated
assert(m_buf[i + 2] == '\0');
o.write(".0", 2);
}
}
};
......@@ -8566,19 +8546,19 @@ class basic_json
case value_t::number_integer:
{
o << numtostr(m_value.number_integer).c_str();
numtostr(m_value.number_integer, o);
return;
}
case value_t::number_unsigned:
{
o << numtostr(m_value.number_unsigned).c_str();
numtostr(m_value.number_unsigned, o);
return;
}
case value_t::number_float:
{
o << numtostr(m_value.number_float).c_str();
numtostr(m_value.number_float, o);
return;
}
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment