std

The <charconv> header from C++17's standard library. More...

Collaboration diagram for std:

Functions
template<typename value_type > requires std::is_floating_point_v<value_type>
to_chars_result	sharg::contrib::charconv_float::to_chars_floating_point (char *first, char *last, value_type value) noexcept
	std::to_chars implementation for floating point via a std::stringstream for default base = 10.
template<typename value_type > requires std::is_floating_point_v<value_type>
from_chars_result	sharg::contrib::charconv_float::from_chars_floating_point (char const *first, char const *last, value_type &value, chars_format fmt=chars_format::general) noexcept
	Delegates to functions strto[d/f/ld] for floating point value extraction.
template<typename floating_point_type > requires std::is_floating_point_v<floating_point_type>
to_chars_result	sharg::contrib::charconv_float::to_chars (char *first, char *last, floating_point_type value) noexcept
	std::to_chars overload for floating point via a std::stringstream for default base = 10.
template<typename floating_point_type > requires std::is_floating_point_v<floating_point_type>
from_chars_result	sharg::contrib::charconv_float::from_chars (char const *first, char const *last, floating_point_type &value, chars_format fmt=chars_format::general) noexcept
	Parse a char sequence into an floating point value.

Detailed Description

The <charconv> header from C++17's standard library.

The following table describes what implementation of std::to_chars and std::from_chars will be used

stdlib version	__cpp_lib_to_chars	chars_format	to_chars_result	from_chars_result	to_chars (int)	from_chars (int)	to_chars (float)	from_chars (float)
gcc 10	undefined and <charconv> header	stdlib	stdlib	stdlib	stdlib	stdlib	shim (ostringstream)	shim (strto[f/d/ld])
gcc 11	undefined (or 201611) and <charconv> header	stdlib	stdlib	stdlib	stdlib	stdlib	stdlib	stdlib

Note: gcc 11 implements float too, but does not define __cpp_lib_to_chars

Function Documentation

from_chars()

template<typename floating_point_type >
requires std::is_floating_point_v<floating_point_type>

from_chars_result sharg::contrib::charconv_float::from_chars ( char const * first, char const * last, floating_point_type & value, chars_format fmt = chars_format::general )

inlinenoexcept

Parse a char sequence into an floating point value.

Template Parameters

floating_point_type

The type to parse the string into; Must model std::is_floating_point_v.

Parameters

[in]	first	The start of the string to parse.
[in]	last	The end of the string to parse.
[in,out]	value	The value to store the parsed result in.
[in]	fmt	The std::chars_format that alters the behaviour of parsing.

Returns

A std::from_char_result. See detail section return value for more information.

Analyzes the character sequence [first,last) for a pattern described below. If no characters match the pattern or if the value obtained by parsing the matched characters is not representable in the type of value, value is unmodified, otherwise the characters matching the pattern are interpreted as a text representation of an arithmetic value, which is stored in value.

Floating-point parsers: Expects the pattern identical to the one used by std::strtod in the default ("C") locale, except that:

• the plus sign is not recognized outside of the exponent (only the minus sign is permitted at the beginning)
• if fmt has std::chars_format::scientific set but not std::chars_format::fixed, the exponent part is required (otherwise it is optional)
• if fmt has std::chars_format::fixed set but not std::chars_format::scientific, the optional exponent is not permitted
• if fmt is std::chars_format::hex, the prefix "0x" or "0X" is not permitted (the string "0x123" parses as the value "0" with unparsed remainder "x123").

Attention

This implementation is a workaround until the function is supported by the compiler. It falls back to use the functions strto[d/f/ld] before checking the above limitations

Return value

This function is workaround until the function is supported by the compiler. It falls back to use the functions strto[d/f/ld] so the return value is NOT as documented here https://en.cppreference.com/w/cpp/utility/from_chars but:

On success, std::from_chars_result::ec is value-initialized. On error, std::from_chars_result::ec is either an std::errc::invalid_argument if an illegal character or format has been encountered, or std::errc::out_of_range if parsing the value would cause an overflow. The std::from_chars_result::ptr value is always set to last.

The locale issue

std::from_chars is documented to be locale independent. The accepted patterns are identical to the one used by strtod in the defailt ("C") locale.

The functions strto[d/f/ld] used here are locale dependent but setting the locale manually by std::setlocale is not thread safe. So for the time being this workaround is locale dependent.

See also

https://en.cppreference.com/w/cpp/utility/from_chars