class StringToDoubleConverter

Performs the conversion.

The output parameter 'processed_characters_count' is set to the number

of characters that have been processed to read the number.

Spaces than are processed with ALLOW_{LEADING|TRAILING}_SPACES are included

in the 'processed_characters_count'. Trailing junk is never included.

Same as StringToDouble above but for 16 bit characters.

Flags should be a bit-or combination of the possible Flags-enum.

- NO_FLAGS: no special flags.

- ALLOW_HEX: recognizes the prefix "0x". Hex numbers may only be integers.

Ex: StringToDouble("0x1234") -> 4660.0

In StringToDouble("0x1234.56") the characters ".56" are trailing

junk. The result of the call is hence dependent on

the ALLOW_TRAILING_JUNK flag and/or the junk value.

With this flag "0x" is a junk-string. Even with ALLOW_TRAILING_JUNK,

the string will not be parsed as "0" followed by junk.

- ALLOW_OCTALS: recognizes the prefix "0" for octals:

If a sequence of octal digits starts with '0', then the number is

read as octal integer. Octal numbers may only be integers.

Ex: StringToDouble("01234") -> 668.0

StringToDouble("012349") -> 12349.0 // Not a sequence of octal

// digits.

In StringToDouble("01234.56") the characters ".56" are trailing

junk. The result of the call is hence dependent on

the ALLOW_TRAILING_JUNK flag and/or the junk value.

In StringToDouble("01234e56") the characters "e56" are trailing

junk, too.

- ALLOW_TRAILING_JUNK: ignore trailing characters that are not part of

a double literal.

- ALLOW_LEADING_SPACES: skip over leading whitespace, including spaces,

new-lines, and tabs.

- ALLOW_TRAILING_SPACES: ignore trailing whitespace.

- ALLOW_SPACES_AFTER_SIGN: ignore whitespace after the sign.

Ex: StringToDouble("- 123.2") -> -123.2.

StringToDouble("+ 123.2") -> 123.2

- ALLOW_CASE_INSENSITIVITY: ignore case of characters for special values:

infinity and nan.

- ALLOW_HEX_FLOATS: allows hexadecimal float literals.

This *must* start with "0x" and separate the exponent with "p".

Examples: 0x1.2p3 == 9.0

0x10.1p0 == 16.0625

ALLOW_HEX and ALLOW_HEX_FLOATS are indendent.

empty_string_value is returned when an empty string is given as input.

If ALLOW_LEADING_SPACES or ALLOW_TRAILING_SPACES are set, then a string

containing only spaces is converted to the 'empty_string_value', too.

junk_string_value is returned when

a) ALLOW_TRAILING_JUNK is not set, and a junk character (a character not

part of a double-literal) is found.

b) ALLOW_TRAILING_JUNK is set, but the string does not start with a

double literal.

infinity_symbol and nan_symbol are strings that are used to detect

inputs that represent infinity and NaN. They can be null, in which case

they are ignored.

The conversion routine first reads any possible signs. Then it compares the

following character of the input-string with the first character of

the infinity, and nan-symbol. If either matches, the function assumes, that

a match has been found, and expects the following input characters to match

the remaining characters of the special-value symbol.

This means that the following restrictions apply to special-value symbols:

- they must not start with signs ('+', or '-'),

- they must not have the same first character.

- they must not start with digits.

If the separator character is not kNoSeparator, then that specific

character is ignored when in between two valid digits of the significant.

It is not allowed to appear in the exponent.

It is not allowed to lead or trail the number.

It is not allowed to appear twice next to each other.

Examples:

flags = ALLOW_HEX | ALLOW_TRAILING_JUNK,

empty_string_value = 0.0,

junk_string_value = NaN,

infinity_symbol = "infinity",

nan_symbol = "nan":

StringToDouble("0x1234") -> 4660.0.

StringToDouble("0x1234K") -> 4660.0.

StringToDouble("") -> 0.0 // empty_string_value.

StringToDouble(" ") -> NaN // junk_string_value.

StringToDouble(" 1") -> NaN // junk_string_value.

StringToDouble("0x") -> NaN // junk_string_value.

StringToDouble("-123.45") -> -123.45.

StringToDouble("--123.45") -> NaN // junk_string_value.

StringToDouble("123e45") -> 123e45.

StringToDouble("123E45") -> 123e45.

StringToDouble("123e+45") -> 123e45.

StringToDouble("123E-45") -> 123e-45.

StringToDouble("123e") -> 123.0 // trailing junk ignored.

StringToDouble("123e-") -> 123.0 // trailing junk ignored.

StringToDouble("+NaN") -> NaN // NaN string literal.

StringToDouble("-infinity") -> -inf. // infinity literal.

StringToDouble("Infinity") -> NaN // junk_string_value.

flags = ALLOW_OCTAL | ALLOW_LEADING_SPACES,

empty_string_value = 0.0,

junk_string_value = NaN,

infinity_symbol = NULL,

nan_symbol = NULL:

StringToDouble("0x1234") -> NaN // junk_string_value.

StringToDouble("01234") -> 668.0.

StringToDouble("") -> 0.0 // empty_string_value.

StringToDouble(" ") -> 0.0 // empty_string_value.

StringToDouble(" 1") -> 1.0

StringToDouble("0x") -> NaN // junk_string_value.

StringToDouble("0123e45") -> NaN // junk_string_value.

StringToDouble("01239E45") -> 1239e45.

StringToDouble("-infinity") -> NaN // junk_string_value.

StringToDouble("NaN") -> NaN // junk_string_value.

flags = NO_FLAGS,

separator = ' ':

StringToDouble("1 2 3 4") -> 1234.0

StringToDouble("1 2") -> NaN // junk_string_value

StringToDouble("1 000 000.0") -> 1000000.0

StringToDouble("1.000 000") -> 1.0

StringToDouble("1.0e1 000") -> NaN // junk_string_value

Same as StringToDouble but reads a float.

Note that this is not equivalent to static_cast

<float

>(StringToDouble(...))

due to potential double-rounding.

Same as StringToFloat above but for 16 bit characters.

Enumeration for allowing octals and ignoring junk when converting

strings to numbers.