Function minimal_lexical::parse_float

pub fn parse_float<'a, F, Iter1, Iter2>(
    integer: Iter1,
    fraction: Iter2,
    exponent: i32,
) -> F
where
    F: Float,
    Iter1: Iterator<Item = &'a u8> + Clone,
    Iter2: Iterator<Item = &'a u8> + Clone,

Parse float from extracted float components.

• integer - Cloneable, forward iterator over integer digits.
• fraction - Cloneable, forward iterator over integer digits.
• exponent - Parsed, 32-bit exponent.

Preconditions

1. The integer should not have leading zeros.
2. The fraction should not have trailing zeros.
3. All bytes in integer and fraction should be valid digits, in the range [`b’0’, b’9’].

Panics

Although passing garbage input will not cause memory safety issues, it is very likely to cause a panic with a large number of digits, or in debug mode. The big-integer arithmetic without the alloc feature assumes a maximum, fixed-width input, which assumes at maximum a value of 10^(769 + 342), or ~4000 bits of storage. Passing in nonsensical digits may require up to ~6000 bits of storage, which will panic when attempting to add it to the big integer. It is therefore up to the caller to validate this input.

We cannot efficiently remove trailing zeros while only accepting a forward iterator.