Add parse_float facilities

Cargo.toml

@@ -1,8 +1,8 @@
 [package]
 name = "num-parse"
-version = "0.1.2"
+version = "0.2.0"
 edition = "2021"
-description = "Generic, JavaScript-like parseInt() functions for Rust."
+description = "parseInt() and parseFloat() as known from JavaScript, but generic, and in Rust"
 authors = [
     "Jan Max Meyer <[email protected]>"
 ]
@@ -13,6 +13,9 @@ categories = [
 ]
 keywords = [
     "parseint",
+    "parsefloat",
+    "javascript",
+    "ecmascript",
     "numbers"
 ]
 

LICENSE

@@ -1,4 +1,4 @@
-Copyright © 2022 by Jan Max Meyer, Phorward Software Technologies.
+Copyright © 2024 by Jan Max Meyer, Phorward Software Technologies.
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -5,11 +5,9 @@
 [![crates.io](https://img.shields.io/crates/v/num-parse)](https://crates.io/crates/num-parse)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
 
-Generic, JavaScript-like parseInt() functions for Rust.
+Generic, JavaScript-style parseInt() and parseFloat() functions for Rust.
 
-This crate is intended to provide a fast and generic `parseInt()`-like implementation for Rust, which mostly follows the specification described in the [MDN parseInt() documentation](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/parseInt).
-
-Likewise in JavaScript, a `parseFloat()`-like implementation for float-types is planned as well, therefore the crate has been named `num-parse` already, althought it currently provides `parse_int()` and variative functions only.
+This crate is intended to provide a fast and generic `parseInt()`- and `parseFloat()`-like implementation for Rust, which mostly follows the specification described in the MDN documentation for [parseInt()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/parseInt) and [parseFloat()](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/parseFloat).
 
 ## parse_int(), parse_uint()
 
@@ -33,3 +31,11 @@ assert_eq!(
     Some(3405691582usize)
 );
 ```
+
+## parse_float()
+
+TODO
+
+## PeekableIterator
+
+This crate is required by and implemented together with the [Tokay programming language](https://tokay.dev) to parse and calculate numerical values from a `PeekableIterator`-trait, which is also defined here. A JavaScript-like numerical parsing was thought to be useful for other projects as well.

src/float.rs

@@ -0,0 +1,184 @@
+/*! Generic, JavaScript-like parseFloat() function for parsing floating point numbers
+from any character-emitting resource. */
+
+use super::*;
+use num;
+
+/** Parse float values from a PeekableIterator.
+
+Trailing `whitespace` is accepted, when set to `true`.
+*/
+pub fn parse_float_from_iter<T: num::Float + num::FromPrimitive + std::fmt::Display>(
+    chars: &mut dyn PeekableIterator<Item = char>,
+    whitespace: bool,
+) -> Option<T> {
+    let mut neg = false;
+    let mut int: Option<u32> = None;
+    let mut dec: Option<u32> = None;
+
+    // Skip over whitespace
+    if whitespace {
+        while let Some(ch) = chars.peek() {
+            if !ch.is_whitespace() {
+                break;
+            }
+
+            chars.next();
+        }
+    }
+
+    // Match sign
+    match chars.peek() {
+        Some(ch) if *ch == '-' || *ch == '+' => {
+            neg = chars.next().unwrap() == '-';
+        }
+        _ => {}
+    }
+
+    // Integer part (optional)
+    while let Some(dig) = chars.peek() {
+        int = match dig.to_digit(10) {
+            Some(digit) => {
+                let mut int = int.unwrap_or(0);
+
+                int = int * 10 + digit;
+
+                chars.next();
+                Some(int)
+            }
+            None => break,
+        }
+    }
+
+    // Decimal point (this *is* mandatory!)
+    match chars.peek() {
+        Some(ch) if *ch == '.' => {
+            chars.next();
+        }
+        _ => return None,
+    }
+
+    // Decimal part (optional)
+    while let Some(dig) = chars.peek() {
+        dec = match dig.to_digit(10) {
+            Some(digit) => {
+                let mut dec = dec.unwrap_or(0);
+
+                dec = dec * 10 + digit;
+
+                chars.next();
+                Some(dec)
+            }
+            None => break,
+        }
+    }
+
+    // Either integer or decimal part must be given, otherwise reject
+    if int.is_none() && dec.is_none() {
+        return None;
+    }
+
+    // Turn integer and decimal part into floating point number
+    let int = T::from_u32(int.unwrap_or(0)).unwrap();
+    let dec = T::from_u32(dec.unwrap_or(0)).unwrap();
+    let ten = T::from_u32(10).unwrap();
+
+    let mut precision =
+        std::iter::successors(Some(dec), |&n| (n >= ten).then(|| n / ten)).count() as u32;
+    let mut ret = int + dec / ten.powi(precision as i32);
+
+    // Parse optionally provided exponential notation
+    match chars.peek() {
+        Some('e') | Some('E') => {
+            chars.next();
+
+            let mut neg = false;
+
+            match chars.peek() {
+                Some(ch) if *ch == '-' || *ch == '+' => {
+                    neg = chars.next().unwrap() == '-';
+                }
+                _ => {}
+            }
+
+            let mut exp: u32 = 0;
+
+            while let Some(dig) = chars.peek() {
+                match dig.to_digit(10) {
+                    Some(digit) => {
+                        exp = exp * 10;
+                        exp = exp + digit;
+
+                        chars.next();
+                    }
+                    None => break,
+                }
+            }
+
+            if neg {
+                precision += exp;
+            } else if precision < exp {
+                precision = 0;
+            }
+
+            for _ in 0..exp {
+                if neg {
+                    ret = ret / ten;
+                } else {
+                    ret = ret * ten;
+                }
+            }
+        }
+        _ => {}
+    }
+
+    let factor = ten.powf(T::from_u32(precision).unwrap());
+
+    ret = (ret * factor).round() / factor;
+
+    // Negate when necessary
+    if neg {
+        Some(-ret)
+    } else {
+        Some(ret)
+    }
+}
+
+/// Parse float values from a &str, ignoring trailing whitespace.
+pub fn parse_float<T: num::Float + num::FromPrimitive + std::fmt::Display>(s: &str) -> Option<T> {
+    parse_float_from_iter::<T>(&mut s.chars().peekable(), true)
+}
+
+#[test]
+fn test_parse_float_f32() {
+    assert_eq!(parse_float::<f32>(" -123.hello "), Some(-123f32));
+    assert_eq!(parse_float::<f32>(" -13.37.hello "), Some(-13.37f32));
+    assert_eq!(parse_float::<f32>(" -13.37e2.hello "), Some(-1337f32));
+    assert_eq!(parse_float::<f32>(" -13.37e-2.hello "), Some(-0.1337f32));
+    assert_eq!(
+        parse_float::<f32>(" -13.37e-16 "),
+        Some(-0.000000000000001337f32)
+    );
+    assert_eq!(parse_float::<f32>(" -1337.0e-30f32 "), Some(-1337.0e-30f32));
+}
+
+#[test]
+fn test_parse_float_f64() {
+    assert_eq!(parse_float::<f64>(" -123.hello "), Some(-123f64));
+    assert_eq!(parse_float::<f64>(" -13.37.hello "), Some(-13.37f64));
+    assert_eq!(parse_float::<f64>(" -13.37e2.hello "), Some(-1337f64));
+    assert_eq!(parse_float::<f64>(" -13.37e-2.hello "), Some(-0.1337f64));
+    assert_eq!(
+        parse_float::<f64>(" -13.37e-16 "),
+        Some(-0.000000000000001337f64)
+    );
+    assert_eq!(parse_float::<f64>(" -1337.0e-30f64 "), Some(-1337.0e-30f64));
+    assert_eq!(
+        parse_float::<f64>(" -1337.0e-296f64 "),
+        Some(-1337.0e-296f64)
+    ); // OK
+    assert_ne!(
+        parse_float::<f64>(" -1337.0e-297f64 "),
+        Some(-1337.0e-297f64)
+    ); // fails due precision error
+}

src/parseint.rs → src/int.rs

@@ -1,4 +1,5 @@
-//! JavaScript-style parseInt-like parsing of numbers from strings in Rust.
+/** Generic, JavaScript-like parseInt() function for parsing integer numbers
+with custom bases from any character-emitting resource. */
 use super::*;
 use num;
 

src/lib.rs

@@ -1,14 +1,16 @@
 // num-parse
-// Copyright © 2022 by Jan Max Meyer, Phorward Software Technologies.
+// Copyright © 2023 by Jan Max Meyer, Phorward Software Technologies.
 // Licensed under the MIT license. See LICENSE for more information.
 
 /*! num-parse
 
-    Generic, JavaScript-like parseInt() functions for Rust.
+    parseInt() and parseFloat() as known from JavaScript, but generic, and in Rust!
 */
 
-mod parseint;
-pub use parseint::*;
+mod float;
+mod int;
+pub use float::*;
+pub use int::*;
 
 /// Trait defining an iterator that implements a peek method on its own.
 pub trait PeekableIterator: std::iter::Iterator {