(icu_datagen 1.4.1) Remove fraction dep from icu_datagen (unicode-org#4472)

Backports unicode-org#4459 and
unicode-org#4458

Backport approved in unicode-org#4458

Unlike that PR, this makes these deps dev-deps since they're not yet
used by an experimental feature

Author: Manishearth

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## icu4x 1.4.x
+ - [Remove icu_datagen's dep on `fractional`](https://github.com/unicode-org/icu4x/pull/4472)
+   - `[email protected]`
+
 ## icu4x 1.4 (Nov 16, 2023)
 
 - General

Cargo.lock

@@ -26,7 +26,7 @@ edition.workspace = true
 homepage.workspace = true
 repository.workspace = true
 rust-version.workspace = true
-version.workspace = true
+version = "1.4.1"
 
 [package.metadata.docs.rs]
 all-features = true
@@ -87,8 +87,6 @@ zip = { version = ">=0.5, <0.7", default-features = false, features = ["deflate"
 rayon = { version = "1.5", optional = true }
 ureq = { version = "2", optional = true }
 
-fraction = {version = "0.14.0", default-features = false }
-num-bigint = { version = "0.4.4", default-features = false }
 
 
 # Dependencies for "bin" feature
@@ -102,6 +100,8 @@ crlify = { path = "../../utils/crlify" }
 icu = { path = "../../components/icu" }
 postcard = "1"
 simple_logger = { version = "4.1.0", default-features = false }
+num-bigint = { version = "0.4.4", default-features = false }
+num-rational = { version = "0.4", default-features = false }
 
 # Pre-experimental components with limited data generation
 icu_singlenumberformatter = { path = "../../experimental/singlenumberformatter", features = ["datagen"] }

provider/datagen/Cargo.toml

@@ -6,10 +6,10 @@ use core::ops::{Div, Mul};
 use core::str::FromStr;
 use std::collections::{BTreeMap, VecDeque};
 
-use fraction::GenericFraction;
 use icu_provider::DataError;
 use icu_unitsconversion::provider::{ConversionInfo, Exactness, Sign};
-use num_bigint::BigUint;
+use num_bigint::BigInt;
+use num_rational::Ratio;
 
 use crate::transform::cldr::cldr_serde::units::units_constants::Constant;
 
@@ -273,9 +273,7 @@ pub fn process_constants<'a>(
 /// - " 1.5 E -2 " is converted to 15/1000
 /// - " 1.5 E - 2" is an invalid scientific notation number
 /// - "1.5E-2.5" is an invalid scientific notation number
-pub fn convert_scientific_notation_to_fraction(
-    number: &str,
-) -> Result<GenericFraction<BigUint>, DataError> {
+pub fn convert_scientific_notation_to_fraction(number: &str) -> Result<Ratio<BigInt>, DataError> {
     let mut parts = number.split('E');
     let base = parts.next().unwrap_or("1").trim();
     let exponent = parts.next().unwrap_or("0").trim();
@@ -285,51 +283,70 @@ pub fn convert_scientific_notation_to_fraction(
         ));
     }
 
-    let base = GenericFraction::<BigUint>::from_str(base)
-        .map_err(|_| DataError::custom("the base is not a valid decimal number"))?;
+    let mut split = base.split('.');
+    let base_whole = split.next().ok_or(DataError::custom("the base is empty"))?;
+    let base_whole = if base_whole.is_empty() {
+        BigInt::from(0)
+    } else {
+        BigInt::from_str(base_whole).map_err(|_| {
+            DataError::custom("the whole part of the base is not a valid whole number")
+        })?
+    };
+
+    let base_dec = if let Some(dec_str) = split.next() {
+        let dec = BigInt::from_str(dec_str).map_err(|_| {
+            DataError::custom("the decimal part of the base is not a valid whole number")
+        })?;
+        let exp = dec_str.chars().filter(char::is_ascii_digit).count();
+        Ratio::new(dec, BigInt::from(10u32).pow(exp as u32))
+    } else {
+        Ratio::new(BigInt::from(0), BigInt::from(1))
+    };
+
+    let base = base_dec + base_whole;
+
     let exponent = i32::from_str(exponent)
         .map_err(|_| DataError::custom("the exponent is not a valid integer"))?;
 
     let result = if exponent >= 0 {
-        base.mul(GenericFraction::new(
-            BigUint::from(10u32).pow(exponent as u32),
-            BigUint::from(1u32),
+        base.mul(Ratio::new(
+            BigInt::from(10u32).pow(exponent as u32),
+            BigInt::from(1u32),
         ))
     } else {
-        base.div(GenericFraction::new(
-            BigUint::from(10u32).pow((-exponent) as u32),
-            BigUint::from(1u32),
+        base.div(Ratio::new(
+            BigInt::from(10u32).pow((-exponent) as u32),
+            BigInt::from(1u32),
         ))
     };
-
     Ok(result)
 }
 
 // TODO: move this to the comment above.
 #[test]
 fn test_convert_scientific_notation_to_fraction() {
     let input = "1E2";
-    let expected = GenericFraction::<BigUint>::new(BigUint::from(100u32), BigUint::from(1u32));
+    let expected = Ratio::<BigInt>::new(BigInt::from(100u32), BigInt::from(1u32));
     let actual = convert_scientific_notation_to_fraction(input).unwrap();
     assert_eq!(expected, actual);
 
     let input = "1E-2";
-    let expected = GenericFraction::<BigUint>::new(BigUint::from(1u32), BigUint::from(100u32));
+    let expected = Ratio::<BigInt>::new(BigInt::from(1u32), BigInt::from(100u32));
     let actual = convert_scientific_notation_to_fraction(input).unwrap();
     assert_eq!(expected, actual);
 
     let input = "1.5E2";
-    let expected = GenericFraction::<BigUint>::new(BigUint::from(150u32), BigUint::from(1u32));
+    let expected = Ratio::<BigInt>::new(BigInt::from(150u32), BigInt::from(1u32));
     let actual = convert_scientific_notation_to_fraction(input).unwrap();
     assert_eq!(expected, actual);
 
     let input = "1.5E-2";
-    let expected = GenericFraction::<BigUint>::new(BigUint::from(15u32), BigUint::from(1000u32));
+    let expected = Ratio::<BigInt>::new(BigInt::from(15u32), BigInt::from(1000u32));
     let actual = convert_scientific_notation_to_fraction(input).unwrap();
     assert_eq!(expected, actual);
 
     let input = " 1.5 E -2 ";
-    let expected = GenericFraction::<BigUint>::new(BigUint::from(15u32), BigUint::from(1000u32));
+    let expected = Ratio::<BigInt>::new(BigInt::from(15u32), BigInt::from(1000u32));
     let actual = convert_scientific_notation_to_fraction(input).unwrap();
     assert_eq!(expected, actual);
 
@@ -340,6 +357,11 @@ fn test_convert_scientific_notation_to_fraction() {
     let input = "1.5E-2.5";
     let actual = convert_scientific_notation_to_fraction(input);
     assert!(actual.is_err());
+
+    let input = "0.308";
+    let expected = Ratio::<BigInt>::new(BigInt::from(308), BigInt::from(1000u32));
+    let actual = convert_scientific_notation_to_fraction(input).unwrap();
+    assert_eq!(expected, actual);
 }
 
 /// Determines if a string contains any alphabetic characters.
@@ -390,28 +412,19 @@ pub fn is_scientific_number(s: &str) -> bool {
 }
 
 /// Transforms a fractional number into byte numerators, byte denominators, and a sign.
-pub fn flatten_fraction(
-    fraction: GenericFraction<BigUint>,
-) -> Result<(Vec<u8>, Vec<u8>, Sign), DataError> {
-    let numerator = match fraction.numer() {
-        Some(numerator) => numerator.to_bytes_le(),
-        None => return Err(DataError::custom("the numerator is too large")),
-    };
-
-    let denominator = match fraction.denom() {
-        Some(denominator) => denominator.to_bytes_le(),
-        None => return Err(DataError::custom("the denominator is too large")),
-    };
-
-    let sign = match fraction.sign() {
-        Some(fraction::Sign::Plus) => Sign::Positive,
-        Some(fraction::Sign::Minus) => Sign::Negative,
-        None => {
-            return Err(DataError::custom("the sign is not defined"));
-        }
+pub fn flatten_fraction(fraction: Ratio<BigInt>) -> Result<(Vec<u8>, Vec<u8>, Sign), DataError> {
+    let (n_sign, numer) = fraction.numer().to_bytes_le();
+    let (d_sign, denom) = fraction.denom().to_bytes_le();
+
+    // Ratio's constructor sets denom > 0 but it's worth
+    // checking in case we decide to switch to new_raw() to avoid reducing
+    let sign = if n_sign * d_sign == num_bigint::Sign::Minus {
+        Sign::Negative
+    } else {
+        Sign::Positive
     };
 
-    Ok((numerator, denominator, sign))
+    Ok((numer, denom, sign))
 }
 
 /// Converts slices of numerator and denominator strings to a fraction.
@@ -425,8 +438,8 @@ pub fn flatten_fraction(
 pub fn convert_slices_to_fraction(
     numerator_strings: &[&str],
     denominator_strings: &[&str],
-) -> Result<GenericFraction<BigUint>, DataError> {
-    let mut fraction = GenericFraction::new(BigUint::from(1u32), BigUint::from(1u32));
+) -> Result<Ratio<BigInt>, DataError> {
+    let mut fraction = Ratio::new(BigInt::from(1u32), BigInt::from(1u32));
 
     for numerator in numerator_strings {
         let num_fraction = convert_scientific_notation_to_fraction(numerator)?;
@@ -446,19 +459,19 @@ pub fn convert_slices_to_fraction(
 fn test_convert_array_of_strings_to_fraction() {
     let numerator: Vec<&str> = vec!["1"];
     let denominator: Vec<&str> = vec!["2"];
-    let expected = GenericFraction::new(BigUint::from(1u32), BigUint::from(2u32));
+    let expected = Ratio::new(BigInt::from(1i32), BigInt::from(2i32));
     let actual = convert_slices_to_fraction(&numerator, &denominator).unwrap();
     assert_eq!(expected, actual);
 
     let numerator = vec!["1", "2"];
     let denominator = vec!["3", "1E2"];
-    let expected = GenericFraction::new(BigUint::from(2u32), BigUint::from(300u32));
+    let expected = Ratio::new(BigInt::from(2i32), BigInt::from(300i32));
     let actual = convert_slices_to_fraction(&numerator, &denominator).unwrap();
     assert_eq!(expected, actual);
 
     let numerator = vec!["1", "2"];
     let denominator = vec!["3", "1E-2"];
-    let expected = GenericFraction::new(BigUint::from(200u32), BigUint::from(3u32));
+    let expected = Ratio::new(BigInt::from(200i32), BigInt::from(3i32));
     let actual = convert_slices_to_fraction(&numerator, &denominator).unwrap();
     assert_eq!(expected, actual);
 
@@ -469,13 +482,13 @@ fn test_convert_array_of_strings_to_fraction() {
 
     let numerator = vec!["1E2"];
     let denominator = vec!["2"];
-    let expected = GenericFraction::new(BigUint::from(50u32), BigUint::from(1u32));
+    let expected = Ratio::new(BigInt::from(50i32), BigInt::from(1i32));
     let actual = convert_slices_to_fraction(&numerator, &denominator).unwrap();
     assert_eq!(expected, actual);
 
     let numerator = vec!["1E2", "2"];
     let denominator = vec!["3", "1E2"];
-    let expected = GenericFraction::new(BigUint::from(2u32), BigUint::from(3u32));
+    let expected = Ratio::new(BigInt::from(2i32), BigInt::from(3i32));
     let actual = convert_slices_to_fraction(&numerator, &denominator).unwrap();
     assert_eq!(expected, actual);
 }

provider/datagen/src/transform/cldr/units/helpers.rs

@@ -5,7 +5,6 @@
 use std::{borrow::Cow, collections::BTreeMap};
 
 use crate::transform::cldr::cldr_serde::{self};
-use fraction::{GenericFraction, Zero};
 use icu_provider::{
     datagen::IterableDataProvider, DataError, DataLocale, DataPayload, DataProvider, DataRequest,
     DataResponse,
@@ -116,6 +115,7 @@ fn test_basic() {
     use icu_provider::prelude::*;
     use icu_unitsconversion::provider::*;
     use num_bigint::BigUint;
+    use num_rational::Ratio;
     use zerofrom::ZeroFrom;
     use zerovec::maps::ZeroVecLike;
 
@@ -157,7 +157,7 @@ fn test_basic() {
             factor_num: ZeroVec::from(big_one.to_bytes_le()),
             factor_den: ZeroVec::from(big_one.to_bytes_le()),
             offset_sign: Sign::Positive,
-            offset_num: ZeroVec::from(BigUint::zero().to_bytes_le()),
+            offset_num: ZeroVec::from(BigUint::from(0u32).to_bytes_le()),
             offset_den: ZeroVec::from(big_one.to_bytes_le()),
             exactness: Exactness::Exact,
         }
@@ -167,17 +167,17 @@ fn test_basic() {
     let foot_convert_index = foot.convert_info.get().unwrap().as_unsigned_int() as usize;
     let foot_convert_ule = convert_units.zvl_get(foot_convert_index).unwrap();
     let foot_convert: ConversionInfo = ZeroFrom::zero_from(foot_convert_ule);
-    let ft_to_m = GenericFraction::<BigUint>::new(BigUint::from(3048u32), BigUint::from(10000u32));
+    let ft_to_m = Ratio::new(BigUint::from(3048u32), BigUint::from(10000u32));
 
     assert_eq!(
         foot_convert,
         ConversionInfo {
             base_unit: Cow::Borrowed("meter"),
             factor_sign: Sign::Positive,
-            factor_num: ZeroVec::from(ft_to_m.numer().unwrap().to_bytes_le()),
-            factor_den: ZeroVec::from(ft_to_m.denom().unwrap().to_bytes_le()),
+            factor_num: ZeroVec::from(ft_to_m.numer().to_bytes_le()),
+            factor_den: ZeroVec::from(ft_to_m.denom().to_bytes_le()),
             offset_sign: Sign::Positive,
-            offset_num: ZeroVec::from(BigUint::zero().to_bytes_le()),
+            offset_num: ZeroVec::from(BigUint::from(0u32).to_bytes_le()),
             offset_den: ZeroVec::from(big_one.to_bytes_le()),
             exactness: Exactness::Exact,
         }