Merge pull request #207 from Rust-Data-Science/wip-rand-float-32

Wip rand float 32

Author: tushushu

tests/test_constructors.py

@@ -23,28 +23,6 @@
         (ul.arange, (5,), {"step": 2}, [0, 2, 4],),
         (ul.arange, (0,), {"stop": 5, "step": 2}, [0, 2, 4],),
 
-        (ul.repeat, (0, 3), {}, [0, 0, 0],),
-        (ul.repeat, (1.0, 3), {}, [1.0, 1.0, 1.0],),
-        (ul.repeat, (False, 3), {}, [False, False, False],),
-        (ul.repeat, ('foo', 3), {}, ['foo', 'foo', 'foo'],),
-
-        (ul.from_seq, (range(3), "float"), {}, [0.0, 1.0, 2.0],),
-        (ul.from_seq, (range(3), "int"), {}, [0, 1, 2],),
-        (ul.from_seq, (range(3), "int32"), {}, [0, 1, 2],),
-        (ul.from_seq, (range(3), "int64"), {}, [0, 1, 2],),
-
-        (ul.from_seq, ([False, True], "bool"), {}, [False, True],),
-        (ul.from_seq, ([0.0, 1.0, 2.0], "float"), {}, [0.0, 1.0, 2.0],),
-        (ul.from_seq, ([0, 1, 2], "int"), {}, [0, 1, 2],),
-        (ul.from_seq, (['foo', 'bar'], "string"), {}, ['foo', 'bar'],),
-
-        (ul.from_seq, ((False, True), "bool"), {}, [False, True],),
-        (ul.from_seq, ((0.0, 1.0, 2.0), "float"), {}, [0.0, 1.0, 2.0],),
-        (ul.from_seq, ((0, 1, 2), "int"), {}, [0, 1, 2],),
-        (ul.from_seq, (('foo', 'bar'), "string"), {}, ['foo', 'bar'],),
-        (ul.from_seq, (('foo', 'bar', None), "string"),
-         {}, ['foo', 'bar', None],),
-
         (ul.cycle, (range(3), 1, 'float'), {}, [0.0],),
         (ul.cycle, (range(3), 1, 'int32'), {}, [0],),
         (ul.cycle, (range(3), 1, 'int64'), {}, [0],),
@@ -64,6 +42,29 @@
         (ul.cycle, ((0.0, 1.0), 3, 'float'), {}, [0.0, 1.0, 0.0],),
         (ul.cycle, ((0, 1), 3, 'int'), {}, [0, 1, 0],),
         (ul.cycle, (('foo', 'bar'), 3, 'string'), {}, ['foo', 'bar', 'foo'],),
+
+
+        (ul.from_seq, (range(3), "float"), {}, [0.0, 1.0, 2.0],),
+        (ul.from_seq, (range(3), "int"), {}, [0, 1, 2],),
+        (ul.from_seq, (range(3), "int32"), {}, [0, 1, 2],),
+        (ul.from_seq, (range(3), "int64"), {}, [0, 1, 2],),
+
+        (ul.from_seq, ([False, True], "bool"), {}, [False, True],),
+        (ul.from_seq, ([0.0, 1.0, 2.0], "float"), {}, [0.0, 1.0, 2.0],),
+        (ul.from_seq, ([0, 1, 2], "int"), {}, [0, 1, 2],),
+        (ul.from_seq, (['foo', 'bar'], "string"), {}, ['foo', 'bar'],),
+
+        (ul.from_seq, ((False, True), "bool"), {}, [False, True],),
+        (ul.from_seq, ((0.0, 1.0, 2.0), "float"), {}, [0.0, 1.0, 2.0],),
+        (ul.from_seq, ((0, 1, 2), "int"), {}, [0, 1, 2],),
+        (ul.from_seq, (('foo', 'bar'), "string"), {}, ['foo', 'bar'],),
+        (ul.from_seq, (('foo', 'bar', None), "string"),
+         {}, ['foo', 'bar', None],),
+
+        (ul.repeat, (0, 3), {}, [0, 0, 0],),
+        (ul.repeat, (1.0, 3), {}, [1.0, 1.0, 1.0],),
+        (ul.repeat, (False, 3), {}, [False, False, False],),
+        (ul.repeat, ('foo', 3), {}, ['foo', 'foo', 'foo'],),
     ],
 )
 def test_constructors(
@@ -84,3 +85,21 @@ def test_constructors(
     else:
         raise TypeError(f"Unexpected type {type(expected_value[0])}!")
     check_test_result(dtype, test_method, result, expected_value)
+
+
+@pytest.mark.parametrize(
+    "test_method, args, kwargs",
+    [
+        (ul.random, (), {"size": 3, "dtype": "float32"},),
+    ],
+)
+def test_rand(
+    test_method: Callable,
+    args: tuple,
+    kwargs: dict,
+) -> None:
+    arr1 = test_method(*args, **kwargs)
+    arr2 = test_method(*args, **kwargs)
+    assert arr1.not_equal(arr2).all()
+    assert len(arr1) == len(arr2)
+    assert len(arr1) == kwargs["size"]

ulist/Cargo.toml

@@ -15,6 +15,7 @@ crate-type = ["cdylib"]
 
 [dependencies]
 csv = "1.1"
+rand = "0.8.5"
 
 [dependencies.pyo3]
 version = "0.16.4"

ulist/python/ulist/__init__.py

@@ -1,4 +1,4 @@
-from .constructor import arange, cycle, from_seq, repeat  # noqa:F401
+from .constructor import arange, cycle, from_seq, random, repeat  # noqa:F401
 from .control_flow import select  # noqa:F401
 from .core import UltraFastList  # noqa:F401
 from .io import read_csv  # noqa:F401

ulist/python/ulist/constructor.py

@@ -192,6 +192,27 @@ def from_seq(obj: Sequence, dtype: str) -> UltraFastList:
     return result
 
 
+def random(size: int, dtype: str) -> UltraFastList:
+    """Return a new ulist of random number in [0.0, 1.0).
+
+    Args:
+        size (int):
+            size (int): Size of the new ulist.
+        dtype (str):
+            The type of the output ulist. 'float', 'float32' or 'float64'.
+    """
+    if dtype == "float" or dtype == "float64":
+        raise NotImplementedError("Not implemented for float64!")
+        # result = UltraFastList(FloatList64.random(size))
+    elif dtype == "float32":
+        result = UltraFastList(FloatList32.random(size))
+    else:
+        raise ValueError(
+            "Parameter dtype should be 'float', 'float32' or 'float64'!"
+        )
+    return result
+
+
 def repeat(elem: ELEM, size: int) -> UltraFastList:
     """Return a new ulist of given size, filled with elem.
 

ulist/python/ulist/ulist.pyi

@@ -89,6 +89,8 @@ class FloatList32:
     def pop(self) -> None: ...
     def pow_scala(self, elem: NUM) -> FloatList32: ...
     @staticmethod
+    def random(size: int) -> FloatList32: ...
+    @staticmethod
     def repeat(elem: float, size: int) -> FloatList32: ...
     def replace(self, old: ELEM_OPT, new: ELEM_OPT) -> FloatList32: ...
     def set(self, index: int, elem: ELEM_OPT) -> None: ...

ulist/src/floatings/float32.rs

@@ -12,6 +12,8 @@ use crate::types::AsIntegerList32;
 use crate::types::AsIntegerList64;
 use crate::types::AsStringList;
 use pyo3::prelude::*;
+use rand::distributions::Uniform;
+use rand::Rng;
 use std::cell::Ref;
 use std::cell::RefCell;
 use std::cell::RefMut;
@@ -135,8 +137,8 @@ impl FloatList32 {
     pub fn greater_than_scala(&self, elem: f32) -> BooleanList {
         NumericalList::greater_than_scala(self, elem)
     }
-    
-    pub fn has_zero(&self) -> bool{
+
+    pub fn has_zero(&self) -> bool {
         NumericalList::has_zero(self)
     }
 
@@ -188,6 +190,13 @@ impl FloatList32 {
         NumericalList::pow_scala(self, elem)
     }
 
+    #[staticmethod]
+    fn random(size: usize) -> Self {
+        let range: Uniform<f32> = Uniform::from(0.0..1.0);
+        let v: Vec<f32> = rand::thread_rng().sample_iter(&range).take(size).collect();
+        FloatList32::new(v, HashSet::new())
+    }
+
     #[staticmethod]
     pub fn repeat(elem: f32, size: usize) -> Self {
         List::repeat(elem, size)

ulist/src/floatings/float64.rs

@@ -297,7 +297,7 @@ impl List<f64> for FloatList64 {
 impl NumericalList<f64, i32, f64> for FloatList64 {
     fn argmax(&self) -> PyResult<usize> {
         self._check_empty()?;
-self._check_all_na()?;
+        self._check_all_na()?;
         let vec = self.values();
         let hset = self.na_indexes();
         let val_0 = &f64::NEG_INFINITY;
@@ -311,7 +311,7 @@ self._check_all_na()?;
 
     fn argmin(&self) -> PyResult<usize> {
         self._check_empty()?;
-self._check_all_na()?;
+        self._check_all_na()?;
         let vec = self.values();
         let hset = self.na_indexes();
         let val_0 = &f64::INFINITY;
@@ -339,7 +339,7 @@ self._check_all_na()?;
 
     fn max(&self) -> PyResult<f64> {
         self._check_empty()?;
-self._check_all_na()?;
+        self._check_all_na()?;
         let hset = self.na_indexes();
         Ok(*self
             .values()
@@ -353,7 +353,7 @@ self._check_all_na()?;
 
     fn min(&self) -> PyResult<f64> {
         self._check_empty()?;
-self._check_all_na()?;
+        self._check_all_na()?;
         let hset = self.na_indexes();
         Ok(*self
             .values()