Add initial test framework for utils functions

Added the initial test framework that will eventually encompass all of the functions in the package. Started with the primary utility functions since those are widely used and also simple enough to get used to writing tests. All current tests pass 😁

tests/test_utils.py

@@ -3,6 +3,9 @@ from typing import Literal
 from pathlib import Path
 from rc_gpfs import utils
 
+from polars.testing import assert_series_equal
+import polars as pl
+
 
 @pytest.mark.parametrize("path", ["/data/rc/gpfs-policy", Path("/data/rc/gpfs-policy")])
 def test_as_path_valid(path: Path | Literal["/data/rc/gpfs-policy"]):
@@ -34,8 +37,102 @@ def test_convert_si(value,unit,to_unit,use_binary,expected):
 def test_as_bytes(val,default,expected):
     assert utils.as_bytes(val,default) == expected
 
-class TestSizeDistribution:
-    inputs = [
+class TestSizeGrouping:
+    input_sizes = pl.Series(
+        name="size",
+        values = [
+            0,
+            2048,
+            4096,
+            1024**2,  # 1 MiB
+            1024**3,  # 1 GiB
+            20*1024**3,  # 20 GiB
+            1024**5   # 1 PiB
+        ],
+        dtype = pl.Int128()
+    )
+
+    expected_size_groups = [
+        pl.Series(
+            name="size",
+            values=[
+                "0 B-4 KiB",
+                "0 B-4 KiB",
+                "0 B-4 KiB",
+                "4 KiB-4 MiB",
+                "4 MiB-1 GiB",
+                "10 GiB-100 GiB",
+                ">1 TiB",
+            ],
+            dtype=pl.Enum(
+                [
+                    "0 B-4 KiB",
+                    "4 KiB-4 MiB",
+                    "4 MiB-1 GiB",
+                    "1 GiB-10 GiB",
+                    "10 GiB-100 GiB",
+                    "100 GiB-1 TiB",
+                    ">1 TiB",
+                ]
+            ),
+        ),
+        pl.Series(
+            name="size",
+            values=[
+                "0 B-4 KiB",
+                "0 B-4 KiB",
+                "0 B-4 KiB",
+                "4 KiB-4 MiB",
+                "4 MiB-1 GiB",
+                "10 GiB-100 GiB",
+                ">1 TiB",
+            ],
+            dtype=pl.Enum(
+                [
+                    "0 B-4 KiB",
+                    "4 KiB-4 MiB",
+                    "4 MiB-1 GiB",
+                    "1 GiB-10 GiB",
+                    "10 GiB-100 GiB",
+                    "100 GiB-1 TiB",
+                    ">1 TiB",
+                ]
+            ),
+        ),
+        pl.Series(
+            name="size",
+            values=[
+                "0 B-1 KiB",
+                "1 KiB-4 KiB",
+                "1 KiB-4 KiB",
+                "4 KiB-10 GiB",
+                "4 KiB-10 GiB",
+                ">10 GiB",
+                ">10 GiB",
+            ],
+            dtype=pl.Enum(["0 B-1 KiB", "1 KiB-4 KiB", "4 KiB-10 GiB", ">10 GiB"]),
+        ),
+        pl.Series(
+            name="size",
+            values=[
+                "0 B-1 MiB",
+                "0 B-1 MiB",
+                "0 B-1 MiB",
+                "0 B-1 MiB",
+                ">1 MiB",
+                ">1 MiB",
+                ">1 MiB",
+            ],
+            dtype=pl.Enum(
+                [
+                    "0 B-1 MiB",
+                    ">1 MiB",
+                ]
+            ),
+        ),
+    ]
+
+    input_bins = [
         ["4 KiB", "4 MiB", "1 GiB", "10 GiB", "100 GiB", "1 TiB"],
         [4096, 4194304, 1073741824, 10737418240, 107374182400, 1099511627776],
         ["10 GiB", 1024, 4096, "1 KiB",0],
@@ -60,6 +157,10 @@ class TestSizeDistribution:
     def test_create_size_bin_labels(self,bins,expected):
         assert utils.create_size_bin_labels(bins) == expected
     
-    @pytest.mark.parametrize("bins,expected",list(zip(inputs,list(zip(expected_bins,expected_labels)))))
+    @pytest.mark.parametrize("bins,expected",list(zip(input_bins,list(zip(expected_bins,expected_labels)))))
     def test_prep_size_distribution(self,bins,expected):
-        assert utils.prep_size_distribution(size_bins=bins) == expected
\ No newline at end of file
+        assert utils.prep_size_distribution(size_bins=bins) == expected
+    
+    @pytest.mark.parametrize("bins,expected",list(zip(input_bins,expected_size_groups)),ids=[1,2,3,4])
+    def test_calculate_size_distribution(self, bins, expected):
+        assert_series_equal(utils.calculate_size_distribution(self.input_sizes, size_bins=bins),expected)
\ No newline at end of file