Remove redundant size from INT32_{FILL,MISSING}

Author: tomwhite

sgkit/io/utils.py

@@ -8,7 +8,7 @@
 from ..typing import ArrayLike, DType
 from ..utils import encode_array, max_str_len
 
-INT32_MISSING, INT32_FILL = -1, -2
+INT_MISSING, INT_FILL = -1, -2
 
 FLOAT32_MISSING, FLOAT32_FILL = np.array([0x7F800001, 0x7F800002], dtype=np.int32).view(
     np.float32

sgkit/io/vcf/vcf_reader.py

@@ -25,7 +25,7 @@
 
 from sgkit import variables
 from sgkit.io.dataset import load_dataset
-from sgkit.io.utils import FLOAT32_FILL, INT32_FILL, INT32_MISSING, STR_FILL
+from sgkit.io.utils import FLOAT32_FILL, INT_FILL, INT_MISSING, STR_FILL
 from sgkit.io.vcf import partition_into_regions
 from sgkit.io.vcf.utils import build_url, chunks, temporary_directory, url_filename
 from sgkit.io.vcfzarr_reader import (
@@ -136,7 +136,7 @@ def _vcf_type_to_numpy_type_and_fill_value(
     if vcf_type == "Flag":
         return "bool", False
     elif vcf_type == "Integer":
-        return "i4", INT32_FILL
+        return "i4", INT_FILL
     # the VCF spec defines Float as 32 bit, and in BCF is stored as 32 bit
     elif vcf_type == "Float":
         return "f4", FLOAT32_FILL
@@ -265,7 +265,7 @@ def update_dataset(self, ds: xr.Dataset) -> None:
         # cyvcf2 represents missing Integer values as the minimum int32 value
         # so change these to be the missing value
         if self.array.dtype == np.int32:
-            self.array[self.array == np.iinfo(np.int32).min] = INT32_MISSING
+            self.array[self.array == np.iinfo(np.int32).min] = INT_MISSING
 
         ds[self.variable_name] = (self.dims, self.array)
         if len(self.description) > 0:

sgkit/io/vcfzarr_reader.py

@@ -24,7 +24,7 @@
 from fsspec import get_mapper
 from scipy.special import comb
 
-from sgkit.io.utils import INT32_FILL, concatenate_and_rechunk, str_is_int
+from sgkit.io.utils import INT_FILL, concatenate_and_rechunk, str_is_int
 
 from ..model import DIM_SAMPLE, DIM_VARIANT, create_genotype_call_dataset
 from ..typing import ArrayLike, PathType
@@ -300,7 +300,7 @@ def _add_field_to_dataset(
 
 def _replace_fill_values(arr: ArrayLike) -> ArrayLike:
     if arr.dtype == np.int32:
-        arr[arr == -1] = INT32_FILL
+        arr[arr == -1] = INT_FILL
     return arr
 
 

sgkit/tests/io/vcf/test_vcf_reader.py

@@ -8,7 +8,7 @@
 from numpy.testing import assert_allclose, assert_array_equal
 
 from sgkit import load_dataset
-from sgkit.io.utils import FLOAT32_FILL, INT32_FILL, INT32_MISSING
+from sgkit.io.utils import FLOAT32_FILL, INT_FILL, INT_MISSING
 from sgkit.io.vcf import (
     MaxAltAllelesExceededWarning,
     partition_into_regions,
@@ -751,7 +751,7 @@ def test_vcf_to_zarr__fields(shared_datadir, tmp_path):
     )
     ds = xr.open_zarr(output)
 
-    missing, fill = INT32_MISSING, INT32_FILL
+    missing, fill = INT_MISSING, INT_FILL
     assert_array_equal(ds["variant_DP"], [fill, fill, 14, 11, 10, 13, 9, fill, fill])
     assert ds["variant_DP"].attrs["comment"] == "Total Depth"
 
@@ -832,7 +832,7 @@ def test_vcf_to_zarr__field_defs(shared_datadir, tmp_path):
     )
     ds = xr.open_zarr(output)
 
-    fill = INT32_FILL
+    fill = INT_FILL
     assert_array_equal(ds["variant_DP"], [fill, fill, 14, 11, 10, 13, 9, fill, fill])
     assert ds["variant_DP"].attrs["comment"] == "Combined depth across samples"
 
@@ -862,7 +862,7 @@ def test_vcf_to_zarr__field_number_A(shared_datadir, tmp_path):
     )
     ds = xr.open_zarr(output)
 
-    fill = INT32_FILL
+    fill = INT_FILL
     assert_array_equal(
         ds["variant_AC"],
         [
@@ -901,7 +901,7 @@ def test_vcf_to_zarr__field_number_R(shared_datadir, tmp_path):
         variants=slice(10002764, 10002793)
     )
 
-    fill = INT32_FILL
+    fill = INT_FILL
     ad = np.array(
         [
             [[40, 14, 0, fill]],
@@ -930,7 +930,7 @@ def test_vcf_to_zarr__field_number_G(shared_datadir, tmp_path):
         variants=slice(10002764, 10002793)
     )
 
-    fill = INT32_FILL
+    fill = INT_FILL
     pl = np.array(
         [
             [[319, 0, 1316, 440, 1358, 1798, fill, fill, fill, fill]],
@@ -973,7 +973,7 @@ def test_vcf_to_zarr__field_number_fixed(shared_datadir, tmp_path):
     )
     ds = xr.open_zarr(output)
 
-    missing, fill = INT32_MISSING, INT32_FILL
+    missing, fill = INT_MISSING, INT_FILL
     assert_array_equal(
         ds["call_HQ"],
         [

sgkit/tests/io/vcf/test_vcf_roundtrip.py

@@ -29,7 +29,7 @@
 from xarray import Dataset
 
 import sgkit as sg
-from sgkit.io.utils import INT32_FILL, INT32_MISSING
+from sgkit.io.utils import INT_FILL, INT_MISSING
 from sgkit.io.vcf import vcf_to_zarr
 
 
@@ -78,7 +78,7 @@ def fix_missing_fields(ds: Dataset) -> Dataset:
     # scikit-allel doesn't distinguish between missing and fill fields, so set all to fill
     for var in ds.data_vars:
         if ds[var].dtype == np.int32:  # type: ignore[comparison-overlap]
-            ds[var] = ds[var].where(ds[var] != INT32_MISSING, INT32_FILL)
+            ds[var] = ds[var].where(ds[var] != INT_MISSING, INT_FILL)
     return ds