extract helper functions

apis/python/src/tiledbsoma/_util.py

@@ -10,6 +10,7 @@
 from itertools import zip_longest
 from typing import Any, Optional, Tuple, Type, TypeVar
 
+import pandas as pd
 import pyarrow as pa
 import somacore
 from somacore import options
@@ -284,3 +285,29 @@ def pa_types_is_string_or_bytes(dtype: pa.DataType) -> bool:
         or pa.types.is_string(dtype)
         or pa.types.is_binary(dtype)
     )
+
+
+def anndata_dataframe_unmodified(old: pd.DataFrame, new: pd.DataFrame) -> bool:
+    """
+    Checks that we didn't mutate the object while ingesting. Intended for unit tests.
+    """
+    try:
+        return (old == new).all().all()
+    except ValueError:
+        # Can be thrown when columns don't match -- which is what we check for
+        return False
+
+
+def anndata_dataframe_unmodified_nan_safe(old: pd.DataFrame, new: pd.DataFrame) -> bool:
+    """
+    Same as anndata_dataframe_unmodified, except it works with NaN data.
+    A key property of NaN is it's not equal to itself: x != x.
+    """
+
+    if old.index.name != new.index.name:
+        return False
+    if len(old) != len(new):
+        return False
+    if any(old.keys() != new.keys()):
+        return False
+    return True

apis/python/tests/test_basic_anndata_io.py

@@ -15,6 +15,10 @@
 import tiledbsoma
 import tiledbsoma.io
 from tiledbsoma import _constants, _factory
+from tiledbsoma._util import (
+    anndata_dataframe_unmodified,
+    anndata_dataframe_unmodified_nan_safe,
+)
 
 HERE = Path(__file__).parent
 
@@ -89,30 +93,6 @@ def adata(h5ad_file):
     return anndata.read_h5ad(h5ad_file)
 
 
-def _anndata_dataframe_unmodified(old, new):
-    """Checks that we didn't mutate the object while ingesting"""
-    try:
-        return (old == new).all().all()
-    except ValueError:
-        # Can be thrown when columns don't match -- which is what we check for
-        return False
-
-
-def _anndata_dataframe_unmodified_nan_safe(old, new):
-    """
-    Same as _anndata_dataframe_unmodified, except it works with NaN data.
-    A key property of NaN is it's not equal to itself: x != x.
-    """
-
-    if old.index.name != new.index.name:
-        return False
-    if len(old) != len(new):
-        return False
-    if any(old.keys() != new.keys()):
-        return False
-    return True
-
-
 @pytest.mark.parametrize(
     "ingest_modes",
     [
@@ -158,8 +138,8 @@ def test_import_anndata(adata, ingest_modes, X_kind):
         if ingest_mode != "schema_only":
             have_ingested = True
 
-        assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-        assert _anndata_dataframe_unmodified(original.var, adata.var)
+        assert anndata_dataframe_unmodified(original.obs, adata.obs)
+        assert anndata_dataframe_unmodified(original.var, adata.var)
 
         exp = tiledbsoma.Experiment.open(uri)
 
@@ -447,8 +427,8 @@ def test_ingest_uns(tmp_path: pathlib.Path, h5ad_file_extended, ingest_uns_keys)
         uns_keys=ingest_uns_keys,
     )
 
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     with tiledbsoma.Experiment.open(uri) as exp:
         uns = exp.ms["hello"]["uns"]
@@ -517,8 +497,8 @@ def test_add_matrix_to_collection(adata):
 
     uri = tiledbsoma.io.from_anndata(output_path, adata, measurement_name="RNA")
 
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     exp = tiledbsoma.Experiment.open(uri)
     with _factory.open(output_path) as exp_r:
@@ -644,8 +624,8 @@ def add_matrix_to_collection(
 
     uri = tiledbsoma.io.from_anndata(output_path, adata, measurement_name="RNA")
 
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     exp = tiledbsoma.Experiment.open(uri)
     with _factory.open(output_path) as exp_r:
@@ -703,8 +683,8 @@ def test_export_anndata(adata):
 
     tiledbsoma.io.from_anndata(output_path, adata, measurement_name="RNA")
 
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     with _factory.open(output_path) as exp:
         with pytest.raises(ValueError):
@@ -753,8 +733,8 @@ def test_null_obs(adata, tmp_path: Path):
     uri = tiledbsoma.io.from_anndata(
         output_path, adata, "RNA", ingest_mode="write", X_kind=tiledbsoma.SparseNDArray
     )
-    assert _anndata_dataframe_unmodified_nan_safe(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified_nan_safe(original.var, adata.var)
+    assert anndata_dataframe_unmodified_nan_safe(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified_nan_safe(original.var, adata.var)
 
     exp = tiledbsoma.Experiment.open(uri)
     with tiledb.open(exp.obs.uri, "r") as obs:
@@ -782,8 +762,8 @@ def test_export_obsm_with_holes(h5ad_file_with_obsm_holes, tmp_path):
     output_path = tmp_path.as_posix()
     tiledbsoma.io.from_anndata(output_path, adata, "RNA")
 
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     exp = tiledbsoma.Experiment.open(output_path)
 
@@ -929,8 +909,8 @@ def test_id_names(tmp_path, obs_id_name, var_id_name, indexify_obs, indexify_var
         obs_id_name=obs_id_name,
         var_id_name=var_id_name,
     )
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     with tiledbsoma.Experiment.open(uri) as exp:
         assert obs_id_name in exp.obs.keys()
@@ -1015,8 +995,8 @@ def test_uns_io(tmp_path, outgest_uns_keys):
     soma_uri = tmp_path.as_posix()
 
     tiledbsoma.io.from_anndata(soma_uri, adata, measurement_name="RNA")
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     with tiledbsoma.Experiment.open(soma_uri) as exp:
         bdata = tiledbsoma.io.to_anndata(
@@ -1067,8 +1047,8 @@ def test_string_nan_columns(tmp_path, adata, write_index):
     uri = tmp_path.as_posix()
     original = adata.copy()
     tiledbsoma.io.from_anndata(uri, adata, measurement_name="RNA")
-    assert _anndata_dataframe_unmodified_nan_safe(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified_nan_safe(original.var, adata.var)
+    assert anndata_dataframe_unmodified_nan_safe(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified_nan_safe(original.var, adata.var)
 
     # Step 3
     with tiledbsoma.open(uri, "r") as exp:
@@ -1126,8 +1106,8 @@ def test_index_names_io(tmp_path, obs_index_name, var_index_name):
 
     original = adata.copy()
     tiledbsoma.io.from_anndata(soma_uri, adata, measurement_name)
-    assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata.var)
+    assert anndata_dataframe_unmodified(original.obs, adata.obs)
+    assert anndata_dataframe_unmodified(original.var, adata.var)
 
     with tiledbsoma.Experiment.open(soma_uri) as exp:
         bdata = tiledbsoma.io.to_anndata(exp, measurement_name)

apis/python/tests/test_platform_config.py

@@ -8,6 +8,7 @@
 import tiledbsoma
 import tiledbsoma.io
 import tiledbsoma.options._tiledb_create_options as tco
+from tiledbsoma._util import anndata_dataframe_unmodified
 
 HERE = Path(__file__).parent
 
@@ -25,15 +26,6 @@ def adata(h5ad_file):
     return anndata.read_h5ad(h5ad_file)
 
 
-def _anndata_dataframe_unmodified(old, new):
-    """Checks that we didn't mutate the object while ingesting"""
-    try:
-        return (old == new).all().all()
-    except ValueError:
-        # Can be thrown when columns don't match -- which is what we check for
-        return False
-
-
 def test_platform_config(adata):
     # Set up anndata input path and tiledb-group output path
     original = adata.copy()
@@ -63,8 +55,8 @@ def test_platform_config(adata):
                 }
             },
         )
-        assert _anndata_dataframe_unmodified(original.obs, adata.obs)
-        assert _anndata_dataframe_unmodified(original.var, adata.var)
+        assert anndata_dataframe_unmodified(original.obs, adata.obs)
+        assert anndata_dataframe_unmodified(original.var, adata.var)
 
         with tiledbsoma.Experiment.open(output_path) as exp:
             x_data = exp.ms["RNA"].X["data"]

apis/python/tests/test_registration_mappings.py

@@ -12,15 +12,7 @@
 
 import tiledbsoma.io
 import tiledbsoma.io._registration as registration
-
-
-def _anndata_dataframe_unmodified(old, new):
-    """Checks that we didn't mutate the object while ingesting"""
-    try:
-        return (old == new).all().all()
-    except ValueError:
-        # Can be thrown when columns don't match -- which is what we check for
-        return False
+from tiledbsoma._util import anndata_dataframe_unmodified
 
 
 def _create_anndata(
@@ -755,8 +747,8 @@ def test_append_items_with_experiment(soma1, h5ad2):
             registration_mapping=rd,
         )
 
-    assert _anndata_dataframe_unmodified(original.obs, adata2.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata2.var)
+    assert anndata_dataframe_unmodified(original.obs, adata2.obs)
+    assert anndata_dataframe_unmodified(original.var, adata2.var)
 
     expect_obs_soma_joinids = list(range(6))
     expect_var_soma_joinids = list(range(5))
@@ -858,8 +850,8 @@ def test_append_with_disjoint_measurements(
         registration_mapping=rd,
     )
 
-    assert _anndata_dataframe_unmodified(original.obs, anndata2.obs)
-    assert _anndata_dataframe_unmodified(original.var, anndata2.var)
+    assert anndata_dataframe_unmodified(original.obs, anndata2.obs)
+    assert anndata_dataframe_unmodified(original.var, anndata2.var)
 
     # exp/obs, use_same_cells=True:                       exp/obs, use_same_cells=False:
     #    soma_joinid obs_id cell_type  is_primary_data       soma_joinid obs_id cell_type  is_primary_data

apis/python/tests/test_registration_signatures.py

@@ -6,6 +6,7 @@
 
 import tiledbsoma.io
 import tiledbsoma.io._registration.signatures as signatures
+from tiledbsoma._util import anndata_dataframe_unmodified
 
 HERE = Path(__file__).parent
 
@@ -21,15 +22,6 @@ def canned_anndata(canned_h5ad_file):
     return ad.read_h5ad(canned_h5ad_file)
 
 
-def _anndata_dataframe_unmodified(old, new):
-    """Checks that we didn't mutate the object while ingesting"""
-    try:
-        return (old == new).all().all()
-    except ValueError:
-        # Can be thrown when columns don't match -- which is what we check for
-        return False
-
-
 def test_signature_serdes(canned_h5ad_file, canned_anndata):
     sig = signatures.Signature.from_h5ad(canned_h5ad_file.as_posix())
     text1 = sig.to_json()
@@ -39,8 +31,8 @@ def test_signature_serdes(canned_h5ad_file, canned_anndata):
 
     original = canned_anndata.copy()
     sig = signatures.Signature.from_anndata(canned_anndata)
-    assert _anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
-    assert _anndata_dataframe_unmodified(original.var, canned_anndata.var)
+    assert anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
+    assert anndata_dataframe_unmodified(original.var, canned_anndata.var)
 
     text2 = sig.to_json()
     assert sig == signatures.Signature.from_json(text2)
@@ -51,8 +43,8 @@ def test_signature_serdes(canned_h5ad_file, canned_anndata):
     output_path = tempdir.name
 
     uri = tiledbsoma.io.from_anndata(output_path, canned_anndata, "RNA")
-    assert _anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
-    assert _anndata_dataframe_unmodified(original.var, canned_anndata.var)
+    assert anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
+    assert anndata_dataframe_unmodified(original.var, canned_anndata.var)
 
     sig = signatures.Signature.from_soma_experiment(uri)
     text3 = sig.to_json()
@@ -67,14 +59,14 @@ def test_compatible(canned_anndata):
 
     original = canned_anndata.copy()
     sig1 = signatures.Signature.from_anndata(canned_anndata)
-    assert _anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
-    assert _anndata_dataframe_unmodified(original.var, canned_anndata.var)
+    assert anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
+    assert anndata_dataframe_unmodified(original.var, canned_anndata.var)
 
     tempdir = tempfile.TemporaryDirectory()
     output_path = tempdir.name
     uri = tiledbsoma.io.from_anndata(output_path, canned_anndata, "RNA")
-    assert _anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
-    assert _anndata_dataframe_unmodified(original.var, canned_anndata.var)
+    assert anndata_dataframe_unmodified(original.obs, canned_anndata.obs)
+    assert anndata_dataframe_unmodified(original.var, canned_anndata.var)
     sig2 = signatures.Signature.from_soma_experiment(uri)
 
     # Check that single inputs result in zero incompatibility
@@ -98,8 +90,8 @@ def test_compatible(canned_anndata):
 
     original = adata3.copy()
     sig3 = signatures.Signature.from_anndata(adata3)
-    assert _anndata_dataframe_unmodified(original.obs, adata3.obs)
-    assert _anndata_dataframe_unmodified(original.var, adata3.var)
+    assert anndata_dataframe_unmodified(original.obs, adata3.obs)
+    assert anndata_dataframe_unmodified(original.var, adata3.var)
 
     with pytest.raises(ValueError):
         signatures.Signature.check_compatible({"orig": sig1, "anndata3": sig3})