iterate on string vs large_string

single-cell-data · Oct 11, 2022 · 324654b · 324654b
1 parent 124e275
commit 324654b
Show file tree

Hide file tree

Showing 7 changed files with 21 additions and 9 deletions.
diff --git a/apis/python/src/tiledbsoma/util_arrow.py b/apis/python/src/tiledbsoma/util_arrow.py
@@ -24,7 +24,9 @@
     # IMPORTANT: ALL non-primitive types supported by TileDB must be in this table.
     #
     pa.string(): "ascii",  # XXX TODO: temporary work-around until UTF8 support is native. GH #338.
+    pa.large_string(): "ascii",  # XXX TODO: temporary work-around until UTF8 support is native. GH #338.
     pa.binary(): np.dtype("S"),
+    pa.large_binary(): np.dtype("S"),
     pa.timestamp("s"): "datetime64[s]",
     pa.timestamp("ms"): "datetime64[ms]",
     pa.timestamp("us"): "datetime64[us]",
@@ -90,7 +92,7 @@ def get_arrow_type_from_tiledb_dtype(tiledb_dtype: Union[str, np.dtype]) -> pa.D
     """
     if tiledb_dtype == "ascii" or tiledb_dtype.name == "bytes":
         # XXX TODO: temporary work-around until UTF8 support is native. GH #338.
-        return pa.string()
+        return pa.large_string()
     else:
         return pa.from_numpy_dtype(tiledb_dtype)
 

diff --git a/apis/python/tests/test_soma_collection.py b/apis/python/tests/test_soma_collection.py
@@ -15,7 +15,7 @@ def create_and_populate_dataframe(dataframe: soma.SOMADataFrame) -> None:
         [
             ("foo", pa.int32()),
             ("bar", pa.float64()),
-            ("baz", pa.string()),
+            ("baz", pa.large_string()),
         ]
     )
 
@@ -108,7 +108,7 @@ def soma_object(request, tmp_path):
 
     elif class_name == "SOMADataFrame":
         so = soma.SOMADataFrame(uri=uri)
-        so.create(pa.schema([("A", pa.int32()), ("B", pa.string())]))
+        so.create(pa.schema([("A", pa.int32()), ("B", pa.large_string())]))
 
     elif class_name == "SOMAIndexedDataFrame":
         so = soma.SOMAIndexedDataFrame(uri=uri)

diff --git a/apis/python/tests/test_soma_dataframe.py b/apis/python/tests/test_soma_dataframe.py
@@ -13,7 +13,7 @@ def test_soma_dataframe_non_indexed(tmp_path):
         [
             ("foo", pa.int32()),
             ("bar", pa.float64()),
-            ("baz", pa.string()),
+            ("baz", pa.large_string()),
         ]
     )
     sdf.create(schema=asch)
@@ -120,7 +120,7 @@ def simple_soma_data_frame(tmp_path):
             ("soma_rowid", pa.uint64()),
             ("A", pa.int64()),
             ("B", pa.float64()),
-            ("C", pa.string()),
+            ("C", pa.large_string()),
         ]
     )
     sdf = t.SOMADataFrame(uri=tmp_path.as_posix())

diff --git a/apis/python/tests/test_soma_experiment_basic.py b/apis/python/tests/test_soma_experiment_basic.py
@@ -14,7 +14,7 @@ def create_and_populate_obs(obs: soma.SOMADataFrame) -> soma.SOMADataFrame:
         [
             ("foo", pa.int32()),
             ("bar", pa.float64()),
-            ("baz", pa.string()),
+            ("baz", pa.large_string()),
         ]
     )
 
@@ -37,7 +37,7 @@ def create_and_populate_var(var: soma.SOMADataFrame) -> soma.SOMADataFrame:
 
     var_arrow_schema = pa.schema(
         [
-            ("quux", pa.string()),
+            ("quux", pa.large_string()),
             ("xyzzy", pa.float64()),
         ]
     )

diff --git a/apis/python/tests/test_soma_indexed_dataframe.py b/apis/python/tests/test_soma_indexed_dataframe.py
@@ -21,6 +21,14 @@ def _schema():
 def test_soma_indexed_dataframe(tmp_path, arrow_schema):
     sdf = t.SOMAIndexedDataFrame(uri=tmp_path.as_posix())
 
+    asch = pa.schema(
+        [
+            ("foo", pa.int32()),
+            ("bar", pa.float64()),
+            ("baz", pa.large_string()),
+        ]
+    )
+
     # Create
     asch = arrow_schema()
     sdf.create(schema=asch, index_column_names=["foo"])
@@ -72,7 +80,7 @@ def simple_soma_indexed_data_frame(tmp_path):
             ("index", pa.uint64()),
             ("A", pa.int64()),
             ("B", pa.float64()),
-            ("C", pa.string()),
+            ("C", pa.large_string()),
         ]
     )
     index_column_names = ["index"]

diff --git a/apis/python/tests/test_soma_metadata.py b/apis/python/tests/test_soma_metadata.py
@@ -33,7 +33,7 @@ def soma_object(request, tmp_path):
 
     elif class_name == "SOMADataFrame":
         so = soma.SOMADataFrame(uri=uri)
-        so.create(pa.schema([("A", pa.int32()), ("B", pa.string())]))
+        so.create(pa.schema([("A", pa.int32()), ("B", pa.large_string())]))
 
     elif class_name == "SOMAIndexedDataFrame":
         so = soma.SOMAIndexedDataFrame(uri=uri)

diff --git a/apis/python/tests/test_type_system.py b/apis/python/tests/test_type_system.py
@@ -25,7 +25,9 @@
     pa.timestamp("us"),
     pa.timestamp("ns"),
     pa.string(),
+    pa.large_string(),
     pa.binary(),
+    pa.large_binary(),
 ]