Allow Setting ascii in column_type For from_pandas/from_csv (#999)

Author: nguyenv

HISTORY.md

@@ -6,6 +6,7 @@
 ## API Changes
 * Addition of `ArraySchema.version` to get version of array schema [#949](https://github.com/TileDB-Inc/TileDB-Py/pull/949)
 * Deprecate `coords_filters` from `ArraySchema` [#993](https://github.com/TileDB-Inc/TileDB-Py/pull/993)
+* Allow setting `ascii` in `column_type` for `from_pandas`/`from_csv` [#999](https://github.com/TileDB-Inc/TileDB-Py/pull/999)
 
 # TileDB-Py 0.13.1 Release Notes
 

tiledb/dataframe_.py

@@ -112,6 +112,9 @@ def from_values(cls, array_like, varlen_types=()):
     def from_dtype(cls, dtype, varlen_types=()):
         from pandas.api import types as pd_types
 
+        if dtype == "ascii":
+            return cls("ascii", var=True)
+
         dtype = pd_types.pandas_dtype(dtype)
         # Note: be careful if you rearrange the order of the following checks
 

tiledb/tests/test_pandas_dataframe.py

@@ -1334,6 +1334,21 @@ def test_tile_and_cell_order(self, tile_order, cell_order):
         with tiledb.open(uri_from_csv) as A:
             tm.assert_frame_equal(df, A.df[:])
 
+    def test_set_ascii_dtype(self):
+        df = make_dataframe_basic1()
+
+        uri_char = self.path("test_set_ascii_dtype_char")
+        tiledb.from_pandas(uri_char, df, sparse=True)
+        with tiledb.open(uri_char) as A:
+            assert A.attr("x").dtype == np.dtype("S")
+            assert not A.attr("x").isascii
+
+        uri_ascii = self.path("test_set_ascii_dtype_ascii")
+        tiledb.from_pandas(uri_ascii, df, sparse=True, column_types={"x": "ascii"})
+        with tiledb.open(uri_ascii) as A:
+            assert A.attr("x").dtype == np.dtype("S")
+            assert A.attr("x").isascii
+
 
 class TestFromPandasOptions(DiskTestCase):
     def test_filters_options(self):