Implement binary and boolean types (and test them)

Signed-off-by: Deepyaman Datta <[email protected]>

Author: deepyaman

pandera/engines/ibis_engine.py

@@ -87,6 +87,28 @@ def dtype(cls, data_type: Any) -> dtypes.DataType:
         return engine.Engine.dtype(cls, np_dtype)
 
 
+###############################################################################
+# boolean
+###############################################################################
+
+
+@Engine.register_dtype(
+    equivalents=[
+        bool,
+        np.bool_,
+        dtypes.Bool,
+        dtypes.Bool(),
+        dt.Boolean,
+        dt.boolean,
+    ]
+)
+@immutable
+class Bool(DataType, dtypes.Bool):
+    """Semantic representation of a :class:`dt.Boolean`."""
+
+    type = dt.boolean
+
+
 ###############################################################################
 # signed integer
 ###############################################################################
@@ -262,6 +284,23 @@ class String(DataType, dtypes.String):
     type = dt.string
 
 
+@Engine.register_dtype(
+    equivalents=[
+        bytes,
+        np.bytes_,
+        dtypes.Binary,
+        dtypes.Binary(),
+        dt.Binary,
+        dt.binary,
+    ]
+)
+@immutable
+class Binary(DataType, dtypes.Binary):
+    """Semantic representation of a :class:`dt.Binary`."""
+
+    type = dt.binary
+
+
 ###############################################################################
 # temporal
 ###############################################################################

tests/ibis/test_ibis_builtin_checks.py

@@ -157,6 +157,12 @@ class TestEqualToCheck(BaseClass):
         "test_expression": datetime.timedelta(100, 10, 1),
     }
 
+    sample_boolean_data = {
+        "test_pass_data": [("foo", True), ("bar", True)],
+        "test_fail_data": [("foo", False), ("bar", False)],
+        "test_expression": True,
+    }
+
     def pytest_generate_tests(self, metafunc):
         """This function passes the parameter for each function based on parameter from get_data_param function"""
         # called once per each test function
@@ -184,6 +190,12 @@ def get_data_param(self):
                 {"datatype": dt.Int32, "data": self.sample_numeric_data},
                 {"datatype": dt.Int64, "data": self.sample_numeric_data},
                 {"datatype": dt.String, "data": self.sample_string_data},
+                {
+                    "datatype": dt.Binary,
+                    "data": self.convert_data(
+                        self.sample_string_data, "binary"
+                    ),
+                },
                 {
                     "datatype": dt.Float32,
                     "data": self.convert_data(
@@ -216,6 +228,7 @@ def get_data_param(self):
                     "datatype": dt.Interval(unit="us"),
                     "data": self.sample_duration_data,
                 },
+                {"datatype": dt.Boolean, "data": self.sample_boolean_data},
             ]
         }
 
@@ -302,6 +315,12 @@ class TestNotEqualToCheck(BaseClass):
         ),
     }
 
+    sample_boolean_data = {
+        "test_pass_data": [("foo", True), ("bar", True)],
+        "test_fail_data": [("foo", False), ("bar", True)],
+        "test_expression": False,
+    }
+
     def pytest_generate_tests(self, metafunc):
         """This function passes the parameter for each function based on parameter from get_data_param function"""
         # called once per each test function
@@ -361,6 +380,7 @@ def get_data_param(self):
                     "datatype": dt.Interval(unit="us"),
                     "data": self.sample_duration_data,
                 },
+                {"datatype": dt.Boolean, "data": self.sample_boolean_data},
             ]
         }
 

tests/polars/test_polars_builtin_checks.py

@@ -192,12 +192,6 @@ class TestEqualToCheck(BaseClass):
         "test_expression": "a",
     }
 
-    sample_boolean_data = {
-        "test_pass_data": [("foo", True), ("bar", True)],
-        "test_fail_data": [("foo", False), ("bar", False)],
-        "test_expression": True,
-    }
-
     sample_array_data = {
         "test_pass_data": [("foo", ["a"]), ("bar", ["a"])],
         "test_fail_data": [("foo", ["a"]), ("bar", ["b"])],
@@ -216,6 +210,12 @@ class TestEqualToCheck(BaseClass):
         "test_expression": datetime.timedelta(100, 10, 1),
     }
 
+    sample_boolean_data = {
+        "test_pass_data": [("foo", True), ("bar", True)],
+        "test_fail_data": [("foo", False), ("bar", False)],
+        "test_expression": True,
+    }
+
     def pytest_generate_tests(self, metafunc):
         """This function passes the parameter for each function based on parameter from get_data_param function"""
         # called once per each test function
@@ -456,7 +456,7 @@ def get_data_param(self):
                     "datatype": List(Utf8),
                     "data": self.sample_array_data,
                 },
-            ],
+            ]
         }
 
     @pytest.mark.parametrize("check_fn", [pa.Check.not_equal_to, pa.Check.ne])

tests/polars/test_polars_components.py

@@ -65,7 +65,7 @@ def test_column_schema_regex(column_kwargs):
             invalid_data.pipe(schema.validate).collect()
 
 
-def test_get_columnd_backend():
+def test_get_column_backend():
     assert isinstance(pa.Column.get_backend(pl.LazyFrame()), ColumnBackend)
     assert isinstance(
         pa.Column.get_backend(check_type=pl.LazyFrame), ColumnBackend