[Python] add tests for writing and reading float16 with Parquet to Py…

…Arrow

python/pyarrow/tests/interchange/test_conversion.py

@@ -326,7 +326,7 @@ def test_pandas_roundtrip_datetime(unit):
 
 @pytest.mark.pandas
 @pytest.mark.parametrize(
-    "np_float", [np.float32, np.float64]
+    "np_float", [np.float32, np.float64] # float16 operations not yet supported
 )
 def test_pandas_to_pyarrow_with_missing(np_float):
     if Version(pd.__version__) < Version("1.5.0"):

python/pyarrow/tests/parquet/common.py

@@ -128,6 +128,7 @@ def _test_dataframe(size=10000, seed=0):
         'int16': _random_integers(size, np.int16),
         'int32': _random_integers(size, np.int32),
         'int64': _random_integers(size, np.int64),
+        'float16': np.arange(size, dtype=np.float16),
         'float32': np.random.randn(size).astype(np.float32),
         'float64': np.arange(size, dtype=np.float64),
         'bool': np.random.randn(size) > 0,
@@ -169,6 +170,7 @@ def alltypes_sample(size=10000, seed=0, categorical=False):
         'int16': np.arange(size, dtype=np.int16),
         'int32': np.arange(size, dtype=np.int32),
         'int64': np.arange(size, dtype=np.int64),
+        'float16': np.arange(size, dtype=np.float16),
         'float32': np.arange(size, dtype=np.float32),
         'float64': np.arange(size, dtype=np.float64),
         'bool': np.random.randn(size) > 0,

python/pyarrow/tests/parquet/test_basic.py

@@ -101,7 +101,7 @@ def test_set_dictionary_pagesize_limit(use_legacy_dataset):
 
 @pytest.mark.pandas
 @parametrize_legacy_dataset
-def test_chunked_table_write(use_legacy_dataset):
+def test_chunked_table_write(use_legacy_dataset): #
     # ARROW-232
     tables = []
     batch = pa.RecordBatch.from_pandas(alltypes_sample(size=10))

python/pyarrow/tests/parquet/test_metadata.py

@@ -169,6 +169,11 @@ def test_parquet_metadata_lifetime(tempdir):
         ([-1, 2, 2, None, 4], pa.int16(), 'INT32', -1, 4, 1, 4, 0),
         ([-1, 2, 2, None, 4], pa.int32(), 'INT32', -1, 4, 1, 4, 0),
         ([-1, 2, 2, None, 4], pa.int64(), 'INT64', -1, 4, 1, 4, 0),
+        #(
+        #    [-1.1, 2.2, 2.3, None, 4.4], pa.float16(),
+        #    'HALFFLOAT', -1.1, 4.4, 1, 4, 0
+        #),
+        # float16 operations are not yet implemented
         (
             [-1.1, 2.2, 2.3, None, 4.4], pa.float32(),
             'FLOAT', -1.1, 4.4, 1, 4, 0

python/pyarrow/tests/parquet/test_pandas.py

@@ -247,6 +247,7 @@ def test_pandas_parquet_pyfile_roundtrip(tempdir, use_legacy_dataset):
     size = 5
     df = pd.DataFrame({
         'int64': np.arange(size, dtype=np.int64),
+        'float16': np.arange(size, dtype=np.float16),
         'float32': np.arange(size, dtype=np.float32),
         'float64': np.arange(size, dtype=np.float64),
         'bool': np.random.randn(size) > 0,
@@ -279,6 +280,7 @@ def test_pandas_parquet_configuration_options(tempdir, use_legacy_dataset):
         'int16': np.arange(size, dtype=np.int16),
         'int32': np.arange(size, dtype=np.int32),
         'int64': np.arange(size, dtype=np.int64),
+        'float16': np.arange(size, dtype=np.float16),
         'float32': np.arange(size, dtype=np.float32),
         'float64': np.arange(size, dtype=np.float64),
         'bool': np.random.randn(size) > 0

python/pyarrow/tests/test_array.py

@@ -1447,6 +1447,7 @@ def test_cast_integers_unsafe():
 
 def test_floating_point_truncate_safe():
     safe_cases = [
+        # float16 does not support casts yet
         (np.array([1.0, 2.0, 3.0], dtype='float32'), 'float32',
          np.array([1, 2, 3], dtype='i4'), pa.int32()),
         (np.array([1.0, 2.0, 3.0], dtype='float64'), 'float64',