Remove legacy JSON reader from Python (#15538)

This PR removes the `engine="cudf_legacy"` option from Python.

This is a part of #15537.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

URL: #15538

python/cudf/cudf/io/json.py

@@ -38,33 +38,14 @@ def read_json(
             f"or a bool, or None. Got {type(dtype)}"
         )
 
-    if engine == "cudf_experimental":
-        raise ValueError(
-            "engine='cudf_experimental' support has been removed, "
-            "use `engine='cudf'`"
-        )
-
-    if engine == "cudf_legacy":
-        # TODO: Deprecated in 23.02, please
-        # give some time until(more than couple of
-        # releases from now) `cudf_legacy`
-        # support can be removed completely.
-        warnings.warn(
-            "engine='cudf_legacy' is a deprecated engine."
-            "This will be removed in a future release."
-            "Please switch to using engine='cudf'.",
-            FutureWarning,
-        )
-    if engine == "cudf_legacy" and not lines:
-        raise ValueError(f"{engine} engine only supports JSON Lines format")
     if engine == "auto":
         engine = "cudf" if lines else "pandas"
     if engine != "cudf" and keep_quotes:
         raise ValueError(
             "keep_quotes='True' is supported only with engine='cudf'"
         )
 
-    if engine == "cudf_legacy" or engine == "cudf":
+    if engine == "cudf":
         if dtype is None:
             dtype = True
 
@@ -117,7 +98,7 @@ def read_json(
             lines,
             compression,
             byte_range,
-            engine == "cudf_legacy",
+            False,
             keep_quotes,
             mixed_types_as_string,
         )

python/cudf/cudf/tests/test_json.py

@@ -495,9 +495,6 @@ def test_json_lines_compression(tmpdir, ext, out_comp, in_comp):
 
 
 @pytest.mark.filterwarnings("ignore:Using CPU")
-@pytest.mark.filterwarnings(
-    "ignore:engine='cudf_legacy' is a deprecated engine."
-)
 def test_json_engine_selection():
     json = "[1, 2, 3]"
 
@@ -519,10 +516,6 @@ def test_json_engine_selection():
     for col_name in df.columns:
         assert isinstance(col_name, int)
 
-    # should raise an exception
-    with pytest.raises(ValueError):
-        cudf.read_json(StringIO(json), lines=False, engine="cudf_legacy")
-
 
 def test_json_bool_values():
     buffer = "[true,1]\n[false,false]\n[true,true]"
@@ -541,30 +534,6 @@ def test_json_bool_values():
     np.testing.assert_array_equal(pd_df.dtypes, cu_df.dtypes)
 
 
-@pytest.mark.filterwarnings(
-    "ignore:engine='cudf_legacy' is a deprecated engine."
-)
-@pytest.mark.parametrize(
-    "buffer",
-    [
-        "[1.0,]\n[null, ]",
-        '{"0":1.0,"1":}\n{"0":null,"1": }',
-        '{ "0" : 1.0 , "1" : }\n{ "0" : null , "1" : }',
-        '{"0":1.0}\n{"1":}',
-    ],
-)
-def test_json_null_literal(buffer):
-    df = cudf.read_json(StringIO(buffer), lines=True, engine="cudf_legacy")
-
-    # first column contains a null field, type should be set to float
-    # second column contains only empty fields, type should be set to int8
-    np.testing.assert_array_equal(df.dtypes, ["float64", "int8"])
-    np.testing.assert_array_equal(
-        df["0"].to_numpy(na_value=np.nan), [1.0, np.nan]
-    )
-    np.testing.assert_array_equal(df["1"].to_numpy(na_value=0), [0, 0])
-
-
 def test_json_bad_protocol_string():
     test_string = StringIO('{"field": "s3://path"}')
 
@@ -739,14 +708,8 @@ def test_default_integer_bitwidth(default_integer_bitwidth, engine):
 @pytest.mark.parametrize(
     "engine",
     [
-        pytest.param(
-            "cudf_legacy",
-            marks=pytest.mark.skip(
-                reason="cannot partially set dtypes for cudf json engine"
-            ),
-        ),
-        "pandas",
         "cudf",
+        "pandas",
     ],
 )
 def test_default_integer_bitwidth_partial(default_integer_bitwidth, engine):

python/cudf/cudf/utils/ioutils.py

@@ -543,7 +543,7 @@
     function or `StringIO`). Multiple inputs may be provided as a list. If a
     list is specified each list entry may be of a different input type as long
     as each input is of a valid type and all input JSON schema(s) match.
-engine : {{ 'auto', 'cudf', 'cudf_legacy', 'pandas' }}, default 'auto'
+engine : {{ 'auto', 'cudf', 'pandas' }}, default 'auto'
     Parser engine to use. If 'auto' is passed, the engine will be
     automatically selected based on the other parameters. See notes below.
 orient : string