remove option for multiple backends & simplify tests

Author: jreback

doc/source/getting_started/install.rst

@@ -258,7 +258,7 @@ matplotlib                2.2.2              Visualization
 openpyxl                  2.4.8              Reading / writing for xlsx files
 pandas-gbq                0.8.0              Google Big Query access
 psycopg2                                     PostgreSQL engine for sqlalchemy
-pyarrow                   0.12.0             Parquet, ORC, and feather reading / writing
+pyarrow                   0.12.0             Parquet, ORC (requires 0.13.0), and feather reading / writing
 pymysql                   0.7.11             MySQL engine for sqlalchemy
 pyreadstat                                   SPSS files (.sav) reading
 pytables                  3.4.2              HDF5 reading / writing

pandas/core/config_init.py

@@ -568,22 +568,6 @@ def use_inf_as_na_cb(key):
         validator=is_one_of_factory(["auto", "pyarrow", "fastparquet"]),
     )
 
-
-# Set up the io.orc specific configuration.
-orc_engine_doc = """
-: string
-    The default orc reader/writer engine. Available options:
-    'auto', 'pyarrow', the default is 'auto'
-"""
-
-with cf.config_prefix("io.orc"):
-    cf.register_option(
-        "engine",
-        "auto",
-        orc_engine_doc,
-        validator=is_one_of_factory(["auto", "pyarrow"]),
-    )
-
 # --------
 # Plotting
 # ---------

pandas/io/orc.py

@@ -1,40 +1,16 @@
 """ orc compat """
 
 import distutils
-from typing import List, Optional
+from typing import TYPE_CHECKING, List, Optional
 
 from pandas.compat._optional import import_optional_dependency
 
-from pandas import DataFrame, get_option
 from pandas._typing import FilePathOrBuffer
 
 from pandas.io.common import get_filepath_or_buffer
 
-
-def get_engine(engine: str) -> "PyArrowImpl":
-    """ return our implementation """
-
-    if engine == "auto":
-        engine = get_option("io.orc.engine")
-
-    if engine == "auto":
-        # try engines in this order
-        try:
-            return PyArrowImpl()
-        except ImportError:
-            pass
-
-        raise ImportError(
-            "Unable to find a usable engine; "
-            "tried using: 'pyarrow'.\n"
-            "pyarrow is required for orc "
-            "support"
-        )
-
-    if engine not in ["pyarrow"]:
-        raise ValueError("engine must be 'pyarrow'")
-
-    return PyArrowImpl()
+if TYPE_CHECKING:
+    from pandas import DataFrame
 
 
 class PyArrowImpl:
@@ -55,22 +31,15 @@ def __init__(self):
 
     def read(
         self, path: FilePathOrBuffer, columns: Optional[List[str]] = None, **kwargs
-    ) -> DataFrame:
+    ) -> "DataFrame":
         path, _, _, _ = get_filepath_or_buffer(path)
-
-        py_file = self.api.input_stream(path)
-        orc_file = self.api.orc.ORCFile(py_file)
-
+        orc_file = self.api.orc.ORCFile(path)
         result = orc_file.read(columns=columns, **kwargs).to_pandas()
-
         return result
 
 
 def read_orc(
-    path: FilePathOrBuffer,
-    engine: str = "auto",
-    columns: Optional[List[str]] = None,
-    **kwargs,
+    path: FilePathOrBuffer, columns: Optional[List[str]] = None, **kwargs,
 ):
     """
     Load an ORC object from the file path, returning a DataFrame.
@@ -91,18 +60,15 @@ def read_orc(
         By file-like object, we refer to objects with a ``read()`` method,
         such as a file handler (e.g. via builtin ``open`` function)
         or ``StringIO``.
-    engine : {'auto', 'pyarrow'}, default 'auto'
-        ORC library to use. If 'auto', then the option ``io.orc.engine`` is
-        used. The default ``io.orc.engine`` behavior is to try 'pyarrow'.
     columns : list, default=None
         If not None, only these columns will be read from the file.
     **kwargs
-        Any additional kwargs are passed to the engine.
+        Any additional kwargs are passed to pyarrow.
 
     Returns
     -------
     DataFrame
     """
 
-    impl = get_engine(engine)
+    impl = PyArrowImpl()
     return impl.read(path, columns=columns, **kwargs)

pandas/tests/io/test_orc.py

@@ -1,27 +1,15 @@
 """ test orc compat """
 import datetime
-import distutils
 import os
 
 import numpy as np
 import pytest
 
-from pandas.compat import is_platform_windows
-
 import pandas as pd
+from pandas import read_orc
 import pandas.util.testing as tm
 
-from pandas.io.orc import PyArrowImpl, get_engine, read_orc
-
-try:
-    import pyarrow  # noqa
-
-    if distutils.version.LooseVersion(pyarrow.__version__) < "0.13.0":
-        raise ImportError("pyarrow must be >= 0.13.0 for read_orc")
-
-    _HAVE_PYARROW = True
-except ImportError:
-    _HAVE_PYARROW = False
+pytest.importorskip("pyarrow", minversion="0.13.0")
 
 pytestmark = pytest.mark.filterwarnings(
     "ignore:RangeIndex.* is deprecated:DeprecationWarning"
@@ -33,35 +21,7 @@ def dirpath(datapath):
     return datapath("io", "data", "orc")
 
 
-@pytest.fixture
-def pa():
-    if not _HAVE_PYARROW:
-        pytest.skip("pyarrow is not installed")
-    if is_platform_windows():
-        pytest.skip("pyarrow orc not available by default on windows")
-    return "pyarrow"
-
-
-def test_options_get_engine(pa):
-    assert isinstance(get_engine("pyarrow"), PyArrowImpl)
-
-    with pd.option_context("io.orc.engine", "pyarrow"):
-        assert isinstance(get_engine("auto"), PyArrowImpl)
-        assert isinstance(get_engine("pyarrow"), PyArrowImpl)
-
-    with pd.option_context("io.orc.engine", "auto"):
-        assert isinstance(get_engine("auto"), PyArrowImpl)
-        assert isinstance(get_engine("pyarrow"), PyArrowImpl)
-
-
-def test_invalid_engine(dirpath):
-    inputfile = os.path.join(dirpath, "TestOrcFile.emptyFile.orc")
-    engine = "foo"
-    with pytest.raises(ValueError):
-        read_orc(inputfile, engine=engine, columns=["boolean1"])
-
-
-def test_orc_reader_empty(dirpath, pa):
+def test_orc_reader_empty(dirpath):
     columns = [
         "boolean1",
         "byte1",
@@ -94,7 +54,7 @@ def test_orc_reader_empty(dirpath, pa):
     tm.assert_equal(expected, got)
 
 
-def test_orc_reader_basic(dirpath, pa):
+def test_orc_reader_basic(dirpath):
     data = {
         "boolean1": np.array([False, True], dtype="bool"),
         "byte1": np.array([1, 100], dtype="int8"),
@@ -114,7 +74,7 @@ def test_orc_reader_basic(dirpath, pa):
     tm.assert_equal(expected, got)
 
 
-def test_orc_reader_decimal(dirpath, pa):
+def test_orc_reader_decimal(dirpath):
     from decimal import Decimal
 
     # Only testing the first 10 rows of data
@@ -143,7 +103,7 @@ def test_orc_reader_decimal(dirpath, pa):
     tm.assert_equal(expected, got)
 
 
-def test_orc_reader_date_low(dirpath, pa):
+def test_orc_reader_date_low(dirpath):
     data = {
         "time": np.array(
             [
@@ -184,7 +144,7 @@ def test_orc_reader_date_low(dirpath, pa):
     tm.assert_equal(expected, got)
 
 
-def test_orc_reader_date_high(dirpath, pa):
+def test_orc_reader_date_high(dirpath):
     data = {
         "time": np.array(
             [
@@ -225,7 +185,7 @@ def test_orc_reader_date_high(dirpath, pa):
     tm.assert_equal(expected, got)
 
 
-def test_orc_reader_snappy_compressed(dirpath, pa):
+def test_orc_reader_snappy_compressed(dirpath):
     data = {
         "int1": np.array(
             [