chore: refactor version checks for pandas library (#1711)

kiraksi · Linchin · web-flow · commit afa9752dba1c · 2023-11-07T13:08:23.000-08:00
* chore: refactor version checks for pandas library

* readded removed importing of pandas

* revert bad commit

* merged from main, added type:ignore tag to get around mypy error

* Added ignore statement for mypy error, removed checking max version of Pandas

* updated docstring error

* Added parameterize to test to test multiple supported versons

---------

Co-authored-by: Lingqing Gan &lt;lingqing.gan@gmail.com&gt;
diff --git a/google/cloud/bigquery/__init__.py b/google/cloud/bigquery/__init__.py
@@ -202,6 +202,7 @@
     # Custom exceptions
     "LegacyBigQueryStorageError",
     "LegacyPyarrowError",
+    "LegacyPandasError",
 ]
 
 
diff --git a/google/cloud/bigquery/_versions_helpers.py b/google/cloud/bigquery/_versions_helpers.py
@@ -24,6 +24,7 @@
 _MIN_PYARROW_VERSION = packaging.version.Version("3.0.0")
 _MIN_BQ_STORAGE_VERSION = packaging.version.Version("2.0.0")
 _BQ_STORAGE_OPTIONAL_READ_SESSION_VERSION = packaging.version.Version("2.6.0")
+_MIN_PANDAS_VERSION = packaging.version.Version("1.1.0")
 
 
 class PyarrowVersions:
@@ -171,3 +172,65 @@ def try_import(self, raise_if_error: bool = False) -> Any:
 
 
 BQ_STORAGE_VERSIONS = BQStorageVersions()
+
+
+class PandasVersions:
+    """Version comparisons for pandas package."""
+
+    def __init__(self):
+        self._installed_version = None
+
+    @property
+    def installed_version(self) -> packaging.version.Version:
+        """Return the parsed version of pandas"""
+        if self._installed_version is None:
+            import pandas  # type: ignore
+
+            self._installed_version = packaging.version.parse(
+                # Use 0.0.0, since it is earlier than any released version.
+                # Legacy versions also have the same property, but
+                # creating a LegacyVersion has been deprecated.
+                # https://github.com/pypa/packaging/issues/321
+                getattr(pandas, "__version__", "0.0.0")
+            )
+
+        return self._installed_version
+
+    def try_import(self, raise_if_error: bool = False) -> Any:
+        """Verify that a recent enough version of pandas extra is installed.
+        The function assumes that pandas extra is installed, and should thus
+        be used in places where this assumption holds.
+        Because `pip` can install an outdated version of this extra despite
+        the constraints in `setup.py`, the calling code can use this helper
+        to verify the version compatibility at runtime.
+        Returns:
+            The ``pandas`` module or ``None``.
+        Raises:
+            exceptions.LegacyPandasError:
+                If the pandas package is outdated and ``raise_if_error`` is
+                ``True``.
+        """
+        try:
+            import pandas
+        except ImportError as exc:  # pragma: NO COVER
+            if raise_if_error:
+                raise exceptions.LegacyPandasError(
+                    "pandas package not found. Install pandas version >="
+                    f" {_MIN_PANDAS_VERSION}"
+                ) from exc
+            return None
+
+        if self.installed_version < _MIN_PANDAS_VERSION:
+            if raise_if_error:
+                msg = (
+                    "Dependency pandas is outdated, please upgrade"
+                    f" it to version >= {_MIN_PANDAS_VERSION}"
+                    f" (version found: {self.installed_version})."
+                )
+                raise exceptions.LegacyPandasError(msg)
+            return None
+
+        return pandas
+
+
+PANDAS_VERSIONS = PandasVersions()
diff --git a/google/cloud/bigquery/client.py b/google/cloud/bigquery/client.py
@@ -115,6 +115,9 @@
 from google.cloud.bigquery.table import RowIterator
 
 pyarrow = _versions_helpers.PYARROW_VERSIONS.try_import()
+pandas = (
+    _versions_helpers.PANDAS_VERSIONS.try_import()
+)  # mypy check fails because pandas import is outside module, there are type: ignore comments related to this
 
 TimeoutType = Union[float, None]
 ResumableTimeoutType = Union[
@@ -124,7 +127,6 @@
 if typing.TYPE_CHECKING:  # pragma: NO COVER
     # os.PathLike is only subscriptable in Python 3.9+, thus shielding with a condition.
     PathType = Union[str, bytes, os.PathLike[str], os.PathLike[bytes]]
-    import pandas  # type: ignore
     import requests  # required by api-core
 
 _DEFAULT_CHUNKSIZE = 100 * 1024 * 1024  # 100 MB
@@ -2488,7 +2490,7 @@ def load_table_from_file(
 
     def load_table_from_dataframe(
         self,
-        dataframe: "pandas.DataFrame",
+        dataframe: "pandas.DataFrame",  # type: ignore
         destination: Union[Table, TableReference, str],
         num_retries: int = _DEFAULT_NUM_RETRIES,
         job_id: Optional[str] = None,
diff --git a/google/cloud/bigquery/exceptions.py b/google/cloud/bigquery/exceptions.py
@@ -29,3 +29,7 @@ class BigQueryStorageNotFoundError(BigQueryError):
     """Raised when BigQuery Storage extra is not installed when trying to
     import it.
     """
+
+
+class LegacyPandasError(BigQueryError):
+    """Raised when too old a version of pandas package is detected at runtime."""
diff --git a/tests/unit/test__versions_helpers.py b/tests/unit/test__versions_helpers.py
@@ -26,6 +26,11 @@
 except ImportError:  # pragma: NO COVER
     bigquery_storage = None
 
+try:
+    import pandas  # type: ignore
+except ImportError:  # pragma: NO COVER
+    pandas = None
+
 from google.cloud.bigquery import _versions_helpers
 from google.cloud.bigquery import exceptions
 
@@ -173,3 +178,49 @@ def test_bqstorage_is_read_session_optional_false():
     bqstorage_versions = _versions_helpers.BQStorageVersions()
     with mock.patch("google.cloud.bigquery_storage.__version__", new="2.5.0"):
         assert not bqstorage_versions.is_read_session_optional
+
+
+@pytest.mark.skipif(pandas is None, reason="pandas is not installed")
+@pytest.mark.parametrize("version", ["1.5.0", "2.0.0", "2.1.0"])
+def test_try_import_raises_no_error_w_recent_pandas(version):
+    versions = _versions_helpers.PandasVersions()
+    with mock.patch("pandas.__version__", new=version):
+        try:
+            pandas = versions.try_import(raise_if_error=True)
+            assert pandas is not None
+        except exceptions.LegacyPandasError:  # pragma: NO COVER
+            raise ("Legacy error raised with a non-legacy dependency version.")
+
+
+@pytest.mark.skipif(pandas is None, reason="pandas is not installed")
+def test_try_import_returns_none_w_legacy_pandas():
+    versions = _versions_helpers.PandasVersions()
+    with mock.patch("pandas.__version__", new="1.0.0"):
+        pandas = versions.try_import()
+        assert pandas is None
+
+
+@pytest.mark.skipif(pandas is None, reason="pandas is not installed")
+def test_try_import_raises_error_w_legacy_pandas():
+    versions = _versions_helpers.PandasVersions()
+    with mock.patch("pandas.__version__", new="1.0.0"):
+        with pytest.raises(exceptions.LegacyPandasError):
+            versions.try_import(raise_if_error=True)
+
+
+@pytest.mark.skipif(pandas is None, reason="pandas is not installed")
+def test_installed_pandas_version_returns_cached():
+    versions = _versions_helpers.PandasVersions()
+    versions._installed_version = object()
+    assert versions.installed_version is versions._installed_version
+
+
+@pytest.mark.skipif(pandas is None, reason="pandas is not installed")
+def test_installed_pandas_version_returns_parsed_version():
+    versions = _versions_helpers.PandasVersions()
+    with mock.patch("pandas.__version__", new="1.1.0"):
+        version = versions.installed_version
+
+    assert version.major == 1
+    assert version.minor == 1
+    assert version.micro == 0

Original file line number	Diff line number	Diff line change
`@@ -202,6 +202,7 @@`
`202`	`202`	`# Custom exceptions`
`203`	`203`	`"LegacyBigQueryStorageError",`
`204`	`204`	`"LegacyPyarrowError",`
	`205`	`+ "LegacyPandasError",`
`205`	`206`	`]`
`206`	`207`
`207`	`208`