Add support for PANDERA_VALIDATION_ENABLED for pandas and Configuration docs

docs/source/configuration.rst

@@ -0,0 +1,17 @@
+.. currentmodule:: pandera
+
+.. _configuration:
+
+Configuration
+===============
+
+*New in version 0.17.3*
+``pandera`` provides a global config `~pandera.config.PanderaConfig`.
+
+This configuration can also be set using environment variables. For instance:
+```
+export PANDERA_VALIDATION_ENABLED=False
+export PANDERA_VALIDATION_DEPTH=DATA_ONLY
+```
+
+Runtime data validation incurs a performance overhead. To mitigate this, you have the option to disable validation globally. This can be achieved by setting the environment variable `PANDERA_VALIDATION_ENABLE=False`. When validation is disabled, any `validate` call will return `None`.

docs/source/index.rst

@@ -369,6 +369,7 @@ page or reach out to the maintainers and pandera community on
    data_format_conversion
    supported_libraries
    integrations
+   configuration
 
 .. toctree::
    :maxdepth: 6

pandera/api/pandas/array.py

@@ -3,7 +3,6 @@
 import copy
 import warnings
 from typing import Any, List, Optional, TypeVar, Union, cast
-
 import pandas as pd
 
 from pandera import errors
@@ -12,6 +11,7 @@
 from pandera.api.checks import Check
 from pandera.api.hypotheses import Hypothesis
 from pandera.api.pandas.types import CheckList, PandasDtypeInputTypes, is_field
+from pandera.config import CONFIG
 from pandera.dtypes import DataType, UniqueSettings
 from pandera.engines import pandas_engine, PYDANTIC_V2
 
@@ -426,6 +426,9 @@ def validate(  # type: ignore [override]
         dtype: float64
 
         """
+        if not CONFIG.validation_enabled:
+            return check_obj
+
         if self._is_inferred:
             warnings.warn(
                 f"This {type(self)} is an inferred schema that hasn't been "

pandera/api/pandas/container.py

@@ -11,6 +11,7 @@
 import pandas as pd
 
 from pandera import errors
+from pandera.config import CONFIG
 from pandera import strategies as st
 from pandera.api.base.schema import BaseSchema, inferred_schema_guard
 from pandera.api.checks import Check
@@ -345,6 +346,9 @@ def validate(
         4         0.80      dog
         5         0.76      dog
         """
+        if not CONFIG.validation_enabled:
+            return check_obj
+
         # NOTE: Move this into its own schema-backend variant. This is where
         # the benefits of separating the schema spec from the backend
         # implementation comes in.

tests/core/test_pandas_config.py

@@ -0,0 +1,71 @@
+"""This module is to test the behaviour change based on defined config in pandera"""
+# pylint:disable=import-outside-toplevel,abstract-method,redefined-outer-name
+
+
+import pandas as pd
+import pytest
+
+import pandera as pa
+from pandera import DataFrameModel, DataFrameSchema, SeriesSchema
+from pandera.config import CONFIG, ValidationDepth
+
+
+@pytest.fixture()
+def disable_validation():
+    """Fixture to disable validation and clean up after the test is finished"""
+    CONFIG.validation_enabled = False
+    yield "resource"
+    CONFIG.validation_enabled = True
+
+
+class TestPandasDataFrameConfig:
+    """Class to test all the different configs types"""
+
+    sample_data = pd.DataFrame(
+        (("Bread", 9), ("Cutter", 15)), columns=["product", "price_val"]
+    )
+    # pylint: disable=unused-argument
+    def test_disable_validation(self, disable_validation):
+        """This function validates that a none object is loaded if validation is disabled"""
+
+        pandera_schema = DataFrameSchema(
+            {
+                "product": pa.Column(
+                    str, pa.Check(lambda s: s.str.startswith("B"))
+                ),
+                "price_val": pa.Column(int),
+            }
+        )
+
+        class TestSchema(DataFrameModel):
+            """Test Schema class"""
+
+            product: str = pa.Field(str_startswith="B")
+            price_val: int = pa.Field()
+
+        expected = {
+            "validation_enabled": False,
+            "validation_depth": ValidationDepth.SCHEMA_AND_DATA,
+        }
+
+        assert CONFIG.dict() == expected
+        assert pandera_schema.validate(self.sample_data) is self.sample_data
+        assert TestSchema.validate(self.sample_data) is self.sample_data
+
+
+class TestPandasSeriesConfig:
+    """Class to test all the different configs types"""
+
+    sample_data = pd.Series([1, 1, 2, 2, 3, 3])
+    # pylint: disable=unused-argument
+    def test_disable_validation(self, disable_validation):
+        """This function validates that a none object is loaded if validation is disabled"""
+        expected = {
+            "validation_enabled": False,
+            "validation_depth": ValidationDepth.SCHEMA_AND_DATA,
+        }
+        pandera_schema = SeriesSchema(
+            int, pa.Check(lambda s: s.value_counts() == 2, element_wise=False)
+        )
+        assert CONFIG.dict() == expected
+        assert pandera_schema.validate(self.sample_data) is self.sample_data