fix validation and coercion on init bug (#1868)

* fix validation and coercion on init bug

Signed-off-by: Niels Bantilan <[email protected]>

* fix test

Signed-off-by: Niels Bantilan <[email protected]>

---------

Signed-off-by: Niels Bantilan <[email protected]>

pandera/typing/common.py

@@ -171,13 +171,17 @@ def __setattr__(self, name: str, value: Any) -> None:
 
             # prevent the double validation problem by preventing checks for
             # dataframes with a defined pandera.schema
-            pandera_accessor = getattr(self, "pandera")
+            pandera_accessor = getattr(self, "pandera", None)
+
             if (
-                pandera_accessor.schema is None
+                pandera_accessor is None
+                or pandera_accessor.schema is None
                 or pandera_accessor.schema != schema_model.to_schema()
             ):
-                pandera_accessor.add_schema(schema_model.to_schema())
                 self.__dict__ = schema_model.validate(self).__dict__
+                if pandera_accessor is None:
+                    pandera_accessor = getattr(self, "pandera")
+                pandera_accessor.add_schema(schema_model.to_schema())
 
 
 # pylint:disable=too-few-public-methods

tests/core/modules/validate_on_init.py

@@ -0,0 +1,17 @@
+"""Module for unit testing validation on initialization."""
+
+import numpy as np
+import pandas as pd
+import pandera as pa
+from pandera.typing import DataFrame
+
+
+class ExampleSchema(pa.DataFrameModel):
+    class Config:
+        coerce = True
+
+    a: np.int64
+
+
+ExampleDataFrame = DataFrame[ExampleSchema]
+validated_dataframe = ExampleDataFrame(pd.DataFrame([], columns=["a"]))

tests/core/test_model.py

@@ -1,10 +1,13 @@
 """Tests schema creation and validation from type annotations."""
 
 # pylint:disable=missing-class-docstring,missing-function-docstring,too-few-public-methods
+import os
 import re
+import runpy
 from copy import deepcopy
 from typing import Any, Generic, Iterable, List, Optional, TypeVar
 
+import numpy as np
 import pandas as pd
 import pytest
 
@@ -1081,6 +1084,18 @@ class Config:
         DataFrame[SchemaNoCoerce](raw_data)
 
 
+def test_validate_on_init_module():
+    """Make sure validation on initialization works when run as a module."""
+    path = os.path.join(
+        os.path.dirname(__file__),
+        "modules",
+        "validate_on_init.py",
+    )
+    result = runpy.run_path(path)
+    expected = pd.DataFrame([], columns=["a"], dtype=np.int64)
+    pd.testing.assert_frame_equal(result["validated_dataframe"], expected)
+
+
 def test_from_records_validates_the_schema():
     """Test that DataFrame[Schema] validates the schema"""