internals rewrite: clean up checks and hypothesis functionality (#1109)

* rename core.pandas.checks to *.builtin_checks

* handle strategies better

* wip re-implement builtin checks

* clean up hypotheses

* add docstrings

* move builtin checks/hypotheses modules

* clean up register_{check, hypothesis}

* minor import cleanup

* clean up check registration, error implementation

* create backends.base subpackage

* remove debugging script

* cleanup, fix codecov

.pylintrc

@@ -15,6 +15,13 @@ good-names=
     fp,
     bar,
     _IS_INFERRED,
+    eq,
+    ne,
+    gt,
+    ge,
+    lt,
+    le,
+    dt
 
 [MESSAGES CONTROL]
 disable=

pandera/__init__.py

@@ -60,10 +60,13 @@
     pandas_version,
 )
 
+import pandera.backends
+
 from pandera.schema_inference.pandas import infer_schema
 from pandera.decorators import check_input, check_io, check_output, check_types
 from pandera.version import __version__
 
+
 if platform.system() != "Windows":
     # pylint: disable=ungrouped-imports
     from pandera.dtypes import Complex256, Float128

pandera/backends/__init__.py

@@ -0,0 +1,7 @@
+"""Pandera backends."""
+
+# ensure that base builtin checks and hypothesis are registered
+import pandera.backends.base.builtin_checks
+import pandera.backends.base.builtin_hypotheses
+
+import pandera.backends.pandas

pandera/backends/base.py → pandera/backends/base/__init__.py

@@ -1,12 +1,12 @@
-"""Base functions for Parsing, Validation, and Error Reporting Backends.
+"""Base classes for parsing, validation, and error Reporting Backends.
 
-This class should implement a common interface of operations needed for
+These classes implement a common interface of operations needed for
 data validation. These operations are exposed as methods that are composed
 together to implement the pandera schema specification.
 """
 
 from abc import ABC
-from typing import Optional
+from typing import Any, Dict, List, Optional
 
 
 class BaseSchemaBackend(ABC):
@@ -100,6 +100,12 @@ def check_dtype(self, check_obj, schema):
         """Core check that checks the data type of a check object."""
         raise NotImplementedError
 
+    def failure_cases_metadata(
+        self, schema_name: str, schema_errors: List[Dict[str, Any]]
+    ):
+        """Get failure cases metadata for lazy validation."""
+        raise NotImplementedError
+
 
 class BaseCheckBackend(ABC):
     """Abstract base class for a check backend implementation."""

pandera/backends/base/builtin_checks.py

@@ -0,0 +1,98 @@
+# pylint: disable=missing-function-docstring
+"""Built-in check functions base implementation.
+
+This module contains check function abstract definitions that correspond to
+the pandera.core.base.checks.Check methods. These functions do not actually
+implement any validation logic and serve as the entrypoint for dispatching
+specific implementations based on the data object type, e.g.
+`pandas.DataFrame`s.
+"""
+
+import re
+from typing import Any, Iterable, TypeVar, Union
+
+from pandera.core.checks import Check
+
+
+T = TypeVar("T")
+
+
+@Check.register_builtin_check_fn
+def equal_to(data: Any, value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def not_equal_to(data: Any, value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def greater_than(data: Any, min_value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def greater_than_or_equal_to(data: Any, min_value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def less_than(data: Any, max_value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def less_than_or_equal_to(data: Any, max_value: Any) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def in_range(
+    data: Any,
+    min_value: T,
+    max_value: T,
+    include_min: bool = True,
+    include_max: bool = True,
+) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def isin(data: Any, allowed_values: Iterable) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def notin(data: Any, forbidden_values: Iterable) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def str_matches(data: Any, pattern: Union[str, re.Pattern]) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def str_contains(data: Any, pattern: Union[str, re.Pattern]) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def str_startswith(data: Any, string: str) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def str_endswith(data: Any, string: str) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def str_length(data: Any, min_value: int = None, max_value: int = None) -> Any:
+    raise NotImplementedError
+
+
+@Check.register_builtin_check_fn
+def unique_values_eq(data: Any, values: Iterable) -> Any:
+    raise NotImplementedError

pandera/backends/base/builtin_hypotheses.py

@@ -0,0 +1,31 @@
+# pylint: disable=missing-function-docstring
+"""Built-in hypothesis functions base implementation.
+
+This module contains hypothesis function abstract definitions that
+correspond to the pandera.core.base.checks.Check methods. These functions do not
+actually implement any validation logic and serve as the entrypoint for
+dispatching specific implementations based on the data object type, e.g.
+`pandas.DataFrame`s.
+"""
+
+from typing import Any, Tuple
+
+from pandera.core.hypotheses import Hypothesis
+
+
+@Hypothesis.register_builtin_check_fn
+def two_sample_ttest(
+    *samples: Tuple[Any, ...],
+    equal_var: bool = True,
+    nan_policy: str = "propagate",
+):
+    raise NotImplementedError
+
+
+@Hypothesis.register_builtin_check_fn
+def one_sample_ttest(
+    *samples: Tuple[Any, ...],
+    popmean: float,
+    nan_policy: str = "propagate",
+):
+    raise NotImplementedError

pandera/backends/pandas/__init__.py

@@ -0,0 +1,38 @@
+"""Pandas backend implementation for schemas and checks."""
+
+import pandas as pd
+
+import pandera.typing
+from pandera.core.checks import Check
+from pandera.core.hypotheses import Hypothesis
+
+from pandera.backends.pandas.checks import PandasCheckBackend
+from pandera.backends.pandas.hypotheses import PandasHypothesisBackend
+from pandera.backends.pandas import builtin_checks, builtin_hypotheses
+
+
+data_types = [pd.DataFrame, pd.Series]
+
+if pandera.typing.dask.DASK_INSTALLED:
+    import dask.dataframe as dd
+
+    data_types.extend([dd.DataFrame, dd.Series])
+
+if pandera.typing.modin.MODIN_INSTALLED:
+    import modin.pandas as mpd
+
+    data_types.extend([mpd.DataFrame, mpd.Series])
+
+if pandera.typing.pyspark.PYSPARK_INSTALLED:
+    import pyspark.pandas as ps
+
+    data_types.extend([ps.DataFrame, ps.Series])
+
+if pandera.typing.geopandas.GEOPANDAS_INSTALLED:
+    import geopandas as gpd
+
+    data_types.extend([gpd.GeoDataFrame, gpd.GeoSeries])
+
+for t in data_types:
+    Check.register_backend(t, PandasCheckBackend)
+    Hypothesis.register_backend(t, PandasHypothesisBackend)

pandera/backends/pandas/array.py

@@ -102,7 +102,9 @@ def validate(
 
         if lazy and error_handler.collected_errors:
             raise SchemaErrors(
-                schema, error_handler.collected_errors, check_obj
+                schema=schema,
+                schema_errors=error_handler.collected_errors,
+                data=check_obj,
             )
         return check_obj
 

pandera/backends/pandas/base.py

@@ -2,6 +2,8 @@
 
 import warnings
 from typing import (
+    Any,
+    Dict,
     FrozenSet,
     Iterable,
     List,
@@ -17,10 +19,12 @@
 from pandera.backends.pandas.error_formatters import (
     format_generic_error_message,
     format_vectorized_error_message,
+    consolidate_failure_cases,
+    summarize_failure_cases,
     reshape_failure_cases,
     scalar_failure_case,
 )
-from pandera.errors import SchemaError
+from pandera.errors import SchemaError, FailureCaseMetadata
 
 
 class ColumnInfo(NamedTuple):
@@ -118,3 +122,19 @@ def run_check(
                 check_output=check_result.check_output,
             )
         return check_result.check_passed
+
+    def failure_cases_metadata(
+        self,
+        schema_name: str,
+        schema_errors: List[Dict[str, Any]],
+    ) -> FailureCaseMetadata:
+        """Create failure cases metadata required for SchemaErrors exception."""
+        failure_cases = consolidate_failure_cases(schema_errors)
+        message, error_counts = summarize_failure_cases(
+            schema_name, schema_errors, failure_cases
+        )
+        return FailureCaseMetadata(
+            failure_cases=failure_cases,
+            message=message,
+            error_counts=error_counts,
+        )