python/pyspark/testing/utils.py - master - githubmirror/spark

python/pyspark/testing/utils.py
1 250 строк47 KB
Yicong Huang
[SPARK-54090][PYTHON] Fix cascading diff in assertDataFrameEqual when row counts differ
24 апр 2026, 07:33
24 апр 2026, 07:33845a1b5
#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

import os
import struct
import sys
import unittest
import difflib
import faulthandler
import functools
from decimal import Decimal
from time import time, sleep
import signal
from typing import (
    Any,
    Optional,
    Union,
    Dict,
    List,
    Callable,
)
from itertools import zip_longest

from pyspark import SparkConf
from pyspark.errors import PySparkAssertionError, PySparkException, PySparkTypeError
from pyspark.errors.exceptions.base import QueryContextType
from pyspark.sql.dataframe import DataFrame
from pyspark.sql import Row
from pyspark.sql.types import StructType, StructField, VariantVal
from pyspark.sql.functions import col, when

__all__ = ["assertDataFrameEqual", "assertSchemaEqual"]


def have_package(name: str) -> bool:
    import importlib

    return importlib.util.find_spec(name) is not None


have_numpy = have_package("numpy")
numpy_requirement_message = "" if have_numpy else "No module named 'numpy'"

have_scipy = have_package("scipy")
scipy_requirement_message = "" if have_scipy else "No module named 'scipy'"

have_sklearn = have_package("sklearn")
sklearn_requirement_message = "" if have_sklearn else "No module named 'sklearn'"

have_torch = have_package("torch")
torch_requirement_message = "" if have_torch else "No module named 'torch'"

have_torcheval = have_package("torcheval")
torcheval_requirement_message = "" if have_torcheval else "No module named 'torcheval'"

have_deepspeed = have_package("deepspeed")
deepspeed_requirement_message = "" if have_deepspeed else "No module named 'deepspeed'"

have_plotly = have_package("plotly")
plotly_requirement_message = "" if have_plotly else "No module named 'plotly'"

have_matplotlib = have_package("matplotlib")
matplotlib_requirement_message = "" if have_matplotlib else "No module named 'matplotlib'"

have_tabulate = have_package("tabulate")
tabulate_requirement_message = "" if have_tabulate else "No module named 'tabulate'"

have_graphviz = have_package("graphviz")
graphviz_requirement_message = "" if have_graphviz else "No module named 'graphviz'"

have_flameprof = have_package("flameprof")
flameprof_requirement_message = "" if have_flameprof else "No module named 'flameprof'"

have_jinja2 = have_package("jinja2")
jinja2_requirement_message = "" if have_jinja2 else "No module named 'jinja2'"

have_openpyxl = have_package("openpyxl")
openpyxl_requirement_message = "" if have_openpyxl else "No module named 'openpyxl'"

have_yaml = have_package("yaml")
yaml_requirement_message = "" if have_yaml else "No module named 'yaml'"

have_grpc = have_package("grpc")
grpc_requirement_message = "" if have_grpc else "No module named 'grpc'"

have_grpc_status = have_package("grpc_status")
grpc_status_requirement_message = "" if have_grpc_status else "No module named 'grpc_status'"

have_zstandard = have_package("zstandard")
zstandard_requirement_message = "" if have_zstandard else "No module named 'zstandard'"

have_kafka = have_package("kafka")
kafka_requirement_message = "" if have_kafka else "No module named 'kafka'"

have_testcontainers = have_package("testcontainers")
testcontainers_requirement_message = (
    "" if have_testcontainers else "No module named 'testcontainers'"
)


googleapis_common_protos_requirement_message = ""

try:
    from google.rpc import error_details_pb2  # noqa: F401
except ImportError as e:
    googleapis_common_protos_requirement_message = str(e)
have_googleapis_common_protos = not googleapis_common_protos_requirement_message

pandas_requirement_message = ""
try:
    from pyspark.sql.pandas.utils import require_minimum_pandas_version

    require_minimum_pandas_version()
except Exception as e:
    # If Pandas version requirement is not satisfied, skip related tests.
    pandas_requirement_message = str(e)

have_pandas = not pandas_requirement_message


pyarrow_requirement_message = ""
try:
    from pyspark.sql.pandas.utils import require_minimum_pyarrow_version

    require_minimum_pyarrow_version()
except Exception as e:
    # If Arrow version requirement is not satisfied, skip related tests.
    pyarrow_requirement_message = str(e)

have_pyarrow = not pyarrow_requirement_message


connect_requirement_message = (
    pandas_requirement_message
    or pyarrow_requirement_message
    or grpc_requirement_message
    or googleapis_common_protos_requirement_message
    or grpc_status_requirement_message
    or zstandard_requirement_message
)

should_test_connect = not connect_requirement_message


is_ansi_mode_test = True
if os.environ.get("SPARK_ANSI_SQL_MODE") == "false":
    is_ansi_mode_test = False

ansi_mode_not_supported_message = "ANSI mode is not supported" if is_ansi_mode_test else ""


def read_int(b):
    return struct.unpack("!i", b)[0]


def write_int(i):
    return struct.pack("!i", i)


def timeout(timeout):
    def decorator(func):
        def handler(signum, frame):
            raise TimeoutError(f"Function {func.__name__} timed out after {timeout} seconds")

        def wrapper(*args, **kwargs):
            signal.alarm(0)
            signal.signal(signal.SIGALRM, handler)
            signal.alarm(timeout)
            try:
                result = func(*args, **kwargs)
            finally:
                signal.alarm(0)
            return result

        return wrapper

    return decorator


def eventually(
    timeout=30.0,
    catch_assertions=False,
    catch_timeout=False,
    quiet=True,
    interval=0.1,
    expected_exceptions=tuple(),
):
    """
    Wait a given amount of time for a condition to pass, else fail with an error.
    This is a helper utility for PySpark tests.

    Parameters
    ----------
    condition : function
        Function that checks for termination conditions. condition() can return:
            - True or None: Conditions met. Return without error.
            - other value: Conditions not met yet. Continue. Upon timeout,
              include last such value in error message.
              Note that this method may be called at any time during
              streaming execution (e.g., even before any results
              have been created).
    timeout : int
        Number of seconds to wait.  Default 30 seconds.
    catch_assertions : bool
        If False (default), do not catch AssertionErrors.
        If True, catch AssertionErrors; continue, but save
        error to throw upon timeout.
    catch_timeout : bool
        If False (default), do not catch TimeoutError.
        If True, catch TimeoutError; continue, but save
        error to throw upon timeout.
    quiet : bool
        If True (default), do not print any output.
        If False, print output.
    interval : float
        Number of seconds to wait between attempts.  Default 0.1 seconds.
    """
    assert timeout > 0
    assert isinstance(catch_assertions, bool)
    assert isinstance(catch_timeout, bool)
    assert isinstance(quiet, bool)
    assert isinstance(interval, float)
    assert isinstance(expected_exceptions, (tuple, list))

    expected_exceptions = list(expected_exceptions)
    if catch_assertions:
        expected_exceptions.append(AssertionError)
    if catch_timeout:
        expected_exceptions.append(TimeoutError)
    expected_exceptions = tuple(expected_exceptions)

    def decorator(condition: Callable) -> Callable:
        assert isinstance(condition, Callable)

        @functools.wraps(condition)
        def wrapper(*args: Any, **kwargs: Any) -> Any:
            start_time = time()
            lastValue = None
            numTries = 0
            while time() - start_time < timeout:
                numTries += 1

                try:
                    lastValue = condition(*args, **kwargs)
                except expected_exceptions as e:
                    lastValue = e

                if lastValue is True or lastValue is None:
                    return

                if not quiet:
                    print(f"\nAttempt #{numTries} failed!\n{lastValue}")
                sleep(interval)

            if isinstance(lastValue, expected_exceptions):
                raise lastValue
            else:
                raise AssertionError(
                    "Test failed due to timeout after %g sec, with last condition returning: %s"
                    % (timeout, lastValue)
                )

        return wrapper

    return decorator


class QuietTest:
    def __init__(self, sc):
        self.log4j = sc._jvm.org.apache.log4j

    def __enter__(self):
        self.old_level = self.log4j.LogManager.getRootLogger().getLevel()
        self.log4j.LogManager.getRootLogger().setLevel(self.log4j.Level.FATAL)

    def __exit__(self, exc_type, exc_val, exc_tb):
        self.log4j.LogManager.getRootLogger().setLevel(self.old_level)


class PySparkBaseTestCase(unittest.TestCase):
    @classmethod
    def setUpClass(cls):
        if os.environ.get("PYSPARK_TEST_TIMEOUT"):
            faulthandler.register(signal.SIGTERM, file=sys.__stderr__, all_threads=True)

    @classmethod
    def tearDownClass(cls):
        if os.environ.get("PYSPARK_TEST_TIMEOUT"):
            faulthandler.unregister(signal.SIGTERM)


class PySparkTestCase(PySparkBaseTestCase):
    def setUp(self):
        from pyspark import SparkContext

        self._old_sys_path = list(sys.path)
        class_name = self.__class__.__name__
        self.sc = SparkContext("local[4]", class_name)

    def tearDown(self):
        self.sc.stop()
        sys.path = self._old_sys_path


class ReusedPySparkTestCase(PySparkBaseTestCase):
    @classmethod
    def conf(cls):
        """
        Override this in subclasses to supply a more specific conf
        """
        return SparkConf()

    @classmethod
    def setUpClass(cls):
        super().setUpClass()

        from pyspark import SparkContext

        cls.sc = SparkContext(cls.master(), cls.__name__, conf=cls.conf())

    @classmethod
    def master(cls):
        return "local[4]"

    @classmethod
    def tearDownClass(cls):
        try:
            cls.sc.stop()
        finally:
            super().tearDownClass()

    def test_assert_classic_mode(self):
        from pyspark.sql import is_remote

        self.assertFalse(is_remote())

    def quiet(self):
        from pyspark.testing.utils import QuietTest

        return QuietTest(self.sc)


class ByteArrayOutput:
    def __init__(self):
        self.buffer = bytearray()

    def write(self, b):
        self.buffer += b

    def close(self):
        pass


def _terminal_color_support():
    try:
        # determine if environment supports color
        script = "$(test $(tput colors)) && $(test $(tput colors) -ge 8) && echo true || echo false"
        return os.popen(script).read()
    except Exception:
        return False


def _context_diff(actual: List[str], expected: List[str], n: int = 3):
    """
    Modified from difflib context_diff API,
    see original code here: https://github.com/python/cpython/blob/main/Lib/difflib.py#L1180
    """

    def red(s: str) -> str:
        red_color = "\033[31m"
        no_color = "\033[0m"
        return red_color + str(s) + no_color

    prefix = dict(insert="+ ", delete="- ", replace="! ", equal="  ")
    for group in difflib.SequenceMatcher(None, actual, expected).get_grouped_opcodes(n):
        yield "*** actual ***"
        if any(tag in {"replace", "delete"} for tag, _, _, _, _ in group):
            for tag, i1, i2, _, _ in group:
                for line in actual[i1:i2]:
                    if tag != "equal" and _terminal_color_support():
                        yield red(prefix[tag] + str(line))
                    else:
                        yield prefix[tag] + str(line)

        yield "\n"

        yield "*** expected ***"
        if any(tag in {"replace", "insert"} for tag, _, _, _, _ in group):
            for tag, _, _, j1, j2 in group:
                for line in expected[j1:j2]:
                    if tag != "equal" and _terminal_color_support():
                        yield red(prefix[tag] + str(line))
                    else:
                        yield prefix[tag] + str(line)


class PySparkErrorTestUtils:
    """
    This util provide functions to accurate and consistent error testing
    based on PySpark error classes.
    """

    def check_error(
        self,
        exception: PySparkException,
        errorClass: str,
        messageParameters: Optional[Dict[str, str]] = None,
        query_context_type: Optional[QueryContextType] = None,
        fragment: Optional[str] = None,
        matchPVals: bool = False,
    ):
        query_context = exception.getQueryContext()
        assert bool(query_context) == (query_context_type is not None), (
            "`query_context_type` is required when QueryContext exists. "
            f"QueryContext: {query_context}."
        )
        # Test if given error is an instance of PySparkException.
        self.assertIsInstance(
            exception,
            PySparkException,
            f"checkError requires 'PySparkException', got '{exception.__class__.__name__}'.",
        )

        # Test error class
        expected = errorClass
        actual = exception.getCondition()
        self.assertEqual(
            expected, actual, f"Expected error class was '{expected}', got '{actual}'."
        )

        # Test message parameters
        expected = messageParameters
        actual = exception.getMessageParameters()
        if matchPVals:
            self.assertEqual(
                len(expected),
                len(actual),
                "Expected message parameters count does not match actual message parameters count"
                f": {len(expected)}, {len(actual)}.",
            )
            for key, value in expected.items():
                self.assertIn(
                    key,
                    actual,
                    f"Expected message parameter key '{key}' was not found "
                    "in actual message parameters.",
                )
                (
                    self.assertRegex(
                        actual[key],
                        value,
                        f"Expected message parameter value '{value}' does not match actual message "
                        f"parameter value '{actual[key]}'.",
                    ),
                )
        else:
            self.assertEqual(
                expected, actual, f"Expected message parameters was '{expected}', got '{actual}'"
            )

        # Test query context
        if query_context:
            expected = query_context_type
            actual_contexts = exception.getQueryContext()
            for actual_context in actual_contexts:
                actual = actual_context.contextType()
                self.assertEqual(
                    expected, actual, f"Expected QueryContext was '{expected}', got '{actual}'"
                )
                if actual == QueryContextType.DataFrame:
                    assert fragment is not None, (
                        "`fragment` is required when QueryContextType is DataFrame."
                    )
                    expected = fragment
                    actual = actual_context.fragment()
                    self.assertEqual(
                        expected,
                        actual,
                        f"Expected PySpark fragment was '{expected}', got '{actual}'",
                    )


def assertSchemaEqual(
    actual: StructType,
    expected: StructType,
    ignoreNullable: bool = True,
    ignoreColumnOrder: bool = False,
    ignoreColumnName: bool = False,
):
    __tracebackhide__ = True
    r"""
    A util function to assert equality between DataFrame schemas `actual` and `expected`.

    .. versionadded:: 3.5.0

    Parameters
    ----------
    actual : StructType
        The DataFrame schema that is being compared or tested.
    expected : StructType
        The expected schema, for comparison with the actual schema.
    ignoreNullable : bool, default True
        Specifies whether a column's nullable property is included when checking for
        schema equality.
        When set to `True` (default), the nullable property of the columns being compared
        is not taken into account and the columns will be considered equal even if they have
        different nullable settings.
        When set to `False`, columns are considered equal only if they have the same nullable
        setting.
        .. versionadded:: 4.0.0
    ignoreColumnOrder : bool, default False
        Specifies whether to compare columns in the order they appear in the DataFrame or by
        column name.
        If set to `False` (default), columns are compared in the order they appear in the
        DataFrames.
        When set to `True`, a column in the expected DataFrame is compared to the column with the
        same name in the actual DataFrame.
        .. versionadded:: 4.0.0
    ignoreColumnName : bool, default False
        Specifies whether to fail the initial schema equality check if the column names in the two
        DataFrames are different.
        When set to `False` (default), column names are checked and the function fails if they are
        different.
        When set to `True`, the function will succeed even if column names are different.
        Column data types are compared for columns in the order they appear in the DataFrames.
        .. versionadded:: 4.0.0

    Notes
    -----
    When assertSchemaEqual fails, the error message uses the Python `difflib` library to display
    a diff log of the `actual` and `expected` schemas.

    Examples
    --------
    >>> from pyspark.sql.types import StructType, StructField, ArrayType, IntegerType, DoubleType
    >>> s1 = StructType([StructField("names", ArrayType(DoubleType(), True), True)])
    >>> s2 = StructType([StructField("names", ArrayType(DoubleType(), True), True)])
    >>> assertSchemaEqual(s1, s2)  # pass, schemas are identical

    Different schemas with `ignoreNullable=False` would fail.

    >>> s3 = StructType([StructField("names", ArrayType(DoubleType(), True), False)])
    >>> assertSchemaEqual(s1, s3, ignoreNullable=False)  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_SCHEMA] Schemas do not match.
    --- actual
    +++ expected
    - StructType([StructField('names', ArrayType(DoubleType(), True), True)])
    ?                                                                 ^^^
    + StructType([StructField('names', ArrayType(DoubleType(), True), False)])
    ?                                                                 ^^^^


    >>> df1 = spark.createDataFrame(data=[(1, 1000), (2, 3000)], schema=["id", "number"])
    >>> df2 = spark.createDataFrame(data=[("1", 1000), ("2", 5000)], schema=["id", "amount"])
    >>> assertSchemaEqual(df1.schema, df2.schema)  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_SCHEMA] Schemas do not match.
    --- actual
    +++ expected
    - StructType([StructField('id', LongType(), True), StructField('number', LongType(), True)])
    ?                               ^^                               ^^^^^
    + StructType([StructField('id', StringType(), True), StructField('amount', LongType(), True)])
    ?                               ^^^^                              ++++ ^

    Compare two schemas ignoring the column order.

    >>> s1 = StructType(
    ...     [StructField("a", IntegerType(), True), StructField("b", DoubleType(), True)]
    ... )
    >>> s2 = StructType(
    ...     [StructField("b", DoubleType(), True), StructField("a", IntegerType(), True)]
    ... )
    >>> assertSchemaEqual(s1, s2, ignoreColumnOrder=True)

    Compare two schemas ignoring the column names.

    >>> s1 = StructType(
    ...     [StructField("a", IntegerType(), True), StructField("c", DoubleType(), True)]
    ... )
    >>> s2 = StructType(
    ...     [StructField("b", IntegerType(), True), StructField("d", DoubleType(), True)]
    ... )
    >>> assertSchemaEqual(s1, s2, ignoreColumnName=True)
    """
    if not isinstance(actual, StructType):
        raise PySparkTypeError(
            errorClass="NOT_EXPECTED_TYPE",
            messageParameters={
                "arg_name": "actual",
                "expected_type": "struct type",
                "arg_type": type(actual).__name__,
            },
        )
    if not isinstance(expected, StructType):
        raise PySparkTypeError(
            errorClass="NOT_EXPECTED_TYPE",
            messageParameters={
                "arg_name": "expected",
                "expected_type": "struct type",
                "arg_type": type(expected).__name__,
            },
        )

    def compare_schemas_ignore_nullable(s1: StructType, s2: StructType):
        if len(s1) != len(s2):
            return False
        zipped = zip_longest(s1, s2)
        for sf1, sf2 in zipped:
            if not compare_structfields_ignore_nullable(sf1, sf2):
                return False
        return True

    def compare_structfields_ignore_nullable(actualSF: StructField, expectedSF: StructField):
        if actualSF is None and expectedSF is None:
            return True
        elif actualSF is None or expectedSF is None:
            return False
        if actualSF.name != expectedSF.name:
            return False
        else:
            return compare_datatypes_ignore_nullable(actualSF.dataType, expectedSF.dataType)

    def compare_datatypes_ignore_nullable(dt1: Any, dt2: Any):
        # checks datatype equality, using recursion to ignore nullable
        if dt1.typeName() == dt2.typeName():
            if dt1.typeName() == "array":
                return compare_datatypes_ignore_nullable(dt1.elementType, dt2.elementType)
            elif dt1.typeName() == "map":
                return compare_datatypes_ignore_nullable(
                    dt1.keyType, dt2.keyType
                ) and compare_datatypes_ignore_nullable(dt1.valueType, dt2.valueType)
            elif dt1.typeName() == "decimal":
                # Fix for SPARK-51062: Compare precision and scale for decimal types
                return dt1.precision == dt2.precision and dt1.scale == dt2.scale
            elif dt1.typeName() == "struct":
                return compare_schemas_ignore_nullable(dt1, dt2)
            else:
                return True
        else:
            return False

    if ignoreColumnOrder:
        actual = StructType(sorted(actual, key=lambda x: x.name))
        expected = StructType(sorted(expected, key=lambda x: x.name))

    if ignoreColumnName:
        actual = StructType(
            [StructField(str(i), field.dataType, field.nullable) for i, field in enumerate(actual)]
        )
        expected = StructType(
            [
                StructField(str(i), field.dataType, field.nullable)
                for i, field in enumerate(expected)
            ]
        )

    if (ignoreNullable and not compare_schemas_ignore_nullable(actual, expected)) or (
        not ignoreNullable and actual != expected
    ):
        generated_diff = difflib.ndiff(str(actual).splitlines(), str(expected).splitlines())
        error_msg = "\n".join(generated_diff)
        raise PySparkAssertionError(
            errorClass="DIFFERENT_SCHEMA",
            messageParameters={"error_msg": error_msg},
        )


from typing import TYPE_CHECKING

if TYPE_CHECKING:
    import pandas
    import pyspark.pandas


def assertDataFrameEqual(
    actual: Union[DataFrame, "pandas.DataFrame", "pyspark.pandas.DataFrame", List[Row]],
    expected: Union[DataFrame, "pandas.DataFrame", "pyspark.pandas.DataFrame", List[Row]],
    checkRowOrder: bool = False,
    rtol: float = 1e-5,
    atol: float = 1e-8,
    ignoreNullable: bool = True,
    ignoreColumnOrder: bool = False,
    ignoreColumnName: bool = False,
    ignoreColumnType: bool = False,
    maxErrors: Optional[int] = None,
    showOnlyDiff: bool = False,
    includeDiffRows=False,
):
    __tracebackhide__ = True
    r"""
    A util function to assert equality between `actual` and `expected`
    (DataFrames or lists of Rows), with optional parameters `checkRowOrder`, `rtol`, and `atol`.

    Supports Spark, Spark Connect, pandas, and pandas-on-Spark DataFrames.
    For more information about pandas-on-Spark DataFrame equality, see the docs for
    `assertPandasOnSparkEqual`.

    .. versionadded:: 3.5.0

    Parameters
    ----------
    actual : DataFrame (Spark, Spark Connect, pandas, or pandas-on-Spark) or list of Rows
        The DataFrame that is being compared or tested.
    expected : DataFrame (Spark, Spark Connect, pandas, or pandas-on-Spark) or list of Rows
        The expected result of the operation, for comparison with the actual result.
    checkRowOrder : bool, optional
        A flag indicating whether the order of rows should be considered in the comparison.
        If set to `False` (default), the row order is not taken into account.
        If set to `True`, the order of rows is important and will be checked during comparison.
        (See Notes)
    rtol : float, optional
        The relative tolerance, used in asserting approximate equality for float values in actual
        and expected. Set to 1e-5 by default. (See Notes)
    atol : float, optional
        The absolute tolerance, used in asserting approximate equality for float values in actual
        and expected. Set to 1e-8 by default. (See Notes)
    ignoreNullable : bool, default True
        Specifies whether a column's nullable property is included when checking for
        schema equality.
        When set to `True` (default), the nullable property of the columns being compared
        is not taken into account and the columns will be considered equal even if they have
        different nullable settings.
        When set to `False`, columns are considered equal only if they have the same nullable
        setting.

        .. versionadded:: 4.0.0
    ignoreColumnOrder : bool, default False
        Specifies whether to compare columns in the order they appear in the DataFrame or by
        column name.
        If set to `False` (default), columns are compared in the order they appear in the
        DataFrames.
        When set to `True`, a column in the expected DataFrame is compared to the column with the
        same name in the actual DataFrame.

        .. versionadded:: 4.0.0
    ignoreColumnName : bool, default False
        Specifies whether to fail the initial schema equality check if the column names in the two
        DataFrames are different.
        When set to `False` (default), column names are checked and the function fails if they are
        different.
        When set to `True`, the function will succeed even if column names are different.
        Column data types are compared for columns in the order they appear in the DataFrames.

        .. versionadded:: 4.0.0
    ignoreColumnType : bool, default False
        Specifies whether to ignore the data type of the columns when comparing.
        When set to `False` (default), column data types are checked and the function fails if they
        are different.
        When set to `True`, the schema equality check will succeed even if column data types are
        different and the function will attempt to compare rows.

        .. versionadded:: 4.0.0
    maxErrors : bool, optional
        The maximum number of row comparison failures to encounter before returning.
        When this number of row comparisons have failed, the function returns independent of how
        many rows have been compared.
        Set to None by default which means compare all rows independent of number of failures.

        .. versionadded:: 4.0.0
    showOnlyDiff : bool, default False
        If set to `True`, the error message will only include rows that are different.
        If set to `False` (default), the error message will include all rows
        (when there is at least one row that is different).

        .. versionadded:: 4.0.0
    includeDiffRows: bool, False
        If set to `True`, the unequal rows are included in PySparkAssertionError for further
        debugging. If set to `False` (default), the unequal rows are not returned as a data set.

        .. versionadded:: 4.0.0

    Notes
    -----
    When `assertDataFrameEqual` fails, the error message uses the Python `difflib` library to
    display a diff log of each row that differs in `actual` and `expected`.

    For `checkRowOrder`, note that PySpark DataFrame ordering is non-deterministic, unless
    explicitly sorted.

    Note that schema equality is checked only when `expected` is a DataFrame (not a list of Rows).

    For DataFrames with float/decimal values, assertDataFrame asserts approximate equality.
    Two float/decimal values a and b are approximately equal if the following equation is True:

    ``absolute(a - b) <= (atol + rtol * absolute(b))``.

    `ignoreColumnOrder` cannot be set to `True` if `ignoreColumnNames` is also set to `True`.
    `ignoreColumnNames` cannot be set to `True` if `ignoreColumnOrder` is also set to `True`.

    Examples
    --------
    >>> df1 = spark.createDataFrame(data=[("1", 1000), ("2", 3000)], schema=["id", "amount"])
    >>> df2 = spark.createDataFrame(data=[("1", 1000), ("2", 3000)], schema=["id", "amount"])
    >>> assertDataFrameEqual(df1, df2)  # pass, DataFrames are identical

    >>> df1 = spark.createDataFrame(data=[("1", 0.1), ("2", 3.23)], schema=["id", "amount"])
    >>> df2 = spark.createDataFrame(data=[("1", 0.109), ("2", 3.23)], schema=["id", "amount"])
    >>> assertDataFrameEqual(df1, df2, rtol=1e-1)  # pass, DataFrames are approx equal by rtol

    >>> df1 = spark.createDataFrame(data=[(1, 1000), (2, 3000)], schema=["id", "amount"])
    >>> list_of_rows = [Row(1, 1000), Row(2, 3000)]
    >>> assertDataFrameEqual(df1, list_of_rows)  # pass, actual and expected data are equal

    >>> import pyspark.pandas as ps  # doctest: +SKIP
    >>> df1 = ps.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6], 'c': [7, 8, 9]})  # doctest: +SKIP
    >>> df2 = ps.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6], 'c': [7, 8, 9]})  # doctest: +SKIP
    >>> # pass, pandas-on-Spark DataFrames are equal
    >>> assertDataFrameEqual(df1, df2)  # doctest: +SKIP

    >>> df1 = spark.createDataFrame(
    ...     data=[("1", 1000.00), ("2", 3000.00), ("3", 2000.00)], schema=["id", "amount"])
    >>> df2 = spark.createDataFrame(
    ...     data=[("1", 1001.00), ("2", 3000.00), ("3", 2003.00)], schema=["id", "amount"])
    >>> assertDataFrameEqual(df1, df2)  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_ROWS] Results do not match: ( 66.66667 % )
    *** actual ***
    ! Row(id='1', amount=1000.0)
      Row(id='2', amount=3000.0)
    ! Row(id='3', amount=2000.0)
    *** expected ***
    ! Row(id='1', amount=1001.0)
      Row(id='2', amount=3000.0)
    ! Row(id='3', amount=2003.0)

    Example for ignoreNullable

    >>> from pyspark.sql.types import StructType, StructField, StringType, LongType
    >>> df1_nullable = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")],
    ...     schema=StructType(
    ...         [StructField("amount", LongType(), True), StructField("id", StringType(), True)]
    ...     )
    ... )
    >>> df2_nullable = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")],
    ...     schema=StructType(
    ...         [StructField("amount", LongType(), True), StructField("id", StringType(), False)]
    ...     )
    ... )
    >>> assertDataFrameEqual(df1_nullable, df2_nullable, ignoreNullable=True)  # pass
    >>> assertDataFrameEqual(
    ...     df1_nullable, df2_nullable, ignoreNullable=False
    ... )  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_SCHEMA] Schemas do not match.
    --- actual
    +++ expected
    - StructType([StructField('amount', LongType(), True), StructField('id', StringType(), True)])
    ?                                                                                      ^^^
    + StructType([StructField('amount', LongType(), True), StructField('id', StringType(), False)])
    ?                                                                                      ^^^^

    Example for ignoreColumnOrder

    >>> df1_col_order = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")], schema=["amount", "id"]
    ... )
    >>> df2_col_order = spark.createDataFrame(
    ...     data=[("1", 1000), ("2", 5000)], schema=["id", "amount"]
    ... )
    >>> assertDataFrameEqual(df1_col_order, df2_col_order, ignoreColumnOrder=True)

    Example for ignoreColumnName

    >>> df1_col_names = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")], schema=["amount", "identity"]
    ... )
    >>> df2_col_names = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")], schema=["amount", "id"]
    ... )
    >>> assertDataFrameEqual(df1_col_names, df2_col_names, ignoreColumnName=True)

    Example for ignoreColumnType

    >>> df1_col_types = spark.createDataFrame(
    ...     data=[(1000, "1"), (5000, "2")], schema=["amount", "id"]
    ... )
    >>> df2_col_types = spark.createDataFrame(
    ...     data=[(1000.0, "1"), (5000.0, "2")], schema=["amount", "id"]
    ... )
    >>> assertDataFrameEqual(df1_col_types, df2_col_types, ignoreColumnType=True)

    Example for maxErrors (will only report the first mismatching row)

    >>> df1 = spark.createDataFrame([(1, "A"), (2, "B"), (3, "C")])
    >>> df2 = spark.createDataFrame([(1, "A"), (2, "X"), (3, "Y")])
    >>> assertDataFrameEqual(df1, df2, maxErrors=1)  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_ROWS] Results do not match: ( 33.33333 % )
    *** actual ***
      Row(_1=1, _2='A')
    ! Row(_1=2, _2='B')
    *** expected ***
      Row(_1=1, _2='A')
    ! Row(_1=2, _2='X')

    Example for showOnlyDiff (will only report the mismatching rows)

    >>> df1 = spark.createDataFrame([(1, "A"), (2, "B"), (3, "C")])
    >>> df2 = spark.createDataFrame([(1, "A"), (2, "X"), (3, "Y")])
    >>> assertDataFrameEqual(df1, df2, showOnlyDiff=True)  # doctest: +IGNORE_EXCEPTION_DETAIL
    Traceback (most recent call last):
    ...
    PySparkAssertionError: [DIFFERENT_ROWS] Results do not match: ( 66.66667 % )
    *** actual ***
    ! Row(_1=2, _2='B')
    ! Row(_1=3, _2='C')
    *** expected ***
    ! Row(_1=2, _2='X')
    ! Row(_1=3, _2='Y')

    The `includeDiffRows` parameter can be used to include the rows that did not match
    in the PySparkAssertionError. This can be useful for debugging or further analysis.

    >>> df1 = spark.createDataFrame(
    ...     data=[("1", 1000.00), ("2", 3000.00), ("3", 2000.00)], schema=["id", "amount"])
    >>> df2 = spark.createDataFrame(
    ...     data=[("1", 1001.00), ("2", 3000.00), ("3", 2003.00)], schema=["id", "amount"])
    >>> try:
    ...     assertDataFrameEqual(df1, df2, includeDiffRows=True)
    ... except PySparkAssertionError as e:
    ...     spark.createDataFrame(e.data).show()  # doctest: +NORMALIZE_WHITESPACE
    +-----------+-----------+
    |         _1|         _2|
    +-----------+-----------+
    |{1, 1000.0}|{1, 1001.0}|
    |{3, 2000.0}|{3, 2003.0}|
    +-----------+-----------+
    """
    if actual is None and expected is None:
        return True
    elif actual is None:
        raise PySparkAssertionError(
            errorClass="INVALID_TYPE_DF_EQUALITY_ARG",
            messageParameters={
                "expected_type": "Union[DataFrame, ps.DataFrame, List[Row]]",
                "arg_name": "actual",
                "actual_type": None,
            },
        )
    elif expected is None:
        raise PySparkAssertionError(
            errorClass="INVALID_TYPE_DF_EQUALITY_ARG",
            messageParameters={
                "expected_type": "Union[DataFrame, ps.DataFrame, List[Row]]",
                "arg_name": "expected",
                "actual_type": None,
            },
        )

    has_pandas = False
    try:
        # If pandas dependencies are available, allow pandas or pandas-on-Spark DataFrame
        import pandas as pd

        has_pandas = True
    except ImportError:
        # no pandas, so we won't call pandasutils functions
        pass

    has_arrow = False
    try:
        import pyarrow  # noqa: F401

        has_arrow = True
    except ImportError:
        pass

    if has_pandas and has_arrow:
        import pyspark.pandas as ps
        from pyspark.testing.pandasutils import PandasOnSparkTestUtils

        if (
            isinstance(actual, pd.DataFrame)
            or isinstance(expected, pd.DataFrame)
            or isinstance(actual, ps.DataFrame)
            or isinstance(expected, ps.DataFrame)
        ):
            # handle pandas DataFrames
            # assert approximate equality for float data
            return PandasOnSparkTestUtils().assert_eq(
                actual, expected, almost=True, rtol=rtol, atol=atol, check_row_order=checkRowOrder
            )

    if not isinstance(actual, (DataFrame, list)):
        raise PySparkAssertionError(
            errorClass="INVALID_TYPE_DF_EQUALITY_ARG",
            messageParameters={
                "expected_type": "Union[DataFrame, ps.DataFrame, List[Row]]",
                "arg_name": "actual",
                "actual_type": type(actual),
            },
        )
    elif not isinstance(expected, (DataFrame, list)):
        raise PySparkAssertionError(
            errorClass="INVALID_TYPE_DF_EQUALITY_ARG",
            messageParameters={
                "expected_type": "Union[DataFrame, ps.DataFrame, List[Row]]",
                "arg_name": "expected",
                "actual_type": type(expected),
            },
        )

    if ignoreColumnOrder:
        actual = actual.select(*sorted(actual.columns))
        expected = expected.select(*sorted(expected.columns))

    def rename_dataframe_columns(df: DataFrame) -> DataFrame:
        """Rename DataFrame columns to sequential numbers for comparison"""
        renamed_columns = [str(i) for i in range(len(df.columns))]
        return df.toDF(*renamed_columns)

    if ignoreColumnName:
        actual = rename_dataframe_columns(actual)
        expected = rename_dataframe_columns(expected)

    def cast_columns_to_string(df: DataFrame) -> DataFrame:
        """Cast all DataFrame columns to string for comparison"""
        for col_name in df.columns:
            # Add logic to remove trailing .0 for float columns that are whole numbers
            df = df.withColumn(
                col_name,
                when(
                    (col(col_name).cast("float").isNotNull())
                    & (col(col_name).cast("float") == col(col_name).cast("int")),
                    col(col_name).cast("int").cast("string"),
                ).otherwise(col(col_name).cast("string")),
            )
        return df

    if ignoreColumnType:
        actual = cast_columns_to_string(actual)
        expected = cast_columns_to_string(expected)

    def compare_rows(r1: Row, r2: Row):
        def compare_vals(val1, val2):
            if isinstance(val1, list) and isinstance(val2, list):
                return len(val1) == len(val2) and all(
                    compare_vals(x, y) for x, y in zip(val1, val2)
                )
            elif isinstance(val1, Row) and isinstance(val2, Row):
                return all(compare_vals(x, y) for x, y in zip(val1, val2))
            elif isinstance(val1, dict) and isinstance(val2, dict):
                return (
                    len(val1) == len(val2)
                    and val1.keys() == val2.keys()
                    and all(compare_vals(val1[k], val2[k]) for k in val1)
                )
            elif isinstance(val1, float) and isinstance(val2, float):
                if abs(val1 - val2) > (atol + rtol * abs(val2)):
                    return False
            elif isinstance(val1, Decimal) and isinstance(val2, Decimal):
                if abs(val1 - val2) > (Decimal(atol) + Decimal(rtol) * abs(val2)):
                    return False
            elif isinstance(val1, VariantVal) and isinstance(val2, VariantVal):
                return compare_vals(val1.toPython(), val2.toPython())
            else:
                if val1 != val2:
                    return False
            return True

        if r1 is None and r2 is None:
            return True
        elif r1 is None or r2 is None:
            return False

        return compare_vals(r1, r2)

    def assert_rows_equal(
        rows1: List[Row], rows2: List[Row], maxErrors: int = None, showOnlyDiff: bool = False
    ):
        __tracebackhide__ = True
        diff_rows_cnt = 0
        diff_rows = []
        has_diff_rows = False

        rows_str1 = ""
        rows_str2 = ""

        def record_diff(r1, r2):
            nonlocal diff_rows_cnt, has_diff_rows
            diff_rows_cnt += 1
            has_diff_rows = True
            if includeDiffRows:
                diff_rows.append((r1, r2))

        # SPARK-54090: For checkRowOrder=False, the input lists have already
        # been sorted by str(x).  When their lengths differ, the original
        # zip_longest pairing causes a single missing/extra row to cascade
        # into a mismatch on every subsequent row -- inflating both the diff
        # count and the reported percentage.  Walk sorted lists of different
        # lengths as a merge so only truly missing/extra rows contribute.
        # When the lengths match, zip_longest aligns positions correctly and
        # is preserved to keep field-level diffs reported as paired rows.
        use_merge_walk = not checkRowOrder and len(rows1) != len(rows2)

        if not use_merge_walk:
            zipped = list(zip_longest(rows1, rows2))
            for r1, r2 in zipped:
                if not compare_rows(r1, r2):
                    record_diff(r1, r2)
                    rows_str1 += str(r1) + "\n"
                    rows_str2 += str(r2) + "\n"
                    if maxErrors is not None and diff_rows_cnt >= maxErrors:
                        break
                elif not showOnlyDiff:
                    rows_str1 += str(r1) + "\n"
                    rows_str2 += str(r2) + "\n"
            total = len(zipped)
        else:
            i = j = 0
            len1, len2 = len(rows1), len(rows2)
            while i < len1 or j < len2:
                if i < len1 and j < len2:
                    r1, r2 = rows1[i], rows2[j]
                    if compare_rows(r1, r2):
                        if not showOnlyDiff:
                            rows_str1 += str(r1) + "\n"
                            rows_str2 += str(r2) + "\n"
                        i += 1
                        j += 1
                        continue
                    s1, s2 = str(r1), str(r2)
                    if s1 < s2:
                        record_diff(r1, None)
                        rows_str1 += str(r1) + "\n"
                        i += 1
                    elif s1 > s2:
                        record_diff(None, r2)
                        rows_str2 += str(r2) + "\n"
                        j += 1
                    else:
                        # Equal str() but compare_rows said they differ (e.g.
                        # float values outside tolerance that still str-sort
                        # identically).  Treat as a paired mismatch.
                        record_diff(r1, r2)
                        rows_str1 += str(r1) + "\n"
                        rows_str2 += str(r2) + "\n"
                        i += 1
                        j += 1
                elif i < len1:
                    record_diff(rows1[i], None)
                    rows_str1 += str(rows1[i]) + "\n"
                    i += 1
                else:
                    record_diff(None, rows2[j])
                    rows_str2 += str(rows2[j]) + "\n"
                    j += 1
                if maxErrors is not None and diff_rows_cnt >= maxErrors:
                    break
            total = max(len1, len2)

        generated_diff = _context_diff(
            actual=rows_str1.splitlines(), expected=rows_str2.splitlines(), n=total
        )

        if has_diff_rows:
            error_msg = "Results do not match: "
            percent_diff = (diff_rows_cnt / total) * 100 if total else 0
            error_msg += "( %.5f %% )" % percent_diff
            error_msg += "\n" + "\n".join(generated_diff)
            data = diff_rows if includeDiffRows else None
            raise PySparkAssertionError(
                errorClass="DIFFERENT_ROWS", messageParameters={"error_msg": error_msg}, data=data
            )

    # only compare schema if expected is not a List
    if not isinstance(actual, list) and not isinstance(expected, list):
        if ignoreNullable:
            assertSchemaEqual(actual.schema, expected.schema)
        elif actual.schema != expected.schema:
            generated_diff = difflib.ndiff(
                str(actual.schema).splitlines(), str(expected.schema).splitlines()
            )
            error_msg = "\n".join(generated_diff)

            raise PySparkAssertionError(
                errorClass="DIFFERENT_SCHEMA",
                messageParameters={"error_msg": error_msg},
            )

    if not isinstance(actual, list):
        if actual.isStreaming:
            raise PySparkAssertionError(
                errorClass="UNSUPPORTED_OPERATION",
                messageParameters={"operation": "assertDataFrameEqual on streaming DataFrame"},
            )
        actual_list = actual.collect()
    else:
        actual_list = actual

    if not isinstance(expected, list):
        if expected.isStreaming:
            raise PySparkAssertionError(
                errorClass="UNSUPPORTED_OPERATION",
                messageParameters={"operation": "assertDataFrameEqual on streaming DataFrame"},
            )
        expected_list = expected.collect()
    else:
        expected_list = expected

    if not checkRowOrder:
        # rename duplicate columns for sorting
        actual_list = sorted(actual_list, key=lambda x: str(x))
        expected_list = sorted(expected_list, key=lambda x: str(x))

    assert_rows_equal(actual_list, expected_list, maxErrors=maxErrors, showOnlyDiff=showOnlyDiff)


def _test() -> None:
    import doctest
    from pyspark.sql import SparkSession
    import pyspark.testing.utils

    globs = pyspark.testing.utils.__dict__.copy()
    spark = SparkSession.builder.master("local[4]").appName("testing.utils tests").getOrCreate()
    globs["spark"] = spark
    failure_count, test_count = doctest.testmod(
        pyspark.testing.utils,
        globs=globs,
        optionflags=doctest.ELLIPSIS | doctest.NORMALIZE_WHITESPACE,
    )
    spark.stop()
    if failure_count:
        sys.exit(-1)


if __name__ == "__main__":
    _test()