python - 使用 pytest 测试 Spark - 无法在本地模式下运行 Spark

Question

我正在尝试使用来自该站点的 pytest 运行 wordcount 测试 - Unit testing Apache Spark with py.test .问题是我无法启动 spark 上下文。我用来运行 Spark 上下文的代码:

@pytest.fixture(scope="session")
def spark_context(request):
    """ fixture for creating a spark context
    Args:
        request: pytest.FixtureRequest object
    """
    conf = (SparkConf().setMaster("local[2]").setAppName("pytest-pyspark-local-testing"))
    sc = SparkContext(conf=conf)
    request.addfinalizer(lambda: sc.stop())

    quiet_py4j()
    return sc

我使用命令执行此代码:

#first way
pytest spark_context_fixture.py

#second way
python spark_context_fixture.py

输出:

platform linux2 -- Python 2.7.5, pytest-3.0.4, py-1.4.31, pluggy-0.4.0
rootdir: /home/mgr/test, inifile:
collected 0 items

然后我想使用 pytest 运行 wordcount 测试。

pytestmark = pytest.mark.usefixtures("spark_context")

def test_do_word_counts(spark_context):
    """ test word couting
    Args:
        spark_context: test fixture SparkContext
    """
    test_input = [
        ' hello spark ',
        ' hello again spark spark'
    ]

    input_rdd = spark_context.parallelize(test_input, 1)
    results = wordcount.do_word_counts(input_rdd)

    expected_results = {'hello':2, 'spark':3, 'again':1}  
    assert results == expected_results

但是输出是:

________ ERROR at setup of test_do_word_counts _________
file /home/mgrabowski/test/wordcount_test.py, line 5
  def test_do_word_counts(spark_context):
E       fixture 'spark_context' not found
>       available fixtures: cache, capfd, capsys, doctest_namespace, monkeypatch, pytestconfig, record_xml_property, recwarn, tmpdir, tmpdir_factory
>       use 'pytest --fixtures [testpath]' for help on them.

有谁知道这个问题的原因是什么？

Answer 1

我做了一些研究，终于找到了解决方案。我使用 Spark 1.6。

首先，我在 .bashrc 文件中添加了两行。

export SPARK_HOME=/usr/hdp/2.5.0.0-1245/spark
export PYTHONPATH=$SPARK_HOME/python/:$SPARK_HOME/python/lib/py4j-0.9-src.zip:$PYTHONPA‌​TH

然后我创建了文件“conftest.py”。文件名非常重要，您不应该更改它，否则您会看到 spark_context 错误。如果您在本地模式下使用 Spark 并且不使用 YARN，conftest.py 应该如下所示:

import logging
import pytest

from pyspark import HiveContext
from pyspark import SparkConf
from pyspark import SparkContext
from pyspark.streaming import StreamingContext

def quiet_py4j():
    logger = logging.getLogger('py4j')
    logger.setLevel(logging.WARN)

@pytest.fixture(scope="session")
def spark_context(request):
    conf = (SparkConf().setMaster("local[2]").setAppName("pytest-pyspark-local-testing"))
    request.addfinalizer(lambda: sc.stop())

    sc = SparkContext(conf=conf)
    quiet_py4j()
    return sc

@pytest.fixture(scope="session")
def hive_context(spark_context):
    return HiveContext(spark_context)

@pytest.fixture(scope="session")
def streaming_context(spark_context):
    return StreamingContext(spark_context, 1)

现在您可以使用简单的 pytest 命令运行测试。 Pytest 应该运行 Spark 并最终停止它。

如果您使用 YARN，您可以将 conftest.py 更改为:

    import logging
    import pytest

    from pyspark import HiveContext
    from pyspark import SparkConf
    from pyspark import SparkContext
    from pyspark.streaming import StreamingContext

    def quiet_py4j():
        """ turn down spark logging for the test context """
        logger = logging.getLogger('py4j')
        logger.setLevel(logging.WARN)

    @pytest.fixture(scope="session",
                params=[pytest.mark.spark_local('local'),
                        pytest.mark.spark_yarn('yarn')])
    def spark_context(request):
        if request.param == 'local':
            conf = (SparkConf()
                    .setMaster("local[2]")
                    .setAppName("pytest-pyspark-local-testing")
                    )
        elif request.param == 'yarn':
            conf = (SparkConf()
                    .setMaster("yarn-client")
                    .setAppName("pytest-pyspark-yarn-testing")
                    .set("spark.executor.memory", "1g")
                    .set("spark.executor.instances", 2)
                    )
        request.addfinalizer(lambda: sc.stop())

        sc = SparkContext(conf=conf)
        return sc

    @pytest.fixture(scope="session")
    def hive_context(spark_context):
        return HiveContext(spark_context)

    @pytest.fixture(scope="session")
    def streaming_context(spark_context):
        return StreamingContext(spark_context, 1)

现在您可以通过调用 py.test -m spark_local 在本地模式下运行测试，在 YARN 模式下通过调用 py.test -m spark_yarn 运行测试。

字数统计示例

在同一个文件夹中创建三个文件:conftest.py(上面)，wordcount.py:

def do_word_counts(lines):
    counts = (lines.flatMap(lambda x: x.split())
                  .map(lambda x: (x, 1))
                  .reduceByKey(lambda x, y: x+y)
             ) 
    results = {word: count for word, count in counts.collect()}
    return results

和 wordcount_test.py:

import pytest
import wordcount

pytestmark = pytest.mark.usefixtures("spark_context")

def test_do_word_counts(spark_context):
    test_input = [
        ' hello spark ',
        ' hello again spark spark'
    ]

    input_rdd = spark_context.parallelize(test_input, 1)
    results = wordcount.do_word_counts(input_rdd)

    expected_results = {'hello':2, 'spark':3, 'again':1}  
    assert results == expected_results

现在您可以通过调用 pytest 来运行测试。