def test_two_files(self):
runner = MRJobRunner(conf_paths=[])
runner._invoke_sort([self.a, self.b], self.out)
self.assertEqual(
list(open(self.out)),
['A\n', 'B\n', 'alligator\n', 'apple\n', 'ball\n', 'banana\n'])
def test_jobconf(self):
jobconf = {'FOO': 'bar', 'BAZ': 'qux', 'BAX': 'Arnold'}
runner = MRJobRunner(conf_path=False, jobconf=jobconf)
assert_equal(runner._hadoop_conf_args(0, 1),
['-jobconf', 'BAX=Arnold',
'-jobconf', 'BAZ=qux',
'-jobconf', 'FOO=bar',])
def test_cmdenv(self):
cmdenv = {'FOO': 'bar', 'BAZ': 'qux', 'BAX': 'Arnold'}
runner = MRJobRunner(conf_path=False, cmdenv=cmdenv)
assert_equal(runner._hadoop_conf_args(0, 1),
['-cmdenv', 'BAX=Arnold',
'-cmdenv', 'BAZ=qux',
'-cmdenv', 'FOO=bar',])
def test_bad_sort(self):
self.use_bad_sort()
runner = MRJobRunner(conf_paths=[])
with no_handlers_for_logger():
self.assertRaises(CalledProcessError, runner._invoke_sort,
[self.a, self.b], self.out)
def test_one_file(self):
runner = MRJobRunner(conf_paths=[])
runner._invoke_sort([self.a], self.out)
self.assertEqual(list(open(self.out)),
['A\n',
'alligator\n',
'apple\n'])
def test_one_file(self):
runner = MRJobRunner(conf_paths=[])
self.addCleanup(runner.cleanup)
runner._invoke_sort([self.a], self.out)
with open(self.out) as out_f:
self.assertEqual(list(out_f), ['A\n', 'alligator\n', 'apple\n'])
def test_hadoop_output_format(self):
format = 'org.apache.hadoop.mapred.SequenceFileOutputFormat'
runner = MRJobRunner(conf_path=False, hadoop_output_format=format)
assert_equal(runner._hadoop_conf_args(0, 1),
['-outputformat', format])
# test multi-step job
assert_equal(runner._hadoop_conf_args(0, 2), [])
assert_equal(runner._hadoop_conf_args(1, 2),
['-outputformat', format])
def test_bad_sort(self):
self.use_bad_sort()
runner = MRJobRunner(conf_paths=[])
self.addCleanup(runner.cleanup)
with no_handlers_for_logger():
# sometimes we get a broken pipe error (IOError) on PyPy
self.assertRaises((CalledProcessError, IOError),
runner._invoke_sort, [self.a, self.b], self.out)
def test_hadoop_extra_args_comes_first(self):
runner = MRJobRunner(
conf_path=False,
cmdenv={'FOO': 'bar'},
hadoop_input_format='FooInputFormat',
hadoop_output_format='BarOutputFormat',
jobconf={'baz': 'quz'},
hadoop_extra_args=['-libjar', 'qux.jar'])
# hadoop_extra_args should come first
conf_args = runner._hadoop_conf_args(0, 1)
assert_equal(conf_args[:2], ['-libjar', 'qux.jar'])
assert_equal(len(conf_args), 10)
def test_python_bin(self):
runner = MRJobRunner(python_bin=['python', '-v'])
self.assertEqual(runner._interpreter(), ['python', '-v'])
self.assertEqual(runner._interpreter(steps=True), [sys.executable])
def test_bootstrap_mrjob_overrides_interpreter(self):
runner = MRJobRunner(
conf_paths=[], interpreter=['ruby'], bootstrap_mrjob=True)
self.assertEqual(runner._bootstrap_mrjob(), True)
def test_interpreter(self):
runner = MRJobRunner(conf_paths=[], interpreter=['ruby'])
self.assertEqual(runner._bootstrap_mrjob(), False)
def test_no_bootstrap_mrjob(self):
runner = MRJobRunner(conf_paths=[], bootstrap_mrjob=False)
self.assertEqual(runner._bootstrap_mrjob(), False)
def test_default(self):
runner = MRJobRunner(conf_paths=[])
self.assertEqual(runner._bootstrap_mrjob(), True)
def test_hadoop_extra_args(self):
extra_args = ['-foo', 'bar']
runner = MRJobRunner(conf_path=False, hadoop_extra_args=extra_args)
assert_equal(runner._hadoop_conf_args(0, 1), extra_args)
def test_environment_variables_non_windows(self):
runner = MRJobRunner(conf_path=False)
self.environment_variable_checks(runner, ['TEMP', 'TMPDIR'])
def test_environment_variables_windows(self):
runner = MRJobRunner(conf_paths=[])
self.addCleanup(runner.cleanup)
runner._sort_is_windows_sort = True
self.environment_variable_checks(runner, ['TMP'])
def test_default(self):
runner = MRJobRunner()
self.assertEqual(runner._interpreter(),
self.default_python_bin())
self.assertEqual(runner._interpreter(steps=True),
[sys.executable])
def test_bad_sort(self):
self.use_bad_sort()
runner = MRJobRunner(conf_path=False)
assert_raises(CalledProcessError,
runner._invoke_sort, [self.a, self.b], self.out)
def test_no_files(self):
runner = MRJobRunner(conf_path=False)
assert_raises(ValueError,
runner._invoke_sort, [], self.out)
def test_environment_variables_non_windows(self):
runner = MRJobRunner(conf_paths=[])
self.addCleanup(runner.cleanup)
self.environment_variable_checks(runner, ['TEMP', 'TMPDIR'])
def test_steps_python_bin(self):
runner = MRJobRunner(steps_python_bin=['python', '-v'])
self.assertEqual(runner._interpreter(),
self.default_python_bin())
self.assertEqual(runner._interpreter(steps=True), ['python', '-v'])
def test_no_files(self):
runner = MRJobRunner(conf_paths=[])
self.assertRaises(ValueError, runner._invoke_sort, [], self.out)
def test_interpreter(self):
runner = MRJobRunner(interpreter=['ruby'])
self.assertEqual(runner._interpreter(), ['ruby'])
self.assertEqual(runner._interpreter(steps=True), ['ruby'])
def test_steps_interpreter(self):
# including whether steps_interpreter overrides interpreter
runner = MRJobRunner(interpreter=['ruby', '-v'],
steps_interpreter=['ruby'])
self.assertEqual(runner._interpreter(), ['ruby', '-v'])
self.assertEqual(runner._interpreter(steps=True), ['ruby'])
def test_interpreter_overrides_steps_python_bin(self):
runner = MRJobRunner(interpreter=['ruby'],
steps_python_bin=['python', '-v'])
self.assertEqual(runner._interpreter(), ['ruby'])
self.assertEqual(runner._interpreter(steps=True), ['ruby'])
def test_environment_variables_windows(self):
runner = MRJobRunner(conf_path=False)
runner._sort_is_windows_sort = True
self.environment_variable_checks(runner, ['TMP'])
def test_empty(self):
runner = MRJobRunner(conf_path=False)
assert_equal(runner._hadoop_conf_args(0, 1), [])