def test_multi_step_counters(self):
stdin = BytesIO(b'foo\nbar\n')
mr_job = MRCountingJob(['-r', 'local', '-'])
mr_job.sandbox(stdin=stdin)
with mr_job.make_runner() as runner:
runner.run()
self.assertEqual(runner.counters(),
[{'group': {'counter_name': 2}},
{'group': {'counter_name': 2}},
{'group': {'counter_name': 2}}])
def test_multi_step_counters(self):
stdin = StringIO('foo\nbar\n')
mr_job = MRCountingJob(['-c', self.mrjob_conf_path, '-'])
mr_job.sandbox(stdin=stdin)
with mr_job.make_runner() as runner:
runner.run()
self.assertEqual(runner.counters(),
[{'group': {'counter_name': 2}},
{'group': {'counter_name': 2}},
{'group': {'counter_name': 2}}])
def test_output_dir_and_step_output_dir(self):
input1_path = self.makefile('input1')
input2_path = self.makefile('input2')
# this has three steps, which lets us test step numbering
job = MRCountingJob([
'-r',
'hadoop',
'--hadoop-bin',
'false', # shouldn't run; just in case
'--output-dir',
'hdfs:///tmp/output',
'--step-output-dir',
'hdfs://tmp/step-output',
input1_path,
input2_path
])
job.sandbox()
with job.make_runner() as runner:
self.assertEqual(runner._num_steps(), 3)
self.add_files_for_upload(runner)
input_uris_0 = runner._step_input_uris(0)
self.assertEqual([os.path.basename(uri) for uri in input_uris_0],
['input1', 'input2'])
output_uri_0 = runner._step_output_uri(0)
self.assertEqual(output_uri_0, 'hdfs://tmp/step-output/0000')
input_uris_1 = runner._step_input_uris(1)
self.assertEqual(input_uris_1, [output_uri_0])
output_uri_1 = runner._step_output_uri(1)
self.assertEqual(output_uri_1, 'hdfs://tmp/step-output/0001')
input_uris_2 = runner._step_input_uris(2)
self.assertEqual(input_uris_2, [output_uri_1])
output_uri_2 = runner._step_output_uri(2)
self.assertEqual(output_uri_2, 'hdfs:///tmp/output')
def test_local_output_dir_and_step_output_dir(self):
input1_path = self.makefile('input1')
input2_path = self.makefile('input2')
output_dir = self.makedirs('output')
step_output_dir = self.makedirs('step_output')
# this has three steps, which lets us test step numbering
job = MRCountingJob([
'-r', 'local',
'--output-dir', output_dir,
'--step-output-dir', step_output_dir,
input1_path, input2_path])
job.sandbox()
with job.make_runner() as runner:
self.assertEqual(runner._num_steps(), 3)
input_uris_0 = runner._step_input_uris(0)
self.assertEqual([os.path.basename(uri) for uri in input_uris_0],
['input1', 'input2'])
self.assertEqual([uri[:8] for uri in input_uris_0],
['file:///', 'file:///'])
output_uri_0 = runner._step_output_uri(0)
self.assertEqual(output_uri_0,
to_uri(os.path.join(step_output_dir, '0000')))
input_uris_1 = runner._step_input_uris(1)
self.assertEqual(input_uris_1, [output_uri_0])
output_uri_1 = runner._step_output_uri(1)
self.assertEqual(output_uri_1,
to_uri(os.path.join(step_output_dir, '0001')))
input_uris_2 = runner._step_input_uris(2)
self.assertEqual(input_uris_2, [output_uri_1])
output_uri_2 = runner._step_output_uri(2)
self.assertEqual(output_uri_2, to_uri(output_dir))
def test_multi_step_counters(self):
# read from STDIN, a regular file, and a .gz
stdin = StringIO('foo\nbar\n')
mr_job = MRCountingJob(['-c', self.mrjob_conf_path, '-'])
mr_job.sandbox(stdin=stdin)
results = []
with mr_job.make_runner() as runner:
runner.run()
for line in runner.stream_output():
key, value = mr_job.parse_output_line(line)
results.append((key, value))
assert_equal(runner._counters, [{
'group': {
'counter_name': 2
}
}, {
'group': {
'counter_name': 2
}
}, {
'group': {
'counter_name': 2
}
}])
def test_gz_split_regression(self):
gz_path_1 = os.path.join(self.tmp_dir, '1.gz')
gz_path_2 = os.path.join(self.tmp_dir, '2.gz')
path_3 = os.path.join(self.tmp_dir, '3')
input_gz_1 = gzip.GzipFile(gz_path_1, 'wb')
input_gz_1.write(b'x\n')
input_gz_1.close()
input_gz_2 = gzip.GzipFile(gz_path_2, 'wb')
input_gz_2.write(b'y\n')
input_gz_2.close()
with open(path_3, 'wb') as f:
f.write(b'z')
mr_job = MRCountingJob(
['--no-conf', '-r', 'local', gz_path_1, gz_path_2, path_3])
with mr_job.make_runner() as r:
splits = r._get_file_splits([gz_path_1, gz_path_2, path_3], 1)
self.assertEqual(len(set(s['task_num'] for s in splits.values())),
3)
