def test_uniq_combiner(self):
# put data in a .gz to force a single map taxsk
x_gz_path = join(self.tmp_dir, 'data.gz')
with gzip.open(x_gz_path, 'wb') as x_gz:
x_gz.write(b'x\nx\nx\nx\nx\nx\n')
job = MRCmdJob(['--combiner-cmd=uniq', '--runner=local', x_gz_path])
job.sandbox()
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'script',
},
'combiner': {
'type': 'command',
'command': 'uniq'
}
}])
r.run()
# there is only one map task, thus only one combiner,
# thus there should only be one value
self.assertEqual(b''.join(r.cat_output()), b'x\n')
def test_multiple(self):
data = b'x\nx\nx\nx\nx\nx\n'
mapper_cmd = 'cat -e'
reducer_cmd = _bash_wrap('wc -l | tr -Cd "[:digit:]"')
job = MRCmdJob([
'--runner', 'local', '--mapper-cmd', mapper_cmd, '--combiner-cmd',
'uniq', '--reducer-cmd', reducer_cmd
])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'command',
'command': mapper_cmd
},
'combiner': {
'type': 'command',
'command': 'uniq'
},
'reducer': {
'type': 'command',
'command': reducer_cmd
},
}])
r.run()
self.assertEqual(list(r.stream_output()), [b'2'])
def test_multiple(self):
# put data in a .gz to force a single map task
x_gz_path = join(self.tmp_dir, 'data.gz')
with gzip.open(x_gz_path, 'wb') as x_gz:
x_gz.write(b'x\nx\nx\nx\nx\nx\n')
reducer_cmd = '/bin/sh -c \'wc -l | tr -Cd "[:digit:]"\''
job = MRCmdJob([
'--runner', 'local',
'--mapper-cmd', 'cat -e',
'--combiner-cmd', 'uniq',
'--reducer-cmd', reducer_cmd,
x_gz_path])
job.sandbox()
with job.make_runner() as r:
self.assertEqual(
r._get_steps(),
[{
'type': 'streaming',
'mapper': {'type': 'command', 'command': 'cat -e'},
'combiner': {'type': 'command', 'command': 'uniq'},
'reducer': {'type': 'command', 'command': reducer_cmd},
}])
r.run()
self.assertEqual(
sum(int(v) for _, v in job.parse_output(r.cat_output())),
1)
def test_uniq_combiner(self):
# put data in a .gz to force a single map taxsk
x_gz_path = join(self.tmp_dir, 'data.gz')
with gzip.open(x_gz_path, 'wb') as x_gz:
x_gz.write(b'x\nx\nx\nx\nx\nx\n')
job = MRCmdJob(['--combiner-cmd=uniq', '--runner=local', x_gz_path])
job.sandbox()
with job.make_runner() as r:
self.assertEqual(
r._get_steps(),
[{
'type': 'streaming',
'mapper': {
'type': 'script',
},
'combiner': {
'type': 'command',
'command': 'uniq'}}])
r.run()
# there is only one map task, thus only one combiner,
# thus there should only be one value
self.assertEqual(b''.join(r.cat_output()), b'x\n')
def test_multiple_2(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--mapper-cmd=cat', '--reducer-cmd-2', 'wc -l',
'--runner=local', '--no-conf'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
r.run()
self.assertEqual(sum(int(l) for l in to_lines(r.cat_output())), 3)
def test_multiple_2(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--mapper-cmd=cat', '--reducer-cmd-2', 'wc -l',
'--runner=local', '--no-conf'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
r.run()
self.assertEqual(sum(int(l) for l in to_lines(r.cat_output())), 3)
def test_cat_mapper(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--mapper-cmd=cat', '--runner=local'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'command',
'command': 'cat'
}
}])
r.run()
lines = [line.strip() for line in list(r.stream_output())]
self.assertEqual(sorted(lines), sorted(data.split()))
def test_cat_mapper(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--mapper-cmd=cat', '--runner=local'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(
r._get_steps(),
[{
'type': 'streaming',
'mapper': {
'type': 'command',
'command': 'cat'}}])
r.run()
lines = [line.strip() for line in to_lines(r.cat_output())]
self.assertEqual(sorted(lines), sorted(data.split()))
def test_command_streaming_step_without_mr_job_script(self):
# you don't need a script to run commands
steps = MRCmdJob(['--mapper-cmd', 'cat'])._steps_desc()
runner = LocalMRJobRunner(steps=steps, stdin=BytesIO(b'dog\n'))
runner.run()
runner.cleanup()
def test_cat_reducer(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--reducer-cmd', 'cat -e', '--runner=local'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'script',
},
'reducer': {
'type': 'command',
'command': 'cat -e'
}
}])
r.run()
lines = list(r.stream_output())
self.assertEqual(sorted(lines), [b'x$\n', b'y$\n', b'z$\n'])
def test_cat_reducer(self):
data = b'x\ny\nz\n'
job = MRCmdJob(['--reducer-cmd', 'cat -e', '--runner=local'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(
r._get_steps(),
[{
'type': 'streaming',
'mapper': {
'type': 'script',
},
'reducer': {
'type': 'command',
'command': 'cat -e'}}])
r.run()
lines = list(to_lines(r.cat_output()))
self.assertEqual(sorted(lines), [b'x$\n', b'y$\n', b'z$\n'])
def test_multiple(self):
# put data in a .gz to force a single map task
x_gz_path = join(self.tmp_dir, 'data.gz')
with gzip.open(x_gz_path, 'wb') as x_gz:
x_gz.write(b'x\nx\nx\nx\nx\nx\n')
reducer_cmd = _bash_wrap('wc -l | tr -Cd "[:digit:]"')
job = MRCmdJob([
'--runner', 'local', '--mapper-cmd', 'cat -e', '--combiner-cmd',
'uniq', '--reducer-cmd', reducer_cmd, x_gz_path
])
job.sandbox()
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'command',
'command': 'cat -e'
},
'combiner': {
'type': 'command',
'command': 'uniq'
},
'reducer': {
'type': 'command',
'command': reducer_cmd
},
}])
r.run()
self.assertEqual(
sum(int(v) for _, v in job.parse_output(r.cat_output())), 1)
def test_uniq_combiner(self):
data = b'x\nx\nx\nx\nx\nx\n'
job = MRCmdJob(['--combiner-cmd=uniq', '--runner=local'])
job.sandbox(stdin=BytesIO(data))
with job.make_runner() as r:
self.assertEqual(r._get_steps(), [{
'type': 'streaming',
'mapper': {
'type': 'script',
},
'combiner': {
'type': 'command',
'command': 'uniq'
}
}])
r.run()
# there are 2 map tasks, each of which has 1 combiner, and all rows
# are the same, so we should end up with just 2 values
self.assertEqual(b''.join(r.stream_output()), b'x\nx\n')
def test_passthrough_options(self):
MRCmdJob(['--help'])
self.exit.assert_called_once_with(0)
output = self.stdout.getvalue()
self.assertIn('--reducer-cmd-2', output)
def test_no_command_steps(self):
job = MRCmdJob(['-r', 'inline', '--mapper-cmd', 'cat'])
job.sandbox()
self.assertRaises(NotImplementedError, job.make_runner)
def test_no_command_steps(self):
job = MRCmdJob(['-r', 'inline', '--mapper-cmd', 'cat'])
job.sandbox()
self.assertRaises(NotImplementedError, job.make_runner)
