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 = CmdJob([
'--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_2(self):
data = 'x\ny\nz\n'
job = CmdJob(['--mapper-cmd=cat', '--reducer-cmd-2', 'wc -l',
'--runner=local', '--no-conf'])
job.sandbox(stdin=StringIO(data))
with job.make_runner() as r:
r.run()
self.assertEqual(sum(int(l) for l in r.stream_output()), 3)
def test_cat_mapper(self):
data = b'x\ny\nz\n'
job = CmdJob(['--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 = 'x\ny\nz\n'
job = CmdJob(['--mapper-cmd=cat', '--runner=local'])
job.sandbox(stdin=StringIO(data))
with job.make_runner() as r:
self.assertEqual(
r._get_steps(),
[{
'type': 'streaming',
'mapper': {
'type': 'command',
'command': 'cat'}}])
r.run()
self.assertEqual(''.join(r.stream_output()), data)
def test_cat_reducer(self):
data = b'x\ny\nz\n'
job = CmdJob(['--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_uniq_combiner(self):
data = b'x\nx\nx\nx\nx\nx\n'
job = CmdJob(['--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):
cmd_job = CmdJob(['--help'])
output = self.stdout.getvalue()
self.assertIn('--reducer-cmd-2', output)
def test_passthrough_options(self):
CmdJob(['--help'])
self.exit.assert_called_once_with(0)
output = self.stdout.getvalue()
self.assertIn('--reducer-cmd-2', output)
