def _pick_spark_tmp_dir(self):
if self._opts['spark_tmp_dir']:
return self.fs.join(self._opts['spark_tmp_dir'], self._job_key)
else:
master = self._spark_master() or 'local'
if master.startswith('local'): # including local-cluster
# need a local temp dir
# add "-spark" so we don't collide with default local temp dir
return os.path.join(gettempdir(), self._job_key + '-spark')
else:
# use HDFS (same default as HadoopJobRunner)
return posixpath.join(fully_qualify_hdfs_path('tmp/mrjob'),
self._job_key)
def _pick_spark_tmp_dir(self):
if self._opts['spark_tmp_dir']:
return self.fs.join(self._opts['spark_tmp_dir'], self._job_key)
else:
master = self._spark_master() or 'local'
if master.startswith('local'): # including local-cluster
# need a local temp dir
# add "-spark" so we don't collide with default local temp dir
return os.path.join(
gettempdir(), self._job_key + '-spark')
else:
# use HDFS (same default as HadoopJobRunner)
return posixpath.join(
fully_qualify_hdfs_path('tmp/mrjob'), self._job_key)
def _pick_spark_tmp_dir(self):
if self._opts['spark_tmp_dir']:
if is_uri(self._opts['spark_tmp_dir']):
return posixpath.join(
self._opts['spark_tmp_dir'], self._job_key)
else:
return os.path.join(
self._opts['spark_tmp_dir'], self._job_key)
elif self._spark_master_is_local():
# need a local temp dir
# add "-spark" so we don't collide with default local temp dir
return os.path.join(
gettempdir(), self._job_key + '-spark')
else:
# use HDFS (same default as HadoopJobRunner)
return posixpath.join(
fully_qualify_hdfs_path('tmp/mrjob'), self._job_key)
def test_hdfs_uri(self):
self.assertEqual(fully_qualify_hdfs_path('hdfs://host/path/'),
'hdfs://host/path/')
def test_absolute_path(self):
self.assertEqual(fully_qualify_hdfs_path('/path/to/cheese'),
'hdfs:///path/to/cheese')
def test_relative_path(self):
with patch('getpass.getuser') as getuser:
getuser.return_value = 'dave'
self.assertEqual(fully_qualify_hdfs_path('path/to/chocolate'),
'hdfs:///user/dave/path/to/chocolate')
def test_empty(self):
with patch('getpass.getuser') as getuser:
getuser.return_value = 'dave'
self.assertEqual(fully_qualify_hdfs_path(''), 'hdfs:///user/dave/')
def test_other_uri(self):
self.assertEqual(fully_qualify_hdfs_path('foo://bar/baz'),
'foo://bar/baz')
def test_s3n_uri(self):
self.assertEqual(fully_qualify_hdfs_path('s3n://bucket/oh/noes'),
's3n://bucket/oh/noes')
def test_absolute_path(self):
self.assertEqual(fully_qualify_hdfs_path('/path/to/cheese'),
'hdfs:///path/to/cheese')
def test_relative_path(self):
with patch('getpass.getuser') as getuser:
getuser.return_value = 'dave'
self.assertEqual(fully_qualify_hdfs_path('path/to/chocolate'),
'hdfs:///user/dave/path/to/chocolate')
def test_empty(self):
with patch('getpass.getuser') as getuser:
getuser.return_value = 'dave'
self.assertEqual(fully_qualify_hdfs_path(''), 'hdfs:///user/dave/')
def test_relative_path(self):
with patch("getpass.getuser") as getuser:
getuser.return_value = "dave"
self.assertEqual(fully_qualify_hdfs_path("path/to/chocolate"), "hdfs:///user/dave/path/to/chocolate")
def test_empty(self):
with patch("getpass.getuser") as getuser:
getuser.return_value = "dave"
self.assertEqual(fully_qualify_hdfs_path(""), "hdfs:///user/dave/")
def test_s3n_uri(self):
self.assertEqual(fully_qualify_hdfs_path('s3n://bucket/oh/noes'),
's3n://bucket/oh/noes')
def test_other_uri(self):
self.assertEqual(fully_qualify_hdfs_path('foo://bar/baz'),
'foo://bar/baz')
def test_hdfs_uri(self):
self.assertEqual(fully_qualify_hdfs_path('hdfs://host/path/'),
'hdfs://host/path/')