def test_loading_boostrapped_mrjob_library(self):
# track the dir we're loading mrjob from rather than the full path
# to deal with edge cases where we load from the .py file,
# and the script loads from the .pyc compiled from that .py file.
our_mrjob_dir = os.path.dirname(os.path.realpath(mrjob.__file__))
with mrjob_conf_patcher():
mr_job = MRJobWhereAreYou(['-r', 'local'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()['bootstrap_mrjob'], True)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
runner.run()
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should load mrjob from its working dir
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertNotEqual(our_mrjob_dir, script_mrjob_dir)
assert script_mrjob_dir.startswith(local_tmp_dir)
def test_can_turn_off_bootstrap_mrjob(self):
with mrjob_conf_patcher({"runners": {"local": {"bootstrap_mrjob": False}}}):
mr_job = MRJobWhereAreYou(["-r", "local"])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()["bootstrap_mrjob"], False)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
try:
with no_handlers_for_logger():
runner.run()
except Exception as e:
# if mrjob is not installed, script won't be able to run
self.assertIn("ImportError", str(e))
return
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should not load mrjob from local_tmp_dir
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertFalse(script_mrjob_dir.startswith(local_tmp_dir))
def test_can_turn_off_bootstrap_mrjob(self):
# track the dir we're loading mrjob from rather than the full path
# to deal with edge cases where we load from the .py file,
# and the script loads from the .pyc compiled from that .py file.
our_mrjob_dir = os.path.dirname(os.path.realpath(mrjob.__file__))
self.mrjob_conf_path = os.path.join(self.tmp_dir, 'mrjob.conf')
dump_mrjob_conf({'runners': {'local': {'bootstrap_mrjob': False}}},
open(self.mrjob_conf_path, 'w'))
mr_job = MRJobWhereAreYou(['-c', self.mrjob_conf_path])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()['bootstrap_mrjob'], False)
runner.run()
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should load mrjob from the same place our test does
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertEqual(our_mrjob_dir, script_mrjob_dir)
def test_can_turn_off_bootstrap_mrjob(self):
with mrjob_conf_patcher(
{'runners': {'local': {'bootstrap_mrjob': False}}}):
mr_job = MRJobWhereAreYou(['-r', 'local'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner._opts['bootstrap_mrjob'], False)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
try:
runner.run()
except StepFailedException:
# this is what happens when mrjob isn't installed elsewhere
return
# however, if mrjob is installed, we need to verify that
# we're using the installed version and not a bootstrapped copy
output = list(mr_job.parse_output(runner.cat_output()))
self.assertEqual(len(output), 1)
# script should not load mrjob from local_tmp_dir
_, script_mrjob_dir = output[0]
self.assertFalse(script_mrjob_dir.startswith(local_tmp_dir))
def test_can_turn_off_bootstrap_mrjob(self):
with mrjob_conf_patcher(
{'runners': {
'local': {
'bootstrap_mrjob': False
}
}}):
mr_job = MRJobWhereAreYou(['-r', 'local'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()['bootstrap_mrjob'], False)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
try:
with no_handlers_for_logger():
runner.run()
except Exception as e:
# if mrjob is not installed, script won't be able to run
self.assertIn('ImportError', str(e))
return
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should not load mrjob from local_tmp_dir
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertFalse(script_mrjob_dir.startswith(local_tmp_dir))
def test_loading_boostrapped_mrjob_library(self):
# track the dir we're loading mrjob from rather than the full path
# to deal with edge cases where we load from the .py file,
# and the script loads from the .pyc compiled from that .py file.
our_mrjob_dir = os.path.dirname(os.path.realpath(mrjob.__file__))
with mrjob_conf_patcher():
mr_job = MRJobWhereAreYou(['-r', 'local', '--bootstrap-mrjob'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner._bootstrap_mrjob(), True)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
runner.run()
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should load mrjob from its working dir
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertNotEqual(our_mrjob_dir, script_mrjob_dir)
self.assertTrue(script_mrjob_dir.startswith(local_tmp_dir))
def test_can_turn_off_bootstrap_mrjob(self):
with mrjob_conf_patcher(
{'runners': {
'local': {
'bootstrap_mrjob': False
}
}}):
mr_job = MRJobWhereAreYou(['-r', 'local'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()['bootstrap_mrjob'], False)
local_tmp_dir = os.path.realpath(runner._get_local_tmp_dir())
try:
with no_handlers_for_logger():
runner.run()
except StepFailedException:
# this is what happens when mrjob isn't installed elsewhere
return
# however, if mrjob is installed, we need to verify that
# we're using the installed version and not a bootstrapped copy
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should not load mrjob from local_tmp_dir
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertFalse(script_mrjob_dir.startswith(local_tmp_dir))
def test_can_turn_off_bootstrap_mrjob(self):
# track the dir we're loading mrjob from rather than the full path
# to deal with edge cases where we load from the .py file,
# and the script loads from the .pyc compiled from that .py file.
our_mrjob_dir = os.path.dirname(os.path.realpath(mrjob.__file__))
self.mrjob_conf_path = os.path.join(self.tmp_dir, 'mrjob.conf')
dump_mrjob_conf({'runners': {
'local': {
'bootstrap_mrjob': False
}
}}, open(self.mrjob_conf_path, 'w'))
mr_job = MRJobWhereAreYou(['-c', self.mrjob_conf_path, '-r', 'local'])
mr_job.sandbox()
with mr_job.make_runner() as runner:
# sanity check
self.assertEqual(runner.get_opts()['bootstrap_mrjob'], False)
runner.run()
output = list(runner.stream_output())
self.assertEqual(len(output), 1)
# script should load mrjob from the same place our test does
_, script_mrjob_dir = mr_job.parse_output_line(output[0])
self.assertEqual(our_mrjob_dir, script_mrjob_dir)