def test_parse_heka_message(data_dir, heka_format):
filename = "{}/test_telemetry_{}.heka".format(data_dir, heka_format)
reference_filename = filename + '.json'
# enable this to regenerate the expected json representation of the ping
if False:
with open(filename, "rb") as f:
if "gzip" in heka_format:
f = streaming_gzip_wrapper(f)
# deep copy the parsed message so lazy-parsed json gets vivified
msg = copy.deepcopy(message_parser.parse_heka_message(f).next())
open(reference_filename, 'w').write(json.dumps(msg, indent=4,
sort_keys=True))
reference = json.load(open(reference_filename))
with open(filename, "rb") as f:
if "gzip" in heka_format:
f = streaming_gzip_wrapper(f)
# deep copy the parsed message so lazy-parsed json gets vivified
msg = copy.deepcopy(message_parser.parse_heka_message(f).next())
assert msg == reference
def test_telemetry(data_dir):
filename = "{}/test_telemetry_gzip.heka".format(data_dir)
with open(filename, "rb") as o:
for r in message_parser.parse_heka_message(streaming_gzip_wrapper(o)):
assert set(r.keys()) == top_keys
assert set(r["payload"].keys()) == payload_keys
filename = "{}/test_telemetry_snappy.heka".format(data_dir)
with open(filename, "rb") as o:
for r in message_parser.parse_heka_message(o):
assert set(r.keys()) == top_keys
assert set(r["payload"].keys()) == payload_keys
def test_unpack(data_dir):
for t in ["plain", "snappy", "mixed", "gzip", "gzip_mixed"]:
filename = "{}/test_{}.heka".format(data_dir, t)
with open(filename, "rb") as o:
if "gzip" in t:
o = streaming_gzip_wrapper(o)
msg = 0
for r, b in message_parser.unpack(o, try_snappy=True):
j = json.loads(r.message.payload)
assert msg == j["seq"]
msg += 1
assert 10 == msg
def test_unpack_nosnappy(data_dir):
expected_counts = {"plain": 10, "snappy": 0, "mixed": 5,
"gzip": 10, "gzip_mixed": 5}
for t in expected_counts.keys():
count = 0
filename = "{}/test_{}.heka".format(data_dir, t)
with open(filename, "rb") as o:
if "gzip" in t:
o = streaming_gzip_wrapper(o)
try:
for r, b in message_parser.unpack(o, try_snappy=False):
count += 1
except:
pass
assert expected_counts[t] == count
def test_unpack(data_dir, heka_format, try_snappy, strict, expected_count,
expected_exception):
count = 0
threw_exception = False
filename = "{}/test_{}.heka".format(data_dir, heka_format)
with open(filename, "rb") as o:
if "gzip" in heka_format:
o = streaming_gzip_wrapper(o)
try:
for r, b in message_parser.unpack(o, try_snappy=try_snappy, strict=strict):
j = json.loads(r.message.payload)
assert count == j["seq"]
count += 1
except DecodeError:
threw_exception = True
assert count == expected_count
assert threw_exception == expected_exception
def test_lazy_parsing(data_dir, monkeypatch):
mock_parse_json = MagicMock(name='_parse_json',
wraps=message_parser._parse_json)
monkeypatch.setattr(message_parser, '_parse_json', mock_parse_json)
# this heka message has 20 json fields, only one of which should be parsed
# on first load
filename = "{}/test_telemetry_gzip.heka".format(data_dir)
with open(filename, "rb") as o:
heka_message = message_parser.parse_heka_message(streaming_gzip_wrapper(o)).next()
# should only have parsed json *once* to get the payload field (other
# json/dictionary fields of the message should be parsed lazily)
assert mock_parse_json.call_count == 1
# deep copying the heka message should cause the lazily evaluated fields
# to be evaluated
copy.deepcopy(heka_message)
assert mock_parse_json.call_count == 20 # 19 lazily instantiated fields + original call
def test_unpack_strict(data_dir):
expected_exceptions = {"plain": False, "snappy": True, "mixed": True,
"gzip": False, "gzip_mixed": True}
for t in expected_exceptions.keys():
count = 0
filename = "{}/test_{}.heka".format(data_dir, t)
threw = False
got_err = False
with open(filename, "rb") as o:
if "gzip" in t:
o = streaming_gzip_wrapper(o)
try:
for r, b in message_parser.unpack(o, strict=True, try_snappy=False):
if r.error is not None:
got_err = True
count += 1
except Exception as e:
threw = True
assert expected_exceptions[t] == threw
