class TestAnonymizer__input_callback(TestCaseMixin, unittest.TestCase):
def setUp(self):
self.event_type = 'bl-update'
self.event_data = {'some...': 'content...', 'id': 'some id...'}
self.routing_key = self.event_type + '.filtered.*.*'
self.body = json.dumps(self.event_data)
self.resource_to_org_ids = {}
self.mock = MagicMock(__class__=Anonymizer)
self.meth = MethodProxy(Anonymizer, self.mock, '_process_input')
self.mock._get_resource_to_org_ids.return_value = self.resource_to_org_ids
self.mock._get_result_dicts_and_output_body.return_value = (
sen.raw_result_dict,
sen.cleaned_result_dict,
sen.output_body,
)
self.force_exit_on_any_remaining_entered_contexts_mock = self.patch(
'n6.utils.anonymizer.force_exit_on_any_remaining_entered_contexts')
@foreach(
param(resource_to_org_ids_items={
'foo': [sen.o1, sen.o2],
}),
param(resource_to_org_ids_items={
'foo': [sen.o1, sen.o2],
'bar': [],
}),
param(resource_to_org_ids_items={
'foo': [],
'bar': [sen.o3, sen.o4, sen.o5],
}),
param(resource_to_org_ids_items={
'foo': [sen.o1, sen.o2],
'bar': [sen.o3, sen.o4, sen.o5],
}),
)
def test_with_some_org_ids(self, resource_to_org_ids_items):
self.resource_to_org_ids.update(resource_to_org_ids_items)
self.meth.input_callback(self.routing_key, self.body, sen.properties)
self.assertEqual(
self.force_exit_on_any_remaining_entered_contexts_mock.mock_calls,
[
call(self.mock.auth_api),
])
self.assertEqual(self.mock.mock_calls, [
call.setting_error_event_info(self.event_data),
call.setting_error_event_info().__enter__(),
call._check_event_type(self.event_type, self.event_data),
call.auth_api.__enter__(),
call._get_resource_to_org_ids(self.event_type, self.event_data),
call._get_result_dicts_and_output_body(
self.event_type, self.event_data, self.resource_to_org_ids),
call._publish_output_data(
self.event_type, self.resource_to_org_ids, sen.raw_result_dict,
sen.cleaned_result_dict, sen.output_body),
call.auth_api.__exit__(None, None, None),
call.setting_error_event_info().__exit__(None, None, None),
])
@foreach(
param(resource_to_org_ids_items={}),
param(resource_to_org_ids_items={
'foo': [],
}),
param(resource_to_org_ids_items={
'foo': [],
'bar': [],
}),
)
def test_without_org_ids(self, resource_to_org_ids_items):
self.resource_to_org_ids.update(resource_to_org_ids_items)
self.meth.input_callback(self.routing_key, self.body, sen.properties)
self.assertEqual(
self.force_exit_on_any_remaining_entered_contexts_mock.mock_calls,
[
call(self.mock.auth_api),
])
self.assertEqual(self.mock.mock_calls, [
call.setting_error_event_info(self.event_data),
call.setting_error_event_info().__enter__(),
call._check_event_type(self.event_type, self.event_data),
call.auth_api.__enter__(),
call._get_resource_to_org_ids(self.event_type, self.event_data),
call.auth_api.__exit__(None, None, None),
call.setting_error_event_info().__exit__(None, None, None),
])
def test_with_some_error(self):
self.resource_to_org_ids.update({
'foo': [sen.o1, sen.o2],
'bar': [sen.o3, sen.o4, sen.o5],
})
exc_type = ZeroDivisionError # (just an example exception class)
self.mock._get_result_dicts_and_output_body.side_effect = exc_type
with self.assertRaises(exc_type) as exc_context:
self.meth.input_callback(self.routing_key, self.body,
sen.properties)
self.assertEqual(
self.force_exit_on_any_remaining_entered_contexts_mock.mock_calls,
[
call(self.mock.auth_api),
])
self.assertEqual(self.mock.mock_calls, [
call.setting_error_event_info(self.event_data),
call.setting_error_event_info().__enter__(),
call._check_event_type(self.event_type, self.event_data),
call.auth_api.__enter__(),
call._get_resource_to_org_ids(self.event_type, self.event_data),
call._get_result_dicts_and_output_body(
self.event_type, self.event_data, self.resource_to_org_ids),
call.auth_api.__exit__(exc_type, exc_context.exception, ANY),
call.setting_error_event_info().__exit__(
exc_type, exc_context.exception, ANY),
])
class TestBaseParser(unittest.TestCase):
def setUp(self):
self.mock = Mock(__class__=BaseParser, allow_empty_results=False)
self.meth = MethodProxy(BaseParser, self.mock)
def _asserts_of_proper__new__instance_adjustment(self, instance):
# BaseQueued.__new__() ensures that
self.assertIsNot(instance.input_queue, BaseParser.input_queue)
def _asserts_of_proper_preinit_hook_instance_adjustment(
self, instance, binding_key):
# for classes with `default_binding_key`
# BaseParser.preinit_hook() ensures that
self.assertEqual(
instance.input_queue, {
'exchange': 'raw',
'exchange_type': 'topic',
'queue_name': binding_key,
'binding_keys': [binding_key],
})
self.assertEqual(BaseParser.input_queue, {
'exchange': 'raw',
'exchange_type': 'topic',
})
def _basic_init_related_asserts(self, instance, subclass, super_mock,
super_cls_mock, expected_config,
expected_config_full):
# assert that an instance of the proper type has been returned
self.assertIsInstance(instance, subclass)
# assert that super used properly
super_mock.assert_called_once_with(BaseParser, instance)
super_cls_mock.__init__.assert_called_once_with(a=sentinel.a,
bb=sentinel.bb)
# assert that configuration stuff has been obtained properly
self.assertEqual(instance.config, expected_config)
self.assertIsInstance(instance.config, ConfigSection)
self.assertEqual(instance.config_full, expected_config_full)
self.assertIsInstance(instance.config_full, Config)
def test_basics(self):
self.assertTrue(issubclass(BaseParser, QueuedBase))
self.assertTrue(hasattr(BaseParser, 'default_binding_key'))
self.assertTrue(hasattr(BaseParser, 'config_spec_pattern'))
self.assertTrue(hasattr(BaseParser, 'constant_items'))
self.assertTrue(hasattr(BaseParser, 'record_dict_class'))
self.assertTrue(hasattr(BaseParser, 'event_type'))
def test_config_spec_pattern(self):
config_spec = BaseParser.config_spec_pattern.format(
parser_class_name='example_foo')
config_spec_parsed = parse_config_spec(config_spec)
prefetch_count_opt_spec = config_spec_parsed.get_opt_spec(
'example_foo.prefetch_count')
self.assertEqual(prefetch_count_opt_spec.name, 'prefetch_count')
self.assertEqual(prefetch_count_opt_spec.converter_spec, 'int')
def test_initialization_without_default_binding_key(self):
class SomeParser(BaseParser):
pass # no `default_binding_key` defined => it's an abstract class
with self.assertRaises(NotImplementedError):
SomeParser()
unready_instance = SomeParser.__new__(SomeParser)
self._asserts_of_proper__new__instance_adjustment(unready_instance)
# for classes without `default_binding_key`
# `queue_name` and `binding_keys` items are *not* added...
self.assertEqual(unready_instance.input_queue, BaseParser.input_queue)
self.assertEqual(BaseParser.input_queue, {
'exchange': 'raw',
'exchange_type': 'topic',
})
@foreach(
param(
mocked_conf_from_files={},
expected_config=ConfigSection('SomeParser', {'prefetch_count': 1}),
expected_config_full=Config.make(
{'SomeParser': {
'prefetch_count': 1
}}),
),
param(
mocked_conf_from_files={
'SomeParser': {
'prefetch_count': '42'
},
'another_section': {
'another_opt': '123.456'
},
},
expected_config=ConfigSection('SomeParser',
{'prefetch_count': 42}),
expected_config_full=Config.make(
{'SomeParser': {
'prefetch_count': 42
}}),
),
param(
custom_config_spec_pattern=concat_reducing_indent(
BaseParser.config_spec_pattern,
'''
some_opt = [-3, null] :: json
[another_section]
another_opt :: float
yet_another_opt = Foo Bar Spam Ham
''',
),
mocked_conf_from_files={
'SomeParser': {
'prefetch_count': '42'
},
'another_section': {
'another_opt': '123.456'
},
},
expected_config=ConfigSection('SomeParser', {
'prefetch_count': 42,
'some_opt': [-3, None],
}),
expected_config_full=Config.make({
'SomeParser': {
'prefetch_count': 42,
'some_opt': [-3, None],
},
'another_section': {
'another_opt': 123.456,
'yet_another_opt': 'Foo Bar Spam Ham',
},
}),
),
)
@foreach(
param(binding_key='foo.bar'),
param(binding_key='foo.bar.33'),
)
def test_initialization_with_default_binding_key(
self,
binding_key,
mocked_conf_from_files,
expected_config,
expected_config_full,
custom_config_spec_pattern=None):
class SomeParser(BaseParser):
default_binding_key = binding_key # => it's a concrete class
if custom_config_spec_pattern is not None:
SomeParser.config_spec_pattern = custom_config_spec_pattern
unready_instance = SomeParser.__new__(SomeParser)
self._asserts_of_proper__new__instance_adjustment(unready_instance)
self._asserts_of_proper_preinit_hook_instance_adjustment(
unready_instance, binding_key)
super_cls_mock = SimpleNamespace(__init__=Mock())
with patch_always('n6.parsers.generic.super',
return_value=super_cls_mock) as super_mock, \
patch('n6.parsers.generic.Config._load_n6_config_files',
return_value=mocked_conf_from_files):
# instantiation
instance = SomeParser(a=sentinel.a, bb=sentinel.bb)
self._asserts_of_proper__new__instance_adjustment(instance)
self._asserts_of_proper_preinit_hook_instance_adjustment(
instance, binding_key)
self._basic_init_related_asserts(instance, SomeParser, super_mock,
super_cls_mock, expected_config,
expected_config_full)
def test__make_binding_keys(self):
self.mock.default_binding_key = 'fooo.barr'
binding_keys = self.meth.make_binding_keys()
self.assertEqual(binding_keys, ['fooo.barr'])
self.assertEqual(self.mock.mock_calls, [])
def test__make_binding_keys_with_raw_format_version_tag(self):
self.mock.default_binding_key = 'fooo.barr.33'
binding_keys = self.meth.make_binding_keys()
self.assertEqual(binding_keys, ['fooo.barr.33'])
self.assertEqual(self.mock.mock_calls, [])
def test__get_script_init_kwargs(self):
self.assertIsInstance(
vars(BaseParser)['get_script_init_kwargs'], classmethod)
init_kwargs = BaseParser.get_script_init_kwargs.__func__(self.mock)
self.assertEqual(init_kwargs, {})
self.assertEqual(self.mock.mock_calls, [])
def test__run_handling__interrupted(self):
self.mock.configure_mock(**{'run.side_effect': KeyboardInterrupt})
self.meth.run_handling()
self.mock.run.assert_called_once_with()
self.mock.stop.assert_called_once_with()
def test__run_handling__not_interrupted(self):
self.meth.run_handling()
self.mock.run.assert_called_once_with()
self.assertEqual(self.mock.stop.mock_calls, [])
@patch('n6.parsers.generic.FilePagedSequence')
def test__input_callback(self, FilePagedSequence_mock):
FilePagedSequence_mock.return_value = MagicMock()
FilePagedSequence_mock.return_value.__enter__.return_value = sentinel.working_seq
data = MagicMock(**{'get.return_value': sentinel.rid})
self.mock.configure_mock(
**{
'_fix_body.return_value':
sentinel.body,
'prepare_data.return_value':
data,
'setting_error_event_info':
MagicMock(),
'get_output_rk.return_value':
sentinel.output_rk,
'get_output_bodies.return_value':
[sentinel.output_body1, sentinel.output_body2],
})
self.meth.input_callback(sentinel.routing_key, sentinel.body,
sentinel.properties)
self.assertEqual(self.mock.mock_calls, [
call._fix_body(sentinel.body),
call.prepare_data(sentinel.routing_key, sentinel.body,
sentinel.properties),
call.prepare_data().get('properties.message_id'),
call.setting_error_event_info(sentinel.rid),
call.setting_error_event_info().__enter__(),
call.get_output_rk(data),
call.get_output_bodies(data, sentinel.working_seq),
call.publish_output(routing_key=sentinel.output_rk,
body=sentinel.output_body1),
call.publish_output(routing_key=sentinel.output_rk,
body=sentinel.output_body2),
call.setting_error_event_info().__exit__(None, None, None),
])
self.assertEqual(FilePagedSequence_mock.mock_calls, [
call(page_size=1000),
call().__enter__(),
call().__exit__(None, None, None),
])
def test__prepare_data(self):
data = self.meth.prepare_data(routing_key='ham.spam',
body=sentinel.body,
properties=SimpleNamespace(
foo=sentinel.foo,
bar=sentinel.bar,
timestamp=1389348840,
headers={'a': sentinel.a}))
self.assertEqual(
data, {
'a': sentinel.a,
'properties.foo': sentinel.foo,
'properties.bar': sentinel.bar,
'source': 'ham.spam',
'properties.timestamp': '2014-01-10 10:14:00',
'raw_format_version_tag': None,
'raw': sentinel.body,
})
def test__prepare_data__rk__with_raw_format_version_tag(self):
data = self.meth.prepare_data(routing_key='ham.spam.33',
body=sentinel.body,
properties=SimpleNamespace(
foo=sentinel.foo,
bar=sentinel.bar,
timestamp=1389348840,
headers={'a': sentinel.a}))
self.assertEqual(
data, {
'a': sentinel.a,
'properties.foo': sentinel.foo,
'properties.bar': sentinel.bar,
'source': 'ham.spam',
'properties.timestamp': '2014-01-10 10:14:00',
'raw_format_version_tag': '33',
'raw': sentinel.body,
})
def test__get_output_rk(self):
self.mock.configure_mock(**{
'event_type': 'foobar',
})
data = {'source': 'ham.spam'}
output_rk = self.meth.get_output_rk(data)
self.assertEqual(output_rk, 'foobar.parsed.ham.spam')
def test__get_output_bodies(self):
parsed = [
MagicMock(
**{
'__class__':
RecordDict,
'used_as_context_manager':
True,
'get_ready_json.return_value':
getattr(sentinel, 'output_body{}'.format(i))
}) for i in (1, 2)
]
self.mock.configure_mock(
**{
'parse.return_value':
parsed,
'get_output_message_id.side_effect': [
sentinel.msg_A,
sentinel.msg_B,
],
'setting_error_event_info':
MagicMock(),
'postprocess_parsed.side_effect': (
lambda data, parsed, total, item_no: parsed),
})
seq_mock = FilePagedSequence._instance_mock()
output_bodies = self.meth.get_output_bodies(sentinel.data, seq_mock)
self.assertIs(output_bodies, seq_mock)
self.assertEqual(seq_mock._list, [
sentinel.output_body1,
sentinel.output_body2,
])
self.assertEqual(parsed[0].mock_calls, [
call.__setitem__('id', sentinel.msg_A),
call.get_ready_json(),
])
self.assertEqual(parsed[1].mock_calls, [
call.__setitem__('id', sentinel.msg_B),
call.get_ready_json(),
])
self.assertEqual(self.mock.mock_calls, [
call.parse(sentinel.data),
call.get_output_message_id(parsed[0]),
call.delete_too_long_address(parsed[0]),
call.get_output_message_id(parsed[1]),
call.delete_too_long_address(parsed[1]),
call.setting_error_event_info(parsed[0]),
call.setting_error_event_info().__enter__(),
call.postprocess_parsed(sentinel.data, parsed[0], 2, item_no=1),
call.setting_error_event_info().__exit__(None, None, None),
call.setting_error_event_info(parsed[1]),
call.setting_error_event_info().__enter__(),
call.postprocess_parsed(sentinel.data, parsed[1], 2, item_no=2),
call.setting_error_event_info().__exit__(None, None, None),
])
def test__get_output_bodies__record_dict_not_used_as_context_manager(self):
parsed = [
MagicMock(**{
'__class__': RecordDict,
'used_as_context_manager': False
}) for i in (1, 2)
]
self.mock.configure_mock(**{'parse.return_value': parsed})
with self.assertRaises(AssertionError):
self.meth.get_output_bodies(sentinel.data,
FilePagedSequence._instance_mock())
self.assertEqual(self.mock.method_calls, [
call.parse(sentinel.data),
])
def test__get_output_bodies__parse_yielded_no_items(self):
self.mock.configure_mock(**{'parse.return_value': iter([])})
with self.assertRaises(ValueError):
self.meth.get_output_bodies(sentinel.data,
FilePagedSequence._instance_mock())
self.assertEqual(self.mock.method_calls, [
call.parse(sentinel.data),
])
def test__get_output_bodies__parse_yielded_no_items__allow_empty_results(
self):
self.mock.configure_mock(**{
'parse.return_value': iter([]),
'allow_empty_results': True
})
seq_mock = FilePagedSequence._instance_mock()
output_bodies = self.meth.get_output_bodies(sentinel.data, seq_mock)
self.assertIs(output_bodies, seq_mock)
self.assertEqual(seq_mock._list, []) # just empty
self.assertEqual(self.mock.mock_calls, [
call.parse(sentinel.data),
])
def test__delete_too_long_address__address_is_ok(self):
parsed = RecordDict()
parsed['address'] = [{'ip': i + 1} for i in xrange(MAX_IPS_IN_ADDRESS)]
expected = RecordDict()
expected['address'] = [{
'ip': i + 1
} for i in xrange(MAX_IPS_IN_ADDRESS)]
self.meth.delete_too_long_address(parsed)
self.assertEqual(parsed, expected)
def test__delete_too_long_address__address_is_too_long(self):
ips = MAX_IPS_IN_ADDRESS + 1
parsed = RecordDict()
parsed['id'] = '0123456789abcdef0123456789abcdef'
parsed['address'] = [{'ip': i + 1} for i in xrange(ips)]
expected = RecordDict()
expected['id'] = '0123456789abcdef0123456789abcdef'
self.meth.delete_too_long_address(parsed)
self.assertEqual(parsed, expected)
def test__delete_too_long_address__address_is_empty(self):
parsed = RecordDict()
parsed.update({'source': 'foo.bar'})
expected = RecordDict()
expected.update({'source': 'foo.bar'})
self.meth.delete_too_long_address(parsed)
self.assertEqual(parsed, expected)
def test__get_output_message_id(self):
inputs_and_resultant_hash_bases = [
# basics
({
'source': 'foo.bar'
}, 'source,foo.bar'),
({
u'source': u'foo.bar'
}, 'source,foo.bar'),
# proper sorting of multiple values
({
'key1': 2,
'key2': ['value2', 'value3', 'value1']
}, 'key1,2\nkey2,value1,value2,value3'),
# ...and of keys + proper encoding of unicode keys/values
({
u'key2': [u'value3', u'value1', u'value2'],
u'key1': 2L
}, 'key1,2\nkey2,value1,value2,value3'),
# ...as well as proper int/long normalization/representation
({
u'key2': [30, 10, 20L],
u'key1': 9000111222333444555666777888999000L
}, 'key1,9000111222333444555666777888999000\nkey2,10,20,30'),
# non-ascii values
({
'target': 'zażółć',
u'client': [u'jaźń', u'gęślą']
}, 'client,gęślą,jaźń\ntarget,zażółć'),
({
u'target': u'zażółć',
'client': ['jaźń', 'gęślą']
}, 'client,gęślą,jaźń\ntarget,zażółć'),
# subdicts
({
'dip': u'3.3.3.3',
u'address': [{
'ip': '255.255.255.0'
}, {
'ip': '127.0.0.1'
}]
},
"address,{'ip': '127.0.0.1'},{'ip': '255.255.255.0'}\ndip,3.3.3.3"
),
# non-ascii subdict keys/values
({
u'key2': [{
'ką2': 'vą2'
}, {
'ką1': 'vą1'
}],
'key1': {
'ką': 'vą'
}
}, "key1,{'k\\xc4\\x85': 'v\\xc4\\x85'}\n" +
"key2,{'k\\xc4\\x851': 'v\\xc4\\x851'},{'k\\xc4\\x852': 'v\\xc4\\x852'}"
),
# proper encoding of unicode keys/values + proper sorting of whole subdicts
({
'key1': {
u'ką': u'vą'
},
'key2': [{
u'ką2': 'vą2'
}, {
'ką1': u'vą1'
}]
}, "key1,{'k\\xc4\\x85': 'v\\xc4\\x85'}\n" +
"key2,{'k\\xc4\\x851': 'v\\xc4\\x851'},{'k\\xc4\\x852': 'v\\xc4\\x852'}"
),
# ...as well as proper int/long normalization/representation
({
'key1': {
u'k': 2L
},
'key2': [{
'k2': 2L
}, {
u'k1': 1
}]
}, "key1,{'k': 2}\nkey2,{'k1': 1},{'k2': 2}"),
({
u'key2': [{
'k2': 2
}, {
'k1': 1
}],
'key1': {
'k': 3
}
}, "key1,{'k': 3}\nkey2,{'k1': 1},{'k2': 2}"),
({
u'key2': [{
'k2': 2L
}, {
'k1': 1L
}],
'key1': {
'k': 9000111222333444555666777888999000L
}
},
"key1,{'k': 9000111222333444555666777888999000}\nkey2,{'k1': 1},{'k2': 2}"
),
# proper sorting of multiple items in subdicts
({
'key1': {
'c': 2,
u'a': 3L,
u'b': 1L
},
'key2': [{
'c': 2,
u'a': 3L,
u'b': 1L
}, {
'd': 3,
u'a': 2L,
u'b': 1L
}]
}, "key1,{'a': 3, 'b': 1, 'c': 2}\n" +
"key2,{'a': 2, 'b': 1, 'd': 3},{'a': 3, 'b': 1, 'c': 2}"),
]
class _RecordDict(RecordDict):
adjust_key1 = adjust_key2 = None
optional_keys = RecordDict.optional_keys | {'key1', 'key2'}
parser = BaseParser.__new__(BaseParser)
for input_dict, expected_base in inputs_and_resultant_hash_bases:
record_dict = _RecordDict(input_dict)
expected_result = hashlib.md5(expected_base).hexdigest()
result = parser.get_output_message_id(record_dict)
self.assertIsInstance(result, str)
self.assertEqual(result, expected_result)
def test__get_output_message_id__errors(self):
inputs_and_exc_classes = [
# bad subdict key type
(
{
'key1': {
32: 2
}
},
TypeError,
),
(
{
'key1': [{
32: 2
}]
},
TypeError,
),
# bad subdict value type
(
{
'key1': {
'k': 2.3
}
},
TypeError,
),
(
{
'key1': [{
'k': 2.3
}]
},
TypeError,
),
(
{
'key1': {
'k': {
'k': 2
}
}
}, # nesting is illegal
TypeError,
),
(
{
'key1': [{
'k': {
'k': 2
}
}]
}, # nesting is illegal
TypeError,
),
# bad value type
(
{
'key1': 2.3
},
TypeError,
),
(
{
'key1': [2.3]
},
TypeError,
),
(
{
'key1': [[2]]
}, # nesting is illegal
TypeError,
),
]
class _RecordDict(RecordDict):
adjust_key1 = adjust_key2 = None
optional_keys = RecordDict.optional_keys | {'key1', 'key2'}
parser = BaseParser.__new__(BaseParser)
for input_dict, exc_class in inputs_and_exc_classes:
record_dict = _RecordDict(input_dict)
with self.assertRaises(exc_class):
parser.get_output_message_id(record_dict)
def test__postprocess_parsed__without__do_not_resolve_fqdn_to_ip(self):
data = {}
parsed = RecordDict()
self.meth.postprocess_parsed(data, parsed, 1, item_no=1)
self.assertEqual(parsed, {})
def test__postprocess_parsed__with__do_not_resolve_fqdn_to_ip__False(self):
data = {'_do_not_resolve_fqdn_to_ip': False}
parsed = RecordDict()
self.meth.postprocess_parsed(data, parsed, 1, item_no=1)
self.assertEqual(parsed, {})
def test__postprocess_parsed__with__do_not_resolve_fqdn_to_ip__True(self):
data = {'_do_not_resolve_fqdn_to_ip': True}
parsed = RecordDict()
self.meth.postprocess_parsed(data, parsed, 1, item_no=1)
self.assertEqual(parsed, {'_do_not_resolve_fqdn_to_ip': True})
