def setUp(self):
"""
Test set up:
* Build, if necessary, and start a CouchDB container and
connect as a BiscuitTin instance
* Start the HTTP API service on a free port, with the necessary
dependencies injected
* Create an HTTP client connection to the API service
"""
self.couchdb_container = CouchDBContainer()
# Configuration for Cookie Jar
self.HOST = self.couchdb_container.couchdb_fqdn
self.DB = 'elmo-test'
self.jar = BiscuitTin(self.HOST, self.DB, 1, timedelta(0))
# Configuration for HTTP service
self.API_PORT = get_open_port()
self.api = HTTP_API()
self.api.inject(APIDependency.CookieJar, self.jar)
self.api.inject(APIDependency.System, None)
self.api.listen(self.API_PORT)
self.http = HTTPConnection('localhost', self.API_PORT)
self.REQ_HEADER = {'Accept': 'application/json'}
# Block until service is up (or timeout)
start_time = finish_time = datetime.now()
service_up = False
while finish_time - start_time < timedelta(seconds=5):
response = None
try:
self.http.request('HEAD', '/')
response = self.http.getresponse()
except:
sleep(0.1)
finally:
self.http.close()
finish_time = datetime.now()
if isinstance(response, HTTPResponse):
service_up = True
break
if not service_up:
self.tearDown()
raise ConnectionError('Couldn\'t start API service in a reasonable amount of time')
def run(config_location):
# Load config
config = load_config(os.path.join(config_location, "setup.conf"))
# Setup measurement logging
logging_buffer_latency = timedelta(seconds=config.influxdb.buffer_latency)
influxdb_config = InfluxDBConnectionConfig(config.influxdb.host, config.influxdb.port, config.influxdb.username,
config.influxdb.password, config.influxdb.database)
logger = InfluxDBLogger(influxdb_config, buffer_latency=logging_buffer_latency)
# Set HTTP(S) connection pool size (for CouchDB)
# NOTE This is taken from an environment variable, as it's not
# something that would probably need tweaking that much:
pool_size = int(os.environ.get('COOKIEMONSTER_POOL_SIZE', 16))
patch_connection_pools(maxsize=pool_size)
# Setup cookie jar
cookie_jar = RateLimitedBiscuitTin(config.cookie_jar.max_requests_per_second,
config.cookie_jar.url,
config.cookie_jar.database,
config.cookie_jar.buffer_capacity,
config.cookie_jar.buffer_latency,
verify=config.cookie_jar.cacert)
add_cookie_jar_logging(cookie_jar, logger)
add_couchdb_logging(cookie_jar, logger)
# Setup data retrieval manager
update_mapper = BatonUpdateMapper(config.baton.binaries_location, zone=config.baton.zone)
retrieval_manager = PeriodicRetrievalManager(config.retrieval.period, update_mapper, logger)
# # Setup basic Slack client
# slack = BasicSlackClient(config.slack.token, config.slack.default_channel, config.slack.default_username)
slack = None
# Setup rule output log file writer
rule_log_writer = RuleOutputWriter(config.output.log_file)
# # Setup basic message queue (e.g. RabbitMQ) client
# message_queue = BasicMessageQueue(config.message_queue.host, config.message_queue.port,
# config.message_queue.username, config.message_queue.password)
message_queue = None
# Define the context that rules and enrichment loaders has access to
context = HgiContext(cookie_jar, config, rule_log_writer, slack, message_queue)
# Setup rules source
rules_source = RuleSource(config.processing.rules_location, context)
rules_source.start()
# Setup enrichment loader source
enrichment_loader_source = EnrichmentLoaderSource(config.processing.enrichment_loaders_location, context)
enrichment_loader_source.start()
# Setup the data processor manager
processor_manager = BasicProcessorManager(cookie_jar, rules_source, enrichment_loader_source,
config.processing.max_threads, logger)
# Connect components to the cookie jar
_connect_retrieval_manager_to_cookie_jar(retrieval_manager, cookie_jar, config.cookie_jar.max_requests_per_second,
logger)
_connect_retrieval_manager_to_since_file(retrieval_manager, config_location)
_connect_processor_manager_to_cookie_jar(processor_manager, cookie_jar)
# Setup the HTTP API
api = HTTP_API()
api.inject(APIDependency.CookieJar, cookie_jar)
api.inject(APIDependency.System, None)
api.listen(config.api.port)
# Start the retrieval manager from the last known successful
# retrieval time (or invocation time, otherwise)
try:
with open(os.path.join(config_location, "since"), "r") as f:
since_time = datetime.fromtimestamp(int(f.read()))
except:
since_time = datetime.now()
retrieval_manager.start(since_time)
# Start processing of any unprocessed cookies
processor_manager.process_any_cookies()
# Setup monitors
ThreadsMonitor(logger, logging_buffer_latency).start()
CookieJarMonitor(logger, logging_buffer_latency, cookie_jar).start()
class TestElmo(unittest.TestCase):
def setUp(self):
"""
Test set up:
* Build, if necessary, and start a CouchDB container and
connect as a BiscuitTin instance
* Start the HTTP API service on a free port, with the necessary
dependencies injected
* Create an HTTP client connection to the API service
"""
self.couchdb_container = CouchDBContainer()
# Configuration for Cookie Jar
self.HOST = self.couchdb_container.couchdb_fqdn
self.DB = 'elmo-test'
self.jar = BiscuitTin(self.HOST, self.DB, 1, timedelta(0))
# Configuration for HTTP service
self.API_PORT = get_open_port()
self.api = HTTP_API()
self.api.inject(APIDependency.CookieJar, self.jar)
self.api.inject(APIDependency.System, None)
self.api.listen(self.API_PORT)
self.http = HTTPConnection('localhost', self.API_PORT)
self.REQ_HEADER = {'Accept': 'application/json'}
# Block until service is up (or timeout)
start_time = finish_time = datetime.now()
service_up = False
while finish_time - start_time < timedelta(seconds=5):
response = None
try:
self.http.request('HEAD', '/')
response = self.http.getresponse()
except:
sleep(0.1)
finally:
self.http.close()
finish_time = datetime.now()
if isinstance(response, HTTPResponse):
service_up = True
break
if not service_up:
self.tearDown()
raise ConnectionError('Couldn\'t start API service in a reasonable amount of time')
def tearDown(self):
""" Tear down test set up """
self.http.close()
self.api.stop()
self.couchdb_container.tear_down()
def test_queue(self):
"""
HTTP API: GET /queue
"""
self.http.request('GET', '/queue', headers=self.REQ_HEADER)
r = self.http.getresponse()
self.assertEqual(r.status, 200)
self.assertEqual(r.headers.get_content_type(), 'application/json')
data = _decode_json_response(r)
self.assertIn('queue_length', data)
self.assertEqual(data['queue_length'], self.jar.queue_length()) # Should be 0
self.http.close()
# Add item to the queue
self.jar.mark_for_processing('/foo')
self.http.request('GET', '/queue', headers=self.REQ_HEADER)
data = _decode_json_response(self.http.getresponse())
self.assertEqual(data['queue_length'], self.jar.queue_length()) # Should be 1
def test_reprocess(self):
"""
HTTP API: POST /queue/reprocess
"""
# Add mocked update notifier to Cookie Jar
dirty_cookie_listener = MagicMock()
self.jar.add_listener(dirty_cookie_listener)
cookie_identifier = '/foo'
request = {'identifier': cookie_identifier}
self.http.request('POST', '/queue/reprocess', body=json.dumps(request), headers=self.REQ_HEADER)
r = self.http.getresponse()
self.assertEqual(r.status, 200)
self.assertEqual(r.headers.get_content_type(), 'application/json')
data = _decode_json_response(r)
self.assertEqual(data, request)
self.http.close()
# Check queue has been updated
self.assertEqual(self.jar.queue_length(), 1)
self.assertEqual(dirty_cookie_listener.call_count, 1)
@staticmethod
def _url_for_identifier(identifier:str):
""" URL for identifier """
if identifier[0] == "/":
return '/cookiejar?identifier={}'.format(identifier)
else:
return '/cookiejar/{}'.format(identifier)
def _fetch_test(self, identifier:str):
""" Generic fetch test """
source = 'foobar'
timestamp = datetime.now().replace(microsecond=0, tzinfo=timezone.utc)
metadata = Metadata({'foo': 123, 'bar': 'quux'})
enrichment = Enrichment(source, timestamp, metadata)
self.jar.enrich_cookie(identifier, enrichment)
self.http.request('GET', TestElmo._url_for_identifier(identifier), headers=self.REQ_HEADER)
r = self.http.getresponse()
self.assertEqual(r.status, 200)
self.assertEqual(r.headers.get_content_type(), 'application/json')
data = _decode_json_response(r)
fetched_identifier = data['identifier']
fetched_enrichment = json.loads(json.dumps(data['enrichments']), cls=EnrichmentJSONDecoder)[0]
self.assertEqual(fetched_identifier, identifier)
self.assertEqual(fetched_enrichment, enrichment)
def test_fetch_by_qs(self):
"""
HTTP API: GET /cookiejar?identifier=<identifier>
"""
self._fetch_test('/path/to/foo')
def test_fetch_by_route(self):
"""
HTTP API: GET /cookiejar/<identifier>
"""
self._fetch_test('foo_bar')
def _delete_test(self, identifier:str):
""" Generic delete test """
self.jar.mark_for_processing(identifier)
self.jar.mark_as_complete(identifier)
cookie = self.jar.fetch_cookie(identifier)
self.assertIsInstance(cookie, Cookie)
self.http.request('DELETE', TestElmo._url_for_identifier(identifier), headers=self.REQ_HEADER)
r = self.http.getresponse()
self.assertEqual(r.status, 200)
self.assertEqual(r.headers.get_content_type(), 'application/json')
data = _decode_json_response(r)
self.assertEqual(data, {'deleted':identifier})
deleted_cookie = self.jar.fetch_cookie(identifier)
self.assertIsNone(deleted_cookie)
def test_delete_by_qs(self):
"""
HTTP API: DELETE /cookiejar?identifier=<identifier>
"""
self._delete_test('/path/to/foo')
def test_delete_by_route(self):
"""
HTTP API: DELETE /cookiejar/<identifier>
"""
self._delete_test('foo_bar')
def test_thread_dump(self):
"""
HTTP API: GET /debug/threads
Note: This test only proves that the endpoint returns an OK
response and JSON data.
TODO At least validate the returned data's schema
"""
self.http.request('GET', '/debug/threads', headers=self.REQ_HEADER)
r = self.http.getresponse()
self.assertEqual(r.status, 200)
self.assertEqual(r.headers.get_content_type(), 'application/json')
def run(config_location):
# Load config
config = load_config(config_location)
# Setup measurement logging
logging_buffer_latency = timedelta(seconds=config.influxdb.buffer_latency)
influxdb_config = InfluxDBConnectionConfig(config.influxdb.host, config.influxdb.port, config.influxdb.username,
config.influxdb.password, config.influxdb.database)
logger = InfluxDBLogger(influxdb_config, buffer_latency=logging_buffer_latency)
# Set HTTP connection pool size (for CouchDB)
# NOTE This is taken from an environment variable, as it's not
# something that would probably need tweaking that much:
pool_size = int(environ.get('COOKIEMONSTER_HTTP_POOL_SIZE', 16))
patch_http_connection_pool(maxsize=pool_size)
# Setup cookie jar
cookie_jar = RateLimitedBiscuitTin(config.cookie_jar.max_requests_per_second, config.cookie_jar.url,
config.cookie_jar.database)
add_cookie_jar_logging(cookie_jar, logger)
# Setup data retrieval manager
update_mapper = BatonUpdateMapper(config.baton.binaries_location, zone=config.baton.zone)
retrieval_manager = PeriodicRetrievalManager(config.retrieval.period, update_mapper, logger)
# # Setup basic Slack client
# slack = BasicSlackClient(config.slack.token, config.slack.default_channel, config.slack.default_username)
#
# # Setup basic message queue (e.g. RabbitMQ) client
# message_queue = BasicMessageQueue(config.message_queue.host, config.message_queue.port,
# config.message_queue.username, config.message_queue.password)
slack = message_queue = None
# Define the context that rules and enrichment loaders has access to
context = HgiContext(cookie_jar, config, slack, message_queue)
# Setup rules source
rules_source = RuleSource(config.processing.rules_location, context)
rules_source.start()
# Setup enrichment loader source
enrichment_loader_source = EnrichmentLoaderSource(config.processing.enrichment_loaders_location, context)
enrichment_loader_source.start()
# Setup the data processor manager
processor_manager = BasicProcessorManager(cookie_jar, rules_source, enrichment_loader_source,
config.processing.max_threads, logger)
# Connect components to the cookie jar
_connect_retrieval_manager_to_cookie_jar(retrieval_manager, cookie_jar, config.cookie_jar.max_requests_per_second,
logger)
_connect_processor_manager_to_cookie_jar(processor_manager, cookie_jar)
# Setup the HTTP API
api = HTTP_API()
api.inject(APIDependency.CookieJar, cookie_jar)
api.listen(config.api.port)
# Start the retrieval manager
retrieval_manager.start(config.retrieval.since)
# Start processing of any unprocessed cookies
processor_manager.process_any_cookies()
# Setup monitors
ThreadsMonitor(logger, logging_buffer_latency).start()
CookieJarMonitor(logger, logging_buffer_latency, cookie_jar).start()