def __init__(self, ingestion_id: str, ingestion_type: str,
request_content: str):
self.status = GiraffeEventType.STARTED_PROCESSING_REQUEST
self.map_status_to_timestamp: Dict[str, str] = {}
self.id = ingestion_id
self.log = log_helper.get_logger(logger_name=__name__)
self.request_body = request_content
self.start_timestamp = datetime.now()
self.start_time_unix = self.now_as_unix_timestamp()
self.end_timestamp = None
self.end_time_unix = None
self.map_source_to_model = {}
self.map_redis_key_to_cardinality = {}
self.map_redis_key_to_processed_amount = {}
self.counters = {}
self.finished_keys = []
self.errors = []
self.set_status(status=GiraffeEventType.STARTED_PROCESSING_REQUEST)
self.log.info(
f'Processing request-id: {ingestion_id} ({ingestion_type}) {request_content}'
)
self.log.admin({
Field.request_id: ingestion_id,
Field.request_type: ingestion_type,
Field.request: request_content
})
def __init__(self, event_dispatcher: EventDispatcher, config=config_helper.get_config()):
self.config = config
self.event_dispatcher = event_dispatcher
self.log = log_helper.get_logger(logger_name=f'{self.__class__.__name__}_{threading.current_thread().name}')
# Connecting py2neo
self.graph = Graph(
uri=config.neo_host_address,
user=config.neo_username,
password=config.neo_password
)
# Connecting official bolt-driver
self._driver = GraphDatabase.driver(uri=config.neo_bolt_uri,
auth=(config.neo_username, config.neo_password))
try:
db_kernel_start = self.graph.database.kernel_start_time
except ServiceUnavailable as _:
raise TechnicalError(f'Neo4j does not seem to be active at {config.neo_host_address}')
self.log.debug(f'Neo4j is active since {db_kernel_start}.')
atexit.register(self._driver.close)
self.indices_cache_label_property: List[Tuple[str, str]] = []
def __init__(self,
event_dispatcher: EventDispatcher,
config=config_helper.get_config()):
self.lock = threading.Lock()
self.config = config
self.log = log_helper.get_logger(logger_name=__name__)
self.log.debug('Progress-Monitor started.')
self.all_tasks: Dict[str, IngestionRequest] = {}
event_dispatcher.register_callback(callback=self.on_giraffe_event)
def __init__(self, mode: CommunicatorMode, host: str = 'localhost', port: int = 65432):
self.lock = threading.Lock()
self._host = host
self._port = port
self._mode = mode
self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
if self._mode == CommunicatorMode.CLIENT:
self._is_client = True
self.socket.settimeout(30) # seconds
self.log = log_helper.get_logger(logger_name='Client-Communicator')
self.buffer = bytearray
else:
self._is_client = False
self.thread_pool = ThreadPoolExecutor(max_workers=2)
self.listeners = []
self.log = log_helper.get_logger(logger_name='Server-Communicator')
atexit.register(self.stop)
self.__start()
def __init__(self,
monitoring_host: str = 'localhost',
monitoring_port: int = 65432):
self.listeners: List[Callable] = []
self.log = log_helper.get_logger(logger_name='Event-Dispatcher')
self.tcp_server: Communicator = Communicator(
mode=CommunicatorMode.SERVER,
host=monitoring_host,
port=monitoring_port)
def __init__(self, event_dispatcher: EventDispatcher, env: EnvProvider):
self.log = log_helper.get_logger(
logger_name=
f'{self.__class__.__name__}_{threading.current_thread().name}')
self.log.debug(f'Initialising redis driver.')
self.config = env.config
self.driver: Redis = redis.StrictRedis(
host=self.config.redis_host_address,
port=self.config.redis_port,
decode_responses=True)
self.spark_helper: SparkHelper = env.spark_helper
self.event_dispatcher = event_dispatcher
atexit.register(self.driver.close)
def __init__(self, env: EnvProvider, multi_helper: MultiHelper,
event_dispatcher: EventDispatcher):
self.is_ready = False
self.event_dispatcher = event_dispatcher
self.log = log_helper.get_logger(logger_name=self.__class__.__name__)
self.config = env.config
try:
self.neo_db: NeoDB = NeoDB(config=self.config,
event_dispatcher=self.event_dispatcher)
self.redis_db: RedisDB = RedisDB(
env=env, event_dispatcher=self.event_dispatcher)
self.multi_helper: MultiHelper = multi_helper
self.is_ready = True
except Exception as the_exception:
self.log.error(the_exception, exc_info=True)
self.is_ready = False
IngestionManager.supported_operations = (
self.config.nodes_ingestion_operation,
self.config.edges_ingestion_operation)
def test_delete_nodes_by_property(config_helper, neo, logger):
prepare_neo(config_helper=config_helper, log=logger, neo=neo)
log: Logger = log_helper.get_logger(logger_name='Timing...')
db: neo_db.NeoDB = neo
test_label = config_helper.test_labels[0]
sleep(3)
query = f"MATCH(n:{test_label}) return count(n) as count"
before_deletion_count = db.pull_query(query=query).value()[0]
log.info(f'Before deletion there are {before_deletion_count} items.')
timer = Timer()
timer.start()
result: dict = db.delete_nodes_by_properties(label=test_label, property_name_value_tuples=[('name', 'Arafat')])
sec = timer.stop()
log.info(f'Time elapsed: {sec} seconds')
after_deletion_count = db.pull_query(query=query).value()[0]
log.info(f'After deletion there are {after_deletion_count} items.')
deleted_count = result['total']
assert deleted_count > 0
assert before_deletion_count - after_deletion_count == deleted_count
def __init__(self, env: EnvProvider, progress_monitor: ProgressMonitor,
event_dispatcher: EventDispatcher):
self.event_dispatcher = event_dispatcher
self.progress_monitor = progress_monitor
self.is_ready = False
self.log = log_helper.get_logger(logger_name=__name__)
# noinspection PyBroadException
try:
self.redis_db = RedisDB(event_dispatcher=self.event_dispatcher,
env=env)
self.data_and_model_provider = env.data_and_model_provider
self.thread_pool = ThreadPool()
self.data_to_graph_translator = env.data_to_graph_entities_provider
self.multi_helper = MultiHelper(config=env.config)
self.im = IngestionManager(env=env,
multi_helper=self.multi_helper,
event_dispatcher=self.event_dispatcher)
self.is_ready = True
except Exception as the_exception:
self.log.error(the_exception, exc_info=True)
self.is_ready = False
def __init__(self, cmd_line_args=None):
self.log = log_helper.get_logger(logger_name=__name__)
configuration_ini_file_path = None if cmd_line_args is None else cmd_line_args[0].config_ini
if configuration_ini_file_path is None:
self.config = config_helper.get_config()
else:
config_ini_file = configuration_ini_file_path
validate_is_file(file_path=config_ini_file)
self.config = config_helper.get_config(configurations_ini_file_path=config_ini_file)
# -------
self.logging_file_path = os.path.join(self.config.logs_storage_folder, 'giraffe.log')
file_handler = RotatingFileHandler(filename=self.logging_file_path,
mode='a',
maxBytes=1_000_000,
backupCount=3,
encoding='utf-8',
delay=False)
file_handler.setFormatter(log_helper.log_row_format)
log_helper.add_handler(handler=file_handler)
atexit.register(log_helper.stop_listener)
# -------
self.execution_env = self.config.execution_environment
# -------
if self.execution_env == 'dev':
self.data_and_model_provider: DataAndModelProvider = MockDataAndModelProvider()
self.data_to_graph_entities_provider: DataToGraphEntitiesProvider = MockDataToGraphEntitiesProvider()
self.spark_helper = DevSparkHelper(config=self.config)
elif self.execution_env == 'cortex':
raise NotImplementedError('Implemented in cortex')
else:
self.log.info(f'Unexpected value in configuration file for execution_environment: {self.execution_env}')
sys.exit(1)
from waitress import serve
coordinator: Coordinator
progress_monitor: ProgressMonitor
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Front Desk')
parser.add_argument('--config_ini', type=str, required=False, default=None)
args = parser.parse_known_args()
env = EnvProvider(cmd_line_args=args)
event_dispatcher = EventDispatcher()
log = log_helper.get_logger(__name__)
log.info(f'Execution environment: {env.execution_env}')
not_acceptable_error_code = 406
log.debug('Initializing coordinator module.')
progress_monitor = ProgressMonitor(event_dispatcher=event_dispatcher,
config=env.config)
atexit.register(progress_monitor.dump_and_clear_memory)
coordinator = Coordinator(env=env,
progress_monitor=progress_monitor,
event_dispatcher=event_dispatcher)
if not coordinator.is_ready:
log.error('Failed initializing coordinator component - aborting.')
sys.exit(-1)
def logger() -> Logger:
return log_helper.get_logger('Testing-Suite', )
def __init__(self, handlers: List[DbLogHandler]):
self.map_job_id_to_name = {}
self.handlers = handlers
self.log = log_helper.get_logger(logger_name=self.__class__.__name__)
def __init__(self, config: ConfigHelper):
self.config = config
self.thread_executor = ThreadPoolExecutor(
max_workers=config.thread_pool_size)
self.log = log_helper.get_logger(logger_name='Multi-Helper')
self.futures: List[Future] = []
def __init__(
self,
configurations_ini_file_path: str = default_configurations_file):
self.log = log_helper.get_logger(logger_name=__name__)
validate_is_file(file_path=configurations_ini_file_path)
neo4j_section = 'NEO4J'
redis_section = 'REDIS'
testing_section = 'TESTING'
giraffe_section = 'GIRAFFE'
general_section = 'GENERAL'
spark_section = 'SPARK'
elastic_section = 'ELASTICSEARCH'
config_file_path = os.path.abspath(configurations_ini_file_path)
self.log.info(f'Configuration file: {config_file_path}')
self.config = configparser.ConfigParser()
self.config.read(configurations_ini_file_path)
self.log.debug(
f'Found the following configuration sections: {self.config.sections()}'
)
# General settings
if self.config.has_section(section=neo4j_section):
self.string_encoding = self.config[general_section][
'string_encoding']
else:
self.log.warning(
f'No configuration found for section {general_section}')
# Reading Neo4j connection details from configuration-file
if self.config.has_section(section=neo4j_section):
self.neo_host_address = self.config[neo4j_section]['HOST']
self.neo_username = self.config[neo4j_section]['USERNAME']
self.neo_password = self.config[neo4j_section]['PASSWORD']
self.neo_bolt_port = self.config[neo4j_section]['BOLT_PORT']
else:
self.log.warning(
f'No configuration found for section {neo4j_section}')
self.neo_bolt_uri = f'bolt://{self.neo_host_address}:{self.neo_bolt_port}'
if self.config.has_section(section=redis_section):
# Reading REDIS connection details from configuration-file
self.redis_host_address = self.config[('%s' %
redis_section)]['HOST']
self.redis_username = self.config[redis_section]['USERNAME']
self.redis_password = self.config[redis_section]['PASSWORD']
self.redis_port = self.config[redis_section]['PORT']
self.redis_stream_milliseconds_block = self.config[redis_section][
'STREAM_BLOCK_MILLISECONDS']
else:
self.log.warning(
f'No configuration found for section {redis_section}')
if self.config.has_section(section=testing_section):
# Unit-Testing settings
self.test_labels = self.config[testing_section][
'test_labels'].split(',')
self.test_edge_type = self.config[testing_section][
'test_edge_type']
self.test_property = self.config[testing_section]['test_property']
self.number_of_test_nodes = int(
self.config[testing_section]['number_of_test_nodes'])
self.number_of_test_edges = int(
self.config[testing_section]['number_of_test_edges'])
self.test_chunk_size = int(
self.config[testing_section]['test_chunk_size'])
self.test_job_name = self.config[testing_section][
'test_request_name']
self.test_elasticsearch_index = self.config[testing_section][
'test_elasticsearch_index']
self.test_redis_table_prefix = self.config[testing_section][
'test_redis_table_prefix']
self.test_redis_stream_name = self.config[testing_section][
'test_redis_stream_name']
self.test_front_desk_address = self.config[testing_section][
'test_front_desk_address']
else:
self.log.warning(
f'No configuration found for section {testing_section}')
if self.config.has_section(section=giraffe_section):
# Giraffe logic
self.nodes_ingestion_operation = self.config[giraffe_section][
'nodes_ingestion_operation']
self.edges_ingestion_operation = self.config[giraffe_section][
'edges_ingestion_operation']
self.key_separator = self.config[giraffe_section]['key_separator']
self.uid_property = self.config[giraffe_section][
'unique_identifier_property_name']
self.from_uid_property = self.config[giraffe_section][
'from_uid_property_name']
self.to_uid_property = self.config[giraffe_section][
'to_uid_property_name']
self.edge_type_property = self.config[giraffe_section][
'edge_type_property_name']
self.deletion_batch_size = self.config[giraffe_section][
'deletion_batch_size']
self.model_vertex_id_prop_name = self.config[giraffe_section][
'model_vertex_id_prop_name']
self.model_edge_id_prop_name = self.config[giraffe_section][
'model_edge_id_prop_name']
self.edge_to_identifier_name = self.config[giraffe_section][
'edge_to_identifier_name']
self.inception_models_rest_address = self.config[giraffe_section][
'inception_models_rest_address']
self.inception_data_sources_rest_address = self.config[
giraffe_section]['inception_data_sources_rest_address']
self.data_source_parts_separator = self.config[giraffe_section][
'data_source_parts_separator']
self.expected_number_of_source_parts = int(
self.config[giraffe_section]
['expected_number_of_source_parts'])
self.front_desk_port = int(
self.config[giraffe_section]['front_desk_port'])
self.redis_stream_name = self.config[giraffe_section][
'redis_stream_name']
self.ingestion_endpoint = self.config[giraffe_section][
'ingestion_endpoint'].strip()
self.redis_get_all_endpoint = self.config[giraffe_section][
'redis_get_all_endpoint'].strip()
self.request_mandatory_field_names = eval(
self.config[giraffe_section]
['request_type_mandatory_field_name'])
self.logs_storage_folder = self.config[giraffe_section][
'logs_storage_folder']
self.progress_monitor_dump_folder = self.config[giraffe_section][
'progress_monitor_dump_folder']
self.admin_db_table_name = self.config[giraffe_section][
'admin_db_table_name']
self.required_request_fields: Dict[str, Dict] = eval(
self.config[giraffe_section]['required_request_fields'])
self.hash_uid_column = eval(
self.config[giraffe_section]['hash_uid_column'])
self.front_desk_ip = self.config[giraffe_section]['front_desk_ip']
self.execution_environment = self.config[giraffe_section][
'execution_environment']
self.logs_structured_prefix = self.config[giraffe_section][
'logs_structured_prefix']
self.logs_fluentd_host = self.config[giraffe_section][
'logs_fluentd_host']
self.logs_fluentd_port = int(
self.config[giraffe_section]['logs_fluentd_port'])
self.thread_pool_size = int(
self.config[giraffe_section]['thread_pool_size'])
self.property_names_to_index = eval(
self.config[giraffe_section]['property_names_to_index'])
else:
self.log.warning(
f'No configuration found for section {giraffe_section}')
if self.config.has_section(section=spark_section):
# Spark logic
self.external_jars_folder = self.config[spark_section][
'external_jars']
self.spark_app_name = self.config[spark_section]['app_name']
else:
self.log.warning(
f'No configuration found for section {spark_section}')
if self.config.has_section(section=elastic_section):
# ElasticSearch logic
self.es_host_address = self.config[elastic_section]['HOST']
else:
self.log.warning(
f'No configuration found for section {elastic_section}')
