def main(profile: str):
"""
Celery worker main entry point
Args:
profile: profile used to run the app
"""
load_config(profile, CONFIGS_PATH, config, 'NLP_SERVICE')
initialize_summary_service()
load()
publisher = container.get('exchange_publisher')
if not publisher.test_connection():
LOGGER.error('Error connecting to the queue provider. Exiting...')
sys.exit(1)
add_logstash_handler(LOG_CONFIG, config.logstash.host, config.logstash.port)
CELERY_APP.configure(task_queue_name='nlp-worker',
broker_config=config.rabbit,
worker_concurrency=config.celery.concurrency,
result_backend_url=build_redis_url(**config.redis))
apm_client = Client(config={
'SERVICE_NAME': config.elastic_apm.service_name,
'SECRET_TOKEN': config.elastic_apm.secret_token,
'SERVER_URL': config.elastic_apm.url
})
register_instrumentation(apm_client)
register_exception_tracking(apm_client)
CELERY_APP.run()
def shutdown_worker(*_, **__):
"""
Shutdown the celery worker shutting down the exchange publisher
"""
LOGGER.info('Shutting down worker')
exchange_publisher: ExchangePublisher = container.get('exchange_publisher')
exchange_publisher.shutdown()
def main(profile: str):
"""
Celery app main entry point
Args:
profile: profile used to run the app
"""
load_config(profile, CONFIGS_PATH, config, 'NEWS_DISCOVERY')
load()
publisher = container.get('exchange_publisher')
if not publisher.test_connection():
LOGGER.error('Error connecting to the queue provider. Exiting...')
sys.exit(1)
add_logstash_handler(LOG_CONFIG, config.logstash.host,
config.logstash.port)
CELERY_APP.configure(task_queue_name='news-discovery',
broker_config=config.rabbit,
worker_concurrency=config.celery.concurrency)
apm_client = Client(
config={
'SERVICE_NAME': 'news-discovery-app',
'SECRET_TOKEN': config.elastic_apm.secret_token,
'SERVER_URL': config.elastic_apm.url
})
register_instrumentation(apm_client)
register_exception_tracking(apm_client)
CELERY_APP.run()
def hydrate_new(new: dict = None, nlp_doc: dict = None, summary: str = None, sentiment: float = None, **_):
"""
Hydrate the input new with the named entities and noun chunks from the input NLP document and with the input
summary and sentiment
Args:
new: new to hydrate
nlp_doc: new NLP information
summary: new summary
sentiment: new sentiment
Returns: hydrated new
"""
LOGGER.info('Hydrating new %s', new['title'])
new = from_dict(New, new)
if summary is not None:
new.summary = summary
if sentiment is not None:
new.sentiment = sentiment
nlp_service = container.get('nlp_service')
if nlp_doc is not None:
doc = nlp_service.doc_from_json_dict(nlp_doc)
new.entities = list(
set(map(lambda entity: NamedEntity(text=str(entity), type=entity.label_), doc.ents)))
new.noun_chunks = list(map(lambda chunk: str(chunk), doc.noun_chunks))
new.hydrated = True
return asdict(new)
def initialize_worker(*_, **__):
"""
Initialize the celery worker process environment
"""
LOGGER.info('Initializing worker')
exchange_publisher: ExchangePublisher = container.get('exchange_publisher')
exchange_publisher.connect()
exchange_publisher.initialize()
def sentiment_analysis(nlp_doc: dict = None, **_):
"""
Get the sentiment score of the input doc sentences
Args:
nlp_doc: doc to analyze sentiment
Returns: input doc sentences sentiment score
"""
LOGGER.info('Generating sentiment score')
nlp_service = container.get('nlp_service')
sentiment_analyzer = container.get('sentiment_analysis_service')
if sentiment_analyzer is not None:
if nlp_doc is not None:
doc = nlp_service.doc_from_json_dict(nlp_doc)
return sentiment_analyzer(list(doc.sents))
else:
LOGGER.warning('NLP document is missing. Skipping sentiment calculation...')
return None
else:
LOGGER.warning('Sentiment analyzer not initialized. Skipping sentiment calculation...')
return None
def summarize(nlp_doc: dict = None, **_):
"""
Generate the summary for the input NLP doc
Args:
nlp_doc: document to generate summary
Returns: summary of the doc sentences
"""
LOGGER.info('Generating summary')
nlp_service = container.get('nlp_service')
summarizer = container.get('summary_service')
if summarizer is not None:
if nlp_doc is not None:
doc = nlp_service.doc_from_json_dict(nlp_doc)
return summarizer(list(doc.sents))
else:
LOGGER.warning('NLP document is missing. Skipping summary generation...')
return None
else:
LOGGER.warning('Summarizer not initialized. Skipping summary generation...')
return None
def discover_news(definition_name: str):
"""
Discover news task
Args:
definition_name: name of the news discovery definition
"""
if 'rabbit' in config:
LOGGER.info(f'Executing discovery {definition_name}')
definition = DEFINITIONS[definition_name]
definition_instance = definition['class'](definition)
exchange_publisher: ExchangePublisher = container.get(
'exchange_publisher')
for discovered_new in definition_instance():
exchange_publisher(asdict(discovered_new))
else:
LOGGER.error('Worker configuration not initialized')
def process_new_content(new: dict = None, **_):
"""
Apply NLP processing to the input new content
Args:
new: new to process content
Returns: new to hydrate in next tasks, processed new content
"""
LOGGER.info('NLP Processing new %s', new['title'])
nlp_service = container.get('nlp_service')
if nlp_service is not None:
processed_content = nlp_service.process_text(new['content'])
return nlp_service.doc_to_json_dict(processed_content)
else:
LOGGER.warning('NLP service not initialized, skipping NLP processing')
return None
def publish_hydrated_new(new: dict = None, **_):
"""
Publish the the input new updated
Args:
new: new to publish
"""
if new is not None:
LOGGER.info('Publishing hydrated new %s', new['title'])
if config.rabbit is not None:
LOGGER.info('Queue connection initialized, publishing...')
exchange_publisher: ExchangePublisher = container.get('exchange_publisher')
exchange_publisher(new)
LOGGER.info('New published')
else:
LOGGER.warning('Queue connection configuration not initialized, skipping publish...')
else:
LOGGER.warning('Tasks chain services not initialized, skipping publish...')
