class AnnotatorWorker(object):
def __init__(self):
pass
def send_to_broker(self,annotated_article):
publisher = Publisher('data_distributor')
publisher.send_message(annotated_article,'articles')
def handle_delivery(self,channel, method, header, body):
article = json.loads(body)
#1.TODO All annotator work would ideally be done here. Import annotator module being developed
#2.TODO Once the article that was sent is annotated. Send it back to be saved to the database
annotated_article = json.dumps('new article here')
self.send_to_broker(annotated_article)
channel.basic_ack(delivery_tag = method.delivery_tag)
print '--Done Annotating Article--'
def run(self):
self.consumer = Consumer('data_distributor',
'annotator_worker',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
def run(self):
self.consumer = Consumer('data_distributor',
'database_queue',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
def run(self):
self.consumer = Consumer('data_distributor',
'annotator_worker',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
class AnnotatorWorker(object):
def __init__(self):
pass
def send_to_broker(self, annotated_article):
publisher = Publisher('data_distributor')
publisher.send_message(annotated_article, 'articles')
def handle_delivery(self, channel, method, header, body):
article = json.loads(body)
#1.TODO All annotator work would ideally be done here. Import annotator module being developed
#2.TODO Once the article that was sent is annotated. Send it back to be saved to the database
annotated_article = json.dumps('new article here')
self.send_to_broker(annotated_article)
channel.basic_ack(delivery_tag=method.delivery_tag)
print '--Done Annotating Article--'
def run(self):
self.consumer = Consumer('data_distributor',
'annotator_worker',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
def run(self):
self.consumer = Consumer('data_distributor',
'annotator_worker',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
class ElasticSearchWorker(object):
def __init__(self):
self.module_path = os.path.dirname(os.path.realpath(__file__))
self.config_dir = os.path.join(self.module_path,os.path.join('..' + os.sep + 'configs' ))
self.file_path = os.path.abspath(self.config_dir) + os.sep + 'settings.yaml'
self.config_file = open(self.file_path,'r')
self.config_data = yaml.load(self.config_file)
#Should ideally read from config file
self.es = Elasticsearch(timeout=30)
def convert_to_datetime(self,timestamp):
date = datetime.fromtimestamp(timestamp/1000)
return date
def handle_delivery(self,channel, method, header, body):
article = json.loads(body)
article['time_of_crawl'] = self.convert_to_datetime(article['time_of_crawl'])
if 'date_published' in article['meta_information']:
article['meta_information']['date_published'] = self.convert_to_datetime(article['meta_information']['date_published'])
es_result = self.es.index(index="news", doc_type="article", body=article)
self.es.indices.refresh(index="news")
channel.basic_ack(delivery_tag = method.delivery_tag)
print '--Done Inserting To ElasticSearch--'
def run(self):
self.consumer = Consumer('data_distributor',
'database_queue',
'articles',
handle_message=self.handle_delivery)
self.consumer.start()
