def _enqueue(self, resource):
# Discovered URI - capped collection --db.discovered_uri
mongoConnection = Connection()
db = mongoConnection.feed_reading_discovery
# settings
max_content_length = config['discovery']['resource_enqueueing']['max_content_length']
# message queue
mq_client = message_queue_client_from_config(config['message_queue']['client'])
mq_client.connect()
mq_codec = JSONCodec()
queue = 'discovered_resources'
try:
#insert into capped
db.discovered_uri.insert({'_id':makeIdFromURI(resource.uri), 'uri':resource.uri})
#insert into queue
if len(resource.content) > max_content_length:
self._logger.warning('Skipped %s: Content length is %s.' %
(resource, len(resource.content)))
msg_body = mq_codec.encode(resource)
mq_client.put_message(queue, msg_body)
self._logger.debug("Enqueued: %s" % resource._id)
except DuplicateKeyError:
pass
# remove resource from collection
self._resources_collection.remove_model(resource)
# MongoDB
host = 'localhost'
port = 27017
mcm = MongoConnectionManager(host, port, MongoCodec())
database = 'processed'
resource_collection = mcm.get_collection(database, 'resources', Resource)
# message queue
mq_config = {
'transport': 'socket',
'protocol': 'binary',
'host': 'localhost',
'port': 9091
}
mq_client = message_queue_client_from_config(mq_config)
mq_codec = JSONCodec()
processed_resource_queue = 'processed_resources'
# ElasticSearch
es = ES('localhost:9200', timeout=60)
es_index = 'topic_tracking'
# dequeue one resource
mq_client.connect()
message = mq_client.get_message(processed_resource_queue)
resource = mq_codec.decode(message.body, Resource)
mq_client.delete_message(processed_resource_queue, message.id)
mq_client.disconnect()
# save the resource to mongo
resource._id = makeIdFromURI(resource.uri)
EMPTY_QUEUE_TIMEOUT = 1
if __name__ == '__main__':
config_file = sys.argv[1]
config = yaml.load(file(config_file, 'r'))
# logging
logging.config.dictConfig(config['logging'])
logger = logging.getLogger()
# message queue
mq_client = message_queue_client_from_config(config['message_queue']['client'])
mq_codec = JSONCodec()
input_queue = 'discovered_resources'
output_queue = 'processed_resources'
# processing
p_client = processing_client_from_config(config['processing']['client'])
# start service clients
mq_client.connect()
p_client.connect()
# begin processing
while True:
try:
# input
input_message = mq_client.get_message(input_queue)
