def listen():
print "attach"
# Kafka
consumer = KafkaConsumer(bootstrap_servers=os.environ["KAFKA_BOOTSTRAP_SRVS"], group_id=os.environ["KAFKA_GROUP_ID"])
consumer.subscribe([os.environ["KAFKA_SOURCE_TOPIC"]])
# Snowplow
e = Emitter(os.environ["SP_COLLECTOR_URI"],protocol=os.environ["SP_COLLECTOR_PROTOCOL"],port=int(os.environ["SP_COLLECTOR_PORT"]),method=os.environ["SP_COLLECTOR_METHOD"])
t = Tracker(emitters=e,namespace="cf",app_id=str(os.environ["APP_ID"]),encode_base64=True)
for msg in consumer:
#
try:
indata = json.loads(msg.value)
s1 = Subject()
s1.set_platform("app")
s1.set_user_id("??")
s1.set_lang("??")
s1.set_ip_address("0.0.0.0")
s1.set_useragent("??")
t.set_subject(s1)
t.track_self_describing_event(SelfDescribingJson("iglu:com.snowplowanalytics.snowplow/unstruct_event/jsonschema/1-0-0",{
"data":{
"data": indata
},
"schema": "iglu:"+os.environ["OPERATOR_ID"]+"/"+os.environ["APP_ID"]+"/jsonschema/1-0-0"
}))
t.flush()
except Exception,Argument:
print "Error:",str(Argument)
def save_tweet(data):
global tracker
print "save_tweet:", __name__
#print data
indata = data
s1 = Subject()
s1.set_platform("web")
s1.set_user_id(str(indata.get("user_id")))
s1.set_lang(str(indata.get("lang")))
#s1.set_ip_address(str(indata.get("i_ip")))
s1.set_useragent(str(indata.get("source")))
tracker.set_subject(s1)
tracker.track_self_describing_event(
SelfDescribingJson(
"iglu:com.snowplowanalytics.snowplow/unstruct_event/jsonschema/1-0-0",
{
"data": {
"data": indata
},
"schema":
"iglu:com.rbox24/" + args.sp_app_id + "/jsonschema/1-0-0"
}))
tracker.flush()
print "Tweet sent to collector, time:", time.time()
def save_tweet(data):
#print "save_tweet"
#print data
indata = data
e = Emitter(args.sp_collector_uri,
protocol=args.sp_collector_protocol,
port=int(args.sp_collector_port),
method=args.sp_collector_method)
t = Tracker(emitters=e,
namespace="cf",
app_id=args.sp_app_id,
encode_base64=True)
s1 = Subject()
s1.set_platform("web")
s1.set_user_id(str(indata.get("user_id")))
s1.set_lang(str(indata.get("lang")))
#s1.set_ip_address(str(indata.get("i_ip")))
s1.set_useragent(str(indata.get("source")))
t.set_subject(s1)
t.track_self_describing_event(
SelfDescribingJson(
"iglu:com.snowplowanalytics.snowplow/unstruct_event/jsonschema/1-0-0",
{
"data": {
"data": indata
},
"schema":
"iglu:com.rbox24/" + args.sp_app_id + "/jsonschema/1-0-0"
}))
t.flush()
print "Tweet sent to collector, time:", time.time()