class Producer():
def __init__(self):
self.producer = KafkaProducer(bootstrap_servers=["52.41.44.90:9092","52.36.206.57:9092","52.40.205.225:9092"],acks=0,linger_ms=500)
def produce_msgs(self,msg_list):
while True:
index = random.randrange(0,999)
json_msg =json.dumps(msg_list[index]).encode('utf-8')
self.producer.send(topic, json_msg)
class Producer():
def __init__(self):
self.producer = KafkaProducer(bootstrap_servers=["52.41.44.90:9092","52.36.206.57:9092","52.40.205.225:9092"],acks=0,linger_ms=500)
def produce_msgs(self,msg_list):
while True:
index = random.randrange(0,999)
info = person_pb2.PersonInfo()
serialize_protobuf (info.user.add(), msg_list[index])
_msg = user.SerializeToString()
self.producer.send(topic, _msg)
class Producer():
def __init__(self):
self.producer = KafkaProducer(bootstrap_servers = 'localhost:9092')
def produce_msgs(self,source):
for drop in source:
if 'text' in drop:
message = json.dumps(drop)
self.producer.send('Twitter-Stream',message)
print(message)
self.producer.send('message-size',len(message))
def on_status(self, status):
print status.text, "\n"
#~ data ={}
#~ data['text'] = status.text
#~ data['created_at'] = status.created_at
#~ data['geo'] = status.geo
#~ data['source'] = status.source
#~ self.db.Tweets.insert(data)
msg = status.text.encode('utf-8')
producer = KafkaProducer(bootstrap_servers='0.0.0.0:9092')
#~ producer = KafkaProducer(bootstrap_servers=['broker1:1234'])
#print(msg)
try:
producer.send(b'twitterstream', msg)
except Exception as err:
print(err)
return False
return True
def main():
s3 = boto3.resource('s3')
bucket = s3.Bucket('nyc-tlc')
# Iterates through all the objects, doing the pagination for you. Each obj
# is an ObjectSummary, so it doesn't contain the body. You'll need to call
# get to get the whole body.
kafka_params = config('kafka')
dataset_params = config('dataset')
producer = KafkaProducer(bootstrap_servers=kafka_params['broker'])
for obj in bucket.objects.all():
key = obj.key
print(key)
if dataset_params['driver'] not in key:
continue
#building absolute file name
file_name = 's3://nyc-tlc/' + key
#skipping header
firstline = True
# Processing each row in file
for line in smart_open(file_name):
# print(line.decode('utf8'))
if firstline: # skip first line
firstline = False
continue
line_split = line.decode('utf8').split(",")
if len(line_split
) < 20: #Skipping rows with large number of columns
continue
if line_split[5] == '0' or line_split[6] == '0' or line_split[
7] == '0' or line_split[8] == '0':
continue
else:
start_point = (float(line_split[5]), float(line_split[6]))
end_point = (float(line_split[7]), float(line_split[8]))
# print(start_point, end_point)
intermediate_points = getEquidistantPoints(
start_point, end_point, 100)
# print(intermediate_points)
#message when trip is started
trip_id = 'drive:' + str(datetime.now()) + ":" + str(
random.randint(1, 1000))
formatted_message = format_message(trip_id, start_point,
start_point, end_point,
"New")
producer.send(kafka_params['driver_topic'],
formatted_message.encode('utf8 '))
#Simulating moving car by sending intermediate points
for int_point in intermediate_points:
# print(int_point)
formatted_message = format_message(trip_id, start_point,
int_point, end_point,
"In Progress")
producer.send(kafka_params['driver_topic'],
formatted_message.encode('utf8 '))
#Ending the driver trip
formatted_message = format_message(trip_id, start_point,
end_point, end_point,
"Closed")
print(formatted_message.encode('utf8 '))
producer.send(kafka_params['driver_topic'],
formatted_message.encode('utf8 '))
row_i = 0
time.sleep(config.PRODUCER_SLEEP_TIME)
#print('column:', column)
for row in df[column]:
#print('row: ', row)
#print('col:',column)
#timestamp = datetime.now().strftime("%H%M%S%f")
timestamp_s = float(datetime.now().strftime("%M")) * 60 + float(
datetime.now().strftime("%S.%f"))
longitude = row.split(',')[0]
latitude = row.split(',')[1]
#user_id = filenames[row_i]
#message_to_send = str_fmt.format(users[row_i],, timestamp, longitude, latitude, int(column==0))
message_to_send = str_fmt.format(users[row_i], timestamp_s,
longitude, latitude,
int(column == 0))
#print(users[row_i], message_to_send)
producer.send(topic=config.KAFKA_TOPIC,
value=message_to_send,
key=users[row_i].encode('utf-8'))
row_i += 1
end_time = float(datetime.now().strftime("%M")) * 60 + float(
datetime.now().strftime("%S.%f"))
print('end time: ', end_time)
total_time = end_time - start_time
messages_per_second = (num_rows * num_files) / total_time
print('total_time: ', total_time)
print('messages_per_second: ', messages_per_second)
logger.info(
"Start MongoDB Change Streams Service for table {}...".format(col_tmp))
with col.watch(pipeline, full_document='updateLookup') as stream:
for change in stream:
update_operations = list()
#delete_operations = list()
logger.info(change)
msg = str(change.get('documentKey').get('_id')) + ',' + str(
change.get('clusterTime').time)
topic = change.get('ns').get('coll')
producer = KafkaProducer(bootstrap_servers=['172.16.42.3:9092'])
producer.send(topic,
key=bytes(
json.dumps(change.get('ns')).encode('utf-8')),
value=bytes(json.dumps(msg).encode('utf-8')),
partition=0)
producer.close()
record = change.get('fullDocument')
record.pop('_id')
record.pop('createdTime')
update_op = UpdateOne({'id': record['id']}, {
'$set': record,
'$setOnInsert': {
'createdTime': datetime.now().strftime('%Y-%m-%dT%H:%M:%S')
}
},
upsert=True)
#delete_op = DeleteOne({'id':record['id']})
update_operations.append(update_op)
import sys
from kafka.producer import KafkaProducer
config = {
'access_key': 'your_access_key_here',
'access_secret': 'your_access_secret_key_here',
'consumer_key': 'your_consumer_key_here',
'consumer_secret': 'your_consumer_secret_key_here'
}
auth = OAuth(config["access_key"], config["access_secret"],
config["consumer_key"], config["consumer_secret"])
stream = TwitterStream(domain='userstream.twitter.com', auth=auth, secure=True)
search_term = "coffee, cappuccino, espresso, frappuccino, mocha, \
tea, matcha, chai, oolong, pu-erh, tisane, \
milk, dairy, half-and-half, \
soda, coke, cola, fanta, sprite, pepsi, Dr pepper, soft drink, \
juice, oj, cider, \
wine, riesling, merlot, syrah, chardonnay, sauvignon, pinot noir, \
beer, brew, amber, ipa, bud light, budweiser, miller lite, corona extra, heineken, \
liquor, sake, shochu, whisky, tequila, gin, cognac, rum"
tweet_iter = stream.statuses.filter(track=search_term, language='en')
producer = KafkaProducer(bootstrap_servers='your_aws_cluster_public_IP:9092')
for tweet in tweet_iter:
print(tweet)
producer.send('insight_topic', json.dumps(tweet))
class MyKafkaProducer(object):
"""
class that implements Kafka producers that ingest data from S3 bucket
"""
def __init__(self, kafka_configfile, schema_file, s3_configfile):
"""
class constructor that initializes the instance according to the configurations
of the S3 bucket and Kafka
:type kafka_configfile: str path to kafka config file
:type schema_file : str path to schema file
:type s3_configfile : str path to S3 config file
"""
self.kafka_config = utility.parse_config(kafka_configfile)
self.schema = utility.parse_config(schema_file)
self.s3_config = utility.parse_config(s3_configfile)
self.producer = KafkaProducer(
bootstrap_servers=self.kafka_config["BROKERS_IP"])
def get_key(self, msg):
"""
produces key for message to Kafka topic
:type msg: dict message for which to generate the key
:rtype : str key that has to be of type bytes
"""
msgwithkey = utility.add_block_fields(msg)
if msgwithkey is None:
return
x, y = msgwithkey["block_lonid"], msgwithkey["block_latid"]
return str((x * 137 + y) % 77703).encode()
def produce_msgs(self):
"""
produces messages and sends them to topic
"""
msg_cnt = 0
while True:
s3 = boto3.client('s3')
# obj = s3.get_object(Bucket=self.s3_config["BUCKET"],
# Key="{}/{}".format(self.s3_config["FOLDER"],
# self.s3_config["STREAMING_FILE"]))
obj = s3.get_object(Bucket='nyctaxitrip',
Key="{}/{}".format(
'yellow_trip',
'yellow_tripdata_sample.csv'))
for line in lazyreader.lazyread(obj['Body'], delimiter='\n'):
message_info = line.strip()
msg = utility.map_schema(message_info, self.schema)
if msg is not None:
self.producer.send(self.kafka_config["TOPIC"],
value=json.dumps(msg),
key=self.get_key(msg))
msg_cnt += 1
time.sleep(0.001)
class AIMSDownsamplingTCPServerConsumer:
LOG_FORMAT ="{} UTC_TS\t"\
"{}"
INTERVAL = 60
DELAY = 0
MAX_CONNECTION = 32
def __init__(self, kafka_host, kafka_port, tcp_host, tcp_port, topic,
log_topic):
self.kafka_host = kafka_host
self.kafka_port = kafka_port
self.tcp_host = tcp_host
self.tcp_port = tcp_port
self.topic = topic
self.log_topic = log_topic
self.consumer = KafkaConsumer(
topic,
bootstrap_servers=["{}:{}".format(kafka_host, kafka_port)],
enable_auto_commit=False,
max_poll_records=1024 * 1024,
max_partition_fetch_bytes=1024 * 1024 * 100)
self.producer = KafkaProducer(
bootstrap_servers=["{}:{}".format(kafka_host, kafka_port)])
self.connections = {}
self.sample_end_time = self.get_end_time(time())
self.lastPolled = []
def run(self):
self.log("running")
asyncio.run(self._async_run())
async def _async_run(self):
tcpServer = await asyncio.start_server(self.connection_handler,
self.tcp_host, self.tcp_port)
await asyncio.gather(tcpServer.serve_forever(), self.poll_from_kafka())
async def connection_handler(self, reader, writer):
addr = str(writer.get_extra_info("peername"))
addr = str(writer.get_extra_info("peername"))
# A new connection, but we can accept no more
if addr not in self.connections and \
len(self.connections)>=self.MAX_CONNECTION:
self.refuse_client(addr, writer)
return
# Add connection
self.add_client(addr, writer)
# Read data from connection
remaining_data = b""
try:
while True:
data = await reader.read(1) # 1024*8 bytes
if not data:
break
except BrokenPipeError:
"""
Catches connecton reset by peer when we are sending the batched data,
which is also when we cannot check for reader. The broken connection
on the writer side will ultimately lead to BrokenPipeError on the
reader side. Hence
"""
pass
finally:
self.remove_client(addr)
async def poll_from_kafka(self):
polled = self.consumer.poll(timeout_ms=self.INTERVAL * 1000 / 2)
self.lastPolled = polled
while True:
t = time()
if t >= self.sample_end_time + self.DELAY:
polled = self.consumer.poll(timeout_ms=self.INTERVAL * 1000 /
2)
lastPolled = self.lastPolled
start_time = self.sample_end_time - self.INTERVAL
end_time = self.sample_end_time
self.lastPolled = polled
self.sample_end_time = self.get_end_time(time())
if len(self.connections) != 0:
# run on lastPolled first to hit the cache
parsed_records = self.get_parsed_records(lastPolled) + \
self.get_parsed_records(polled)
parsed_records = list(
filter(
lambda rec: rec["observation_date_time"] is
not None, parsed_records))
ds_records = self.down_sample(parsed_records, start_time,
end_time)
messages = [rec["message"] for rec in ds_records]
for addr in self.connections.keys():
await self.send_or_ignore_message(addr, messages)
await asyncio.sleep(0.1)
def get_parsed_records(self, polled):
# Create cache
if "_get_parsed_records__polled" not in self.__dict__:
self._get_parsed_records__polled = []
if "_get_parsed_records__ret" not in self.__dict__:
self._get_parsed_records__ret = []
# Cache hit
if self._get_parsed_records__polled == polled:
return self._get_parsed_records__ret
# Cache not hit
self._get_parsed_records__polled = polled
self._get_parsed_records__ret = []
records = []
for recordList in polled.values():
records.extend([rec.value for rec in recordList])
for rec in records:
self._get_parsed_records__ret.append(self.parse_hl7(rec))
return self._get_parsed_records__ret
def parse_hl7(self, message):
segments = message.decode(errors="ignore") \
.strip() \
.split(MESSAGE_SEGMENT_END_BYTE)
location = None
date_time = None
observation_types = []
observation_type = None
for seg in segments:
fields = seg.split('|')
if fields[0] == "PV1":
try:
location = fields[3]
except IndexError:
pass
if fields[0] == "OBR":
try:
date_time = mktime(strptime(fields[7], "%Y%m%d%H%M%S"))
except IndexError:
pass
if fields[0] == "OBX":
try:
observation_types.append(fields[13])
except IndexError:
observation_types.append(None)
observation_type_set = set(observation_types)
if len(observation_type_set)==1 and \
"APERIODIC" in observation_type_set:
observation_type = "aperiodic"
if len(observation_type_set)==1 and \
None in observation_type_set:
observation_type = "default"
return {
"assigned_patient_location": location,
"observation_date_time": date_time,
"observation_type": observation_type,
"message": message
}
def down_sample(self, parsed_records, start_time, end_time):
dt = [rec["observation_date_time"] for rec in parsed_records]
records = []
tmp = {}
sorted_records = sorted(parsed_records,
key=lambda rec: rec["observation_date_time"])
for rec in sorted_records:
date_time = rec["observation_date_time"]
location = rec["assigned_patient_location"]
observation_type = rec["observation_type"]
message = rec["message"]
if date_time<start_time or \
date_time>=end_time:
continue
tmp[location] = tmp.get(location, {})
tmp[location][observation_type] = rec
for d in tmp.values():
for rec in d.values():
records.append(rec)
return records
def log(self, msg):
self.producer.send( self.log_topic,
self.LOG_FORMAT.format( datetime.now().timestamp(),
msg
) \
.encode()
)
def get_end_time(self, current_time):
interval = self.INTERVAL
return current_time - current_time % interval + interval
async def send_or_ignore_message(self, addr, messages):
writer = self.connections[addr]
try:
for msg in messages:
writer.write(msg)
await writer.drain()
except ConnectionResetError:
"""
The error is not thrown reliably. If a connection is broken, and
one try to
writer.write(record)
await writer.drain()
This error may not manifest. It is thrown more often when one try
to repeatedly write to and drain a broken connection.
"""
self.remove_client(addr)
def refuse_client(self, addr, writer):
self.log("{} refused".format(addr))
writer.close()
def add_client(self, addr, writer):
if addr not in self.connections:
self.log("{} accepted".format(addr))
self.connections[addr] = writer
else:
self.remove_client(addr)
self.add_client(addr)
def remove_client(self, addr):
if addr in self.connections:
self.log("{} closed".format(addr))
writer = self.connections.pop(addr)
try:
writer.close()
except ConnectionResetError:
pass
def cleanup(self):
self.log("shutdown")
for addr in self.connections.keys():
self.remove_client(addr)
self.producer.flush()
self.producer.close()
from kafka.producer import KafkaProducer bootstrap_servers = ['localhost:9092'] topicName = 'myTopic' producer = KafkaProducer(bootstrap_servers=bootstrap_servers) producer = KafkaProducer() ack = producer.send(topicName, b'Hello World!!!!!!!!') metadata = ack.get() print(metadata.topic) print(metadata.partition)
encoding='utf8') as s3_taxi_data:
read_taxi = csv.reader(s3_taxi_data, delimiter=',')
next(read_taxi, None)
for line in read_taxi:
# data cleanup; to reduce the string length of taxi id
key_string = line[1][:10]
line_string = ''
# to skip invalid data
if not line[0] or not line[1] or not line[2] or not line[
3] or not line[4]:
continue
for i in range(5):
if i == 0 or i == 1:
line[i] = line[i][:10]
# to change the time format to be recoganized in KSQL
if i == 2 or i == 3:
line[i] = pd.to_datetime(line[i])
line[i] = line[i].strftime("%Y-%m-%d-%H:%M")
if i == (4):
delimiter = ''
else:
delimiter = ','
line_string = line_string + ''.join(line[i]) + delimiter
print(key_string)
print(line_string)
producer.send('topic_fatigue', value=line_string, key=key_string)
# optional; to control ingestion rate
# time.sleep(1)
# Get observations from AoT
observations = client.list_observations(filters=f)
# Iterate through records
try:
for page in observations:
print(f'Page {page_num}')
# data_stream = []
for obs in page.data:
ts = ciso8601.parse_datetime(obs["timestamp"])
prev_record_timestamp = obs["timestamp"]
data_stream = {
'ts': int(time.mktime(ts.timetuple())),\
'node_id': obs["node_vsn"],\
'sensor_path': obs["sensor_path"],\
'value_hrf': obs["value"]\
}
producer.send(topic, value=data_stream)
# Block until all the messages have been sent
producer.flush()
page_num += 1
except (Exception, HTTPError) as error:
print(error)
finally:
# Write latest processed timestamp to the file
fh = open("state.txt", "w+")
fh.write(prev_record_timestamp)
print(prev_record_timestamp)
fh.close()
import time
import json
import boto3
from kafka.producer import KafkaProducer
if __name__ == '__main__':
s3 = boto3.client('s3')
producer = KafkaProducer(bootstrap_servers="127.0.0.1:9092")
obj = s3.get_object(
Bucket='nyctaxitrip',
Key="{}/{}".format('yellow_trip',
'yellow_tripdata_sample.csv')) # read s3 csv
lines = str(obj['Body'].read())
for line in lines.split("\\n"):
print(json.dumps(line))
producer.send(
"new_topic",
value=line.encode(
), # encode the value to enable kafka consumer to recieve the stream msg
key=b'key'
) # encode the key to enable kafka consumer to recieve the stream msg
time.sleep(0.1)
# coding=utf-8
import logging
from kafka.producer import KafkaProducer
if __name__ == "__main__":
logging.basicConfig(level=logging.DEBUG)
producer = KafkaProducer(bootstrap_servers=["192.168.120.90:9092"])
producer.send("wangybnet", b"Hello, World!")
from kafka.producer import KafkaProducer
import ConfigParser
import socket
if __name__ == "__main__":
config = ConfigParser.ConfigParser()
config.read('configuration.cfg')
urlKafkaProducer = config.get('StreamingProperties', 'URLKafkaProducer')
topicName = config.get('StreamingProperties', 'TopicName')
virtualMachine = 'local'
if socket.gethostname() == 'ubuntu':
virtualMachine = socket.gethostname()
if virtualMachine == 'local':
fileName = config.get('StreamingProperties', 'StreamingFileLocal')
else:
fileName = config.get('StreamingProperties', 'StreamingFileVirtual')
producer = KafkaProducer(bootstrap_servers=urlKafkaProducer)
infile = open (fileName, 'r')
for line in infile:
producer.send (topicName, line)
#time.sleep(0.000000001)
infile.close()
class IBUSStreamingDownsamplingConsumer:
LOG_FORMAT ="{} UTC_TS\t"\
"{}"
def __init__(self, kafkaHost, kafkaPort, tcpHost, tcpPort, group_id, topic,
logTopic, interval):
self.kafkaHost = kafkaHost
self.kafkaPort = kafkaPort
self.tcpHost = tcpHost
self.tcpPort = tcpPort
self.group_id = group_id
self.topic = topic
self.logTopic = logTopic
self.interval = int(interval)
self.consumer = KafkaConsumer(
topic,
bootstrap_servers=["{}:{}".format(kafkaHost, kafkaPort)],
group_id=group_id,
enable_auto_commit=False)
self.producer = KafkaProducer(
bootstrap_servers=["{}:{}".format(kafkaHost, kafkaPort)])
self.tcpWriter = None
def getTopicPartitions(self):
self.consumer.topics() #This ensures local cache is updated with
# information about partitions, offsets etc.
pids = self.consumer.partitions_for_topic(self.topic)
tps = [TopicPartition(self.topic, pid) for pid in pids]
return tps
def getTopicPartitionsCommittedPositions(self):
tps = self.getTopicPartitions()
ret = [(tp, self.consumer.committed(tp)) for tp in tps]
return ret
async def tcp_server_handler(self, reader, writer):
addr = str(writer.get_extra_info("socket").getpeername())
if self.tcpWriter is not None:
self.log("refused " + addr)
writer.write(b"Connection limit reached; connection refused.")
writer.close()
return
self.log("accepted " + addr)
self.tcpWriter = writer
t1 = asyncio.create_task(self.poll_from_Kafka(writer))
try:
while True:
data = await reader.read(1) # 1024*16 bytes
if not data:
break
except BrokenPipeError:
"""
Catches connecton reset by peer when we are sending the batched data,
which is also when we cannot check for reader. The broken connection
on the writer side will ultimately lead to BrokenPipeError on the
reader side. Hence
"""
pass
finally:
t1.cancel()
self.log("closed " + addr)
writer.close()
self.tcpWriter = None
async def poll_from_Kafka(self, writer):
while True:
prevPos = self.getTopicPartitionsCommittedPositions()
polled = self.consumer.poll(timeout_ms=1000)
records = [
record.value for recordList in polled.values()
for record in recordList
]
try:
for record in records:
writer.write(record)
await writer.drain()
except ConnectionResetError:
"""
The error is not thrown reliably. If a connection is broken, and
one try to
writer.write(record)
await writer.drain()
This error may not manifest. It is thrown more often when one try
to repeatedly write to and drain a broken connection.
"""
print("Last batch not fully sent, not commited.")
for tp, pos in prevPos:
self.consumer.seek(tp, pos)
break
else:
self.consumer.commit()
await asyncio.sleep(self.interval)
def log(self, msg):
self.producer.send( self.logTopic,
self.LOG_FORMAT.format( datetime.now().timestamp(),
msg
) \
.encode()
)
def cleanup(self):
self.log("shutdown")
self.consumer.close()
self.producer.flush()
self.producer.close()
def run(self):
self.log("running")
asyncio.run(self._async_run())
async def _async_run(self):
tcpServer = await asyncio.start_server(self.tcp_server_handler,
self.tcpHost, self.tcpPort)
await tcpServer.serve_forever()
DataId = self.increment
MeterCode = random.choice(self.station_list())
TransactionId = self.increment
TransactionDateTime = datetime.now(pacific_time).strftime(fmt)
Amount = random.choice([0.25, 0.5, 1, 1.5, 2, 3])
PaymentMean = random.choice(['CREDIT CARD', 'PHONE', 'CASH'])
max_duration_sec = 60 * 60 * 6
PaidDuration = random.randint(1, max_duration_sec)
ElementKey = MeterCode
record = [
DataId, MeterCode, TransactionId, TransactionDateTime, Amount, '',
PaymentMean, PaidDuration, ElementKey, '2019', '4', ''
]
data_send = ",".join(map(str, record))
self.increment += 1
print(data_send)
key = str(MeterCode).encode()
value = data_send.encode()
return key, value
limit = 0
datagen = GenerateData()
while True:
producer.send('paid-transaction', *datagen.run())
limit += 1
time.sleep(.1)
from time import sleep
from kafka.producer import KafkaProducer
import json
producer = KafkaProducer(
# set host and port that producer should contact to bootstrap initial cluster metadata
bootstrap_servers=['localhost:9092'],
# how data should be serialized before sending to broker (convert the data to a json file and encode it to utf-8)
value_serializer=lambda x: json.dumps(x).encode('utf-8'))
for e in range(1000):
data = {
'number': e
} # key:value pairs to send (nb: this is not the topic key). Use a key for hashed-partitioning
future = producer.send('numtest', value=data)
# How to make sure the message is received by the broker?
print(f'sent {data}')
sleep(5) # option 1: take a break
# result = future.get(timeout=60) # option 2: block until a single message is sent (or timeout)
# option 3: Block until all pending messages are at least put on the network. This does not guarantee delivery
# or success! It is really. Only useful if you configure internal batching using linger_ms
# producer.flush()
import sys
import time
import json
import boto3
import lazyreader
#import helpers
from kafka.producer import KafkaProducer
producer = KafkaProducer(bootstrap_servers=['localhost:9092'])
while True:
s3 = boto3.client('s3')
obj = s3.get_object(Bucket='nyctaxi-trip-data',
Key="{}/{}".format('test_data', 'test1.txt'))
for line in lazyreader.lazyread(obj['Body'], delimiter='\n'):
#message_info = line.strip()
#msg = helpers.map_schema(message_info, self.schema)
# data = {'number' : line}
producer.send('read_s3', value=line)
time.sleep(0.1)
