async def test_predict_ce_avro_binary(self, http_server_client):
schema = avro.schema.parse(test_avsc_schema)
msg = {"name": "foo", "favorite_number": 1, "favorite_color": "pink"}
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(msg, encoder)
data = bytes_writer.getvalue()
event = dummy_cloud_event(data, set_contenttype=True)
# Creates the HTTP request representation of the CloudEvent in binary content mode
headers, body = to_binary(event)
resp = await http_server_client.fetch('/v1/models/TestModel:predict',
method="POST",
headers=headers,
body=body)
assert resp.code == 200
assert resp.headers['content-type'] == "application/json"
assert resp.headers['ce-specversion'] == "1.0"
assert resp.headers["ce-id"] != "36077800-0c23-4f38-a0b4-01f4369f670a"
assert resp.headers['ce-source'] == "io.kserve.kfserver.TestModel"
assert resp.headers['ce-type'] == "io.kserve.inference.response"
assert resp.headers['ce-time'] > "2021-01-28T21:04:43.144141+00:00"
assert resp.body == b'{"predictions": [["foo", 1, "pink"]]}'
def sendPackageTimeout(self, accountId):
message = {
"accountId": accountId,
"host": None,
"item": None,
"severity": "ERROR",
"description": "account %s workflow timeout" % accountId
}
all = {
"timestamp": 1L,
"src": "rundeck",
"host_ip": "10.74.113.101",
"rawdata": json.dumps(message)
}
schema = avro.schema.parse(avro_schema)
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(all, encoder)
try:
self.producer.send_messages(b"%s" % self.zabbix_alert,
bytes_writer.getvalue())
logger.info("send to zabbix sa successfully")
except:
logger.error(
"occur error when send package timeout message to zabbix alert topic"
)
def senddata():
producer = KafkaProducer(bootstrap_servers="localhost:9092")
schema = avro.schema.parse(open("schema/flood.json").read())
data = json.load(open("data/sample.json", "r"))
for item in data:
a_item = {
"construction": str(item.get("construction")),
"county": str(item.get("county")),
"eq_site_deductible": item.get("eq_site_deductible"),
"eq_site_limit": item.get("eq_site_limit"),
"fl_site_deductible": item.get("fl_site_deductible"),
"fl_site_limit": item.get("fl_site_limit"),
"fr_site_deductible": item.get("fr_site_deductible"),
"fr_site_limit": item.get("fr_site_limit"),
"hu_site_deductible": item.get("hu_site_deductible"),
"hu_site_limit": item.get("hu_site_limit"),
"line": str(item.get("line")),
"point_granularity": item.get("point_granularity"),
"point_latitude": item.get("point_latitude"),
"point_longitude": item.get("point_longitude"),
"policyID": item.get("policyID"),
"statecode": str(item.get("statecode")),
"tiv_2011": item.get("tiv_2011"),
"tiv_2012": item.get("tiv_2012")
}
# defines encoding format
writer = avro.io.DatumWriter(schema)
bytes_writes = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writes)
writer.write(a_item, encoder)
raw_bytes = bytes_writes.getvalue()
producer.send("floodinfo", raw_bytes)
producer.flush()
def deserialize(self, test_out, exp_out):
logger = logging.getLogger()
if self.sink["serialize"]:
if not isinstance(exp_out, str):
raise TypeError("'exp_out' must be of type 'str'")
if not isinstance(test_out, bytes):
raise TypeError("'test_out' must be of type 'bytes'")
if self.sink["type"] == "Avro":
py_obj = json.loads(exp_out)
self.tc_drv.store_exp_one(py_obj)
value = bytearray(test_out)
bytes_reader = io.BytesIO(value)
decoder = avro.io.BinaryDecoder(bytes_reader)
py_obj = self.sink["avro_reader"].read(decoder)
self.tc_drv.store_rx_one(py_obj)
return py_obj
elif self.sink["type"] == "Binary":
self.tc_drv.store_exp_one(exp_out)
self.tc_drv.store_rx_one(test_out)
return test_out.decode("utf-8")
else:
self.tc_drv.store_exp_one(exp_out)
return test_out
def handle_avro_client_print_to_file(connection, address):
schema = avro.schema.Parse(open("schema/addressbook.avsc", "rb").read())
data = connection.recv(4)
message_length, = struct.unpack('>I', data)
message = connection.recv(message_length)
message_buf = io.BytesIO(message)
reader = avro.datafile.DataFileReader(message_buf, avro.io.DatumReader())
# Create a data file using DataFileWriter
dataFile = open("schema/addressbook.avro", "wb")
writer = DataFileWriter(dataFile, DatumWriter(), schema)
for thing in reader:
writer.append(thing)
reader.close()
writer.close()
return (len(message))
def serialize_data(self, data, schema):
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(data, encoder)
return bytes_writer.getvalue()
def consumer(self, consumer_group="flume_test_group"):
simple_consumer = self.collect_topic.get_balanced_consumer(
reset_offset_on_start=False,
auto_commit_enable=True,
auto_commit_interval_ms=1000,
consumer_group=consumer_group,
consumer_timeout_ms=10000,
zookeeper_connect=ZOOKEEPER_HOST,
)
# simple_consumer = self.collect_topic.get_simple_consumer(
# reset_offset_on_start=False,
# auto_commit_enable=True,
# auto_commit_interval_ms=1000,
# consumer_group="flume_test_group",
# consumer_timeout_ms=1000,
# )
count = 0
consumer = []
for message in simple_consumer:
# print 'offset: %s' % message.offset, 'data: ' + message.value
bytes_msg = io.BytesIO(message.value[5:])
decode_msg = avro.io.BinaryDecoder(bytes_msg)
recode_msg = self.avro_reader.read(decode_msg)
# print message.offset, recode_msg
# simple_consumer.commit_offsets()
consumer.append(recode_msg)
count += 1
print count
return consumer
def convert(self, obj_map):
writer = avro.io.DatumWriter(self.schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(obj_map, encoder)
val = bytes_writer.getvalue()
return val
def _decodemsg(self,msg):
value = bytearray(msg.value)
bytes_reader = io.BytesIO(value[5:])
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(self.schema)
message = reader.read(decoder)
return message
def divoltecall():
# get configuration parameters from confguration file
conf = Configuration("default.yml")
host,username,password,dbName = conf.getMySQLDetails()
kafka_host,kafka_port = conf.getBrokerDetails()
topic,consumergroup = conf.getConsumerDetails()
schemaAvro = conf.getAvroSchema()
# Kafka Broker Configuration
broker_config=kafka_host+":"+str(kafka_port)
# To consume messages
consumer = KafkaConsumer(topic,
group_id=consumergroup,
bootstrap_servers=[broker_config])
# read Avro schema
schema = avro.schema.parse(open(schemaAvro).read())
# Open database connection
db = MySQLdb.connect(host,username,password,dbName)
# prepare a cursor object using cursor() method
cursor = db.cursor()
for msg in consumer:
bytes_reader = io.BytesIO(msg.value)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
user1 = reader.read(decoder)
insertIntoDatabase(user1)
# disconnect from server
db.close()
def run(self):
ctx = ServiceContext()
config = ctx.getConfigService()
queue = ctx.getQueueService()
self.schema = avro.schema.parse(avro_schema)
constructor="KafkaConsumer(%s,group_id=%s,bootstrap_servers=%s)"
topics = config.get("Input Plugin: kafka_collector","kafka_topics")
group_id = config.get("Input Plugin: kafka_collector","kafka_groupid")
bootstrap_server = config.get("Message","kafka_broker")
str = constructor % (topics,group_id,bootstrap_server)
self.consumer = eval(str)
for msg in self.consumer:
value = bytearray(msg.value)
topic = msg.topic
bytes_reader = io.BytesIO(value[5:])
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(self.schema)
kafkamsg = reader.read(decoder)
try:
jsondata = json.loads(kafkamsg['rawdata'])
eventType = jsondata["eventName"]
jsondata['topic'] = topic
queue.put(EventFactory.getEvent(eventType,jsondata))
except InputError,e:
self.error(str(e))
except:
def avro_decode_message(self, message):
if message:
bytes_from_message = bytearray(message)
# Check ID for coherency
message_id = int.from_bytes(bytes_from_message[1:5], byteorder='big')
if self._schema_id != message_id:
logging.warning("Possible incoherence between message's id (%d) and schema's id (%d), for topic (%s)",
message_id, self._schema_id, self.topic)
# Remove 5-byte header the first byte is reserved for future, 4 bytes for 32 bit number indicating ID
message = bytes(bytes_from_message[5:])
# Parse the rest of the message using the schema
bytes_reader = io.BytesIO(message)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(self.avro_schema)
decoded_messages = []
# We iterate in case there are more than one messages
while bytes_reader.tell() < len(message):
try:
# Here is where the messages are read
decoded_messages.append(reader.read(decoder))
sys.stdout.flush()
except Exception as e:
logging.error(e)
return decoded_messages
def get_tweet(msg):
bytes_reader = io.BytesIO(msg)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
tweet = reader.read(decoder)
return tweet
def deserializeBinaryFromStream(schemaFile, binaryData):
bytes_reader = io.BytesIO(binaryData)
decoder = avro.io.BinaryDecoder(bytes_reader)
schema = parse_schema(schemaFile)
reader = avro.io.DatumReader(schema)
data = reader.read(decoder)
return data
def deserializeBinaryFromFile(schemaFile, inputFile):
bytes_reader = io.BytesIO(open(inputFile, "rb").read())
decoder = avro.io.BinaryDecoder(bytes_reader)
schema = parse_schema(schemaFile)
reader = avro.io.DatumReader(schema)
data = reader.read(decoder)
return data
def send_inventory(self, account_id, module, operation, result, data):
message = {
"accountId": account_id,
"module": module,
"operation": operation,
"result": result,
"data": data
}
all = {
"timestamp": 1L,
"src": "sa_inventory",
"host_ip": "10.74.113.101",
"rawdata": json.dumps(message)
}
schema = avro.schema.parse(avro_schema)
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(all, encoder)
try:
self.producer.send_messages(b"%s" % self.inventory_notify,
bytes_writer.getvalue())
logger.info("send to redis successfully")
except Exception as exp:
logger.error("occur error when send package inventory message to "
"sa inventory(dms.log.inventory) topic" + exp.message)
def create_avro_message(log_device, writer, id):
"""Create message bytes using Avro schema"""
# Initialise with magic byte = 0 and 4 byte schema id
# TODO use id rather than hardcoding id
kafka_magic = io.BytesIO(b'\x00\x00\x00\x00\x01')
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(
{
"name": log_device.get_name(),
"value": log_device.get_value(),
"time": log_device.get_time(),
"datetime": str(datetime.datetime.utcnow()).split('.')[0]
}, encoder)
return kafka_magic.getvalue() + bytes_writer.getvalue()
def _parserequest(self, request):
schema = avro.schema.parse(test_avsc_schema)
raw_bytes = request
bytes_reader = io.BytesIO(raw_bytes)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
record1 = reader.read(decoder)
return record1
def commonToAvroBinarySchema(schema, dictContent):
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(dictContent, encoder)
raw_bytes = bytes_writer.getvalue()
b = bytearray()
b.extend(raw_bytes)
return b
def writeMessageWithId(schema_name, message):
schema = avro.schema.Parse(schema_cache[schema_name])
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
bytes_writer.write(bytes([0]))
bytes_writer.write(id_cache[schema_name].to_bytes(4, byteorder="big"))
writer.write(message, encoder)
return bytes_writer.getvalue()
def retrainClassifier(session, db_config):
'''
Gathers available labeled faces from the database for personas, sub-personas to train a new keras classifier
'''
#Get table names from the config
(keyspace, personaTableName, subPersonaTableName, subPersonaFaceEdgeTableName, faceSubPersonaEdgeTableName, rawImageTableName, faceImageTableName) = getTables(db_config)
#Grab the list of personas to retrieve their labels
persona_list = list(session.execute("SELECT persona_name FROM " + personaTableName))
persona_list = list(map(lambda x: x.persona_name, persona_list))
features_list = []
labels_list = []
classes = len(persona_list)
logging.info("Found {0} personas".format(len(persona_list)))
schema = avro.schema.Parse(open("./VGGFaceFeatures.avsc", "r").read())
for persona in persona_list:
image_id_list = list(session.execute("SELECT sub_persona_name, assoc_face_id, label_v_predict_assoc_flag FROM {0} WHERE sub_persona_name='{1}'".format(subPersonaFaceEdgeTableName, persona),timeout=60))
logging.info("{0} features retrieved for {1}".format(len(image_id_list), persona))
for image_id in image_id_list:
image_features = None
while image_features is None:
try:
image_features = list(session.execute("SELECT face_id, face_bytes, feature_bytes FROM {0} WHERE face_id='{1}'".format(faceImageTableName, image_id.assoc_face_id)))
except:
time.sleep(60)
pass
for image_byte in image_features:
bytes_reader = io.BytesIO(image_byte.feature_bytes)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
features = reader.read(decoder)
features_list.append(features['features'])
labels_list.append(persona)
#time.sleep(60)
#Convert the persona labels into integers for keras, produce reversal dictionary
homogenized_label_list = list(map(lambda x: persona_list.index(x), labels_list))
label_persona_dictionary = dict(map(lambda x: (persona_list.index(x), x), persona_list))
#label_persona_dictionary = dict(zip(homogenized_label_list, persona_list))
logging.info("Generated conversion dictionary")
logging.info(label_persona_dictionary)
model = Sequential()
model.add(Dense(1024, input_dim=512, activation='relu')) #we add dense layers so that the model can learn more complex functions and classify for better results.
model.add(Dense(1024,activation='relu')) #dense layer 2
model.add(Dense(512,activation='relu')) #dense layer 3
model.add(Dense(1,activation='softmax')) #final layer with softmax activation
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
model.fit(x=np.array(features_list), y=np.array(homogenized_label_list), epochs=2, batch_size=1)
scores = model.evaluate(np.array(features_list), np.array(homogenized_label_list))
logging.info("Generated model")
return (model, label_persona_dictionary)
def avro_encoder(schema, value: dict):
"""
Encode dictionary to avro format with designated schema
"""
writer = avro.io.DatumWriter(schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(value, encoder)
raw_bytes = bytes_writer.getvalue()
return raw_bytes
def getBlobPayload(self, blob, block_blob_service):
container_name = self.getPipelineContainer()
blob_bytes = block_blob_service.get_blob_to_bytes(
container_name, blob.name)
avro_content = blob_bytes.content
last_modified = blob_bytes.properties.last_modified.isoformat()
content_bytes = io.BytesIO(avro_content)
all_payloads = avro.datafile.DataFileReader(content_bytes,
avro.io.DatumReader())
return all_payloads, last_modified
def avro_encode_messages(self, json_messages):
bytes_writer = io.BytesIO()
writer = avro.io.DatumWriter(self.avro_schema)
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(json_messages, encoder)
raw_bytes = bytes_writer.getvalue()
# Add 5-byte header the first byte is reserved for future, 4 bytes for 32 bit number indicating ID
return bytes(bytearray(b'\x00') +
bytearray(self._schema_id.to_bytes(4, byteorder='big')) +
bytearray(raw_bytes))
def handle_avro_client(connection):
message = connection.recv()
message_buf = io.BytesIO(message)
reader = avro.datafile.DataFileReader(message_buf, avro.io.DatumReader())
for thing in reader:
print(thing)
reader.close()
return (len(message))
def predictImage(image_features):
schema = avro.schema.Parse(open("./VGGFaceFeatures.avsc", "rb").read())
bytes_reader = io.BytesIO(image_features.feature_bytes)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
features = reader.read(decoder)
#
prediction = model.predict(features)
logging.info(prediction)
return prediction
def encode(self, item, writers_schema=None):
"""Returns encoded data
- ``item``: item to be encoded according to schma
- ``writers_schema``: avro writers schema
"""
writer = avro.io.DatumWriter(writers_schema)
bytes_writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(bytes_writer)
writer.write(item, encoder)
encoded = bytes_writer.getvalue()
return encoded
def deserializeBinaryFromStreamWithHeader(headerSchemaFile, dataSchema,
binaryData):
bytes_reader = io.BytesIO(binaryData)
decoder = avro.io.BinaryDecoder(bytes_reader)
headerSchema = parse_schema(headerSchemaFile)
dataSchema = parse_schema(dataSchema)
reader = avro.io.DatumReader(headerSchema)
header = reader.read(decoder)
datareader = avro.io.DatumReader(dataSchema)
data = datareader.read(decoder)
return {'header': header, 'data': data}
def serializeDataToBinaryFile(schemaFile, outputFile, dataToSerialize):
writer = io.BytesIO()
encoder = avro.io.BinaryEncoder(writer)
schema = parse_schema(schemaFile)
datum_writer = avro.io.DatumWriter(schema)
datum_writer.write(dataToSerialize, encoder)
raw_bytes = writer.getvalue()
newFile = open(outputFile, "wb")
newFile.write(raw_bytes)
newFile.close()
logging.debug("Binary data written to:" + outputFile)
def deserialize_msg(msg, serializer, schema=None):
if serializer == "Avro":
bytes_reader = io.BytesIO(msg.value)
decoder = avro.io.BinaryDecoder(bytes_reader)
reader = avro.io.DatumReader(schema)
msg_data = reader.read(decoder)
return_val = msg_data
elif serializer == "JSON":
return_val = json.loads(msg)
else:
return_val = msg
return return_val
