class SegmentEventService():
def __init__(self):
self.segment = Client(ENV["segment"]["write_key"],
debug=True, on_error=self.on_error, send=True, max_queue_size=100000)
'''
analytics.write_key = ENV["segment"]["write_key"]
analytics.debug = True
analytics.on_error = self.on_error
'''
def on_error(self, error, items):
print("An error occurred:", error)
#
# insert an event log
#
def insert_event(self, event_log):
try:
#print "%s did event %s " % (event_log["anonymousId"], event_log["event"])
#analytics.track(event_log["userId"], event_log["event"], {},
self.segment.track(event_log["userId"], event_log["event"], {},
event_log["context"], event_log["timestamp"], event_log["anonymousId"],
event_log["integrations"])
except Exception, e:
log.exception(e)
def upload_to_segment():
dir_path = dirname(realpath(__file__))
audit_file_path = join(dir_path, "auditing")
client = Client(config.SEGMENT_API_KEY)
for txt_file in [
join(audit_file_path, filename)
for filename in listdir(audit_file_path)
if filename.endswith(".txt")
]:
renamed_to = txt_file.replace(".txt", ".toupload")
rename(txt_file, renamed_to)
with open(renamed_to, "rb") as output_file:
for line in output_file.readlines():
properties = json.loads(line)
user = properties.pop("user")
event = properties.pop("event")
timestamp = parse(properties.pop("timestamp"))
kwargs = {
"user_id": user,
"event": event,
"timestamp": timestamp,
"properties": properties,
}
client.track(**kwargs)
client.flush()
afterwards_to = renamed_to.replace(
".toupload",
datetime.now().strftime(".%Y%m%d%H%M"))
rename(renamed_to, afterwards_to)
def track():
write_key = request.args.get('writeKey')
params = request.get_json(True)
analytics = Client(write_key)
analytics.track(params['userId'], params['event'], params['properties'])
analytics.flush()
return json.dumps({'status': 'success'})
def track():
write_key = request.args.get('writeKey')
params = request.get_json(True)
analytics = Client(write_key)
analytics.track(params['userId'], params['event'], params['properties'])
analytics.flush()
return json.dumps({'status': 'success'})
class SegmentClient:
"""Implements base Segment client that can be extended for domain-specific
use-cases."""
def __init__(self, write_key: str):
is_local = not in_gcp()
# When `send` is set to False, we do not send any logs to Segment.
# We also set `debug` to True locally for more logging during development.
self.client = Client(
write_key,
send=(not is_local),
debug=is_local,
)
def track(self, user_id: Optional[str], event_name: str,
metadata: Dict[str, Any]) -> None:
if user_id:
self.client.track(user_id, event_name, metadata)
# Load dataframe in memory
logger.info("Reading data from GCS")
df = pd.read_csv(f"{gcs_directory}/tables_1.csv")
logger.info(f"Loaded {df.shape[0]} rows")
# Remove rows with empty account IDs
logger.info(f"Removing {df['account_id'].isna().sum()} rows with no account_id")
clean_df = df[~df["account_id"].isna()]
# Generate the proper schema and send it to Segment
logger.info("Sending data to Segment")
for index, row in clean_df.iterrows():
properties = {
"product": row["product"],
"productUserId": row["user_id"],
"subscriptionId": row["subscription_id"],
"trialId": row["trial_id"],
"planId": row["plan_id"],
"modelName": MODEL_NAME,
"modelVersion": MODEL_VERSION,
"predictedAt": predicted_at,
"score": row["converted_true_score"],
}
segment_client.track(row["account_id"], "Trial Conversion Predicted", properties)
# Flush before finishing the script
logger.info("Flushing queue")
segment_client.flush()
print("An error occurred:", error)
writeKey = os.getenv('writeKey', None)
events = os.getenv('events', 500000)
assert writeKey is not None, "Please configure a write key using the writeKey environment variable"
print(f'Sending {events} events to write key "{writeKey}"')
analytics = Client(writeKey,
debug=True,
on_error=on_error,
max_queue_size=math.inf,
upload_size=math.inf,
upload_interval=1)
start = time.time()
for it in range(events):
sys.stdout.write(f"\rProgress: {round(it / events * 100, 1)}%")
analytics.track('test', f'Iterated-{it}', {'plan': it})
print()
print('Shutting down..')
analytics.shutdown()
elapsed = time.time() - start
print(f'elapsed: {elapsed} seconds')
print(f'{round(events / elapsed)}rps average')
