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 alias():
write_key = request.args.get('writeKey')
params = request.get_json(True)
analytics = Client(write_key)
analytics.alias(params['previousId'], params['userId'])
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'})
def alias():
write_key = request.args.get('writeKey')
params = request.get_json(True)
analytics = Client(write_key)
analytics.alias(params['previousId'], params['userId'])
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'})
def identify():
write_key = request.args.get('writeKey')
params = request.get_json(True)
# Shouldn't have any concurrency issue since we are not multi-threading here
analytics = Client(write_key)
analytics.identify(params['userId'], params['traits'])
analytics.flush()
return json.dumps({'status': 'success'})
def identify():
write_key = request.args.get('writeKey')
params = request.get_json(True)
# Shouldn't have any concurrency issue since we are not multi-threading here
analytics = Client(write_key)
analytics.identify(params['userId'], params['traits'])
analytics.flush()
return json.dumps({'status': 'success'})
# 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()
