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 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 __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 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'})
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)
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 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 setup_analytics():
global default_client, development
write_key = getattr(settings, 'ANALYTICS_WRITE_KEY', '')
host = getattr(settings, 'ANALYTICS_HOST', 'https://api.segment.io')
production = getattr(settings, 'ENVIRONMENT_NAME', None) == 'production'
send = not (development
or getattr(settings, 'TESTING', False)) or production
if not default_client:
default_client = Client(write_key=write_key,
host=host,
debug=False,
on_error=handle_analytics_error,
send=send)
def setup_analytics():
global default_client, development
write_key = getattr(settings, "ANALYTICS_WRITE_KEY", "")
host = getattr(settings, "ANALYTICS_HOST", "https://api.segment.io")
production = getattr(settings, "ENVIRONMENT_NAME", None) == "production"
send = not (development or getattr(settings, "TESTING", False)) or production
if not default_client:
default_client = Client(
write_key=write_key,
host=host,
debug=False,
on_error=handle_analytics_error,
send=send,
)
def __init__(self):
self.segment = Client(ENV["segment"]["write_key"],
debug=True, on_error=self.on_error, send=True, max_queue_size=100000)
'''
import os
import analytics
import pandas as pd
from analytics import Client
from google.cloud import automl_v1beta1
import stacklogging
logger = stacklogging.getLogger(__name__)
MODEL_NAME = "trial_prediction_model"
MODEL_VERSION = "v6"
segment_client = Client(os.getenv("SEGMENT_WRITE_KEY"), max_queue_size=100000)
automl_client = automl_v1beta1.TablesClient(project="buffer-data", region="us-central1")
# Run the batch predict
logger.info("Running AutoML Batch predict")
operation = automl_client.batch_predict(
bigquery_input_uri="bq://buffer-data.dbt_buffer.predict_publish_trial_conversion_prediction",
gcs_output_uri_prefix="gs://automl-predictions/",
model_display_name=f"{MODEL_NAME}_{MODEL_VERSION}",
)
operation.result()
logger.info("AutoML Batch prediction finished")
predicted_at = str(pd.Timestamp.now())
gcs_directory = (
operation.metadata.batch_predict_details.output_info.gcs_output_directory
import config
from analytics import Client
segment_client = Client(config.SEGMENT_API_KEY, sync_mode=True)
def _annotation_event(user, action, collection, document, label_type_id, *args,
**kwargs):
properties = {
'collection': collection,
'document': document,
'label_type_id': label_type_id,
}
segment_client.track(user_id=user, event=action, properties=properties)
def _login_event(user, *args, **kwargs):
segment_client.identify(user_id=user)
event_mapper = {'login': _login_event}
class AuditLog:
def log_event(user, action, *args, **kwargs):
if action in event_mapper:
event_mapper[action](user=user, action=action, *args, **kwargs)
else:
_annotation_event(user=user, action=action, *args, **kwargs)
def on_error(error, items):
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
