Votre première lambda en direct !

Données en continu avec AWS Kinesis et Lambda

Maksim Pecherskiy

Data Engineer

Leçon précédente

Cette leçon

Format de journal courant

bc75da5f-1bf6-444c-80ad-49c180e1b8de 23:16:06.000 3FTEX1G5XAK844393 -76.6990172 2.481207 40
ff8e7131-408d-463b-8d07-d016419b0656 20:26:44.000 2LAXX1C8XAK844292 114.39239199999999 36.097577 90
f29a5b3d-d0fa-43c0-9e1a-e2a5cdb8be7a 8:10:47.000 3FTEX1G5XAK844393 108.58068100000001 34.79925 37

Lecture de fichiers S3

import pandas as pd

résulte en :

Response:
{
  "errorMessage": "Unable to import module 'lambda_function': No module named 'pandas'",
  "errorType": "Runtime.ImportModuleError"
}

Mise à jour du gestionnaire

#lambda_function.py
import json, boto3, pandas as pd
# Initialize clients
...

SPEED_ALERT_THRESHOLD = 45
ALERT_PHONE_NUMBER = "+1234567890"

# Helper function to get dataframe from written records
def get_new_data(event):
    pd.read_csv()...
    ...
    return data

# Lambda function handler
def record_created_handler(event, context):
    data = get_new_data(event)
    ...
    sns.publish()
    ...

get_new_data()

def get_new_data(event):
    # Create a list to store new object keys.
    written_objects = []

    # Iterate over each S3 event record.
    for record in event['Records']:

        # Get the variables to check for
        event_name = record['eventName']
        bucket_name = record['s3']['bucket']['name']
        obj_key = record['s3']['object']['key']

get_new_data()

def get_new_data(event):
        ...
        # Verify that event is created from sd-vehicle-data bucket.
        if event_name == 'ObjectCreated:Put' and bucket_name == 'sd-vehicle-data':

            obj = s3.get_object(Bucket=bucket_name, Key = obj_key)
            df = pd.read_csv(obj['Body'], delimiter = " ", 
                   names=["record_id", "timestamp", "vin", "lon", "lat", "speed"])

            written_objects.append(df)

       # Concatenate new records into a single dataframe.
       return pd.concat(written_objects)

record_created_handler()

SPEED_ALERT_THRESHOLD = 45
ALERT_PHONE_NUMBER = "+1234567890"
...
def record_created_handler(event, context):
    # Call the helper method
    data = get_new_data(event)

    ## Get the top speeds
    top_speeds = data.groupby(['vin'])['speed'].max().reset_index()

    ## Get top speeds that exceed the limit of 45
    too_fast = top_speeds.loc[top_speeds.speed > SPEED_ALERT_THRESHOLD, :]

record_created_handler()

SPEED_ALERT_THRESHOLD = 45
ALERT_PHONE_NUMBER = "+1234567890"
...

def record_created_handler(event, context):
    ...
    ## Send SMS
    sns.publish(PhoneNumber=ALERT_PHONE_NUMBER,
        Message="Speeding Alert \n" + too_fast.to_string())

    ## This doesn't go anywhere yet, but we need to return something.
    totals = data.groupby(['vin'])['speed'].max().reset_index()
    return totals.to_csv(sep=" ", index=False)

Tester la fonction Lambda

Ajout de variables d'environnement

import os
os.environ.get("ENV_VARIABLE_NAME", "DEFAULT_VALUE")

import os
SPEED_ALERT_THRESHOLD = os.environ.get("SPEED_ALERT_THRESHOLD", 45)
ALERT_PHONE_NUMBER = os.environ.get("ALERT_PHONE_NUMBER", None)
...

def record_created_handler(event, context):
    ...

Votre première lambda en direct !

Leçon précédente

Cette leçon

Format de journal courant

Lecture de fichiers S3

Mise à jour du gestionnaire

get_new_data()

get_new_data()

record_created_handler()

record_created_handler()

Tester la fonction Lambda

Ajout de variables d'environnement

Ajout de variables d'environnement

Ajout de variables d'environnement

Ajout d'un déclencheur

Ajout d'un déclencheur

Ajout d'un déclencheur

Révision

Révision

Révision

Révision

Passons à la pratique !