​
Overview

This guide walks through how to leverage the transformation script to send data from a source directly to your API during a hotglue job. This should be used instead of a target. We will be using the Python requests library to do this.

​
Reasoning

We recommend using the Python transformation script to connect with your API for more flexibility. You can:
  • handle authorization (for example, you can read/write api keys for your tenants using the tenant config)
  • do any mapping necessary to your internal schema
  • manage the order requests are made in
  • and more!

​
Setup your environment

​
Dependencies

The script we will walk through is tested with the following requirements.txt – if you have a different desired behavior, verify your dependency versions match.
requirements.txt
gluestick==1.0.9
numpy==1.16.2
pandas==0.25.3
requests==2.26.0

​
Sample data

For this example we will be working with Mailchimp list members test data. Download the CSV here to follow along.
list_members.csv
id,web_id,email_address,unique_email_id,email_type,status,unsubscribe_reason,merge_fields,stats,ip_signup,timestamp_signup,ip_opt,timestamp_opt,member_rating,last_changed,language,vip,email_client,location,source,tags_count,tags,list_id
030d546fad14e050174131b92f9dfa4a,532598831,tony@starkindustries.com,41e3a84f8c,html,subscribed,,"{'FNAME': 'Tony', 'LNAME': 'Stark', 'ADDRESS': {'addr1': '', 'addr2': '', 'city': '', 'state': '', 'zip': '', 'country': 'US'}, 'PHONE': '', 'BIRTHDAY': ''}","{'avg_open_rate': 0.0, 'avg_click_rate': 0.0}",,,161.97.109.172,2022-05-16T15:45:03.000000Z,2.0,2022-05-16T15:45:03.000000Z,,False,,"{'latitude': 0.0, 'longitude': 0.0, 'gmtoff': 0, 'dstoff': 0, 'country_code': '', 'timezone': ''}",Admin Add,0,[],3b1252e2c4
f9c278107a89812816ae8769c969bf3c,533108879,fury@avengers.org,211fb11bbb,html,unsubscribed,N/A (Unsubscribed by admin),"{'FNAME': 'Director', 'LNAME': 'Fury', 'ADDRESS': '', 'PHONE': '', 'BIRTHDAY': ''}","{'avg_open_rate': 0.0, 'avg_click_rate': 0.0}",,,188.43.235.177,2022-05-17T08:59:40.000000Z,1.0,2022-05-20T10:09:19.000000Z,,False,,"{'latitude': 0.0, 'longitude': 0.0, 'gmtoff': 0, 'dstoff': 0, 'country_code': '', 'timezone': ''}",Import,1,"[{'id': 12632599, 'name': 'promo'}]",3b1252e2c4

​
Sample Script

​
Prepare the data

Refer to the sample script below which:
  • Reads the list_members CSV as pandas dataframe, and
  • Does some light transformation on the data
etl.py
# Import dependencies
import gluestick as gs
import pandas as pd
import os
import requests

# Establish standard directories for hotglue
ROOT_DIR = os.environ.get("ROOT_DIR", ".")
INPUT_DIR = f"{ROOT_DIR}/sync-output"
OUTPUT_DIR = f"{ROOT_DIR}/etl-output"

# Read input data
input_data = gs.read_csv_folder(INPUT_DIR)

# Get the list members as a dataframe
list_members = input_data.get("list_members", pd.DataFrame())

# Explode personalized data
list_members = gs.explode_json_to_rows(list_members, "merge_fields")

# Select relevant columns
list_members = list_members[['id', 'email_address', 'status', 'source', 'merge_fields.FNAME', 'merge_fields.LNAME', 'last_changed']]

# Rename columns to standard names
list_members.rename(columns={
    "merge_fields.FNAME": "firstName",
    "merge_fields.LNAME": "lastName"
}, inplace=True)

​
Send to the API

Now that the data has been mapped to our schema, we can create a payload and POST it to our API. If you’d like to learn more about the requests library syntax, visit their docs. Refer to the sample script below:
etl.py
# Create a payload that includes the contacts with some context on the hotglue job
payload = {
    "env_id": os.environ.get("ENV_ID"),
    "job_id": os.environ.get("JOB_ID"),
    "flow_id": os.environ.get("FLOW"),
    "tenant": os.environ.get("TENANT"),
    "tap": os.environ.get("TAP", "mailchimp"),
    "contacts": list_members.to_dict(orient="records")
}

# Send to API endpoint
endpoint_url = "https://api.acme.com/contacts"
r = requests.post(endpoint_url, json=payload)

# Log the result of the request
print(r.status_code)