PySpark — Create Spark Data Frame from API

Subham Khandelwal ✅
2 min readOct 5, 2022


We all have been in situations where we have to read data from API and load the same in Spark Data Frames for further operations.

Following is a small snippet of code which reads data from API and generates a Spark Data Frame.

Lets create a Python function to read API data.

# Create Python function to read data from API
import requests, json
def read_api(url: str):
normalized_data = dict()
data = requests.get(api_url).json()
normalized_data["_data"] = data # Normalize payload to handle array situtations
return json.dumps(normalized_data)

Following code generates Spark Data Frame from the json payload of the API response

api_url = r""
# api_url = ""
# Read data into Data Frame
# Create payload rdd
payload = json.loads(read_api(api_url))
payload_rdd = spark.sparkContext.parallelize([payload])
# Read from JSON
df ="_data").printSchema()
Load data into Data Frame

Now in case you want to expand the root element of the data frame

# Expand root element to read Struct Data"_data.*").show(truncate=False)
Expand Struct Elements

If you want to expand further to reach to particular element(in our case say USD)

# Expand further elements to read USD data"_data.*").select("bpi.*").select("USD.*").show(truncate=False)
Expand further to read USD data

We will see to expand such data dynamically(flatten json data) in further posts.

Checkout the iPython notebook on Github —

Wish to Buy me a Coffee: Buy Subham a Coffee