Schema validation in Apache Spark - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of schema validation with missing required field

Given the following Spark DataFrame schema and data, what will be the output when validating the data against the schema?

Apache Spark

from pyspark.sql import SparkSession
from pyspark.sql.types import StructType, StructField, StringType, IntegerType

spark = SparkSession.builder.master('local').appName('SchemaValidation').getOrCreate()

schema = StructType([
    StructField('name', StringType(), nullable=False),
    StructField('age', IntegerType(), nullable=False)
])

data = [('Alice', 30), (None, 25)]

df = spark.createDataFrame(data, schema=schema)
df.show()

A Throws an error during DataFrame creation due to null in non-nullable field

B Shows both rows; second row has null for 'name' without error

C Shows only the first row; second row is dropped silently

D Shows both rows; null in 'name' replaced by empty string

💡 Hint

Think about how Spark handles nullable fields when creating DataFrames.

Spark allows creating DataFrames with nulls in non-nullable fields but may cause errors later during operations like writing to certain formats. The show() method will display both rows including the null value.

❓ data_output

intermediate

Result of schema enforcement with cast

What is the output DataFrame when applying the schema with enforced types to the following data?

Apache Spark

from pyspark.sql import SparkSession
from pyspark.sql.types import StructType, StructField, IntegerType

spark = SparkSession.builder.master('local').appName('SchemaValidation').getOrCreate()

schema = StructType([
    StructField('id', IntegerType(), nullable=False)
])

data = [('1',), ('2',), ('three',)]

df = spark.createDataFrame(data, ['id'])
df_cast = df.selectExpr('cast(id as int) as id')
df_cast.show()

A

+----+
|  id|
+----+
|null|
|null|
|null|
+----+

B Throws a runtime error due to 'three' not castable to int

C

+----+
|  id|
+----+
|   1|
|   2|
|   3|
+----+

D

+----+
|  id|
+----+
|   1|
|   2|
|null|
+----+

💡 Hint

Casting invalid strings to int results in null values in Spark.

When casting 'three' to int, Spark returns null instead of throwing an error. The other values are cast correctly.

🔧 Debug

advanced

Identify the error in schema validation code

What error will this Spark code raise when trying to create a DataFrame with the given schema and data?

Apache Spark

from pyspark.sql import SparkSession
from pyspark.sql.types import StructType, StructField, StringType, IntegerType

spark = SparkSession.builder.master('local').appName('SchemaValidation').getOrCreate()

schema = StructType([
    StructField('name', StringType(), nullable=False),
    StructField('age', IntegerType(), nullable=False)
])

data = [('Bob', 'thirty')]

df = spark.createDataFrame(data, schema=schema)
df.show()

A TypeError: Data type mismatch for field 'age'

B RuntimeException: Failed to convert value 'thirty' to int

C No error; shows 'Bob' and null for age

D ValueError: Cannot cast string 'thirty' to IntegerType

💡 Hint

Consider how Spark handles invalid type values during DataFrame creation.

Spark does not throw an error at DataFrame creation for invalid type values but sets them to null. The show() method will display 'Bob' and null for age.

🧠 Conceptual

advanced

Understanding nullable field behavior in Spark schema

Which statement best describes how Spark handles nullable fields in a schema during DataFrame operations?

A Spark allows nulls in non-nullable fields during DataFrame creation but may fail during write operations

B Spark enforces non-nullable fields strictly and throws errors if nulls appear at any time

C Spark automatically replaces nulls in non-nullable fields with default values

D Spark ignores nullable settings and treats all fields as nullable

💡 Hint

Think about when Spark enforces schema constraints strictly.

Spark does not enforce non-nullable constraints strictly during DataFrame creation or transformations but may throw errors when writing data to external storage that requires strict schema adherence.

🚀 Application

expert

Detecting schema mismatch in streaming data ingestion

You have a Spark Structured Streaming job reading JSON data with a predefined schema. If incoming JSON messages have missing required fields, what will happen during streaming processing?

A Rows with missing required fields will have nulls in those fields without failing the job

B The streaming job will fail immediately with a schema mismatch error

C Spark will skip rows missing required fields silently without warning

D Spark will automatically infer missing fields and fill them with default values

💡 Hint

Consider how Spark handles schema enforcement in streaming ingestion.

In Spark Structured Streaming, missing fields in JSON data are set to null if the schema expects them. The job continues running unless explicit validation or checks are added.