Transpile a SQL Script into PySpark DataFrame API equivalent code

[thecodecache]

Input SQL Script (assume any dialect) :

SELECT b.se10,
       b.se3,
       b.se_aggrtr_indctr,
       b.key_swipe_ind
FROM
  (SELECT se10,
          se3,
          se_aggrtr_indctr,
          ROW_NUMBER() OVER (PARTITION BY SE10
                             ORDER BY se_aggrtr_indctr DESC) AS rn,
          key_swipe_ind
   FROM fraud_details_data_whole
   GROUP BY se10,
            se3,
            se_aggrtr_indctr ,
            key_swipe_ind) b
WHERE b.rn<2

Output PySpark Code using DataFrame API :

from pyspark.sql import SparkSession
from pyspark.sql.window import Window
from pyspark.sql.functions import row_number

# Create a SparkSession
spark = SparkSession.builder \
    .appName("TranspileSQLExample") \
    .getOrCreate()

# Sample data (replace with your actual DataFrame)
data = [
    ("se10_value1", "se3_value1", "aggrtr_value1", "swipe_value1"),
    ("se10_value1", "se3_value2", "aggrtr_value2", "swipe_value2"),
    ("se10_value2", "se3_value3", "aggrtr_value3", "swipe_value3"),
    ("se10_value2", "se3_value4", "aggrtr_value4", "swipe_value4"),
    ("se10_value3", "se3_value5", "aggrtr_value5", "swipe_value5"),
    ("se10_value3", "se3_value6", "aggrtr_value6", "swipe_value6")
]

# Create DataFrame
fraud_details_data_whole = spark.createDataFrame(data, ["se10", "se3", "se_aggrtr_indctr", "key_swipe_ind"])

# Define Window specification for row_number() function
windowSpec = Window.partitionBy("se10").orderBy(fraud_details_data_whole["se_aggrtr_indctr"].desc())

# Add row number column
fraud_details_data_whole = fraud_details_data_whole.withColumn("rn", row_number().over(windowSpec))

# Filter rows where rn < 2
result_df = fraud_details_data_whole.filter("rn < 2")

# Select required columns
result_df = result_df.select("se10", "se3", "se_aggrtr_indctr", "key_swipe_ind")

# Show the result DataFrame
result_df.show()

Is there any way to translate the above Sql Query into its equivalent PySpark DataFrame API code? The result must be equal when we execute Sql Script and its transpiled PySpark code separately.

Kindly suggest

https://stackoverflow.com/q/78446543/6842300

[Kaniz_Fatma]

Hi @thecodecache, When executed separately, the PySpark code mentioned above should produce the same result as your original SQL query.

[thecodecache]

Hi @Kaniz_Fatma, Thanks for your response. 
I'm looking for a utility or an automated way of translating any generic SQL into PySpark DataFrame code.

So, the translate function should look like below:

def translate(input_sql):
    # translate/convert it into pyspark dataframe code
    return pyspark_code

Please suggest

[Kaniz_Fatma]

Hi @thecodecache, Here’s a simplified example of how you might approach this (note that this is a basic illustration, and you can expand it to handle more complex SQL queries):

from pyspark.sql import SparkSession

def translate(input_sql):
    # Create a SparkSession (you can adjust this based on your environment)
    spark = SparkSession.builder.appName("SQLToPySpark").getOrCreate()

    # Parse the input SQL query (you'll need a more robust parser for production use)
    # Example: SELECT col1, col2 FROM my_table WHERE col3 > 10
    # Extract relevant components (SELECT, FROM, WHERE, etc.)

    # Create a DataFrame (replace with your actual data source)
    df = spark.read.csv("path/to/your/csv/file.csv", header=True)

    # Apply transformations based on parsed components
    # Example: df.select("col1", "col2").filter(df["col3"] > 10)

    # Generate the PySpark code
    pyspark_code = "df.select(\"col1\", \"col2\").filter(df[\"col3\"] > 10)"

    return pyspark_code

# Example usage
input_sql_query = "SELECT col1, col2 FROM my_table WHERE col3 > 10"
translated_code = translate(input_sql_query)
print("Translated PySpark code:")
print(translated_code)

If you have any further questions or need additional assistance, feel free to ask.