Python Source

Applies To:	Pipeline Bundle
Configuration Scope:	Data Flow Spec
Databricks Docs:	NA

Overview

You can specify a Python function as a source type in your Data Flow Specs. These allow for flexibility and more complex data retrieval to be supported, as needed, without overly complicating the Framework.

There are two approaches to defining Python sources:

1. Pipeline logic modules: Define functions in the src/python/ directory and reference them by module name using pythonModule

2. File Path: Define functions in ./python_functions/ directories and reference by file path using functionPath

Sample Bundle

Samples are available in the bronze_sample bundle in the src/dataflows/feature_samples folder.

Configuration

Using Pipeline Logic Modules (src/python/)

Place your Python source functions in src/python/ — the framework adds this directory to sys.path at pipeline initialisation so spec strings resolve without extra configuration.

Deprecation Notice

The legacy src/extensions/ directory is deprecated as of v0.13.0 and will be removed in v1.0.0. Move .py files to src/python/ — existing pythonModule strings in Data Flow Specs are unchanged.

1. Create a module in ``src/python/``

Create your source functions in the src/python/ directory:

my_pipeline_bundle/
├── src/
│   ├── python/
│   │   └── sources.py      # Your source functions
│   ├── dataflows/
│   │   └── ...

Your module can contain multiple functions. Each function must:

• Accept spark (SparkSession) and tokens (Dict) as parameters

• Return a DataFrame

# src/python/sources.py
from pyspark.sql import DataFrame, SparkSession
from pyspark.sql import functions as F
from typing import Dict

def get_customer_cdf(spark: SparkSession, tokens: Dict) -> DataFrame:
    """
    Get customer data with Change Data Feed enabled.
    """
    source_table = tokens["sourceTable"]
    reader_options = {
        "readChangeFeed": "true"
    }

    df = spark.readStream.options(**reader_options).table(source_table)
    return df.withColumn("TEST_COLUMN", F.lit("testing from extension..."))

def get_orders_batch(spark: SparkSession, tokens: Dict) -> DataFrame:
    """
    Get orders data as a batch read.
    """
    source_table = tokens["sourceTable"]
    return spark.read.table(source_table)

2. Reference in Data Flow Spec

Use pythonModule in sourceDetails to reference your function:

JSONYAML

{
    "dataFlowId": "feature_python_extension_source",
    "dataFlowGroup": "feature_samples",
    "dataFlowType": "standard",
    "sourceSystem": "testSystem",
    "sourceType": "python",
    "sourceViewName": "v_feature_python_extension_source",
    "sourceDetails": {
        "tokens": {
            "sourceTable": "{staging_schema}.customer"
        },
        "pythonModule": "sources.get_customer_cdf"
    },
    "mode": "stream",
    "targetFormat": "delta",
    "targetDetails": {
        "table": "feature_python_extension_source",
        "tableProperties": {
            "delta.enableChangeDataFeed": "true"
        }
    }
}

Using File Path

To define a python source function using file paths, create a python_functions folder under the base folder for your dataflowspec:

my_pipeline_bundle/
├── src/
│   ├── dataflows/
│   │   ├── use_case_1/
│   │   │   ├── dataflowspec/
│   │   │   │   └── my_data_flow_spec_main.json
│   │   │   ├── python_functions/
│   │   │   │   └── my_source_function.py
│   │   │   └── schemas/

Your file must contain a function called get_df that:

• Accepts spark (SparkSession) and tokens (Dict) as parameters

• Returns a DataFrame

from pyspark.sql import DataFrame, SparkSession
from pyspark.sql import functions as F
from typing import Dict

def get_df(spark: SparkSession, tokens: Dict) -> DataFrame:
    """
    Get a DataFrame from the source details with applied transformations.
    """
    source_table = tokens["sourceTable"]
    reader_options = {
        "readChangeFeed": "true"
    }

    df = spark.readStream.options(**reader_options).table(source_table)
    return df.withColumn("TEST_COLUMN", F.lit("testing..."))

Reference using functionPath:

JSONYAML

{
    "dataFlowId": "feature_python_function_source",
    "dataFlowGroup": "feature_samples",
    "dataFlowType": "standard",
    "sourceSystem": "testSystem",
    "sourceType": "python",
    "sourceViewName": "v_feature_python_function_source",
    "sourceDetails": {
        "tokens": {
            "sourceTable": "{staging_schema}.customer"
        },
        "functionPath": "my_source_function.py"
    },
    "mode": "stream",
    "targetFormat": "delta",
    "targetDetails": {
        "table": "feature_python_function_source"
    }
}

sourceDetails Schema for Python Source

When using sourceType: "python", the sourceDetails object supports the following properties:

Property	Required	Description
pythonModule	One of pythonModule/functionPath	Module and function reference (e.g., sources.get_customer_cdf). The module must be in the src/python/ directory.
functionPath	One of pythonModule/functionPath	Path to a Python file containing a get_df function. Resolved relative to the ./python_functions/ directory.
tokens	No	Dictionary of token values to pass to the source function. Supports substitution variables like {staging_schema}.

Function Signatures

For Extensions (pythonModule)

The function name can be anything, but it must accept spark and tokens:

def my_source_function(spark: SparkSession, tokens: Dict) -> DataFrame:
    ...

For File Path (functionPath)

The function must be named get_df:

def get_df(spark: SparkSession, tokens: Dict) -> DataFrame:
    ...

Additional Resources

Refer to the Data Flow Spec - Source Details section of the Data Flow Spec Reference documentation for more information on source configuration.