simpleetl 1.0.3

SimpleETL - ETL Processing by Simple Specifications

SimpleETL er an ETL tool developed by FlexDanmark to easily handling processing of data from user-defined data sources and automatically generates a dimensionally modelled data warehouse.

SimpleETL is developed to work with the PostgreSQL DBMS backend with psycopg2 as database adapter.

Features

• Automatically generates data warehouse dimensional model (star schema)
• Can track changes of facts
• User-defined automatic fact table partitioning
• Handle deleted facts
• Ensures data quality by type and value checking
• Provides a wide range of default data types and allows user to define their own

Installation

SimpleETL can be installed in multiple ways. The simples is to install from pypi:

$ pip install simpleetl

Dependencies

SimpleETL requires the psycopg2-binary and the pygrametl package for database PostgreSQL database connections and table handling.

Example usages

From the source repository multiple code examples can be found in the examples folder.

A simple example could be:

from simpleetl import FactTable, runETL, datatypes as dt

factobj = FactTable(schema="testschema", table="userdata",
                    migrate_updates=True,
                    # Updated to data will be processed. Can be set to False if only appending (will speed things up)
                    store_history=False,  # Create a seperate userdata_historic table for storing changes to facts.
                    track_last_updated=True,
                    # Adds an _updated attribute which keeps track of when data was last updated.
                    lookupatts=["userid"]  # List of attributes uniquely defining a fact
                    )

factobj.handle_deleted_rows("mark")
# Tells ETL to mark deleted rows from source with an _deleted timestamp attribute

factobj.add_column_mapping("userid", dt.bigint, "userid")
# Map userid from source data to database with same name

factobj.add_column_mapping("sys_username", dt.text, "username")
# Rename "sys_username" from source data to "username" in database

datafeeder = [{"userid": 42, "sys_username": "Jens"}, {"userid": 56, "sys_username": "Svend"}]
# datafeeder can be a generator or simple a iterable of dictionaries

stats = runETL(facttable=factobj, datafeeder=datafeeder,
               db_host="localhost", db_port="5432", db_name="test_db", db_user="dbuser", db_pass="dbpass")

Publications

Ove Andersen, Christian Thomsen, Kristian Torp: SimpleETL: ETL Processing by Simple Specifications. DOLAP 2018 http://ceur-ws.org/Vol-2062/paper10.pdf