Spark Safe Delta

Combination of tools that allow more convenient use of PySpark within Azure DataBricks environment.

I. Package contents:

1.delta_write_safe

Tool that allows to automatically update schema of DataBricks Delta in case of Changes in data structure

2.write_data_mysql

Method writes data into MySQL and takes care of repartitioning in case if it's necessary.

Dependencies:

1. MySQL connector Java 8_0_13
dbfs:/FileStore/jars/7b863f06_67cf_4a51_8f3b_67d414d808b3-Barnymysql_connector_java_8_0_13_4ac45-2f7c7.jar

http://dev.mysql.com/doc/connector-j/en/
https://mvnrepository.com/artifact/mysql/mysql-connector-java

By default, it relies on constant variables outside of method that define MySQL credentials, that can be also specified as a parameters:

* MYSQL_URL
* MYSQL_DRIVER
* MYSQL_USER
* MYSQL_PASSWORD
* MYSQL_SSL_CA_PATH
* MYSQL_QUERY_TIMEOUT

Method Parameters:

* p_spark_dataframe - dataframe to write
* p_mysql_db_name - name of database to write to
* p_mysql_table_name - name of table to write to
* p_num_partitions - amount of partitions, if -1, runs with default amount of partitions defined in spark environment or specific delta

Method default parameters:

p_num_partitions=-1
url=MYSQL_URL,
driver=MYSQL_DRIVER,
user=MYSQL_USER,
password=MYSQL_PASSWORD,
ssl_ca=MYSQL_SSL_CA_PATH,
queryTimeout=MYSQL_QUERY_TIMEOUT

Usage example:

#MySQL settings defined outside of a method below:
MYSQL_DRIVER = "com.mysql.jdbc.Driver"
MYSQL_URL = "jdbc:mysql://hostname:port/database?useUnicode=true&characterEncoding=utf-8&useJDBCCompliantTimezoneShift=true&useLegacyDatetimeCode=false"
MYSQL_QUERY_TIMEOUT = 0

MYSQL_USER = "user@namespace"
MYSQL_PASSWORD = "example_password"
MYSQL_SSL_CA_PATH = "/mnt/alex-experiments-blob/certs/cert.txt"

#Method execution itself
write_data_mysql(p_spark_dataframe=target_data, p_mysql_dbtable=destination_db_name_column_name_construct)

3.remove_columns

remove_columns() method removes columns from a specified dataframe.
It will silently return a result even if user specifies column that doesn't exist.
Usage example: destination_df = remove_columns(source_df, "SequenceNumber;Body;Non-existng-column")

4.read_mysql

Method allows fetch the table, or a query as a Spark DataFrame.
Returnws Spark DataFrame as a result.

# Example usage:
read_mysql(table_name=customers)
read_mysql(table_name=h2.customers)
read_mysql(table_name=h2.customers, url=MYSQL_URL, driver=MYSQL_DRIVER, user=MYSQL_USER, password=MYSQL_PASSWORD, ssl_ca=MYSQL_SSL_CA_PATH, queryTimeout=MYSQL_QUERY_TIMEOUT)

4.list_available_mysql_tables

Method allows to list all the tables that available to a particular user.
Returnws Spark DataFrame as a result

Package sample usage:

#!/usr/bin/env python

from sparksafedelta import sparksafedelta
sparksafedelta.delta_write_safe(sp_df_to_write, SP_CONTEXT, DATABRICKS_TABLE_NAME)