Snowpark Migration Accelerator: Supported Platforms¶

SQL Dialects¶

The Snowpark Migration Accelerator (SMA) can analyze code files to identify SQL elements. Currently, SMA can detect SQL code written in the following formats:

• Spark SQL

• Hive QL

• Databricks SQL

SQL Assessment and Conversion Guidelines¶

While Spark SQL and Snowflake SQL are highly compatible, some SQL code may not convert perfectly.

SQL analysis is only possible when the SQL is received in the following ways:

• A SQL cell within a supported notebook file

• A .sql or .hql file

• A complete string passed to a spark.sql statement.

  Some variable substitutions are not supported. Here are a few examples:

  • Parsed:

    spark.sql("select * from TableA")
    

    Copy

    New SMA scenarios supported include the following:

    # explicit concatenation
    spark.sql("select * from TableA" + ' where col1 = 1')
    
    # implicit concatenation (juxtaposition)
    spark.sql("select * from TableA" ' where col1 = 1')
    
    # var initialized with sql in previous lines before execution on same scope
    sql = "select * from TableA"
    spark.sql(sql)
    
    # f-string interpolation:
    spark.sql(f"select * from {varTableA}")
    
    # format kindof interpolation
    spark.sql("select * from {}".format(varTableA))
    
    # mix var with concat and f-string interpolation
    sql = f"select * from {varTableA} " + f'where {varCol1} = 1'
    spark.sql(sql)
    

    Copy

  • Not Parsed:

    some_variable = "TableA"
    spark.sql("select * from" + some_variable)
    

    Copy

  SQL elements are accounted for in the object inventories, and a readiness score is generated specifically for SQL.