As we speak, Databricks proclaims help for the ANSI SQL/PSM scripting language!
SQL Scripting is now accessible in Databricks, bringing procedural logic like looping and control-flow instantly into the SQL you already know. Scripting in Databricks is predicated on open requirements and totally suitable with Apache Spark™.
For SQL-first customers, this makes it simpler to work instantly on the Lakehouse whereas profiting from Databricks’ scalability and AI capabilities.
For those who already use Databricks, you’ll discover SQL scripting particularly helpful for constructing administrative logic and ELT duties. Key options embody:
- Scoped native variables
- Native exception dealing with based mostly on symbolic error situations
- IF-THEN-ELSE and CASE help
- A number of loop constructs, together with FOR loops over queries
- Loop management with ITERATE and LEAVE
- Dynamic SQL execution by way of EXECUTE IMMEDIATE
Sufficient with the function record — let’s stroll by way of some actual examples. You’ll be able to use this pocket book to comply with alongside.
Information administration
Administrative duties and information cleanup are a continuing in enterprise information administration — needed, routine, and unimaginable to keep away from. You’ll want to wash up historic information, standardize blended codecs, apply new naming conventions, rename columns, widen information varieties, and add column masks. The extra you may automate these duties, the extra dependable and manageable your techniques shall be over time. One frequent instance: implementing case-insensitive conduct for all STRING columns in a desk.
Let’s stroll by way of how SQL scripting could make this sort of schema administration repeatable and simple.
Schema administration: make all STRING columns in a desk case-insensitive
On this instance, we wish to apply a brand new coverage for string sorting and comparability for each relevant column within the desk referred to as workers. We are going to use a regular collation kind, UTF8_LCASE, to make sure that sorting and evaluating the values on this desk will at all times be case-insensitive. Making use of this normal permits customers to profit from the efficiency advantages of utilizing collations, and simplifies the code as customers now not have to use LOWER() of their queries.
We are going to use widgets to specify which desk and collation kind to change. Utilizing the knowledge schema, we’ll then discover all present columns of kind STRING in that desk and alter their collation. We are going to gather the column names into an array. Lastly, we’ll gather new statistics for the altered columns, multi function script.
A pure extension of the above script is to increase it to all tables in a schema, and refresh views to select up the collation change.
Information cleaning: repair grammar in free-form textual content fields
Is there any difficulty extra frequent on this planet of knowledge than ‘soiled information’? Information from completely different techniques, units, and people, will inevitably have variations or errors that should be corrected. If information isn’t cleaned up, you will have flawed outcomes and miss an essential perception. You’ll be able to count on a rubbish response in the event you feed rubbish into an LLM.
Let’s take a look at an instance that features the bane of each publication, together with this weblog: typos. We’ve got a desk that features free-text entries in a column referred to as description. The problems within the textual content, which embody spelling and grammar errors, can be obvious to anybody who is aware of English. Leaving the info on this state will undoubtedly result in points later if attempting to research or examine the textual content. Let’s repair it with SQL Scripting! First, we extract tables holding this column identify from the knowledge schema. Then repair any spelling errors utilizing ai_fix_grammar(). This operate is non-deterministic. So we use MERGE to attain our purpose.
An attention-grabbing enchancment may very well be to let ai_classify() deduce whether or not a column accommodates free-form textual content from the column identify or pattern information. SQL Scripting makes administrative duties and cleansing up messy information environment friendly and simple.
ETL
Clients use SQL for ETL in the present day. Why? As a result of SQL helps a strong set of knowledge transformation capabilities, together with joins, aggregations, filtering, with intuitive syntax, making pipeline code simple for any Information Engineer to increase, replace, and keep. Now, with SQL Scripting, clients can simplify beforehand complicated approaches or deal with extra complicated logic with pure SQL.
Updating a number of tables
Anybody who sells bodily merchandise could have a course of for monitoring gross sales and monitoring shipments. A typical information administration sample is to mannequin a number of tables to trace transactions, shipments, deliveries, and returns. Transaction monitoring is enterprise crucial, and like all crucial course of, it requires the dealing with of surprising values. With SQL Scripting, it’s simple to leverage a conditional CASE assertion to parse transactions into their acceptable desk, and if an error is encountered, to catch the exception.
On this instance, we take into account a uncooked transactions desk for which rows should be routed right into a recognized set of goal tables based mostly on the occasion kind. If the script encounters an unknown occasion, a user-defined exception is raised. A session variable tracks how far the script received earlier than it completed or encountered an exception.
This instance script may very well be prolonged with an outer loop that retains polling for extra information. With SQL Scripting, you will have each the facility and adaptability to handle and replace information throughout your information property. SQL Scripting provides you energy to deal with any information administration process and effectively management the circulation of knowledge processing.
Keep tuned to the Databricks weblog and the SQL periods on the upcoming Information + AI Summit, as we put together to launch help for Temp Tables, SQL Saved Procedures, and extra!
What to do subsequent
Whether or not you might be an present Databricks consumer doing routine upkeep or orchestrating a large-scale migration, SQL Scripting is a functionality you need to exploit. SQL Scripting is described intimately in SQL Scripting | Databricks Documentation.
You’ll be able to attempt these examples instantly on this SQL Scripting Pocket book. For extra particulars, keep tuned for Half 2 of this sequence, which dives into SQL Scripting constructs and the right way to use them.