SQL Scripting for ETL
This page describes how to use the SQL-based scripting language to extract data from source systems and load data into QPR ProcessAnalyzer, i.e. how to do ETL (extract, transform and load). The supported commands are described in SQL Scripting Commands. Scripts can be written in the Manage Scripts dialog. The SQL scripts consists of SQL statements and QPR ProcessAnalyzer commands that take the preceding SQL statements as parameters.
Introduction to ETL Scripting
The QPR ProcessAnalyzer ETL system enables data extraction, transformation and loading based on SQL queries implemented by scripts. Scripts are written in SQL and consist of standard SQL commands, QPR ProcessAnalyzer commands and special parameters related to QPR ProcessAnalyzer. Scripts are useful for performing several database related operations. More specifically, with the QPR ProcessAnalyzer ETL system it is possible to, for example:
- extract data from a source system to data tables
- load data from data tables and create events and case attributes to new or existing models based on this data
- read data from existing models and create reports
- import data from existing models and create new models extending the current features of QPR ProcessAnalyzer
SQL scripts have the below listed variables available.
|Id of the project in which context the script is run. Undefined if project context has not been specified in script execution parameters.
|Id of the model in which context the script is run. Undefined if model context has not been specified in script execution parameters.
|Id of the filter in which context the script is run. Undefined if filter context has not been specified in script execution parameters.
|Id of the user running the script.
|Id of the script that originally started the script run.
|Id of the script where the execution currently is. If scripts call other scripts, @_ScriptId doesn't change, whereas @_CurrentScriptId changes when the parent script calls other script.
|If there was an exception when running the script, the value is 1, otherwise 0. INT
|If there was an exception when running the script, contains the message of the exception.
|All passed parameters are available is variables with name @_Parameter_<ParameterName> where ParameterName is the name of the variable. For example, variable myVariable1 can be used with name @_Parameter_myVariable1.
Script Variable Examples
The following script command defines ProjectId, ModelId, and FilterId variables by using the script variables:
SELECT @_ProjectId as ProjectId, @_ModelId as ModelId, @_FilterId as FilterId;
The following script gets various information about the environment:
SELECT @_QPRProcessAnalyzerVersion as QPRProcessAnalyzerVersion, @_UserId as Userid (SELECT 'SheetName', 'Info') --#ShowReport
In general, scripts are meant to be developed in such a way that in the end you can run the scripts without any errors. However, sometimes there may be some system issues (timeouts SAP etc.) that can cause for example data extraction scripts to fail. For these kind of situations and for development time and troubleshooting purposes, you can use the CatchOperationExceptions parameter and the @_ExceptionOccurred, @_ExceptionType, @_ExceptionMessage, and @_ExceptionDetails script variables with the QPR ProcessAnalyzer script commands to handle exceptions in ProcessAnalyzer. Out of these, the @_ExceptionOccurred is handy for example in defining some other command to be run in case there was an exception. For SQL errors, the TRY-CATCH error handling should be used.
Note that the CatchOperationExceptions parameter is in effect only for the command it is used with, i.e. it isn't in effect in child scripts or scripts that are run via the --#Exit command. In addition, when there are multiple ProcessAnalyzer script commands in the script, the @_ExceptionOccurred, @_ExceptionType, @_ExceptionMessage, and @_ExceptionDetails script variables get updated each time, i.e. the variable values are available only until the next ProcessAnalyzer command is executed. To use the same variable values in multiple ProcessAnalyzer commands in the script, place the values into a temporary table:
SELECT @_ExceptionOccurred 'ExceptionOccurred', @_ExceptionMessage 'ExceptionMessage' INTO #PACommandExceptions
Example Script Walkthough
Here is an example of a simple ETL script that prints data taken from the table #Customers:
SELECT * FROM #Customers; (SELECT 'Title', 'Customers') UNION ALL (SELECT 'MaximumCount', '0'); --#ShowReport
This script consists of SQL statements (lines 1-3) and a QPR ProcessAnalyzer command (line 4) that take the results of the preceding SQL SELECT statements as parameters.
Let’s go through this example in more detail (line 1):
SELECT * FROM #Customers;
SELECT is one of the most basic SQL commands you can use. The SELECT statement is used to extract data from a table. You can define the criteria for what data is selected for extraction. On line 1 of the example above, the SQL statement selects all the columns from the #Customers table. In addition, this produces the first input argument for the QPR ProcessAnalyzer command —#ShowReport by defining what to print.
The preceding hash symbol (#) indicates that results are stored to a temporary table. They exist only during the execution of the script so they are not stored permanently.
(SELECT 'Title', 'Customers') UNION ALL (SELECT 'MaximumCount', '0');
These lines create the second input argument for the command --#ShowReport by giving a label for the sheet and defining how many lines to print. ‘0’ means that all rows will be printed.
SQL uses the UNION operator to combine the result-set of two or more SELECT statements. UNION ALL is used to select all the values (including duplicate values) whereas UNION is used to select only the distinct values.
When writing scripts with the QPR ProcessAnalyzer ETL system, the lines in the script that start with "--#" (without the quotes and without any preceding blank spaces at the beginning of the line) are treated as QPR ProcessAnalyzer Commands and SQL ignores these lines as comments. The supported QPR ProcessAnalyzer commands and some examples of how to use them in scripts are listed in here.
Notes for SQL Scripts
- It is important to follow the T-SQL syntax when writing SQL queries.
- SQL is not case sensitive, so you can write either SELECT or select. However, the parameter values and the QPR ProcessAnalyzer commands given as comments are case sensitive.
- Semicolon is used to end SQL statements. It is not always required, but in some cases the missing semicolon may cause SQL errors due to ambiguity. We recommend using a semicolon especially at the end of SELECT statements preceding the QPR ProcessAnalyzer ETL commands to avoid SQL errors.
- Table field names may contain white spaces, special characters, or reserved words but then they must be referred to by using brackets () around them. This situation happens frequently in QPR ProcessAnalyzer ETL scripts since the case and event attribute data contain fields that have extra spaces.
- When transforming data, temporary tables can be used (temporary tables names start with #). Note that global temporary tables (tables names start with ##) is not allowed.
- The Print SQL statement can be used to generate log entries into the script execution log.
- Only those lines in the script that start with "--#" (without the quotes) are treated as QPR ProcessAnalyzer commands, i.e. if there are leading whitespaces before the command, the line is treated as a comment.
- If you don't define a value for the MaximumCount parameter, 1000 will be used as default, i.e. only the 1000 first rows from a given table or model will be used.
- When doing more advanced operations with scripts, you may run into the error messages such as: "The data types sql_variant and varchar are incompatible in the add operation.", "Argument data type sql_variant is invalid for argument 1 of like function.", "Argument data type sql_variant is invalid for argument 1 of left function.". This is because case attributes, event attributes, and data inside datatables are sql_variant type data. In order to use them with more advanced operations (e.g. Add), you need to CONVERT or CAST them into some other data type before the operations. See this example.
- For certain characters in attribute values, you need to use escaping in order to have them interpreted correctly in the script.