Software Development

Easy Mocking of Your Database

Test-driven development is something wonderful! Once you’ve established it in your organisation, you will start to:

  • Greatly improve your quality (things break less often)
  • Greatly improve your processes (things can be changed more easily)
  • Greatly improve your developer atmosphere (things are more fun to do)

The importance of doing the right test-driven development is to find a good ratio of what kind of code is to be covered…
 
 

  • by automated unit tests
  • by automated integration tests
  • by manual “smoke tests”
  • by manual “acceptance tests”
  • not at all

Finding that ratio can be grounds for heated, religious discussions. I will soon blog about my own opinion on that subject. In this post, however, we will focus on the first kind of test: unit tests.

Unit testing your data access

When databases are involved, people will probably quickly jump to writing integration tests, because all they have to do is create a little Derby, H2 or HSQLDB (or other) test database, and run a couple of data-setup queries prior to the actual test. Their code module will then hopefully not notice the difference to a productive environment, and the whole system can be tested as a blackbox. The advantage of this is that your tests can be written in a way to verify your business requirements, your user stories, or whatever you call them. So far, the theory.

When these database integration tests pile up, it starts to become increasingly difficult to shield them off one another. Avoiding inter-dependencies and at the same time, avoiding costly database setups is hard. You won’t be able to run the whole test-suite immediately after building / committing. You need nightly builds, weekly builds. But unit testing the data access layer isn’t that much easier! Because JDBC is an awful API to mock. There are so many different ways of configuring and executing queries through this highly stateful API, your unit tests quickly become unmanageable.

There are a few libraries that help you with database testing. Just to name a few:

  • MockRunner: This one has some JDBC-specific extensions that allow for simulating JDBC ResultSets, as well as for checking whether actual queries are executed
  • jMock: An “ordinary” Java mocking library
  • mockito: An “ordinary” Java mocking library
  • DBUnit: This one doesn’t mock your database, it’s good for testing your database. Another use-case, but still worth mentioning here

Some of the above libraries will not get you around the fact that JDBC is an awkward API to mock, specifically if you need to support several (incompatible!) versions of JDBC at the same time. Some examples can be seen here:

Mocking the database with jOOQ

When you’re using jOOQ in your application, mocking your database just became really easy in jOOQ 3.0. jOOQ now also ships with a Mock JDBC Connection. Unlike with other frameworks, however, you only have to implement a single functional interface with jOOQ, and provide that implementation to your MockConnection: The MockDataProvider. Here’s a simple implementation example:

MockDataProvider provider = new MockDataProvider() {

    // Your contract is to return execution results, given a context
    // object, which contains SQL statement(s), bind values, and some
    // other context values
    @Override
    public MockResult[] execute(MockExecuteContext context) 
    throws SQLException {

        // Use ordinary jOOQ API to create an org.jooq.Result object.
        // You can also use ordinary jOOQ API to load CSV files or
        // other formats, here!
        Result<MyTableRecord> result = executor.newResult(MY_TABLE);
        result.add(executor.newRecord(MY_TABLE));

        // Now, return 1-many results, depending on whether this is
        // a batch/multi-result context
        return new MockResult[] {
            new MockResult(1, result)
        };
    }
};

// Put your provider into a MockConnection and use that connection
// in your application. In this case, with a jOOQ Executor:
Connection connection = new MockConnection(provider);
Executor create = new Executor(connection, dialect);

// Done! just use regular jOOQ API. It will return the values
// that you've specified in your MockDataProvider
assertEquals(1, create.selectOne().fetch().size());

The above implementation acts as a callback for JDBC’s various executeXXX() methods. Through a very simple MockExecuteContext API, you can thus:

  • Get access to the executed SQL and bind values (Use general jOOQ API to inline bind values into the SQL statement)
  • Distinguish between regular SQL statements and both single-statement/multi-bind-value and multi-statement/no-bind-value batch executions
  • Return one or several results using jOOQ’s org.jooq.Result objects (which you can easily import from CSV, XML, JSON, TEXT formats)
  • Return “generated keys” results through the same API
  • Let jOOQ’s MockStatement take care of the serialisation of your mock data through the JDBC API

There is also an experimental implementation of a MockFileDatabase, a text-based mock database that uses the following format:

# This is a sample test database for MockFileDatabase
# Its syntax is inspired from H2's test script files

# When this query is executed...
select 'A' from dual;
# ... then, return the following result
> A
> -
> A
@ rows: 1

# Just list all possible query / result combinations
select 'A', 'B' from dual;
> A B
> - -
> A B
@ rows: 1

select 'TABLE1'.'ID1', 'TABLE1'.'NAME1' from 'TABLE1';
> ID1 NAME1
> --- -----
> 1   X
> 2   Y
@ rows: 2

MockFileDatabase implements MockDataProvider, so it’s dead-simple to provide your unit tests with sample data. Future versions of jOOQ will allow for:

  • Regex pattern-matching SQL statements to provide mock results
  • Load these results from other formats, such as jOOQ’s supported export formats
  • Specify the behaviour of batch statements, multi-result statements, etc.

Using jOOQ’s MockConnection in other contexts

Things don’t stop here. As jOOQ’s MockConnection is the entry point for this mocking sub-API of jOOQ, you can also use it in other environments, such as when running JPA queries, Hibernate queries, iBatis or just your plain old legacy JDBC queries.

jOOQ has just become your preferred JDBC mock framework!

 

Reference: Easy Mocking of Your Database from our JCG partner Lukas Eder at the JAVA, SQL, AND JOOQ blog.

Lukas Eder

Lukas is a Java and SQL enthusiast developer. He created the Data Geekery GmbH. He is the creator of jOOQ, a comprehensive SQL library for Java, and he is blogging mostly about these three topics: Java, SQL and jOOQ.
Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Learn how your comment data is processed.

0 Comments
Inline Feedbacks
View all comments
Back to top button