View on GitHub


JDBC related extensions for Spock

Download this project as a .zip file Download this project as a tar.gz file

Spock JDBC extensions

JDBC related extensions for Spock.

Build Status Bintray License


repositories {

dependencies {
  testCompile "co.freeside:spock-jdbc:1.0.0"

Automatic cleanup of test data

If you have tests that insert data to a database it's important to ensure the data is cleaned up between tests. Unfortunately, writing a cleanup method that carefully deletes things from the database in the right order is really boring. Also, you and I both know you're going to do slightly different things in different tests and 6 months from now something will start leaking and it will take you all morning to figure out why.

Instead you can use the @TruncateTables annotation.

The annotation can be applied to a java.sql.Connection, javax.sql.DataSource or groovy.sql.Sql property of the specification. In the cleanup phase the extension will use the annotated field to connect to the database and simply delete everything from all tables.

For example you could use the annotation in a Spring integration test with a dependency-injected DataSource instance like this:

@TruncateTables @Autowired DataSource dataSource

Foreign key constraints

The extension will analyze foreign key constraints on the tables it finds and delete data in an order that will not cause constraint violation exceptions.

Using a custom connection source

If you want to use @TruncateTables with something other than a Connection, DataSource or Sql field you can write an implementation of Connector or TypedConnector and specify it on the annotation.

For example, if you were using JDBI and have a org.skife.jdbi.v2.DBI field you could annotate the field with:

@TruncateTables(DBIConnector) DBI dbi

… and implement a connector like this:

static class DBIConnector extends TypedConnector<DBI> {
  DBIConnector() { super(DBI) }

  protected Connection apply(DBI source) {

Connection state

The connection will be closed after data is deleted.

If you're annotating a raw Connection field that might not be ideal. DataSource or something like it that can be used to acquire a fresh connection is really the optimal use-case.

Logging activity

To log what the extension does:

@TruncateTables(verbose = true)

Ignoring exceptions

To ignore any exceptions encountered when deleting data:

@TruncateTables(quiet = true)


Can I make it only delete some data?

Not at the moment. If you have a set of "baseline" data you will need to re-insert it in your setup() method.

Is this useful for tests that run in their own transaction like if I'm using Spring Boot's @DataJpaTest or a Grails integration tests?

Not really. I have no idea whether the deletion or the transaction rollback would happen first. Either way would be bad.

Shouldn't I just drop the entire database schema and recreate it for each test?

That's also a valid approach. @TruncateTables is for people who don't want to do that for some reason.

I have a circular foreign key constraint, will that be a problem?

Yes. Yes it will. You monster.