Exercise_8_Part2_UseRepositoryToAccessDatabase.md

Exercise 8 (Part 2): Use Repository to Access Database

Learning Goal

After this exercise you know how to access our database using the convenience CRUD repository interface.

The task of this exercise is to persist the advertisements via the corresponding repository instead of storing the data temporarily into a Map and ensure that the JUnit tests are still running successfully. That's why you also need to configure the in-memory database (H2) in context of the JUnit tests.

Prerequisite

Continue with your solution of the last exercise. If this does not work, you can checkout the branch origin/solution-8-1-Configure-Persistence.

Step 1: Inject Repository Into Controller

In the previous exercise we used the Spring Data JPA plugin to create an instance of the AdvertisementRepository interface. In this step we inject such an instance into the AdvertisementController.

• As constructor injection is the preferred approach, we create in the AdvertisementController class, a constructor with an AdvertisementRepository typed argument, and annotate the constructor with @Inject.

Step 2: Use CRUD Repository in Controller

Refactor the AdvertisementController class so that instead of using the HashMap, the advertisements are persisted using the AdvertisementRepository instance (using the CrudRepository methods findOne, findAll, and save).

Note: The CRUD repository automatically decides whether to create a new database entry, or update an existing one when using its save method. To trigger updates, you need to pass an entity object containing the ID in addition to all the (updated) fields.

Step 3: Run and Test the Service Locally

Run the service as described here: Exercise 1: Getting Started

Now you can test the service manually in the browser using the Postman chrome plugin. Please also test that the data is persisted. To test this you could create some advertisements, relaunch the application and see whether the advertisements created in the previous session are returned by new GET requests.

Note: Currently the tests are failing (which will be fixed in the next steps). If you want to build using Maven, run mvn clean package -DskipTests.

Step 4: Run Component Tests

The service tests from Exercise 4 are broken now, because the test provides no information about how to inject the AdvertisementRepository instance.

Furthermore the tests should run independently from a concrete database, so we provide an alternative database configuration which uses an in-memory database named H2.

• Add the h2 dependency to your pom.xml:

<!-- H2 (in-memory) database implementations -->
<dependency>
    <groupId>com.h2database</groupId>
    <artifactId>h2</artifactId>
    <version>1.4.184</version>
    <scope>test</scope>
</dependency>

Note: After you've changed the Maven settings, don't forget to update your Eclipse project! To do so right click on your Eclipse project and select Maven - Update Project ... (ALT-F5)

• Create a new class EmbeddedDatabaseConfig in the test package com.sap.bulletinboard.ads.config and copy the code from here.
• Now you can run the JUnit tests as described in Exercise 4.

Explanation: When the tests are started, the application is not initialized using the AppInitializer. The application context is initialized with all classes that are added as part of the @ContextConfiguration annotation (@see AdvertisementControllerTest class). As consequence the "cloud" profile isn't set in context of the test context.

You might have the question why you need an extra database configuration namely EmbeddedDatabaseConfig for your tests instead of providing different configuration settings via system environment variables? We decided to use a separate configuration class for the tests, as the tests do not run in a cloud-like environment. Even though the configuration classes look very similar, the main difference is, that the CloudDatabaseConfig extends from the AbstractCloudConfig class that expects a cloud environment. A cloud environment is only detected if the VCAP_APPLICATION and VCAP_SERVICES variables are set. To make that more explicit, we have annotated the CloudDatabaseConfig class with @Profile("cloud") i.e. it gets only loaded when "cloud" is set as active profile.

[Optional] Step 5: Check ID for Updates

If you have implemented the PUT Request according to Exercise 4-2, make sure that the ID contained in the URL is the same ID contained in the advertisement entity. Therefore add a test that sends a PUT request for an entity with a different ID than what is specified in the URL. In the test assert that the server responds with a Bad Request (400) status code. You may add a setId method to the Advertisement class so that, in the tests, you can create advertisement entity instances with a specific ID.

Fix the code according to the test by introducing a throwIfInconsistent method comparing the two IDs. Also make sure that the tests in AdvertisementControllerTest are still passing, especially updateNotFound.

[Optional] Step 6: Debug Entity ID Generation

For new advertisements the code does not directly set the ID. However, the advertisement object returned by the POST endpoint has a non-null ID. Use the Eclipse debugger to find out when exactly the ID is set in the object instance.

• In Eclipse, add a breakpoint to the first line of the add method (for that you can double-click on the bar to the left of your code)
• Deploy the application and start Tomcat in debug mode
• Use Postman chrome plugin to create a new advertisement
• In the Debug perspective and in the Variables view, observe that advertisement.id is null
• Using "Step over (F6)" advance to the next line (stepping over the save method invocation)
• Observe that advertisement.id is now set
• Continue to run the application (F9), or stop Tomcat

[Optional] Step 7: Introduce Pagination

Assume you have hundreds or more entries in a table. As part of a GET-all request the application needs to fetch all these entries from the database, buffer them, convert them into JSON in order to provide it as part of the response etc. This costs CPU, memory and network resources while most of the data is usually neither required nor consumable by the client / UI. To improve the performance of the mass GET query it is common practice to introduce pagination, sorting and filtering. Accordingly, you should also differentiate on REST API level which kind of pagination you like to implement:

• client-driven pagination normally chosen for consumer flexibility: the consumer decides what portions of data are to be retrieved.
• server-driven pagination for optimizations (caching) and protection (e.g. against denial of service by selecting a too large page size): the server decides on page cutting and provides additional information about the locations of the previous and next page; the client still has a possibility to configure the page size.

For paging / pagination, you can simply make use of the PagingAndSortingRepository, which is an extension of CrudRepository. As part of your readAll request handler method you just need to introduce additional RequestParams pageId and pageSize with some meaningful defaults.

Have a look into the last commit of the solution-8-3-IntroducePaging branch to see how an implementation could look like.

Used Frameworks and Tools

• H2 in-memory Database

Further Reading

• JDoc CrudRepository
• JDoc PagingAndSortingRepository

---

© 2018 SAP SE

---