aws-movie-search

Rhyme - AWS Lambda Movie Search Example

Summary

This is an example project that shows how Rhyme can be used to implement a HAL+JSON API as an AWS Lambda function.

It implements a simple movie search functionality that is built on top of Kai Toedter's Hypermedia Movie Demo:

• https://hypermedia-movies-demo.herokuapp.com
• https://github.com/toedter/movies-demo

The project contains @HalApiInterface definitions for this external API, and uses them to load the available movies via HTTP using proxies created by Rhyme's HalApiClient.

A simple search logic is then applied, and responses are rendered by the Rhyme framework based on additional API interfaces defined for the search functionality.

Explore

The example movie search API can be accessed directly at https://088zuopqd6.execute-api.eu-central-1.amazonaws.com/default or through the HAL Explorer.

The Doc icons in the HAL Explorer will link to an HTML documentation of the API (which is generated from the annotated interfaces and embedded in the Lambda's JAR file).

Embedded Metadata

You can have a look at the embedded rhyme:metadata to get an idea of what's happening in the background. For example you can see that for search terms which have many hits from the top ranked movies, only the first page of movies from the upstream API has to be loaded:

• https://088zuopqd6.execute-api.eu-central-1.amazonaws.com/default/results?page=0&searchTerm=the

For search terms where there is no hit at all, you can see that all pages of movies had to be loaded before the (empty) result can be rendered:

• https://088zuopqd6.execute-api.eu-central-1.amazonaws.com/default/results?page=0&searchTerm=foo

Performance

Note that the first load of these resources can take up to 30 seconds, as this may trigger a cold start of the lambda function and the Heroku upstream server. Subsequent searches will be much faster, as they will be using responses that are cached in memory for up to one hour.

The JVM will also further optimize the compiled code after repeated requests, so it's worth to reload the search results multiple times (with caching disabled in your browser) if you are interested in the performance details.

Source Code

• the API package defines a @HalApiInterface for each type of resource from the original Hypermedia Movie Demo API, and the new resources added by this example
• the impl package contains
  • a simple class that contains the search logic and all interaction with the upstream service (using streams of the annotated API interfaces)
  • two server-side resource implementation classes to generate the links and embedded resources
  • an AWS Lambda RequestHandler that wires everything together, and configures the caching for upstream resources
• integration tests show how the implementation is tested using either static responses from the file system, or with a dynamic stub implementation of the HAL API interfaces.
• the awslambda folder contains additional integration code that is not specific to this example (and may be moved to its own integration module in the future)

Documentation

• See the main README for a general introduction to the usage of the Rhyme framework
• Javadocs for the annotations and interfaces used in the @HalApiInterface definitions: https://caravan.wcm.io/rhyme/api-interfaces/apidocs/
• Javadocs of Rhyme's core module: https://caravan.wcm.io/rhyme/core/apidocs/