Building a HTTP API using request handlers and processors
This tutorial builds a simple application consisting of these pieces:
- A custom REST API - implemented in a request handler.
- Two pieces of request/response processing logic - implemented as two chained processors.
- A component shared by the above processors.
- A custom output format - a renderer.
http://hostname:port/demo?terms=something%20completely%20differentinto a nested structure response produced by the processors and serialized by the renderer. Use the sample application found at http-api-using-request-handlers-and-processors.
The custom request handler is required to implement a custom API. In many cases it is not necessary to add a custom handler as the Processors can access the request data directly. However it is needed if e.g your application wants more control over exactly which parameters are used to route to a particular processing chain.
In this case the request handler will simply add the request URI as a property and then forward to the build-in processing handler for processing. Refer to selection of base classes to choose the correct starting point for your own application, including adding an access log.
Review the code in DemoHandler.java
This application contains two processors, one for annotating the incoming request (using default values from config) and checking the result, and one for creating the result using the shared component.
Review the code in AnnotatingProcessor.java
The other processor creates some structured Response Data from data handled to it in the request. This is done in cases where the web service is a processing service. In cases where the service is implementing some middleware on top of other services, similar processors will instead make outgoing requests to downstream web services to produce Response Data.
Review the code in DataProcessor.java
Notice how the task of the server is decomposed into separate Processing steps which can be composed by chaining at configuration time and which communicates through the Request and Response only. This structure enhances sharing, reuse and modularity and makes it easy to create variations where some logic encapsulated in a Processor is added, removed or modified.
The order of the processors is decided by the @Before and @After annotations - refer to chained components.
The default terms used by the AnnotatingProcessor are placed in user configuration, where the definition is in demo.def:
package=com.mydomain.demo demo.term stringIn other words, a configuration class containing a single array named demo, containing a class Demo which only contains single string named term.
The responsibility of the renderer is to serialize the structured result into bytes for transport back to the client.
Rendering works by first creating a single instance of the renderer,
invoking the constructor, then cloning a new renderer for each result set to be rendered.
init() will be invoked once on each new clone
render() is invoked.
Review the code in DemoRenderer.java
The responsibility of this custom component is to decouple some parts of the application from the searcher. This makes it possible to reconfigure the searcher without rebuilding the potentially costly custom component.
In this case, what the component does is more than a little silly. More typical use would be an FSA or complex, shared helper functionality.
Review the code in DemoComponent.java
Review the application's configuration in services.xml
OK Renderer initialized: 1369733374898 http://localhost:8080/demo?terms=1%202%203%204 1 2 3 4 Rendering finished work: 1369733374902