Christian Oestreich

   two guys in design - software.development.professional

Allowing Users to Inject Custom Beans Into Your Grails Plugin

| Comments

Recently with my [cxf-client][1] plugin I wanted to give users a way to inject custom beans into the auto-wired client services. At first I was having a little trouble figuring out how to get each configured service to have a uniquely injected security interceptor. Burt Beckwith does something similar with the [spring-security-core][2] plugin. I borrowed heavily on his method for using [filters][3]. The key to getting this working was knowing that the service client factory I use is a singleton as are each unique instance of the service that is configured. Given the way I have the apache cxf code wrapped, any clients created hand off their actual creation of the service proxy to the service client factory. The security interceptor that is injected to the client has to be a unique bean as it will be wired up as a singleton as well and be handed off to the factory during creation. What this ultimately means is that I can't just have one global property that can be overridden by a common bean name as this would be shared across clients and not give each client the flexibility to define it's own unique security interceptor property values such as username and password Here is a snippet of the code from my plugin's doWithSpring block. This code is within a loop that iterates all the clients set up with in the application's Config.groovy (see [plugin configuration][4] in the docs). ``` groovy if(client?.secured && !client?.securityInterceptor) { "securityInterceptor${cxfClientName}"( { username = client?.username ?: "" password = client?.password ?: "" } } "${cxfClientName}"(DynamicWebServiceClient) { webServiceClientFactory = ref("webServiceClientFactory") if(client?.secured || client?.securityInterceptor) { if(client?.securityInterceptor) { securityInterceptor = ref("${client.securityInterceptor}") } else { securityInterceptor = ref("securityInterceptor${cxfClientName}") } } clientInterface = client.clientInterface ?: "" serviceName = cxfClientName serviceEndpointAddress = client?.serviceEndpointAddress ?: "" secured = (client?.secured || client?.securityInterceptor) ?: false } ``` With the code above the security interceptor is either referenced by name directly from the client config block if provided or the default one is used if the service is marked as secure. A default interceptor is created for each client as the username and password will most likely be different per configured client. For creating a mock custom interceptor I am just modifying the default one slightly below and using a different name for the user and password fields to keep it simple. As a convenience to the user, I created an interface to inherit from that allows you to customize the specifics of the interceptor without having to inherit all the contract methods for a cxf interceptor. You simply have to inherit from SecurityInterceptor in the package and use an existing interceptor type and return it in the create method. Here is the custom interceptor I created for the [demo project][5]. ``` groovy package import import import import import import import class CustomSecurityInterceptor implements SecurityInterceptor { def pass def user WSS4JOutInterceptor create() { Map outProps = [:] outProps.put(WSHandlerConstants.ACTION, outProps.put(WSHandlerConstants.USER, user) outProps.put(WSHandlerConstants.PASSWORD_TYPE, outProps.put(WSHandlerConstants.PW_CALLBACK_REF, new CallbackHandler() { void handle(Callback[] callbacks) throws IOException, UnsupportedCallbackException { WSPasswordCallback pc = (WSPasswordCallback) callbacks[0] pc.password = pass } }) new WSS4JOutInterceptor(outProps) } } ``` You have to make sure your create method returns an object that already inherits from the appropriate cxf interceptor classes such as the WSS4JOutInterceptor I used here. It is technically possible for your interceptor to extend something like SoapHeaderInterceptor, but by doing so you will be responsible for overriding all the appropriate interceptor methods yourself. To see how to setup security with the cxf plugin you can see the [following example][6] on how to define a basic authentication in interceptor on the server side for testing your clients. More specifically refer to [this file][7] for sample code on create your own in interceptor or to the [demo project file][8] that injects a server side in interceptor. Perhaps the **best documentation** on writing a complex interceptor can be found at the [Apache CXF][9] site. These in interceptors are what handle the security request that comes to the server. Most likely if you are dealing with secured service, an in interceptor already exists on the server and you only care about creating a custom out interceptor to use with your client. In the case of the above CustomSecurityInterceptor, you would then place the following in your projects resources.groovy. ``` groovy beans = { myCustomInterceptor({ user = "wsuser" pass = "secret" } } ``` The last step to hooking up the custom interceptor is to define the securityInterceptor for the client config block. The myCustomInterceptor bean can be hooked up by adding the line in the config below. ``` groovy cxf { client { customSecureServiceClient { wsdl = "docs/SecureService.wsdl" //only used for wsdl2java script target namespace = "" clientInterface = //secured = true //implied when you define a value for securityInterceptor securityInterceptor = 'myCustomInterceptor' serviceEndpointAddress = "${}" } } } ``` This is added as a 'string' as the plugin will essentially use this as the name of the bean it will use for injecting into the client. I suppose if the bean wasn't found the code might hiccup when trying to run it. I will have to check that out! :) I hope this can help anyone wanting to write more extensible plugins in the future overcome the challenges I faced. It wasn't that much code in the end, but putting all the pieces of the puzzle together was a bit challenging. [1]: [2]: [3]: [4]: [5]: [6]: [7]: [8]: [9]: