I have an external validator library which needs access to both the request and the response in a test. The response is easy - I just need to write an implicit conversion to the Java interface the library wants - but I would like to write code that lets me call the external validator on a request/response pair as well as asserting other things, without having to mention the request twice (which would be a potential source of bugs - I could accidentally make the requests different). How can I do that?
I am thinking of maybe something like
Post(url) ~> validateRAML(sealRoute(myRoute)) ~> check { ... }
but what would validateRAML look like?
I couldn't figure out a way of using the Spray Testkit DSL for this, so I just ended up doing the obvious thing of creating a method:
def validateRAML(req: HttpRequest, response: HttpResponse) = {
// This invokes implicit methods that I have defined to convert req and response
apiDef.testAgainst(req, response) should be('empty)
}
def validateRAML[T](req: HttpRequest, r: Route)(body: => T): Unit = {
req ~> r ~> check {
body
validateRAML(req, response)
}
}
Related
I know that in GRAILS/GROOVY
def content=urlrestservicestring.toURL().getBytes(requestProperties: ['User-Accepted': username])
is a short form to have all the byte content (for example for PDF donwload), but I don't know all the request properties available for URL for richer connections, for example for POST method (this is a GET call) with payload in json. Is it possible? In which way?
It looks like per requestProperties you can set request headers only which might help for simple cases.
On the other hand if you want to do something more complex, like a POST, you have to use a propper HTTP-client.
In Groovy there's an idiomatic HTTPBuilder which is very straight-forward and easy to use, or Grails own RESTBuilder
UPDATE:
Using HTTPBuilder the download could look like:
import static groovyx.net.http.HttpBuilder.configure
configure {
request.uri = "http://example.org/download"
request.contentType = 'application/json'
}.post {
request.headers.username = 'Scarpanti'
request.body = [ some:'json' ]
Download.toStream delegate, response.outputStream // from grails
// or
// Download.toStream delegate, file
}
see also ref-doc
I have a rest and a microservice.In microservice i have a table and i want that table data to be fetched to rest and i have written the below way in a rest demoController.
def result = restBuilder().post("http://localhost:2222/api/microservice/fetchData"){
header 'authorization', 'fdgtertddfgfdgfffffff'
accept("application/json")
contentType("application/json")
json "{'empId':1,'ename':'test1'}"
}
But it throws an error "No signature of method: demoController.restBuilder() is applicable for argument types: () values: []".How should i fetch data from a microservice to rest?
You are calling a method named restBuilder() and that method does not exist. If you want that to work, you will need to implement that method and have it return something that can deal with a call to post(String, Closure).
You probably are intending to use the RestBuilder class. The particulars will depend on which version of Grails you are using but you probably want is something like this...
RestBuilder restBuilder = new RestBuilder()
restBuilder.post('http://localhost:2222/api/microservice/fetchData'){
header 'authorization', 'fdgtertddfgfdgfffffff'
accept 'application/json'
json {
empId = 1
name = 'test1'
}
}
You may need to add a dependency on grails-datastore-rest-client in your build.gradle.
compile "org.grails:grails-datastore-rest-client"
I hope that helps.
I'm using Citrus to automate our RESTful services where service1's input param is passed onto service2' input after URLEncoding.
Below is the code snippet, to explain the ways I tried but somehow unable to URLEncode, on top of designer.traceVariables().
Owing to the limitation (possible known!) that the response params can't be passed outside of context, I had to call multiple send(), receive() calls, defined in the same Gherkin method as below.
I'm able to print the variable myVar (extracted from payLoad) using designer.traceVariables() but unable to process & persist it for further use, to pass onto the next service call.
#When("^I call CPAuth service for ChasePay$")
public void i_call_CPAuth_service_for_ChasePay() throws Throwable {
designer.send("CitrusContext")
.messageType(MessageType.JSON)
.header(X_JPMC_CSRF_TOKEN_HEADER, csrfToken)
.header(HTTP.CONTENT_TYPE, APPLICATION_JSON)
.header(COOKIE, ConnRoutePNames.DEFAULT_PROXY + "=" + proxy)
.header("citrus_endpoint_uri", authUrl +"/auth/fcc/login")
.message(new HttpMessage("<large_string=no_issues_here>").method(HttpMethod.POST));
designer.receive("CitrusContext").messageType(MessageType.JSON)
.validate("$.response", "secauth")
.validate("$.newstoken", false)
.validate("$.smtoken", "void")
.extractFromPayload("$.spid", "myVar");
String request = URLEncoder.encode("appStoreBundleId=com.testmerchant.sampleapp&version=APP%2BBUILD%2BVERSION&channelId=MPD&spid=**${myVar}**&type=json&chasePayMerchantId=1390xxx", CharEncoding.UTF_8);
System.out.println("request: "+request);
designer.send("CitrusContext")
.messageType(MessageType.JSON)
.header(X_JPMC_CSRF_TOKEN_HEADER, csrfToken)
.header(HTTP.CONTENT_TYPE, APPLICATION_JSON)
.header(COOKIE, ConnRoutePNames.DEFAULT_PROXY + "=" + proxy)
.header("citrus_endpoint_uri", hostUrl +"/pwc/provisioning/pos/otp/contact/v20160313/list.action")
.message(new HttpMessage(request).method(HttpMethod.POST));
Result:
Before encoding: MSZp2V/czcKsxej+Q04Da/QeVlo=MCwCFAqrBN/6/J8WZENecE7JQlEODnecAhQCcnXGBjMoiQ7deunlyXKacFo/lQ== (This should be the string, to be encoded.)
After encoding: %3D%24%7BmyVar%7D%26
Any help would be greatly appreciated, Thanks.
The encode operation destroys the "${myVar}" expression before Citrus is able to parse and interpret the expression as part of the message payload. You need to execute the encode operation as a Citrus function. How to write this function is described here: http://citrusframework.org/docs/custom-functions/
You end up having something like:
designer.send("CitrusContext")
.messageType(MessageType.JSON)
.message(new HttpMessage("custom:urlEncode('appStoreBundleId=com.testmerchant.sampleapp&version=APP%2BBUILD%2BVERSION&channelId=MPD&spid=**${myVar}**&type=json&chasePayMerchantId=1390xxx')").method(HttpMethod.POST));
I'm building an app that needs to make a GET request to the API endpoint https://thecountedapi.com/api/counted using the Siesta framework. The endpoint returns a JSON array, just like an endpoint like https://api.github.com/users/ebelinski/repos, which is used in the Siesta example Github Browser. As a result, I'm trying to make my app use Siesta in the say way that one does. I create a service:
let API = Service(baseURL: "https://thecountedapi.com/api")
Then a transformer for my endpoint in application:didFinishLaunchingWithOptions:
API.configureTransformer("/counted") {
($0.content as JSON).arrayValue.map(Incident.init)
}
Where Incident is a struct with an initializer that takes in a JSON object.
Then in my view controller, I create a resource:
let resource = API.resource("/counted")
and in viewDidLoad:
resource.addObserver(self)
and in viewWillAppear:
resource.loadIfNeeded()
Then I have the following function in my VC to listen to changes:
func resourceChanged(resource: Resource, event: ResourceEvent) {
print(resource.jsonArray)
if let error = resource.latestError {
print(error.userMessage)
return
}
if let content: [Incident] = resource.typedContent() {
print("content exists")
incidents = content
}
print(incidents.count)
}
But when I run my app, I get mixed results. print(resource.jsonArray) just prints [], I have an error message Cannot parse server response, and if I set Siesta.enabledLogCategories = LogCategory.detailed, I can see the error mesage [Siesta:StateChanges] Siesta.Resource(https://thecountedapi.com/api/counted)[] received error: Error(userMessage: "Cannot parse server response", httpStatusCode: nil, entity: nil, cause: Optional(Siesta.Error.Cause.WrongTypeInTranformerPipeline(expectedType: "JSON", actualType: "__NSCFArray", transformer: Siesta.ResponseContentTransformer<SwiftyJSON.JSON, Swift.Array<TheCountedViewer.Incident….
If I comment out the whole transformer, I have some success in that print(resource.jsonArray) prints out the correct array from the endpoint. So my transformer must be wrong in some way, but I think I'm using basically the same transformer as in Github Browser:
service.configureTransformer("/users/*/repos") {
($0.content as JSON).arrayValue.map(Repository.init)
}
Am I missing something?
The key clue to your problem is buried in that (perhaps not ideally helfpul) log message:
Siesta.Error.Cause.WrongTypeInTranformerPipeline
expectedType: "JSON"
actualType: "__NSCFArray"
It’s saying that your transformer expected an input type of JSON, which makes sense — you said as much with ($0.content as JSON). However, it got the type __NSCFArray, which is the secret internal backing type for NSArray. In other words, it expected a SwiftyJSON value, but it got the raw output of NSJSONSerialization instead.
Why? The GithubBrowser project includes an NSDict/NSArray → SwiftyJSON transformer which it configures in the parsing stage. The model transformers in that project all depend on it.
To use SwiftyJSON in the same way in your project, you’ll need to include that transformer from the example project in yours:
private let SwiftyJSONTransformer =
ResponseContentTransformer
{ JSON($0.content as AnyObject) }
And then when setting up your service:
service.configure {
$0.config.pipeline[.parsing].add(SwiftyJSONTransformer, contentTypes: ["*/json"])
}
(Note that you might want to create the ResponseContentTransformer with transformErrors: true if you are interested in errors.)
An alternative way to use SwiftyJSON, which is not quite as pretty but requires less setup, is to manually wrap things in JSON in each individual response transformer:
service.configureTransformer("/users/*/repos") {
JSON($0.content as AnyObject).arrayValue.map(Incident.init)
}
I'm building an app which is architected as a Rails server app providing RESTful api's to the client. The Rails server uses RABL. The client is an Angular JS client performing standard $http calls (gets, puts, etc).
Occasionally my Rails server will produce an error (let's say validation error attached to the object) or even no error in which case I would want to display something to the user - either the errror e.g., "The record did not save because..." or "The record was updated successfully".
I'm trying to map out a pattern on both the Rails side and the Angular/client side to handle this.
As for Rails:
I can certainly pass back a node in each of my RABL files to contain error arrays
I can also return different RABL by checking in the controller before returning
Most suggest using http codes (which makes sense) as per here (although there doesn't seem to be a consistent usages of the codes for something like a validation error).
As for Angular:
I suppose I can write a response interceptor but not sure how that would fully get flushed out.
I guess I'm hoping that I don't have to reinvent the wheel here and someone can point me to a pattern that's currently used and suggested (and localized).
I went ahead and implemented what I thought needed to be done. Thanks for digger69 for some help with this.
On the Rails side, I went with using an http status code. As per here I agreed with using a 400 http status code for error validation.
In my controllers I now have something like the following:
def create
my_obj = MyObj.build_with_params(params)
if my_obj.save
respond_with(my_obj) # regular RABL response
else
respond_with_errors(my_obj.errors)
end
end
In my application_controller.rb I defined a common method respond_with_errors
# respond back to the client an http 400 status plus the errors array
def respond_with_errors(errors)
render :json => {:errors => errors}, :status => :bad_request
end
Note that the :bad_request symbol is already defined for Rails as per here
On the client side I needed to intercept http calls (not only for validation but for authentication failures too (and probably more). Here is an example of my code in Angular (thanks to this post for the help with that):
var interceptor = ['$rootScope', '$q', function (scope, $q) {
function success(response) {
return response;
}
function error(response) {
var status = response.status;
if (status == 401) { // unauthorized - redirect to login again
window.location = "/";
} else if (status == 400) { // validation error display errors
alert(JSON.stringify(response.data.errors)); // here really we need to format this but just showing as alert.
} else {
// otherwise reject other status codes
return $q.reject(response);
}
}
return function (promise) {
return promise.then(success, error);
}
}];
$httpProvider.responseInterceptors.push(interceptor);
I now can be consistent with my rails code and deal with success returns from http calls on the client. I'm sure I have some more to do, but I think this gives a localized solution.
Use an HTTP response interceptor. I am currently using that successfully in an application.
http://docs.angularjs.org/api/ng.$http
From the documentation:
$provide.factory('myHttpInterceptor', function($q, dependency1, dependency2) {
return function(promise) {
return promise.then(function(response) {
// do something on success
}, function(response) {
// do something on error
if (canRecover(response)) {
return responseOrNewPromise
}
return $q.reject(response);
});
}
});
$httpProvider.responseInterceptors.push('myHttpInterceptor');
In my case I created a feedback service, which displays either success or error messages globally. An other option would be to broadcast the responses on the rootscope.