I use Retrofit 2.3.0 to interact with OAI-PMH endpoints.
I now happen to interact with an endpoint which is picky about whether its base URL ends with a slash or not:
Without slash:
http://www.relacionesinternacionales.info/ojs/oai.html?verb=Identify works as intended.
With slash:
http://www.relacionesinternacionales.info/ojs/oai.html/?verb=Identify causes a redirect to a 404 page.
Now the problem is that Retrofit 2.3.0 demands base URLs to end with a slash.
Providing the Retrofit builder with the no-ending-slash base URL makes Retrofit complain.
Providing the Retrofit builder with the ending-with-a-slash base URL causes Retrofit to build an incorrect URL, causing the 404 error.
How can I circumvent this limitation?
As a workaround, I use Java Reflection to manipulate the Retrofit object's baseUrl field.
First, the code checks if the provided baseUrl ends with a slash. If so, nothing particular has to happen.
In case the provided baseUrl does not end with a slash, first the retrofit object is created with a trailing slash to baseUrl, then later this baseUrl object is replaced with the original non-slash-ending baseUrl:
String baseUrl = "..."; // can end with slash or not
Retrofit retrofit = new Retrofit.Builder()
.baseUrl( baseUrl.endsWith("/") ? baseUrl : baseUrl + "/" )
.addConverterFactory(ScalarsConverterFactory.create())
.build();
// workaround for https://stackoverflow.com/q/47331753/923560
if ( ! baseUrl.endsWith("/") ) {
try {
Field baseUrlField = retrofit.getClass().getDeclaredField("baseUrl");
baseUrlField.setAccessible(true);
HttpUrl newHttpUrl = HttpUrl.parse(baseUrl);
baseUrlField.set(retrofit, newHttpUrl);
baseUrlField.setAccessible(false);
}
catch (NoSuchFieldException | SecurityException | IllegalArgumentException | IllegalAccessException e) {
LOG.error("Exception while manipulating baseUrl=" + baseUrl + " to not end with a slash", e);
throw new RuntimeException(e);
}
}
service = retrofit.create(OaiPmhService.class);
// ...
Related
I'm trying to use Groovy to script a GET request to our GitLab server to retrieve a file. The API URI format is:
https://githost/api/v4/projects/<namespace>%2F<repo>/files/<path>?ref=<branch>
Note that there is an encoded '/' between namespace and repo. The final URI needs to look like the following to work properly:
https://githost/api/v4/projects/mynamespace%2Fmyrepo/files/myfile.json?ref=master
I have the following code:
File f = HttpBuilder.configure {
request.uri.scheme = scheme
request.uri.host = host
request.uri.path = "/api/v4/projects/${apiNamespace}%2F${apiRepoName}/repository/files/${path}/myfile.json"
request.uri.query.put("ref", "master")
request.contentType = 'application/json'
request.accept = 'application/json'
request.headers['PRIVATE-TOKEN'] = apiToken
ignoreSslIssues execution
}.get {
Download.toFile(delegate as HttpConfig, new File("${dest}/myfile.json"))
}
However, the %2F is re-encoded as %252F. I've tried multiple ways to try and create the URI so that it doesn't encode the %2F in between the namespace and repo, but I can't get anything to work. It either re-encodes the '%' or decodes it to the literal "/".
How do I do this using Groovy + http-builder-ng to set the URI in a way that will preserve the encoded "/"? I've searched but can't find any examples that have worked.
Thanks!
As of the 1.0.0 release you can handle requests with encoded characters in the URI. An example would be:
def result = HttpBuilder.configure {
request.raw = "http://localhost:8080/projects/myteam%2Fmyrepo/myfile.json"
}.get()
Notice, the use of raw rather than uri in the example. Using this approach requires you to do any other encoding/decoding of the Uri yourself.
Possible Workaround
The Gitlab API allows you to query via project id or project name. Look up the project id first, then query the project.
Lookup the project id first. See https://docs.gitlab.com/ee/api/projects.html#list-all-projects
def projects = // GET /projects
def project = projects.find { it['path_with_namespace'] == 'diaspora/diaspora-client' }
Fetch Project by :id, See https://docs.gitlab.com/ee/api/projects.html#get-single-project
GET /projects/${project.id}
base on Retrofit #Field doc, when making a post request
a combination of using #FormUrlEncoded and #Field will yields a request body of: paramName=paramValue¶mName=paramValue.
but what I am not getting field paramemters included in RequestBody.
my interface definition as below:
(I have no endpoint, and jake Wharton says use ./ as explicit intent that you want to use the path of the base URL and add nothing to it, but I tried #POST("./") it's not work, i got 404 not found error, so I add full url address to bypass this error temporarily)
public interface BannerService {
#FormUrlEncoded
#POST("http://10.10.20.190:6020/router")
Flowable<List<BannerBeanList.BannerBean>> getBannerData(#Field("method") String method, #Field("adspaceId") String adspaceId);
}
and this is how I make calls to interface service:
public class RemoteListDataSource implements RemoteDataSource {
#Override
public Flowable<List<BannerBeanList.BannerBean>> getBannerListData(ADFilterType adFilterType) {
BannerService bannerService = RetrofitHttpManger.getInstance().create(BannerService.class);
return bannerService.getBannerData("mz.app.ad.list", String.valueOf(adFilterType.getValue()));
}
}
below is retrofit instance in it's private constructor
retrofit = new Retrofit.Builder()
.client(httpClientBuilder.build())
.addConverterFactory(GsonConverterFactory.create())
.addCallAdapterFactory(RxJava2CallAdapterFactory.create())
//TODO baseurl tempororily hard code for test purpose
.baseUrl("http://10.10.20.190:6020/router/")
.build();
this is the result I got:
the current request body that I am logging is the common parameters that I added from FromBody in interceptor, only except the parameters that I passed in from #Field annoation, and server side info tells the same thing.
I have solved this issue, thanks to #iagreen's comment.
the request body was replaced by FormBody.Builder().add().build() which passed into chain.request().newBuilder().post().build() in my interceptor.
then the question turns out to be how to append paramemters in RequestBody, and the solution can refers to Retrofit2: Modifying request body in OkHttp Interceptor
I'm using Volley to send an http post request with parameters from my android app to my local server running in http://192.168.1.4:3000/battery_signal_report
I'm pretty sure the server is running properly (I checked it with Postman successfully).
also, I successfully sent the request through Android Studio's Emulator using ip 10.0.2.2
Trying to make it work, i used various request implementations including JsonObjectRequest, StringRequest and the custom request described here: Volley JsonObjectRequest Post request not working
Also, I've read somewhere that Volley post requests have some problems with the request header, so i tried to override it in different ways.
Nothing works. onErrorResponse is called every time with an empty VolleyError input.
I've fairly new to android development, so any insights would be much appreciated.
Thanks in advance.
For anyone else coming across this, you need to forget about the header override and setup your own getBodyContentType() and getBody() methods. Follow this pattern:
StringRequest stringRequest = new StringRequest(Request.Method.POST, url, successListener, errorListener) {
#Override
public String getBodyContentType() {
return "application/json; charset=utf-8";//set here instead
}
#Override
public byte[] getBody() {
try {
Map<String, String> params = yourObject.getMappedParams();
JSONObject json = new JSONObject(params);
String requestBody = json.toString();
return requestBody == null ? null : requestBody.getBytes("utf-8");
} catch (UnsupportedEncodingException uee) {
return null;
}
}
};
I have some HTTPBuilder code that behaves differently depending on whether or not I reuse the same HTTPBuilder object to perform two different requests to the same REST service:
def http = new HTTPBuilder( 'https://myBaseURI/' )
http.auth.basic username, password.getPlainText()
http.ignoreSSLIssues()
http.request(GET,JSON) { req ->
uri.path = 'some/api/path/'
headers.'User-Agent' = 'Mozilla/5.0'
} // this request always behaves as expected
http.request(POST, JSON) { req ->
uri.path = 'some/other/api/path'
headers.'User-Agent' = 'Mozilla/5.0'
body = {
// Request body elided for brevity
}
}
The 'correct' behavior is for the POST to return a 201 - Created, but the response comes back as 200 OK unless I create a new HTTPBuilder to handle issuing the second request, in which case, the API call behaves as expected.
Certainly, the cause of the different results could be elsewhere, but I first wanted to make sure I wasn't misusing this object. Are there problems to be aware of when reusing the HTTPBuilder to issue multiple HTTP requests?
Try removing that forward slash at the end when you set uri.path in the GET request.
Looking at the documentation for setPath in URIBuilder you get:
//Set the path component of this URI. The value may be absolute or relative to the current path. e.g.
def uri = new URIBuilder( 'http://localhost/p1/p2?a=1' )
uri.path = '/p3/p2'
assert uri.toString() == 'http://localhost/p3/p2?a=1'
uri.path = 'p2a'
assert uri.toString() == 'http://localhost/p3/p2a?a=1'
uri.path = '../p4'
assert uri.toString() == 'http://localhost/p4?a=1&b=2&c=3#frag'
I understand this to mean that if you set the uri.path of an httpbuilder object with a slash at the end, you have essentially updated the working path so any subsequent relative path updates to uri.path will result in a concatenation of the path. Therefore, your POST in that example ends up pointing at https://myBaseURI/some/api/path/some/other/api/path
I've followed instructions on how creating a ServiceStack here at:
https://github.com/ServiceStack/ServiceStack/wiki/Create-your-first-webservice
I'm sure I have followed it to the letter, but as soon as I run the web application. I get a 'Snapshot' view of my response. I understand this happens when I don't have a default view/webpage. I set up the project as a ASP.net website, not a ASP.net MVC website. Could that be the problem?
I also wrote a test console application with the following C# code. It got the response as a HTML webpage rather than as a plain string e.g. "Hello, John".
static void sendHello()
{
string contents = "john";
string url = "http://localhost:51450/hello/";
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
request.Method = "POST";
request.ContentLength = contents.Length;
request.ContentType = "application/x-www-form-urlencoded";
// SEND TO WEBSERVICE
using (StreamWriter writer = new StreamWriter(request.GetRequestStream()))
{
writer.Write(contents);
}
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
string result = string.Empty;
using (StreamReader reader = new StreamReader(response.GetResponseStream()))
{
result = reader.ReadToEnd();
}
Console.WriteLine(result);
}
How can I switch off the 'snapshot' view? What am I doing wrong?
The browser is requesting html so ServiceStack is returning the html snapshot.
There are a couple of ways to stop the snapshot view:
First is to use the ServiceClient classes provided by servicestack. These also have the advantage of doing automatic routing and strongly typing the response DTOs.
Next way would be to set the Accept header of the request to something like application/json or application/xml which would serialize the response into json or xml respectively. This is what the ServiceClients do internally
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
request.Accept = "application/json";
...
Another method would be to add a query string parameter called format and set it to json or xml
string url = "http://localhost:51450/hello/?format=json";
Putting the specific format requesting is the practical way to do this
string url = "http://localhost:51450/hello/?format=json";
I suggest simply deleting this feature.
public override void Configure(Container container)
{
//...
this.Plugins.RemoveAll(p => p is ServiceStack.Formats.HtmlFormat);
//...
}
Now all requests with the Content-Type=text/html will be ignored.