A modern programming language that makes developers happier.

kotlinlang

Hi all, how can I create a client plugin/interceptor that reads the response with `bodyAsText` without interfering with other plugins that might want to do the same? At the moment I see that the body can only be read once, after which it gets consumed. Is there a way to prevent that? Thanks

I believe this is what you want: <https://ktor.io/docs/double-receive.html>

I'm already using `call.save()` following the implementation in ktor's default RequestValidator, I think that should have a similar effect, as it reads the body and copies it in a `SavedHttpCall` , which also sets `allowDoubleReceive: Boolean = true` on the call itself, but I'm still unable to read the body twice

my understanding is that `SavedHttpResponse` creates a new `ByteReadChannel` based on the saved body every time `response.content` is called, but it doesn't seem to be working for me. When I try to read `bodyAsText` again, I get a `CancellationException` saying that the parent job was completed

the response coroutineContext's job is completed as soon as the body is read and I can't find a way to prevent that

Unless I'm missing something I'm starting to think there's something wrong in the api, the `finally` block for `bodyNullable` always completes the response coroutine job, and there's no other way to access the saved responseBody in the `SavedHttpCall` because `getResponseContent` is protected and that's the only one creating a new ByteReadChannel with the saved content

Can you describe what problem is to be solved by the interceptor/plugin?

<@U017TSBJ2C9> The usecase is the combination of 2 interceptors handling failure responses:
1. The first one is ready the body via `bodyAsText()` and logging it to our backend logging service
2. The second one is my version of the `HttpResponseValidator` where I convert the response into my internal exception which contains the deserialized error (via `body&lt;MyError&gt;()` + the raw response body via `bodyAsText()`).
For instance, for interceptor no.2, reading the body via `bodyAsText()` makes the following `body&lt;MyError&gt;()` throw the cancelled exception even if I transformed my call in a `SavedHttpCall` by calling `save()` in advance, and I assume that's because `bodyNullable()`  completes the job even for `SavedHttpResponse`

From what I understand, being able to do something like that appears to be the point of `SavedHttpCall`, and I can see that it overrides `getResponseContent()` by always creating a new `ByteReadChannel` with the saved response content for that purpose, but then the fact that executing `bodyNullable` always completes the scope's job defeats its purpose.

And from what I can see using `bodyNullable` is the only way I can access the saved response content (because of the `getResponseContent`), as the other APIs are not public

Calling the `body` method in a `HttpResponseValidator`  consumes a response body as described in <https://youtrack.jetbrains.com/issue/KTOR-4225|KTOR-4225>.

Thanks, good to know it's being worked on, I'm not using HttpResponseValidator directly, but I think this still applies in my case as I'm following a similar implementation