A scriptable/customizable web server for testing HTTP clients using OAuth2/OpenID Connect or applications with a dependency to a running OAuth2 server (i.e. APIs requiring signed JWTs from a known issuer). The server also provides the necessary endpoints for token validation (endpoint for JWKS) and ID Provider metadata discovery ("well-known" endpoints providing server metadata)
mock-oauth2-server is written in Kotlin using the great OkHttp MockWebServer as the underlying server library and can be used in unit/integration tests in both Java and Kotlin or in any language as a standalone server in e.g. docker-compose.
Even though the server aims to be compliant with regards to the supported OAuth2/OpenID Connect specifications, you should never use it for anything else than tests. That being said, when developing OAuth2 clients you should always verify that the expected requests are being made in your tests.
The motivation behind this library is to provide a setup such that application developers don't feel the need to disable security in their apps when running tests! If you have any issues with regards to OAuth2 and tokens et. al. and consider to disable "security" when running tests please submit an issue or a PR so that we can all help developers and security to live in harmony once again (if ever..)!
Gradle Kotlin DSL
testImplementation("no.nav.security:mock-oauth2-server:$mockOAuth2ServerVersion")
Maven
<dependency>
<groupId>no.nav.security</groupId>
<artifactId>mock-oauth2-server</artifactId>
<version>${mock-oauth2-server.version}</version>
<scope>test</scope>
</dependency>
Docker
docker pull ghcr.io/navikt/mock-oauth2-server:$MOCK_OAUTH2_SERVER_VERSION
The mock-oauth2-server will supply different configurations depending on the url used against the server, more specifically the first path (or context root) element in your request url will specify the issuerId
.
A request to http://localhost:8080/default/.well-known/openid-configuration
will yield an issuerId
of default
with the following configuration:
{
"issuer":"http://localhost:8080/default",
"authorization_endpoint":"http://localhost:8080/default/authorize",
"end_session_endpoint" : "http://localhost:8080/default/endsession",
"revocation_endpoint" : "http://localhost:8080/default/revoke",
"token_endpoint":"http://localhost:8080/default/token",
"userinfo_endpoint":"http://localhost:8080/default/userinfo",
"jwks_uri":"http://localhost:8080/default/jwks",
"introspection_endpoint":"http://localhost:8080/default/introspect",
"response_types_supported":[
"code",
"none",
"id_token",
"token"
],
"response_modes_supported":[
"query",
"fragment",
"form_post"
],
"subject_types_supported":[
"public"
],
"id_token_signing_alg_values_supported":[
"ES256",
"ES384",
"RS256",
"RS384",
"RS512",
"PS256",
"PS384",
"PS512"
],
"code_challenge_methods_supported":[
"plain",
"S256"
]
}
The actual issuer value in a JWT will be iss: "http://localhost:8080/default"
To use another issuer with id anotherissuer
simply make a request to http://localhost:8080/anotherissuer/.well-known/openid-configuration
and the configuration will change accordingly.
val server = MockOAuth2Server()
server.start()
// Can be anything you choose - should uniquely identify your issuer if you have several
val issuerId = "default"
// Discovery url to authorization server metadata
val wellKnownUrl = server.wellKnownUrl(issuerId).toString()
// ......
// Setup your app with metadata from wellKnownUrl and do your testing here
// ......
server.shutdown()
If you need to get a login for a specific user you can use the OAuth2TokenCallback
interface to provide your own or set values in the DefaultOAuth2TokenCallback
@Test
fun loginWithIdTokenForSubjectFoo() {
server.enqueueCallback(
DefaultOAuth2TokenCallback(
issuerId = issuerId,
subject = "foo"
)
)
// Invoke your app here and assert user foo is logged in
}
If you need specific claims in the resulting id_token
- e.g. acr
or a custom claim you can also use the OAuth2TokenCallback
:
@Test
fun loginWithIdTokenForAcrClaimEqualsLevel4() {
server.enqueueCallback(
DefaultOAuth2TokenCallback(
issuerId = issuerId,
claims = mapOf("acr" to "Level4")
)
)
// Invoke your app here and assert acr=Level4 is present in id_token
}
val token: SignedJWT = oAuth2Server.issueToken(issuerId, "someclientid", DefaultOAuth2TokenCallback())
//use your favourite HTTP client to invoke your API and attach the serialized token
val request = // ....
request.addHeader("Authorization", "Bearer ${token.serialize()}")
If you for some reason need to manipulate the system time/clock you can configure the OAuth2TokenProvider to use a specific time, resulting in the iat
claim being set to that time:
@Test
fun testWithSpecificTime() {
val server = MockOAuth2Server(
config = OAuth2Config(
tokenProvider = OAuth2TokenProvider(systemTime = Instant.parse("2020-01-21T00:00:00Z")
)
)
val token = server.issueToken(issuerId = "issuer1")
// do whatever token testing you need to do here and assert the token has iat=2020-01-21T00:00:00Z
}
Have a look at some examples in both Java and Kotlin in the src/test directory:
You can retrieve URLs from the server with the correct port and issuerId etc. by invoking one of the fun *Url(issuerId: String): HttpUrl
functions/methods:
val server = MockOAuth2Server()
server.start()
val wellKnownUrl = server.wellKnownUrl("yourissuer")
// will result in the following url:
// http://localhost:<a random port>/yourissuer/.well-known/openid-configuration
The standalone server will default to port 8080
and can be started by invoking main()
in StandaloneMockOAuth2Server.kt
(in kotlin) or StandaloneMockOAuth2ServerKt
(in Java)
On Windows, it's easier to run the server in docker while specifying the host as localhost, e.g. docker run -p 8080:8080 -h localhost $IMAGE_NAME
Note If you want to check if the server is up and running you can visit
/isalive
and see if you get a 200 in return.
The standalone server supports the following configuration by ENV
variables:
Variable | Description |
---|---|
SERVER_HOSTNAME |
Lets the standalone server bind to a specific hostname, by default it binds to 0.0.0.0 |
SERVER_PORT or PORT |
The port that the standalone server will listen to, defaults to 8080 . The PORT environment variable may be used to run the Docker image on Heroku as per the documentation here. |
JSON_CONFIG_PATH |
The absolute path to a json file containing configuration about the OAuth2 part of the server (OAuth2Config ). More details on the format below. |
JSON_CONFIG |
The actual JSON content of OAuth2Config , this ENV var takes precedence over the JSON_CONFIG_PATH var. More details on the format below. |
LOG_LEVEL |
How verbose the root logging output is, defaults to INFO |
LOGBACK_CONFIG |
You can override the default logging config in logback-standalone.xml with a path to your own logback xml file. |
The JSON_CONFIG lets you configure the contents of the OAuth2Config
class using JSON.
Example:
{
"interactiveLogin": true,
"httpServer": "NettyWrapper",
"tokenCallbacks": [
{
"issuerId": "issuer1",
"tokenExpiry": 120,
"requestMappings": [
{
"requestParam": "code",
"match": "code1",
"claims": {
"sub": "subByCode",
"aud": [
"audByCode"
]
}
}
]
},
{
"issuerId": "issuer2",
"requestMappings": [
{
"requestParam": "someparam",
"match": "somevalue",
"claims": {
"sub": "subBySomeParam",
"aud": [
"audBySomeParam"
]
}
}
]
}
]
}
A token provider can support different signing
algorithms. Configure your token provider and
add this to your config with preferred JWS algorithm
:
{
"tokenProvider" : {
"keyProvider" : {
"algorithm" : "ES256"
}
}
}
A token provider can also support a static "systemTime", i.e. the time for when the token is issued (iat
claim) if you have tests that require a specific time.
The following configuration will set the system time to 2020-01-21T00:00:00Z
:
{
"tokenProvider" : {
"systemTime" : "2020-01-21T00:00:00Z"
}
}
Property | Description | |
---|---|---|
interactiveLogin |
true or false , enables login screen when redirecting to server /authorize endpoint |
|
loginPagePath |
An optional string refering to a html file that is served as login page. This page needs to contain a form that posts a username and optionally a claims field. See src/test/resource/login.example.html as an example. |
|
staticAssetsPath |
The path to a directory containing static resources/assets. Lets you serve your own static resources from the server. Resources are served under the /static URL path. E.g. http://localhost:8080/static/myimage.svg or by reference /static/myimage.svg from the login page. |
|
rotateRefreshToken |
true or false , setting to true will generate a new unique refresh token when using the refresh_token grant. |
|
httpServer |
A string identifying the httpserver to use. Must match one of the following enum values: MockWebServerWrapper or NettyWrapper |
|
tokenCallbacks |
A list of RequestMappingTokenCallback that lets you specify which token claims to return when a token request matches the specified condition. |
From the first JSON example above:
A token request to http://localhost:8080/issuer1/token
with parameter code
equal to code1
will match the first tokenCallback
:
{
"issuerId": "issuer1",
"tokenExpiry": 120,
"requestMappings": [
{
"requestParam": "code",
"match": "code1",
"claims": {
"sub": "subByCode",
"aud": [
"audByCode"
]
}
}
]
}
and return a token response containing a token with the following claims:
{
"sub": "subByCode",
"aud": "audByCode",
"nbf": 1616416942,
"iss": "http://localhost:54905/issuer1",
"exp": 1616417062,
"iat": 1616416942,
"jti": "28697333-6f25-4b1f-b2c2-409ce010933a"
}
Earlier versions of this documentation used scope
as requestParam
in the example. The scope request parameter is no logner supported by nimbus due to c960757, so the call back needs to be configured with a different key.
Use variable clientId
to set sub
claim for Client Credentials Grant dynamically.
A token request with client credentials where clientId = myClientId
and tokenCallback
:
{
"issuerId": "issuer1",
"tokenExpiry": 120,
"requestMappings": [
{
"requestParam": "code",
"match": "code1",
"claims": {
"sub": "${clientId}",
"aud": [
"audByCode"
]
}
}
]
}
will return a token response containing a token with the following claims:
{
"sub": "myClientId",
"aud": "audByCode",
"nbf": 1616416942,
"iss": "http://localhost:54905/issuer1",
"exp": 1616417062,
"iat": 1616416942,
"jti": "28697333-6f25-4b1f-b2c2-409ce010933a"
}
Build to local docker daemon
./gradlew -Djib.from.platforms=linux/amd64 jibDockerBuild # or alternatively -Djib.from.platforms=linux/arm64 for ARM64
Run container
docker run -p 8080:8080 $IMAGE_NAME
In order to get container-to-container networking to work smoothly alongside browser interaction you must specify a host entry in your hosts
file, 127.0.0.1 host.docker.internal
and set hostname
in the mock-oauth2-server service in your docker-compose.yaml
file:
version: '3.7'
services:
your_app:
build: .
ports:
- 8080:8080
mock-oauth2-server:
image: ghcr.io/navikt/mock-oauth2-server:$MOCK_OAUTH2_SERVER_VERSION
ports:
- 8080:8080
hostname: host.docker.internal
The debugger is a OAuth2 client implementing the authorization_code
flow with a UI for debugging (e.g. request parameters).
Point your browser to http://localhost:8080/default/debugger to check it out.
In order to enable HTTPS you can either provide your own keystore or let the server generate one for you.
You need to supply the server with an SSL config, in order to do that you must specify your chosen server type in OAuth2Config
and
pass in the SSL config to your server.
Generate keystore:
val ssl = Ssl()
val server = MockOAuth2Server(
OAuth2Config(httpServer = MockWebServerWrapper(ssl))
)
This will generate a SSL certificate for localhost
and can be added to your client's truststore by getting the ssl config:
ssl.sslKeystore.keyStore
Bring your own:
val ssl = Ssl(
SslKeystore(
keyPassword = "",
keystoreFile = File("src/test/resources/localhost.p12"),
keystorePassword = "",
keystoreType = SslKeystore.KeyStoreType.PKCS12
)
)
val server = MockOAuth2Server(
OAuth2Config(httpServer = MockWebServerWrapper(ssl))
)
In order to enable HTTPS for the server in Docker or standalone mode you can either make the server generate the keystore or bring your own.
Generate keystore:
{
"httpServer" : {
"type" : "NettyWrapper",
"ssl" : {}
}
}
Bring your own:
{
"httpServer" : {
"type" : "NettyWrapper",
"ssl" : {
"keyPassword" : "",
"keystoreFile" : "src/test/resources/localhost.p12",
"keystoreType" : "PKCS12",
"keystorePassword" : ""
}
}
}
Find the newest version of gradle here: https://gradle.org/releases/
Then run this command:
./gradlew wrapper --gradle-version $gradleVersion
Remeber to also update the gradle version in the build.gradle.kts file
gradleVersion = "$gradleVersion"
This project is currently maintained by the organisation @navikt.
If you need to raise an issue or question about this library, please create an issue here and tag it with the appropriate label.
For contact requests within the @navikt org, you can use the Slack channel #pig_sikkerhet
If you need to contact anyone directly, please see contributors.
To get started, please fork the repo and checkout a new branch. You can then build the library with the Gradle wrapper
./gradlew build
See more info in CONTRIBUTING.md
This library is licensed under the MIT License