Add dind test to validate docker build (#94)

* Add dind test to validate docker build

* Updates to see if we can get this to work

* Add fake model for testing

* Fix all the things, move tests to a separate file

* Whoops

* Rename

* Fix

* Fix name collision in tests

.github/workflows/deploy.yml

@@ -12,9 +12,13 @@ jobs:
     if: github.repository == 'JoshuaC215/agent-service-toolkit'
     uses: ./.github/workflows/test.yml
 
+  test-docker:
+    if: github.repository == 'JoshuaC215/agent-service-toolkit'
+    uses: ./.github/workflows/test-docker.yml
+
   build:
     runs-on: ubuntu-latest
-    needs: test
+    needs: [test, test-docker]
 
     steps:
       - uses: actions/checkout@v4

.github/workflows/test-docker.yml

@@ -11,20 +11,35 @@ permissions:
   contents: read
 
 jobs:
-  docker-build:
+  test-docker:
     runs-on: ubuntu-latest
 
+    services:
+      dind:
+        image: docker:dind
+        ports:
+          - 2375:2375
+        options: >-
+          --privileged
+          --health-cmd "docker info"
+          --health-interval 10s
+          --health-timeout 5s
+          --health-retries 5
+
     steps:
       - uses: actions/checkout@v4
 
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
+        with:
+          driver-opts: network=host
 
       - name: Build service image
         uses: docker/build-push-action@v3
         with:
           context: .
           push: false
+          load: true
           tags: agent-service-toolkit.service:${{ github.sha }}
           file: docker/Dockerfile.service
 
@@ -33,5 +48,55 @@ jobs:
         with:
           context: .
           push: false
+          load: true
           tags: agent-service-toolkit.app:${{ github.sha }}
           file: docker/Dockerfile.app
+
+      - name: Start service container
+        run: docker run -d --name service-container --network host -e USE_FAKE_MODEL=true agent-service-toolkit.service:${{ github.sha }}
+
+      - name: Confirm service starts correctly
+        run: |
+          timeout 30 bash -c '
+            while ! curl -s http://localhost/health; do
+              echo "Waiting for service to be ready..."
+              docker logs service-container
+              sleep 2
+            done
+          '
+
+      - name: Run app container
+        run: docker run -d --name app-container --network host -e AGENT_URL=http://localhost agent-service-toolkit.app:${{ github.sha }}
+
+      - name: Confirm app starts correctly
+        run: |
+          timeout 30 bash -c '
+            while ! curl -s http://localhost:8501/healthz; do
+              echo "Waiting for app to be ready..."
+              docker logs app-container
+              sleep 2
+            done
+          '
+
+      - name: Set up Python 3.12
+        uses: actions/setup-python@v5
+        with:
+          python-version-file: "pyproject.toml"
+      - name: Install uv
+        uses: astral-sh/setup-uv@v2
+        with:
+          version: "0.4.12"
+      - name: Install ONLY CLIENT dependencies with uv
+        run: |
+          uv pip install -r requirements.txt
+          uv pip install pytest pytest-env
+        env:
+          UV_SYSTEM_PYTHON: 1
+      - name: Run integration tests
+        run: pytest tests/integration -v --run-docker
+
+      - name: Clean up containers
+        if: always()
+        run: |
+          docker stop service-container app-container || true
+          docker rm service-container app-container || true

src/agents/models.py

@@ -2,6 +2,7 @@
 
 from langchain_anthropic import ChatAnthropic
 from langchain_aws import ChatBedrock
+from langchain_community.chat_models import FakeListChatModel
 from langchain_core.language_models.chat_models import BaseChatModel
 from langchain_google_genai import ChatGoogleGenerativeAI
 from langchain_groq import ChatGroq
@@ -10,6 +11,7 @@
 # NOTE: models with streaming=True will send tokens as they are generated
 # if the /stream endpoint is called with stream_tokens=True (the default)
 models: dict[str, BaseChatModel] = {}
+
 if os.getenv("OPENAI_API_KEY") is not None:
     models["gpt-4o-mini"] = ChatOpenAI(model="gpt-4o-mini", temperature=0.5, streaming=True)
 if os.getenv("GROQ_API_KEY") is not None:
@@ -26,6 +28,8 @@
     models["bedrock-haiku"] = ChatBedrock(
         model_id="anthropic.claude-3-5-haiku-20241022-v1:0", temperature=0.5
     )
+if os.getenv("USE_FAKE_MODEL") == "true":
+    models["fake"] = FakeListChatModel(responses=["This is a test response from the fake model."])
 
 if not models:
     print("No LLM available. Please set environment variables to enable at least one LLM.")

src/service/service.py

@@ -237,4 +237,10 @@ def history(input: ChatHistoryInput) -> ChatHistory:
         raise HTTPException(status_code=500, detail="Unexpected error")
 
 
+@app.get("/health")
+async def health_check():
+    """Health check endpoint."""
+    return {"status": "ok"}
+
+
 app.include_router(router)

tests/conftest.py

@@ -0,0 +1,19 @@
+import pytest
+
+
+def pytest_addoption(parser):
+    parser.addoption(
+        "--run-docker", action="store_true", default=False, help="run docker integration tests"
+    )
+
+
+def pytest_configure(config):
+    config.addinivalue_line("markers", "docker: mark test as requiring docker containers")
+
+
+def pytest_collection_modifyitems(config, items):
+    if not config.getoption("--run-docker"):
+        skip_docker = pytest.mark.skip(reason="need --run-docker option to run")
+        for item in items:
+            if "docker" in item.keywords:
+                item.add_marker(skip_docker)

tests/integration/test_service_integration.py

@@ -0,0 +1,15 @@
+import pytest
+
+from client import AgentClient
+
+
+@pytest.mark.docker
+def test_service_with_fake_model():
+    """Test the service using the fake model.
+
+    This test requires the service container to be running with USE_FAKE_MODEL=true
+    """
+    client = AgentClient("http://localhost", agent="chatbot")
+    response = client.invoke("Tell me a joke?", model="fake")
+    assert response.type == "ai"
+    assert response.content == "This is a test response from the fake model."