feat(testing): add mcp integration tests

.github/workflows/integration-tests.yml

@@ -0,0 +1,31 @@
+name: Integration Tests
+
+on:
+  push:
+  pull_request:
+
+jobs:
+  integration-tests:
+    runs-on: ubuntu-24.04-arm
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+          cache: pip
+
+      - name: Install test dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -r mcp-local/tests/requirements.txt
+
+      - name: Build MCP Docker image
+        run:  docker buildx build -f mcp-local/Dockerfile -t arm-mcp .
+
+      - name: Run integration tests
+        env:
+          MCP_IMAGE: arm-mcp:latest
+        run: pytest -v mcp-local/tests/test_mcp.py

.gitignore

@@ -0,0 +1,27 @@
+# Virtual environments
+venv/
+.venv/
+env/
+.env/
+embedding-generation/venv/
+embedding-generation/yaml_data
+embedding-generation/info
+
+# Python
+__pycache__/
+*.py[cod]
+*.egg-info/
+dist/
+build/
+.eggs/
+*.egg
+
+# IDE
+.idea/
+.vscode/
+*.swp
+*.swo
+
+# OS
+.DS_Store
+Thumbs.db

README.md

@@ -28,13 +28,13 @@ If you would prefer to use a pre-built, multi-arch image, the official image can
 From the root of this repository:
 
 ```bash
-docker buildx build --platform linux/arm64,linux/amd64 -f mcp-local/Dockerfile -t arm-mcp mcp-local
+docker buildx build --platform linux/arm64,linux/amd64 -f mcp-local/Dockerfile -t arm-mcp .
 ```
 
 For a single-platform build (faster):
 
 ```bash
-docker buildx build -f mcp-local/Dockerfile -t arm-mcp mcp-local
+docker buildx build -f mcp-local/Dockerfile -t arm-mcp .
 ```
 
 ### 2. Configure Your MCP Client
@@ -119,6 +119,19 @@ After updating the configuration, restart your MCP client to load the Arm MCP se
   - `Dockerfile`: Multi-stage Docker build
 - **`embedding-generation/`**: Scripts for regenerating the knowledge base from source documents
 
+## Integration Testing
+
+### Pre-requisites
+
+- Build the mcp server docker image
+- Install the required test packages using - `pip install -r tests/requirements.txt` within the `mcp_local` directory.
+
+### Testing Steps
+
+- Run the test script - `python -m pytest -s tests/test_mcp.py`
+- Check if following 2 docker containers have started - **mcp server** & **testcontainer**
+- All tests should pass without any errors. Warnings can be ignored.
+
 ## Troubleshooting
 
 ### Accessing the Container Shell

embedding-generation/generate-chunks.py

@@ -424,17 +424,29 @@ def createLearningPathChunks():
     learn_url = "https://learn.arm.com/"
     response = http_session.get(learn_url, timeout=60)
     soup = BeautifulSoup(response.text, 'html.parser')
-    for card in soup.find_all(class_='main-topic-card'):
-        if 'tool-install' == card.get('id'): 
-            ig_rel_path = card.get('link')
-            processLearningPath(ig_rel_path,"Install Guide")
+
+    # Process Install Guides separately (directly from /install-guides page)
+    processLearningPath("/install-guides", "Install Guide")
+
+    # Find category links - main-topic-card elements are now wrapped in <a> tags
+    # Look for <a> tags that contain main-topic-card divs
+    for a_tag in soup.find_all('a', href=True):
+        card = a_tag.find(class_='main-topic-card')
+        if card:
+            cat_rel_path = a_tag.get('href')
+            if cat_rel_path is None or cat_rel_path.startswith('http'):
+                continue
+            # Skip non-learning-path links (like /tag/ml/ or install guides button)
+            if not cat_rel_path.startswith('/learning-paths/'):
+                continue
 
-        else:         
-            cat_rel_path = card.get('link')
-            cat_response = http_session.get(learn_url+cat_rel_path, timeout=60)
+            cat_response = http_session.get(learn_url.rstrip('/') + cat_rel_path, timeout=60)
             cat_soup = BeautifulSoup(cat_response.text, 'html.parser')
             for lp_card in cat_soup.find_all(class_="path-card"):
-                lp_url = learn_url + lp_card.get('link')
+                lp_link = lp_card.get('link')
+                if lp_link is None:
+                    continue
+                lp_url = learn_url.rstrip('/') + lp_link
                 # Chunking step
                 processLearningPath(lp_url, "Learning Path")
 

mcp-local/Dockerfile

@@ -43,11 +43,16 @@ RUN rm -f /usr/local/bin/aperf \
            /opt/arm-migration-tools/processwatch \
            /opt/arm-migration-tools/papi
 
+COPY embedding-generation/generate-chunks.py /tmp/embedding-generation/generate-chunks.py
+COPY embedding-generation/local_vectorstore_creation.py /tmp/embedding-generation/local_vectorstore_creation.py
+COPY embedding-generation/requirements.txt /tmp/embedding-generation/requirements.txt
+COPY embedding-generation/urls-to-chunk.csv /tmp/embedding-generation/urls-to-chunk.csv
+
 # Copy embedding data and scripts from the embedding base image
-COPY --from=joestech324/mcp-embedding-base:latest /embedding-data /tmp/embedding-data
+COPY --from=joestech324/mcp-embedding-base:latest /embedding-data/intrinsic_chunks /tmp/embedding-generation/intrinsic_chunks
 
 # Install dependencies for vector database generation
-WORKDIR /tmp/embedding-data
+WORKDIR /tmp/embedding-generation
 RUN pip3 install --no-cache-dir --break-system-packages -r requirements.txt
 
 # Generate vector database
@@ -61,16 +66,16 @@ ENV VIRTUAL_ENV=/app/.venv \
 
 RUN python3 -m venv "$VIRTUAL_ENV" && pip install --upgrade pip
 
-COPY requirements.txt .
+COPY mcp-local/requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
 # Copy generated vector database files
 RUN mkdir -p ./data
-RUN cp /tmp/embedding-data/metadata.json ./data/ && \
-    cp /tmp/embedding-data/usearch_index.bin ./data/
+RUN cp /tmp/embedding-generation/metadata.json ./data/ && \
+    cp /tmp/embedding-generation/usearch_index.bin ./data/
 
-COPY utils/ ./utils/
-COPY server.py .
+COPY mcp-local/utils/ ./utils/
+COPY mcp-local/server.py .
 
 FROM ubuntu:24.04 AS runtime