Fix Docker Compose models attribute and tests

.github/workflows/smoke-tests.yml

@@ -18,8 +18,10 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
 
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+      - name: Set up Docker Compose
+        uses: docker/[email protected]
+        with:
+          version: v2.40.3
 
       - name: Install Docker Model Plugin
         run: |
@@ -86,12 +88,13 @@ jobs:
             -H "Content-Type: application/json" \
             -d '{"message": "Hello"}' \
             --max-time 30)
-
-          if [[ $? -eq 0 ]] && [[ "$RESPONSE" == *"response"* ]]; then
+          if [[ -n "$RESPONSE" ]] && jq -e '.response' <<< "$RESPONSE" > /dev/null 2>&1; then
             echo "✅ Chat API test passed"
+            echo "Response: $(jq -r '.response' <<< "$RESPONSE" | head -c 200)..."
           else
-            echo "❌ Chat API test failed: $RESPONSE"
-            # Don't fail the test for now as model interaction might be flaky
+            echo "❌ Chat API test failed - empty response, invalid JSON, or missing 'response' key"
+            echo "Raw response: $RESPONSE"
+            exit 1
           fi
 
           # Test model info endpoint
@@ -100,11 +103,13 @@ jobs:
             -H "Content-Type: application/json" \
             -d '{"message": "!modelinfo"}' \
             --max-time 10)
-
-          if [[ $? -eq 0 ]] && [[ "$MODEL_INFO" == *"model"* ]]; then
+          if [[ -n "$MODEL_INFO" ]] && jq -e '.model' <<< "$MODEL_INFO" > /dev/null 2>&1; then
             echo "✅ Model info test passed"
+            echo "Model: $(jq -r '.model' <<< "$MODEL_INFO")"
           else
-            echo "❌ Model info test failed: $MODEL_INFO"
+            echo "❌ Model info test failed - empty response, invalid JSON, or missing 'model' key"
+            echo "Raw response: $MODEL_INFO"
+            exit 1
           fi
 
           echo "✅ Smoke tests completed for ${{ matrix.project }}"

go-genai/docker-compose.yml

@@ -19,4 +19,4 @@ services:
 models:
   llama:
     model: ai/llama3.2:1B-Q8_0
-    context_size: 2048
+    context_size: 2048

node-genai/docker-compose.yml

@@ -19,4 +19,4 @@ services:
 models:
   llama:
     model: ai/llama3.2:1B-Q8_0
-    context_size: 2048
+    context_size: 2048

py-genai/docker-compose.yml

@@ -20,4 +20,4 @@ services:
 models:
   llama:
     model: ai/llama3.2:1B-Q8_0
-    context_size: 2048
+    context_size: 2048

rust-genai/src/handlers.rs

@@ -63,6 +63,11 @@ pub struct ChatResponse {
     pub response: String,
 }
 
+#[derive(Serialize)]
+pub struct ModelInfoResponse {
+    pub model: String,
+}
+
 #[post("/api/chat")]
 pub async fn chat_api(
     req: HttpRequest,
@@ -79,6 +84,12 @@ pub async fn chat_api(
     if message.len() > 4000 {
         return HttpResponse::BadRequest().json(serde_json::json!({"error": "Message too long (max 4000 chars)"}));
     }
+    // Special command for getting model info
+    if message == "!modelinfo" {
+        return HttpResponse::Ok().json(ModelInfoResponse {
+            model: config.llm_model_name.clone()
+        });
+    }
     if let Some(resp) = cache.get(message) {
         return HttpResponse::Ok().json(ChatResponse { response: resp });
     }