feat(language-gate): detect non-English queries and return polite response

- Added language detection as first step in AgentRouterUseCase (before intent classification)
- Non-English queries now get a polite bilingual response instead of hallucinating
- Native language messages for: Arabic, Amharic, Somali, Swahili, French
- Generic English fallback for all other languages
- Both execute_chat and execute_stream paths protected
- Includes 3 English follow-up question suggestions to guide user
- Zero LLM calls for non-English queries (fast, no hallucination risk)

src/core/use_cases/agent_router_use_case.py

@@ -6,6 +6,50 @@ from src.core.use_cases.rag_chat_use_case import RagChatUseCase
 from src.core.use_cases.account_use_case import AccountUseCase
 from src.infrastructure.adapters.intent_classifier import intent_classifier
 
+import json
+import uuid
+import logging
+
+logger = logging.getLogger(__name__)
+
+# Languages we support for search (backend retrieval)
+SUPPORTED_SEARCH_LANGUAGES = {"en"}
+
+# Human-readable names for polite response
+LANGUAGE_NAMES = {
+    "ar": "Arabic (العربية)",
+    "am": "Amharic (አማርኛ)",
+    "so": "Somali (Soomaali)",
+    "sw": "Swahili (Kiswahili)",
+    "fr": "French (Français)",
+    "zh": "Chinese (中文)",
+    "es": "Spanish (Español)",
+    "de": "German (Deutsch)",
+    "pt": "Portuguese (Português)",
+    "hi": "Hindi (हिन्दी)",
+    "tr": "Turkish (Türkçe)",
+    "ru": "Russian (Русский)",
+    "it": "Italian (Italiano)",
+    "ja": "Japanese (日本語)",
+    "ko": "Korean (한국어)",
+}
+
+# Polite "not yet supported" message in each language
+LANGUAGE_NOT_SUPPORTED_MESSAGES = {
+    "ar": "مرحباً! أنا ARKI AI. حالياً أدعم اللغة الإنجليزية فقط. دعم اللغة العربية قادم قريباً! 🌍\n\nHello! I'm ARKI AI. I currently support English only. Arabic support is coming soon!",
+    "am": "ሰላም! እኔ ARKI AI ነኝ። አሁን እንግሊዝኛ ቋንቋ ብቻ እጠቀማለሁ። አማርኛ ድጋፍ በቅርቡ ይመጣል! 🌍\n\nHello! I'm ARKI AI. I currently support English only. Amharic support is coming soon!",
+    "so": "Salaan! Waxaan ahay ARKI AI. Hadda waxaan taageeraa Ingiriisiga kaliya. Taageerada Soomaaliga waxay timaadaa dhawaan! 🌍\n\nHello! I'm ARKI AI. I currently support English only. Somali support is coming soon!",
+    "sw": "Habari! Mimi ni ARKI AI. Kwa sasa ninasaidia Kiingereza tu. Msaada wa Kiswahili unakuja hivi karibuni! 🌍\n\nHello! I'm ARKI AI. I currently support English only. Swahili support is coming soon!",
+    "fr": "Bonjour! Je suis ARKI AI. Je supporte actuellement l'anglais uniquement. Le support du français arrive bientôt! 🌍\n\nHello! I'm ARKI AI. I currently support English only. French support is coming soon!",
+}
+
+DEFAULT_NOT_SUPPORTED_MESSAGE = (
+    "Hello! I'm ARKI AI, a real-time news assistant.\n\n"
+    "I currently support **English** queries only.\n"
+    "Support for {lang_name} is coming soon! 🌍\n\n"
+    "Please try asking your question in English and I'll do my best to help."
+)
+
 
 class AgentRouterUseCase:
     """
@@ -23,13 +67,62 @@ class AgentRouterUseCase:
         print(f"DEBUG: Intent Classification: {intent} for query: '{query[:80]}'")
         return intent
 
+    def _detect_language(self, query: str) -> str:
+        """
+        Detect query language. Returns ISO 639-1 code (en, ar, am, so, sw, fr...).
+        Falls back to 'en' if detection fails.
+        """
+        try:
+            from src.infrastructure.adapters.language_detector import language_detector
+            if language_detector:
+                result = language_detector.detect(query)
+                logger.debug(
+                    f"Language detected: {result.language} "
+                    f"(confidence={result.confidence:.2f}, method={result.method})"
+                )
+                return result.language
+        except Exception as e:
+            logger.debug(f"Language detection failed: {e}")
+        return "en"
+
+    def _build_not_supported_response(self, lang_code: str, session_id: str) -> Dict[str, Any]:
+        """Build a polite 'language not supported yet' response."""
+        # Use native language message if available, else generic English
+        if lang_code in LANGUAGE_NOT_SUPPORTED_MESSAGES:
+            message = LANGUAGE_NOT_SUPPORTED_MESSAGES[lang_code]
+        else:
+            lang_name = LANGUAGE_NAMES.get(lang_code, f"your language ({lang_code})")
+            message = DEFAULT_NOT_SUPPORTED_MESSAGE.format(lang_name=lang_name)
+
+        return {
+            "answer": message,
+            "sources": [],
+            "follow_up_questions": [
+                "What is the latest news from Ethiopia?",
+                "Tell me about current events in Africa",
+                "What happened in Ethiopia today?"
+            ],
+            "session_id": session_id,
+            "debug": {
+                "intent": "LANGUAGE_NOT_SUPPORTED",
+                "detected_language": lang_code,
+                "routed_to": "LanguageGate"
+            }
+        }
+
     async def execute_chat(self, request: ChatRequest) -> Dict[str, Any]:
-        intent = self._classify_intent(request.query)
-        # Generate a unique session ID if none provided — never use a shared fallback
         if not request.session_id:
-            import uuid
             request.session_id = str(uuid.uuid4())
         session_id = request.session_id
+
+        # ── Language Gate: detect non-English queries first ───────────────────
+        detected_lang = self._detect_language(request.query)
+        if detected_lang not in SUPPORTED_SEARCH_LANGUAGES:
+            logger.info(f"Non-English query detected: {detected_lang} — returning language gate response")
+            print(f"DEBUG: Language gate triggered: {detected_lang} for query: '{request.query[:60]}'")
+            return self._build_not_supported_response(detected_lang, session_id)
+
+        intent = self._classify_intent(request.query)
         
         if intent == "OTHER":
             print(f"DEBUG: Routing to OTHER (Direct LLM Response)")
@@ -58,13 +151,22 @@ Response:"""
             return await self.rag_chat.execute_chat(request)
 
     async def execute_stream(self, request: ChatRequest, is_guest: bool = False, user_id: int = None) -> AsyncGenerator[str, None]:
-        intent = self._classify_intent(request.query)
-        # Generate a unique session ID if none provided — never use a shared fallback
         if not request.session_id:
-            import uuid
             request.session_id = str(uuid.uuid4())
         session_id = request.session_id
 
+        # ── Language Gate: detect non-English queries first ───────────────────
+        detected_lang = self._detect_language(request.query)
+        if detected_lang not in SUPPORTED_SEARCH_LANGUAGES:
+            logger.info(f"Non-English query detected: {detected_lang} — returning language gate response")
+            print(f"DEBUG: Language gate triggered: {detected_lang} for query: '{request.query[:60]}'")
+            response = self._build_not_supported_response(detected_lang, session_id)
+            yield f"data: {json.dumps(response)}\n\n"
+            yield "data: [DONE]\n\n"
+            return
+
+        intent = self._classify_intent(request.query)
+
         if intent == "OTHER":
             full_answer = ""
             # Identity-safe prompt — never reveals underlying model
@@ -84,14 +186,12 @@ Response:"""
                 yield chunk
                 if chunk.startswith("data: "):
                     try:
-                        import json
                         data = json.loads(chunk[6:])
                         if "token" in data:
                             full_answer += data["token"]
                     except:
                         pass
 
-            import json
             final_response = {
                 "answer": full_answer,
                 "sources": [],