Update app.py

app.py

@@ -9,67 +9,102 @@ Changes over v3.5:
   header is not needed and actively harmful in this deployment context.
 """
 
+import asyncio
 import logging
-import os
 import time
 from contextlib import asynccontextmanager
-from typing import Optional
 
-from fastapi import FastAPI, Request, Response
+from fastapi import FastAPI, Request
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import JSONResponse
 
+try:
+    from cachetools import TTLCache
+    _ttl_cache_available = True
+except ImportError:
+    _ttl_cache_available = False
+
 from src.config import get_settings
 from src.services.metrics_service import MetricsService
 from src.services.rag_service import RAGService
 
+logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-# ── Singletons ────────────────────────────────────────────────────────────────
+# ── Module-level singletons (matches original working pattern) ────────────────
+rag_service = RAGService()
+metrics_service = MetricsService(get_settings().metrics_path)
+
+# ── Rate-limit request log ────────────────────────────────────────────────────
+if _ttl_cache_available:
+    _request_log = TTLCache(maxsize=10_000, ttl=120)
+    _request_log_lock = asyncio.Lock()
+else:
+    from collections import defaultdict, deque
+    _request_log = defaultdict(deque)  # type: ignore[assignment]
+    _request_log_lock = asyncio.Lock()
 
-_rag_service: Optional[RAGService] = None
-_metrics_service: Optional[MetricsService] = None
 
+# ── Public accessors ──────────────────────────────────────────────────────────
 
 def get_rag_service() -> RAGService:
-    global _rag_service
-    if _rag_service is None:
-        raise RuntimeError("RAG service not initialised")
-    return _rag_service
+    return rag_service
 
 
 def get_metrics_service() -> MetricsService:
-    global _metrics_service
-    if _metrics_service is None:
-        raise RuntimeError("Metrics service not initialised")
-    return _metrics_service
+    return metrics_service
+
+
+# ── Background metrics flush ──────────────────────────────────────────────────
+
+async def _metrics_flush_loop(interval: int = 30) -> None:
+    while True:
+        await asyncio.sleep(interval)
+        try:
+            await metrics_service.flush()
+        except Exception as exc:
+            logger.warning("Metrics flush error: %s", exc)
 
 
 # ── Lifespan ──────────────────────────────────────────────────────────────────
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    global _rag_service, _metrics_service
-
     settings = get_settings()
     logger.info("LibBee startup — initialising services")
 
-    # Metrics (in-memory)
-    _metrics_service = MetricsService()
-    logger.info("MetricsService ready")
-
-    # RAG service
-    _rag_service = RAGService(settings)
+    # Staff index
     try:
-        await _rag_service.initialise()
-        stats = _rag_service.get_stats()
-        logger.info("RAGService ready — %s", stats)
+        from src.services.staff_service import build_staff_index
+        build_staff_index()
+        logger.info("Staff index built")
     except Exception as exc:
-        logger.error("RAGService init failed: %s", exc)
+        logger.error("Staff index build failed: %s", exc)
+
+    # RAG service
+    if settings.openai_api_key:
+        try:
+            await rag_service.initialize(openai_api_key=settings.openai_api_key)
+            logger.info("RAG service ready — chunks: %d", len(rag_service.bm25_corpus))
+        except Exception as exc:
+            logger.error("RAG initialization failed: %s", exc, exc_info=True)
+    else:
+        logger.warning("OPENAI_API_KEY not set — RAG disabled")
+
+    # Background metrics flush
+    flush_task = asyncio.create_task(_metrics_flush_loop(interval=30))
+    logger.info("LibBee startup complete")
 
     yield
 
-    logger.info("LibBee shutdown")
+    # Shutdown
+    flush_task.cancel()
+    try:
+        await asyncio.wait_for(metrics_service.flush(), timeout=5.0)
+        logger.info("Metrics flushed on shutdown")
+    except Exception as exc:
+        logger.warning("Metrics flush on shutdown failed: %s", exc)
+    logger.info("LibBee shutdown complete")
 
 
 # ── App factory ───────────────────────────────────────────────────────────────
@@ -107,11 +142,10 @@ def create_app() -> FastAPI:
     # ── Security headers middleware ────────────────────────────────────────────
     @app.middleware("http")
     async def security_headers(request: Request, call_next):
-        response: Response = await call_next(request)
-        # X-Frame-Options: SAMEORIGIN is intentionally NOT set here.
-        # HuggingFace Spaces renders the app inside an iframe; setting
-        # SAMEORIGIN would block the HF Space preview panel from loading.
-        # Frame embedding is already controlled by the CSP header below.
+        response = await call_next(request)
+        # X-Frame-Options: SAMEORIGIN intentionally NOT set.
+        # HuggingFace Spaces runs inside an iframe; SAMEORIGIN blocks the
+        # HF Space preview panel. Frame embedding is controlled by CSP below.
         response.headers["X-Content-Type-Options"] = "nosniff"
         response.headers["X-XSS-Protection"] = "1; mode=block"
         response.headers["Referrer-Policy"] = "strict-origin-when-cross-origin"
@@ -128,16 +162,15 @@ def create_app() -> FastAPI:
         )
         return response
 
-    # ── Request timing middleware ─────────────────────────────────────────────
+    # ── Request timing ────────────────────────────────────────────────────────
     @app.middleware("http")
     async def add_process_time_header(request: Request, call_next):
         start_time = time.time()
         response = await call_next(request)
-        process_time = time.time() - start_time
-        response.headers["X-Process-Time"] = str(round(process_time, 4))
+        response.headers["X-Process-Time"] = str(round(time.time() - start_time, 4))
         return response
 
-    # ── Health / root endpoints ───────────────────────────────────────────────
+    # ── Health / utility endpoints ────────────────────────────────────────────
     @app.get("/")
     async def root():
         return {
@@ -165,9 +198,9 @@ def create_app() -> FastAPI:
         )
 
     @app.get("/metrics")
-    async def metrics():
+    async def metrics_endpoint():
         try:
-            return get_metrics_service().get_summary()
+            return get_metrics_service().snapshot()
         except Exception as exc:
             return {"error": str(exc)}