From 906c9bbdf422f17458d0f8d54590c7d5785c4ecc Mon Sep 17 00:00:00 2001
From: Untone <anton.rewin@gmail.com>
Date: Sat, 30 Aug 2025 21:18:48 +0300
Subject: [PATCH] search-index-metric

---
 main.py            |  1 +
 services/search.py | 33 +++++++++++++++++++++++++++++++--
 2 files changed, 32 insertions(+), 2 deletions(-)

diff --git a/main.py b/main.py
index 01a88b75..be7597d5 100644
--- a/main.py
+++ b/main.py
@@ -74,6 +74,7 @@ graphql_app = GraphQL(
 
 
 # Оборачиваем GraphQL-обработчик для лучшей обработки ошибок
+
 async def graphql_handler(request: Request) -> Response:
     """
     Обработчик GraphQL запросов с поддержкой middleware и обработкой ошибок.
diff --git a/services/search.py b/services/search.py
index 77330c6a..a307f45f 100644
--- a/services/search.py
+++ b/services/search.py
@@ -9,6 +9,26 @@ import numpy as np
 from settings import MUVERA_INDEX_NAME, SEARCH_MAX_BATCH_SIZE, SEARCH_PREFETCH_SIZE
 from utils.logger import root_logger as logger
 
+# Простые метрики производительности поиска
+class SearchMetrics:
+    def __init__(self):
+        self.indexing_start_time: float | None = None
+        self.documents_indexed: int = 0
+        
+    def start_indexing(self, doc_count: int):
+        self.indexing_start_time = time.time()
+        self.documents_indexed = doc_count
+        print(f"🔍 Индексация {doc_count} документов...")
+        
+    def end_indexing(self):
+        if self.indexing_start_time:
+            duration = time.time() - self.indexing_start_time
+            rate = self.documents_indexed / duration if duration > 0 else 0
+            print(f"✅ Индексация завершена за {duration:.2f}s ({rate:.1f} doc/s)")
+
+# Глобальный экземпляр метрик
+search_metrics = SearchMetrics()
+
 # Global collection for background tasks
 background_tasks: List[asyncio.Task] = []
 
@@ -127,7 +147,12 @@ class SearchService:
             # Get Muvera service info
             if self.muvera_client:
                 muvera_info = await self.muvera_client.info()
-                return {"status": "enabled", "provider": "muvera", "mode": "local", "muvera_info": muvera_info}
+                return {
+                    "status": "enabled", 
+                    "provider": "muvera", 
+                    "mode": "local", 
+                    "muvera_info": muvera_info
+                }
             return {"status": "error", "message": "Muvera client not available"}
         except Exception:
             logger.exception("Failed to get search info")
@@ -280,6 +305,8 @@ class SearchService:
             )
             return
 
+        # Запускаем метрики индексации
+        search_metrics.start_indexing(len(shouts))
         start_time = time.time()
         logger.info(f"Starting Muvera bulk indexing of {len(shouts)} documents")
 
@@ -291,7 +318,7 @@ class SearchService:
             try:
                 # Prepare document data for Muvera
                 doc_data: Dict[str, Any] = {
-                    "id": str(shout.id),
+                    "id": str(getattr(shout, "id", "")),
                     "title": getattr(shout, "title", "") or "",
                     "body": "",
                     "metadata": {},
@@ -351,6 +378,8 @@ class SearchService:
                     f"Muvera bulk indexing completed in {elapsed:.2f}s: "
                     f"{len(documents)} documents indexed, {total_skipped} shouts skipped"
                 )
+                # Завершаем метрики индексации
+                search_metrics.end_indexing()
             except Exception as e:
                 logger.exception(f"Muvera bulk indexing failed: {e}")
         else: