feat: add streaming learning assistant and knowledge base scaffolding

app/services/embedding_service.py

@@ -0,0 +1,22 @@
+import time
+
+from app.core.config import settings
+from app.integrations.embedding_adapter import OpenAICompatibleEmbeddingClient
+
+
+class EmbeddingService:
+    """Embedding 服务：按配置批量调用向量模型，控制批大小和耗时统计。"""
+
+    def __init__(self, client: OpenAICompatibleEmbeddingClient | None = None) -> None:
+        self.client = client or OpenAICompatibleEmbeddingClient()
+
+    async def embed_texts(self, texts: list[str]) -> tuple[list[list[float]], int]:
+        """批量向量化：按 EMBEDDING_BATCH_SIZE 分批生成向量并返回总耗时。"""
+        start = time.perf_counter()
+        vectors: list[list[float]] = []
+        batch_size = max(1, settings.embedding_batch_size)
+        for index in range(0, len(texts), batch_size):
+            batch = texts[index : index + batch_size]
+            batch_vectors, _usage = await self.client.embed_texts(batch)
+            vectors.extend(batch_vectors)
+        return vectors, int((time.perf_counter() - start) * 1000)