feat: 添加多端点支持和负载均衡功能，增强 LiteLLM 嵌入管理

codex-lens/src/codexlens/semantic/factory.py

@@ -5,7 +5,7 @@ Provides a unified interface for instantiating different embedder backends.
 
 from __future__ import annotations
 
-from typing import Any
+from typing import Any, Dict, List, Optional
 
 from .base import BaseEmbedder
 
@@ -15,6 +15,9 @@ def get_embedder(
     profile: str = "code",
     model: str = "default",
     use_gpu: bool = True,
+    endpoints: Optional[List[Dict[str, Any]]] = None,
+    strategy: str = "latency_aware",
+    cooldown: float = 60.0,
     **kwargs: Any,
 ) -> BaseEmbedder:
     """Factory function to create embedder based on backend.
@@ -29,6 +32,13 @@ def get_embedder(
               Used only when backend="litellm". Default: "default"
         use_gpu: Whether to use GPU acceleration when available (default: True).
                 Used only when backend="fastembed".
+        endpoints: Optional list of endpoint configurations for multi-endpoint load balancing.
+                  Each endpoint is a dict with keys: model, api_key, api_base, weight.
+                  Used only when backend="litellm" and multiple endpoints provided.
+        strategy: Selection strategy for multi-endpoint mode:
+                 "round_robin", "latency_aware", "weighted_random".
+                 Default: "latency_aware"
+        cooldown: Default cooldown seconds for rate-limited endpoints (default: 60.0)
         **kwargs: Additional backend-specific arguments
 
     Returns:
@@ -47,13 +57,40 @@ def get_embedder(
 
         Create litellm embedder:
             >>> embedder = get_embedder(backend="litellm", model="text-embedding-3-small")
+
+        Create rotational embedder with multiple endpoints:
+            >>> endpoints = [
+            ...     {"model": "openai/text-embedding-3-small", "api_key": "sk-..."},
+            ...     {"model": "azure/my-embedding", "api_base": "https://...", "api_key": "..."},
+            ... ]
+            >>> embedder = get_embedder(backend="litellm", endpoints=endpoints)
     """
     if backend == "fastembed":
         from .embedder import Embedder
         return Embedder(profile=profile, use_gpu=use_gpu, **kwargs)
     elif backend == "litellm":
-        from .litellm_embedder import LiteLLMEmbedderWrapper
-        return LiteLLMEmbedderWrapper(model=model, **kwargs)
+        # Check if multi-endpoint mode is requested
+        if endpoints and len(endpoints) > 1:
+            from .rotational_embedder import create_rotational_embedder
+            return create_rotational_embedder(
+                endpoints_config=endpoints,
+                strategy=strategy,
+                default_cooldown=cooldown,
+            )
+        elif endpoints and len(endpoints) == 1:
+            # Single endpoint in list - use it directly
+            ep = endpoints[0]
+            ep_kwargs = {**kwargs}
+            if "api_key" in ep:
+                ep_kwargs["api_key"] = ep["api_key"]
+            if "api_base" in ep:
+                ep_kwargs["api_base"] = ep["api_base"]
+            from .litellm_embedder import LiteLLMEmbedderWrapper
+            return LiteLLMEmbedderWrapper(model=ep["model"], **ep_kwargs)
+        else:
+            # No endpoints list - use model parameter
+            from .litellm_embedder import LiteLLMEmbedderWrapper
+            return LiteLLMEmbedderWrapper(model=model, **kwargs)
     else:
         raise ValueError(
             f"Unknown backend: {backend}. "