feat: wrap LLM clients with TrackedLLMClient in LLMFactory

backend/app/services/llm/llm_factory.py

@@ -7,6 +7,8 @@ from functools import lru_cache
 from .base_client import BaseLLMClient, LLMConfig, LLMProvider, LLMResponse
 from .deepseek_client import DeepSeekClient
 from .qwen_client import QwenClient, QwenVLClient
+from .tracked_client import TrackedLLMClient
+from app.shared.model_usage_tracker import get_model_usage_tracker
 # Keep provider-specific behavior explicit so debugging stays straightforward.
 
 
@@ -45,7 +47,7 @@ class LLMFactory:
         max_tokens: int = 4096,
         temperature: float = 0.7,
         **kwargs
-    ) -> BaseLLMClient:
+    ) -> "BaseLLMClient | TrackedLLMClient":
         """Handle create for the L L M Factory instance."""
         provider_enum = self._parse_provider(provider)
 
@@ -76,11 +78,16 @@ class LLMFactory:
         # Keep provider-specific behavior explicit so debugging stays straightforward.
         client = self._create_client(config)
 
+        # Wrap in TrackedLLMClient so every call site (agentic, HyDE, perception,
+        # compliance, document summarization, main answer generation) is recorded
+        # without each of them needing to know about usage tracking.
+        tracked_client = TrackedLLMClient(client, get_model_usage_tracker())
+
         # Keep provider-specific behavior explicit so debugging stays straightforward.
-        LLMFactory._global_instances[cache_key] = client
+        LLMFactory._global_instances[cache_key] = tracked_client
 
         logger.info(f"LLM客户端创建成功并缓存: {provider} - {model}")
-        return client
+        return tracked_client
 
     def _parse_provider(self, provider: str) -> LLMProvider:
         """Handle parse provider for this module for the L L M Factory instance."""
@@ -137,7 +144,7 @@ class LLMFactory:
 
         return client_class(config)
 
-    def get_cached(self, provider: str, model: Optional[str] = None) -> Optional[BaseLLMClient]:
+    def get_cached(self, provider: str, model: Optional[str] = None) -> "BaseLLMClient | TrackedLLMClient | None":
         """Return cached for the L L M Factory instance."""
         provider_enum = self._parse_provider(provider)
         model = model or DEFAULT_MODELS.get(provider_enum)
@@ -200,7 +207,7 @@ def get_llm_client(
     provider: str = "qwen",
     model: Optional[str] = None,
     **kwargs
-) -> BaseLLMClient:
+) -> "BaseLLMClient | TrackedLLMClient":
     """Return llm client."""
     factory = get_llm_factory()