feat(api): 统一 LLM JSON 调用层 llm_json_call，按域 Schema 迁移 chat/memoir agents

api/app/agents/memoir/story_route_agent.py

@@ -15,7 +15,7 @@ from app.agents.memoir.prompts import (
 )
 from app.agents.memoir.story_route_payload import build_route_candidate_json
 from app.core.config import settings
-from app.core.langchain_llm import invoke_json_object
+from app.core.llm_call import LLMCallError, llm_json_call
 from app.core.logging import get_logger
 from app.features.story.models import Story
 
@@ -132,23 +132,23 @@ class StoryRouteAgent:
             batch_transcript=batch_transcript,
             candidate_stories_json=payload,
         )
-        try:
-            raw = invoke_json_object(
-                llm,
-                prompt,
-                max_tokens=1024,
-                agent="StoryRouteAgent.decide",
-            ).strip()
-            data = json.loads(raw)
-            decision = StoryRouteDecision.model_validate(data)
-        except Exception as e:
-            logger.warning("StoryRouteAgent 解析失败: {}", e)
+
+        def _decide_fallback() -> StoryRouteDecision:
             return StoryRouteDecision(
                 decision="new_story",
                 new_story_title=None,
                 reason="parse_error",
             )
 
+        decision = llm_json_call(
+            llm,
+            prompt,
+            StoryRouteDecision,
+            max_tokens=settings.memoir_story_route_max_tokens,
+            agent="StoryRouteAgent.decide",
+            fallback_factory=_decide_fallback,
+        )
+
         if decision.decision == "append_story":
             tid = decision.target_story_id
             if not tid or tid not in valid_story_ids:
@@ -188,15 +188,14 @@ class StoryRouteAgent:
             candidate_stories_json=payload,
         )
         try:
-            raw = invoke_json_object(
+            plan = llm_json_call(
                 llm,
                 prompt,
-                max_tokens=4096,
+                StoryBatchPlan,
+                max_tokens=settings.memoir_story_batch_plan_max_tokens,
                 agent="StoryRouteAgent.plan_batch",
-            ).strip()
-            data = json.loads(raw)
-            plan = StoryBatchPlan.model_validate(data)
-        except Exception as e:
+            )
+        except LLMCallError as e:
             logger.warning("StoryRouteAgent.plan_batch 解析失败: {}", e)
             return None