处理兼容问题

app/ai/resume_extractor/extractor.py

@@ -1,8 +1,10 @@
 """简历并行提取：将完整简历文本拆分为5个AI任务并行提取"""
 
 import asyncio
+import re
 
-from langchain_core.output_parsers import JsonOutputParser
+from json_repair import repair_json
+from langchain_core.output_parsers import StrOutputParser
 from langchain_core.prompts import ChatPromptTemplate
 
 from app.ai.models import LLM
@@ -13,12 +15,19 @@ from app.ai.resume_extractor.prompts import (
 from app.core.logger import log
 
 
+def _parse_json(text: str) -> dict:
+    """解析 AI 输出的 JSON，自动去除 markdown 代码块包裹，容错处理"""
+    cleaned = re.sub(r"^```(?:json)?\s*\n?", "", text.strip())
+    cleaned = re.sub(r"\n?```\s*$", "", cleaned)
+    return repair_json(cleaned, return_objects=True)
+
+
 def _build_chain(prompt: str):
-    """构建单个提取链：prompt → LLM → JSON解析"""
+    """构建单个提取链：prompt → LLM → 文本输出"""
     return (
         ChatPromptTemplate.from_messages([("system", prompt), ("human", "{text}")])
         | LLM.JIAYU_CLAUDE_SONNET_4_5.create(temperature=0)
-        | JsonOutputParser()
+        | StrOutputParser()
     )
 
 
@@ -55,7 +64,8 @@ async def extract_all(text: str) -> dict:
 async def _safe_invoke(chain, inp: dict, label: str):
     """单个链调用，失败返回空"""
     try:
-        return await chain.ainvoke(inp)
+        raw = await chain.ainvoke(inp)
+        return _parse_json(raw)
     except Exception as e:
         log.warning(f"AI提取[{label}]失败: {e}")
         return {} if "个人信息" in label else []