From b48d5cc8345edff49bce6bdb210b4766bf5f69c9 Mon Sep 17 00:00:00 2001
From: Xiaofeng <1169646434@qq.com>
Date: Wed, 21 May 2025 18:10:26 +0800
Subject: [PATCH] =?UTF-8?q?=E4=BF=AE=E6=94=B9=E6=8E=A8=E7=90=86=E6=A8=A1?=
 =?UTF-8?q?=E5=9E=8B=E5=9C=B0=E5=9D=80?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 apps/expertproducts/views.py | 38 ++++++++++++++++++++++++------------
 1 file changed, 26 insertions(+), 12 deletions(-)

diff --git a/apps/expertproducts/views.py b/apps/expertproducts/views.py
index e192c58..dc725b0 100644
--- a/apps/expertproducts/views.py
+++ b/apps/expertproducts/views.py
@@ -281,17 +281,23 @@ class NegotiationViewSet(viewsets.ModelViewSet):
 
         # 6. 调用大模型
         try:
-            response = client.chat(
-                model="deepseek-r1:70b",
-                messages=messages,
-            )
-            ai_response, current_status_json = self.filter_ai_response(response['message']['content'])
+            vllm_url = "http://localhost:9090/v1/chat/completions"
+            payload = {
+                # "model": "Qwen2.5-14B-Instruct",
+                "messages": messages,
+                "temperature": 0.2,
+                "max_tokens": 2048
+            }
+            resp = requests.post(vllm_url, json=payload)
+            resp.raise_for_status()
+            response_json = resp.json()
+            ai_content = response_json['choices'][0]['message']['content']
+            ai_response, current_status_json = self.filter_ai_response(ai_content)
 
             if negotiation.status == 'draft_ready':
-                if current_status_json['is_resend_email']:
+                if current_status_json and current_status_json.get('is_resend_email'):
                     self._send_contract_to_email(negotiation)
 
-            # print("response: ", response['message']['content'])
         except Exception as e:
             return Response({"error": f"AI服务错误: {str(e)}"}, status=500)
 
@@ -762,11 +768,19 @@ class CreatorSQLSearchAPI(APIView):
         """
 
         # 2. 让大模型生成SQL
-        response = client.chat(
-            model="qwen2.5:32b",
-            messages=[{'role': 'user', 'content': prompt}],
-        )
-        sql = self._extract_sql(response['message']['content'])
+        vllm_url = "http://localhost:9090/v1/chat/completions"
+        payload = {
+            # "model": "Qwen2.5-14B-Instruct",
+            "messages": [
+                {"role": "user", "content": prompt}
+            ],
+            "temperature": 0.2,
+            "max_tokens": 2048
+        }
+        resp = requests.post(vllm_url, json=payload)
+        resp.raise_for_status()
+        response_json = resp.json()
+        sql = self._extract_sql(response_json['choices'][0]['message']['content'])
 
         # 3. 校验SQL（只允许select，防止注入）
         if not sql.strip().lower().startswith('select'):