fix: auto_compact压缩后用摘要+用户问题重新请求模型，而非直接返回摘要

ai/huawei_gateway.py

@@ -412,58 +412,101 @@ def auto_compact(raw_body, real_token, orig_request):
         except Exception:
             final_summary = "\n".join(summaries)
 
-    logger.info(f"auto_compact: 完成, {body_size//1024}KB → {len(final_summary)}字 ({num_batches}批)")
+    logger.info(f"auto_compact: 压缩完成, {body_size//1024}KB → {len(final_summary)}字 ({num_batches}批)")
 
-    # 返回标准 OpenAI 格式（客户端无感知）
+    # ============ 用压缩后的摘要+用户最新问题，重新请求模型 ============
-    if stream:
+    # 提取用户最后一条消息
-        chat_id = str(uuid.uuid4())
+    last_user_msg = None
-        created = int(time.time())
+    for m in reversed(convo_msgs):
+        if m.get('role') == 'user':
+            last_user_msg = m
+            break
 
-        def compact_sse():
+    # 构建压缩后的请求：system + 摘要(作为assistant上下文) + 用户最新问题
-            first = {
+    compact_messages = system_msgs + [
-                "id": chat_id, "object": "chat.completion.chunk", "created": created,
+        {"role": "assistant", "content": f"[上下文压缩摘要]\n{final_summary}"}
-                "model": model,
+    ]
-                "choices": [{"index": 0, "delta": {"role": "assistant", "content": ""}, "finish_reason": None}]
+    if last_user_msg:
-            }
+        compact_messages.append(last_user_msg)
-            yield f"data: {_json.dumps(first, ensure_ascii=False)}\n\n"
 
-            content_chunk = {
+    compact_payload = {
-                "id": chat_id, "object": "chat.completion.chunk", "created": created,
+        "model": model,
-                "model": model,
+        "messages": compact_messages,
-                "choices": [{"index": 0, "delta": {"content": final_summary}, "finish_reason": None}]
+        "max_tokens": max_tokens,
-            }
+        "stream": stream,
-            yield f"data: {_json.dumps(content_chunk, ensure_ascii=False)}\n\n"
+    }
+    # 保留原始请求中的其他参数
+    for k in ('temperature', 'top_p', 'presence_penalty', 'frequency_penalty'):
+        if k in payload:
+            compact_payload[k] = payload[k]
 
-            done_chunk = {
+    compact_body = _json.dumps(compact_payload, ensure_ascii=False).encode('utf-8')
-                "id": chat_id, "object": "chat.completion.chunk", "created": created,
+    compact_headers = {
-                "model": model,
+        'Authorization': f'Bearer {real_token}',
-                "choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}]
+        'Host': TARGET_HOST,
-            }
+        'Content-Type': 'application/json',
-            yield f"data: {_json.dumps(done_chunk, ensure_ascii=False)}\n\n"
+    }
-            yield "data: [DONE]\n\n"
+    target_url = f'https://{TARGET_HOST}/v2/chat/completions'
 
-        return Response(compact_sse(), status=200, headers={
+    logger.info(f"auto_compact: 用压缩上下文重新请求模型 ({len(compact_body)//1024}KB, stream={stream})")
-            'Content-Type': 'text/event-stream;charset=UTF-8',
+
-            'Cache-Control': 'no-cache',
+    try:
-            'X-Auto-Compact': f'batches={num_batches},original_kb={body_size//1024}',
+        resp = http_session.request(
-        })
+            method='POST', url=target_url, headers=compact_headers,
-    else:
+            data=compact_body, allow_redirects=False,
-        return {
+            timeout=UPSTREAM_TIMEOUT_MAX, stream=True
-            "id": f"chatcmpl-compact-{int(time.time())}",
+        )
-            "object": "chat.completion",
+
-            "created": int(time.time()),
+        # 401 重试
-            "model": model,
+        if resp.status_code == 401:
-            "choices": [{
+            resp.close()
-                "index": 0,
+            cache.blacklist_current()
-                "message": {"role": "assistant", "content": final_summary},
+            new_token = find_token_in_memory()
-                "finish_reason": "stop"
+            if new_token:
-            }],
+                compact_headers['Authorization'] = f'Bearer {new_token}'
-            "usage": {
+                resp = http_session.request(
-                "prompt_tokens": body_size // 4,
+                    method='POST', url=target_url, headers=compact_headers,
-                "completion_tokens": len(final_summary),
+                    data=compact_body, allow_redirects=False,
-                "total_tokens": body_size // 4 + len(final_summary)
+                    timeout=UPSTREAM_TIMEOUT_MAX, stream=True
-            }
+                )
-        }
+
+        if resp.status_code != 200:
+            try:
+                err_body = resp.content[:500]
+                logger.error(f"auto_compact: 重新请求模型失败: HTTP {resp.status_code} - {err_body.decode('utf-8', errors='replace')}")
+            except:
+                logger.error(f"auto_compact: 重新请求模型失败: HTTP {resp.status_code}")
+            resp.close()
+            # 降级：返回摘要
+            return {"error": {"message": f"压缩后重新请求失败(HTTP {resp.status_code})，上下文摘要: {final_summary[:500]}", "type": "server_error"}}, 502
+
+        # 流式转发
+        content_type = resp.headers.get('Content-Type', '')
+        if 'text/event-stream' in content_type or stream:
+            skip_h = {'transfer-encoding', 'content-encoding', 'content-length', 'connection', 'keep-alive', 'upgrade'}
+            resp_headers = [(k, v) for k, v in resp.headers.items() if k.lower() not in skip_h]
+            resp_headers.append(('X-Auto-Compact', f'batches={num_batches},original_kb={body_size//1024}'))
+
+            def compact_stream():
+                try:
+                    for chunk in resp.iter_content(chunk_size=16384):
+                        if chunk:
+                            yield chunk
+                finally:
+                    resp.close()
+
+            return Response(compact_stream(), status=200, headers=resp_headers, direct_passthrough=True)
+        else:
+            # 非流式
+            content = resp.content
+            resp.close()
+            # 在响应头中标记经过了压缩
+            return Response(content, status=200,
+                           headers={'Content-Type': 'application/json', 'X-Auto-Compact': f'batches={num_batches},original_kb={body_size//1024}'})
+
+    except Exception as e:
+        logger.error(f"auto_compact: 重新请求模型异常: {e}")
+        return {"error": {"message": f"压缩后请求异常: {str(e)}", "type": "server_error"}}, 500
 
 
 # ================= 全局请求日志（捕获所有请求，包括404） =================