From 4a2a07f2354ee55ff76c0d264c0e95dafdf7c02a Mon Sep 17 00:00:00 2001
From: z30057876 <zhouweitong@huawei.com>
Date: Sat, 22 Feb 2025 15:14:18 +0800
Subject: [PATCH] =?UTF-8?q?=E4=BF=AE=E6=AD=A3=E9=94=99=E8=AF=AF=EF=BC=9B?=
 =?UTF-8?q?=E7=A7=BB=E5=8A=A8=E8=AF=B7=E6=B1=82=E6=88=AA=E6=96=AD=E9=80=BB?=
 =?UTF-8?q?=E8=BE=91=E4=BD=8D=E7=BD=AE?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 apps/constants.py                    | 1 -
 apps/llm/reasoning.py                | 2 +-
 apps/scheduler/call/api/sanitizer.py | 8 --------
 apps/scheduler/executor/flow.py      | 5 ++++-
 apps/scheduler/vector.py             | 2 +-
 5 files changed, 6 insertions(+), 12 deletions(-)

diff --git a/apps/constants.py b/apps/constants.py
index 39d2dfaa..da02aa25 100644
--- a/apps/constants.py
+++ b/apps/constants.py
@@ -10,7 +10,6 @@ CURRENT_REVISION_VERSION = "0.0.0"
 NEW_CHAT = "New Chat"
 SLIDE_WINDOW_TIME = 60
 SLIDE_WINDOW_QUESTION_COUNT = 10
-MAX_API_RESPONSE_LENGTH = 4096
 MAX_SCHEDULER_HISTORY_SIZE = 3
 
 LOGGER = logging.getLogger("gunicorn.error")
diff --git a/apps/llm/reasoning.py b/apps/llm/reasoning.py
index a9844f46..1b29537c 100644
--- a/apps/llm/reasoning.py
+++ b/apps/llm/reasoning.py
@@ -84,7 +84,7 @@ class ReasoningLLM(metaclass=Singleton):
         result = ""
 
         is_first_chunk = True
-        is_reasoning = True
+        is_reasoning = False
         reasoning_type = ""
 
         async for chunk in stream:
diff --git a/apps/scheduler/call/api/sanitizer.py b/apps/scheduler/call/api/sanitizer.py
index 0a767d4d..bfddfccc 100644
--- a/apps/scheduler/call/api/sanitizer.py
+++ b/apps/scheduler/call/api/sanitizer.py
@@ -6,9 +6,6 @@ import json
 from textwrap import dedent
 from typing import Any, Optional
 
-from untruncate_json import untrunc
-
-from apps.constants import MAX_API_RESPONSE_LENGTH
 from apps.entities.plugin import CallResult
 from apps.scheduler.slot.slot import Slot
 
@@ -80,11 +77,6 @@ class APISanitizer:
                 message=f"调用接口{url}成功，但返回值为空。",
             )
 
-        if len(response_data) > MAX_API_RESPONSE_LENGTH:
-            response_data = response_data[:MAX_API_RESPONSE_LENGTH]
-            response_data = response_data[:response_data.rfind(",") - 1]
-            response_data = untrunc.complete(response_data)
-
         response_data = APISanitizer._process_response_schema(response_data, response_schema)
 
         message = dedent(f"""调用API从外部数据源获取了数据。API和数据源的描述为：{usage}""")
diff --git a/apps/scheduler/executor/flow.py b/apps/scheduler/executor/flow.py
index 560d7dd2..dbd961a7 100644
--- a/apps/scheduler/executor/flow.py
+++ b/apps/scheduler/executor/flow.py
@@ -2,9 +2,12 @@
 
 Copyright (c) Huawei Technologies Co., Ltd. 2023-2024. All rights reserved.
 """
+import json
 import traceback
 from typing import Optional
 
+from untruncate_json import untrunc
+
 from apps.constants import LOGGER, MAX_SCHEDULER_HISTORY_SIZE
 from apps.entities.enum import StepStatus
 from apps.entities.plugin import (
@@ -249,7 +252,7 @@ class Executor:
         tool_info = {
             "name": call_name,
             "description": call_description,
-            "output": call_result.output,
+            "output": untrunc.complete(json.dumps(call_result.output)[:6144]),
         }
         # 更新背景
         self.flow_state.thought = await ExecutorThought().generate(
diff --git a/apps/scheduler/vector.py b/apps/scheduler/vector.py
index 5cf2c746..141e4386 100644
--- a/apps/scheduler/vector.py
+++ b/apps/scheduler/vector.py
@@ -27,7 +27,7 @@ def _get_embedding(text: list[str]) -> list[np.ndarray]:
     :param text: 待向量化文本（多条文本组成List）
     :return: 文本对应的向量（顺序与text一致，也为List）
     """
-    api = config["EMBEDDING_URL"].rstrip("/") + "/embedding"
+    api = config["EMBEDDING_URL"].rstrip("/") + "/embeddings"
 
     headers = {
         "Authorization": f"Bearer {config['EMBEDDING_KEY']}",
-- 
Gitee