From ff4b4ff61b1e1a4ef8a0eb21466871a6b9d6b64f Mon Sep 17 00:00:00 2001
From: HuangHai <10402852@qq.com>
Date: Mon, 24 Mar 2025 13:32:29 +0800
Subject: [PATCH] 'commit'

---
 AI/WxMini/Milvus/T5_search_near_data.py | 15 ++++++----
 AI/WxMini/Milvus/X3_insert_data.py      | 39 +++++++++++++++++++++----
 2 files changed, 44 insertions(+), 10 deletions(-)

diff --git a/AI/WxMini/Milvus/T5_search_near_data.py b/AI/WxMini/Milvus/T5_search_near_data.py
index ac32201c..f5edcd8b 100644
--- a/AI/WxMini/Milvus/T5_search_near_data.py
+++ b/AI/WxMini/Milvus/T5_search_near_data.py
@@ -1,4 +1,5 @@
 import time
+import jieba  # 导入 jieba 分词库
 from WxMini.Milvus.Utils.MilvusCollectionManager import MilvusCollectionManager
 from WxMini.Milvus.Utils.MilvusConnectionPool import *
 from WxMini.Milvus.Config.MulvusConfig import *
@@ -12,11 +13,14 @@ print(f"模型加载成功，词向量维度: {model.vector_size}")
 
 # 将文本转换为嵌入向量
 def text_to_embedding(text):
-    # 直接使用全句进行向量计算
-    if text in model:
-        embedding = model[text]
-        print(f"生成的全句向量: {embedding[:5]}...")  # 打印前 5 维
-        return embedding
+    words = jieba.lcut(text)  # 使用 jieba 分词
+    print(f"文本: {text}, 分词结果: {words}")
+    embeddings = [model[word] for word in words if word in model]  # 获取有效词向量
+    print(f"有效词向量数量: {len(embeddings)}")
+    if embeddings:
+        avg_embedding = sum(embeddings) / len(embeddings)  # 计算平均向量
+        print(f"生成的平均向量: {avg_embedding[:5]}...")  # 打印前 5 维
+        return avg_embedding
     else:
         print("未找到有效词，返回零向量")
         return [0.0] * model.vector_size
@@ -40,6 +44,7 @@ input_text = input("请输入一句话：")  # 例如：“我今天心情不太
 
 # 7. 将文本转换为嵌入向量
 current_embedding = text_to_embedding(input_text)
+print(f"当前文本的向量: {current_embedding[:5]}...")  # 打印前 5 维
 
 # 8. 查询与当前对话最相关的历史对话
 search_params = {
diff --git a/AI/WxMini/Milvus/X3_insert_data.py b/AI/WxMini/Milvus/X3_insert_data.py
index d6665500..9b6c499c 100644
--- a/AI/WxMini/Milvus/X3_insert_data.py
+++ b/AI/WxMini/Milvus/X3_insert_data.py
@@ -1,9 +1,18 @@
 from WxMini.Milvus.Config.MulvusConfig import *
 from WxMini.Milvus.Utils.MilvusCollectionManager import MilvusCollectionManager
 from WxMini.Milvus.Utils.MilvusConnectionPool import *
+from WxMini.TtsConfig import *
 from gensim.models import KeyedVectors
+
 import jieba
 import time
+from openai import OpenAI
+
+# 初始化 OpenAI 客户端
+client = OpenAI(
+    api_key=MODEL_API_KEY,
+    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
+)
 
 # 1. 加载预训练的 Word2Vec 模型
 model_path = MS_MODEL_PATH
@@ -36,14 +45,34 @@ collection_manager = MilvusCollectionManager(collection_name)
 
 # 5. 输入一个用户问题
 user_input = input("请输入一句话：")  # 例如：“我今天心情不太好”
-model_response = "我没听懂，能再说一遍吗？"  # 大模型的固定回复
+
+# 6. 调用大模型生成回复
+try:
+    response = client.chat.completions.create(
+        model=MODEL_NAME,
+        messages=[
+            {"role": "system", "content": "你是一个非常好的聊天伙伴，可以疏导用户，帮他解压,一句控制在20字以内。"},
+            {"role": "user", "content": user_input}
+        ],
+        max_tokens=500
+    )
+    if response.choices and response.choices[0].message.content:
+        model_response = response.choices[0].message.content.strip()
+    else:
+        model_response = "大模型未返回有效结果"
+except Exception as e:
+    model_response = f"调用大模型失败: {str(e)}"
+
+print(f"大模型回复: {model_response}")
+
+# 7. 获取当前时间和会话 ID
 timestamp = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())  # 当前时间
 session_id = "session_001"  # 会话 ID（可以根据需要动态生成）
 
-# 6. 将用户问题转换为嵌入向量
+# 8. 将用户问题转换为嵌入向量
 user_embedding = text_to_embedding(user_input)
 
-# 7. 插入数据，确保字段顺序与集合定义一致
+# 9. 插入数据，确保字段顺序与集合定义一致
 entities = [
     [session_id],  # session_id
     [user_input],  # user_input
@@ -54,8 +83,8 @@ entities = [
 collection_manager.insert_data(entities)
 print("数据插入成功。")
 
-# 8. 释放连接
+# 10. 释放连接
 milvus_pool.release_connection(connection)
 
-# 9. 关闭连接池
+# 11. 关闭连接池
 milvus_pool.close()
\ No newline at end of file