You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

52 lines
1.8 KiB

import numpy as np
import time
from WxMini.Milvus.Utils.MilvusCollectionManager import MilvusCollectionManager
from WxMini.Milvus.Utils.MilvusConnectionPool import *
from WxMini.Milvus.Config.MulvusConfig import *
# 1. 使用连接池管理 Milvus 连接
milvus_pool = MilvusConnectionPool(host=MS_HOST, port=MS_PORT, max_connections=MS_MAX_CONNECTIONS)
# 2. 从连接池中获取一个连接
connection = milvus_pool.get_connection()
# 3. 初始化集合管理器
collection_name = MS_COLLECTION_NAME
collection_manager = MilvusCollectionManager(collection_name)
# 4. 加载集合到内存
collection_manager.load_collection()
# 5. 模拟当前对话的嵌入向量
current_embedding = np.random.random(128).tolist() # 随机生成一个 128 维向量
# 6. 查询与当前对话最相关的历史对话
search_params = {
"metric_type": "L2", # 使用 L2 距离度量方式
"params": {"nprobe": 100} # 设置 IVF_FLAT 的 nprobe 参数
}
start_time = time.time()
results = collection_manager.search(current_embedding, search_params, limit=2)
end_time = time.time()
# 7. 输出查询结果
print("当前对话的嵌入向量:", current_embedding)
print("最相关的历史对话:")
if results:
for hits in results:
for hit in hits:
try:
text = collection_manager.query_text_by_id(hit.id)
print(f"- {text} (距离: {hit.distance})")
except Exception as e:
print(f"查询失败: {e}")
else:
print("未找到相关历史对话,请检查查询参数或数据。")
# 8. 输出查询耗时
print(f"查询耗时: {end_time - start_time:.4f}")
# 9. 释放连接
milvus_pool.release_connection(connection)
# 10. 关闭连接池
milvus_pool.close()