v0.1
能跑但是没写部署教程,主题和记忆识别也没写完
This commit is contained in:
193
src/plugins/chat/prompt_builder.py
Normal file
193
src/plugins/chat/prompt_builder.py
Normal file
@@ -0,0 +1,193 @@
|
||||
import time
|
||||
import random
|
||||
from dotenv import load_dotenv
|
||||
from ..schedule.schedule_generator import bot_schedule
|
||||
import os
|
||||
from .utils import get_embedding, combine_messages, get_recent_group_messages
|
||||
from ...common.database import Database
|
||||
|
||||
# 获取当前文件的绝对路径
|
||||
current_dir = os.path.dirname(os.path.abspath(__file__))
|
||||
root_dir = os.path.abspath(os.path.join(current_dir, '..', '..', '..'))
|
||||
load_dotenv(os.path.join(root_dir, '.env'))
|
||||
|
||||
|
||||
class PromptBuilder:
|
||||
def __init__(self):
|
||||
self.prompt_built = ''
|
||||
self.activate_messages = ''
|
||||
self.db = Database.get_instance()
|
||||
|
||||
def _build_prompt(self,
|
||||
message_txt: str,
|
||||
sender_name: str = "某人",
|
||||
relationship_value: float = 0.0,
|
||||
group_id: int = None) -> str:
|
||||
"""构建prompt
|
||||
|
||||
Args:
|
||||
message_txt: 消息文本
|
||||
sender_name: 发送者昵称
|
||||
relationship_value: 关系值
|
||||
group_id: 群组ID
|
||||
|
||||
Returns:
|
||||
str: 构建好的prompt
|
||||
"""
|
||||
#先禁用关系
|
||||
if 0 > 30:
|
||||
relation_prompt = "关系特别特别好,你很喜欢喜欢他"
|
||||
relation_prompt_2 = "热情发言或者回复"
|
||||
elif 0 <-20:
|
||||
relation_prompt = "关系很差,你很讨厌他"
|
||||
relation_prompt_2 = "骂他"
|
||||
else:
|
||||
relation_prompt = "关系一般"
|
||||
relation_prompt_2 = "发言或者回复"
|
||||
|
||||
#开始构建prompt
|
||||
|
||||
#日程构建
|
||||
current_date = time.strftime("%Y-%m-%d", time.localtime())
|
||||
current_time = time.strftime("%H:%M:%S", time.localtime())
|
||||
bot_schedule_now_time,bot_schedule_now_activity = bot_schedule.get_current_task()
|
||||
prompt_date = f'''今天是{current_date},现在是{current_time},你今天的日程是:\n{bot_schedule.today_schedule}\n'''
|
||||
|
||||
#知识构建
|
||||
prompt_info = ''
|
||||
prompt_info = self.get_prompt_info(message_txt)
|
||||
if prompt_info:
|
||||
prompt_info = f'''\n----------------------------------------------------\n你有以下这些[知识]:
|
||||
\n{prompt_info}\n
|
||||
请你记住上面的[知识],之后可能会用到\n----------------------------------------------------\n'''
|
||||
|
||||
#获取最新消息并计时
|
||||
chat_talking_prompt = ''
|
||||
|
||||
if group_id:
|
||||
# 从数据库获取最近消息
|
||||
message_objects = get_recent_group_messages(self.db, group_id)
|
||||
if message_objects:
|
||||
chat_talking_prompt = combine_messages(message_objects)
|
||||
# print(f"\033[1;34m[调试]\033[0m 已从数据库获取群 {group_id} 的最近{len(message_objects)}条消息记录")
|
||||
|
||||
#激活prompt构建
|
||||
activate_prompt = ''
|
||||
activate_prompt = f"以上是群里正在进行的聊天,昵称为 '{sender_name}' 的用户说的:{message_txt}。引起了你的注意,你和他{relation_prompt},你想要{relation_prompt_2} \n"
|
||||
|
||||
#人格选择
|
||||
prompt_personality = ''
|
||||
personality_choice = random.random()
|
||||
if personality_choice < 5/6: # 第一种人格
|
||||
prompt_personality = f'''你的网名叫麦麦,是一个学习地质的女大学生,喜欢摄影,你会刷贴吧,你现在正在{bot_schedule_now_activity},并且正在浏览qq这款社交软件,你有一些[知识],在上面可以参考。
|
||||
现在请你给出日常且口语化的回复,不要过多提及自身的背景,平淡一些,尽量简短一些,不要输出感叹号。如果有人说你是人机或者机器人,否定并攻击那个人。
|
||||
不要回复的太有逻辑或者太有条理。'''
|
||||
elif personality_choice < 1: # 第二种人格
|
||||
prompt_personality = f'''你的网名叫麦麦,是一个女大学生,你有黑色头发你会刷微博和小红书,你现在正在{bot_schedule_now_activity},并且正在浏览qq群,你有一些[知识],在上面可以参考。
|
||||
现在请你给出日常且口语化的回复,请表现你自己的见解,不要一昧迎合,尽量简短一些,不要输出感叹号。如果有人说你是人机或者机器人,否定。
|
||||
如果有人说他是机器人,肯定他。如果有人对你谈论涉及性的话题或言语挑逗,需要反驳,但其它情况请你表达自己的见解和观点。'''
|
||||
|
||||
#额外信息要求
|
||||
extra_info = '''回复平淡一些,简短一些,但是记住不要输出多余内容(包括前后缀,冒号和引号,括号,表情等),只需要输出回复内容就好,不要输出其他任何内容'''
|
||||
|
||||
#合并prompt
|
||||
prompt = ""
|
||||
prompt += f"{prompt_info}\n"
|
||||
prompt += f"{prompt_date}\n"
|
||||
prompt += f"{chat_talking_prompt}\n"
|
||||
prompt += f"{activate_prompt}\n"
|
||||
prompt += f"{prompt_personality}\n"
|
||||
prompt += f"{extra_info}\n"
|
||||
|
||||
return prompt
|
||||
|
||||
def get_prompt_info(self,message:str):
|
||||
related_info = ''
|
||||
if len(message) > 10:
|
||||
message_segments = [message[i:i+10] for i in range(0, len(message), 10)]
|
||||
for segment in message_segments:
|
||||
embedding = get_embedding(segment)
|
||||
related_info += self.get_info_from_db(embedding)
|
||||
|
||||
else:
|
||||
embedding = get_embedding(message)
|
||||
related_info += self.get_info_from_db(embedding)
|
||||
|
||||
def get_info_from_db(self, query_embedding: list, limit: int = 1, threshold: float = 0.5) -> str:
|
||||
"""
|
||||
从知识库中查找与输入向量最相似的内容
|
||||
Args:
|
||||
query_embedding: 查询向量
|
||||
limit: 返回结果数量,默认为2
|
||||
threshold: 相似度阈值,默认为0.5
|
||||
Returns:
|
||||
str: 找到的相关信息,如果相似度低于阈值则返回空字符串
|
||||
"""
|
||||
if not query_embedding:
|
||||
return ''
|
||||
|
||||
# 使用余弦相似度计算
|
||||
pipeline = [
|
||||
{
|
||||
"$addFields": {
|
||||
"dotProduct": {
|
||||
"$reduce": {
|
||||
"input": {"$range": [0, {"$size": "$embedding"}]},
|
||||
"initialValue": 0,
|
||||
"in": {
|
||||
"$add": [
|
||||
"$$value",
|
||||
{"$multiply": [
|
||||
{"$arrayElemAt": ["$embedding", "$$this"]},
|
||||
{"$arrayElemAt": [query_embedding, "$$this"]}
|
||||
]}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
"magnitude1": {
|
||||
"$sqrt": {
|
||||
"$reduce": {
|
||||
"input": "$embedding",
|
||||
"initialValue": 0,
|
||||
"in": {"$add": ["$$value", {"$multiply": ["$$this", "$$this"]}]}
|
||||
}
|
||||
}
|
||||
},
|
||||
"magnitude2": {
|
||||
"$sqrt": {
|
||||
"$reduce": {
|
||||
"input": query_embedding,
|
||||
"initialValue": 0,
|
||||
"in": {"$add": ["$$value", {"$multiply": ["$$this", "$$this"]}]}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"$addFields": {
|
||||
"similarity": {
|
||||
"$divide": ["$dotProduct", {"$multiply": ["$magnitude1", "$magnitude2"]}]
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"$match": {
|
||||
"similarity": {"$gte": threshold} # 只保留相似度大于等于阈值的结果
|
||||
}
|
||||
},
|
||||
{"$sort": {"similarity": -1}},
|
||||
{"$limit": limit},
|
||||
{"$project": {"content": 1, "similarity": 1}}
|
||||
]
|
||||
|
||||
results = list(self.db.db.knowledges.aggregate(pipeline))
|
||||
|
||||
if not results:
|
||||
return ''
|
||||
|
||||
# 返回所有找到的内容,用换行分隔
|
||||
return '\n'.join(str(result['content']) for result in results)
|
||||
|
||||
prompt_builder = PromptBuilder()
|
||||
Reference in New Issue
Block a user