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_detailed_plain_text from ...common.database import Database from .config import global_config from .topic_identifier import topic_identifier from ..memory_system.memory import memory_graph from random import choice # 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 """ memory_prompt = '' start_time = time.time() # 记录开始时间 topic = topic_identifier.identify_topic_jieba(message_txt) # print(f"\033[1;32m[pb主题识别]\033[0m 主题: {topic}") all_first_layer_items = [] # 存储所有第一层记忆 all_second_layer_items = {} # 用字典存储每个topic的第二层记忆 overlapping_second_layer = set() # 存储重叠的第二层记忆 if topic: # 遍历所有topic for current_topic in topic: first_layer_items, second_layer_items = memory_graph.get_related_item(current_topic, depth=2) # if first_layer_items: # print(f"\033[1;32m[前额叶]\033[0m 主题 '{current_topic}' 的第一层记忆: {first_layer_items}") # 记录第一层数据 all_first_layer_items.extend(first_layer_items) # 记录第二层数据 all_second_layer_items[current_topic] = second_layer_items # 检查是否有重叠的第二层数据 for other_topic, other_second_layer in all_second_layer_items.items(): if other_topic != current_topic: # 找到重叠的记忆 overlap = set(second_layer_items) & set(other_second_layer) if overlap: # print(f"\033[1;32m[前额叶]\033[0m 发现主题 '{current_topic}' 和 '{other_topic}' 有共同的第二层记忆: {overlap}") overlapping_second_layer.update(overlap) # 合并所有需要的记忆 if all_first_layer_items: print(f"\033[1;32m[前额叶]\033[0m 合并所有需要的记忆1: {all_first_layer_items}") if overlapping_second_layer: print(f"\033[1;32m[前额叶]\033[0m 合并所有需要的记忆2: {list(overlapping_second_layer)}") all_memories = all_first_layer_items + list(overlapping_second_layer) if all_memories: # 只在列表非空时选择随机项 random_item = choice(all_memories) memory_prompt = f"看到这些聊天，你想起来{random_item}\n" else: memory_prompt = "" # 如果没有记忆，则返回空字符串 end_time = time.time() # 记录结束时间 print(f"\033[1;32m[回忆耗时]\033[0m 耗时: {(end_time - start_time):.3f}秒") # 输出耗时 #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你现在正在{bot_schedule_now_activity}\n''' #知识构建 start_time = time.time() prompt_info = '' promt_info_prompt = '' prompt_info = self.get_prompt_info(message_txt,threshold=0.5) 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''' promt_info_prompt = '你有一些[知识]，在上面可以参考。' end_time = time.time() print(f"\033[1;32m[知识检索]\033[0m 耗时: {(end_time - start_time):.3f}秒") # print(f"\033[1;34m[调试]\033[0m 获取知识库内容结果: {prompt_info}") # print(f"\033[1;34m[调试信息]\033[0m 正在构建聊天上下文") chat_talking_prompt = '' if group_id: chat_talking_prompt = get_recent_group_detailed_plain_text(self.db, group_id, limit=global_config.MAX_CONTEXT_SIZE,combine = True) chat_talking_prompt = f"以下是群里正在聊天的内容：\n{chat_talking_prompt}" # print(f"\033[1;34m[调试]\033[0m 已从数据库获取群 {group_id} 的消息记录:{chat_talking_prompt}") #激活prompt构建 activate_prompt = '' activate_prompt = f"以上是群里正在进行的聊天，{memory_prompt} 现在昵称为 '{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}。" #检测机器人相关词汇 bot_keywords = ['人机', 'bot', '机器', '入机', 'robot', '机器人'] is_bot = any(keyword in message_txt.lower() for keyword in bot_keywords) if is_bot: is_bot_prompt = '有人说你是人机或者机器人，否定这一事实，攻击或反问他。' else: is_bot_prompt = '' #人格选择 prompt_personality = '' personality_choice = random.random() if personality_choice < 4/6: # 第一种人格 prompt_personality = f'''{activate_prompt}你的网名叫{global_config.BOT_NICKNAME}，曾经是一个学习地质的女大学生，现在学习心理学和脑科学，你会刷贴吧，你正在浏览qq群,{promt_info_prompt}, 现在请你给出日常且口语化的回复，平淡一些，尽量简短一些。{is_bot_prompt} 请注意把握群里的聊天内容，不要回复的太有条理，可以有个性。''' elif personality_choice < 1: # 第二种人格 prompt_personality = f'''{activate_prompt}你的网名叫{global_config.BOT_NICKNAME}，是一个女大学生，你有黑色头发，你会刷小红书, 你正在浏览qq群,{promt_info_prompt}, 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。{is_bot_prompt} 请你表达自己的见解和观点。可以有个性。''' #中文高手(新加的好玩功能) prompt_ger = '' if random.random() < 0.04: prompt_ger += '你喜欢用倒装句' if random.random() < 0.02: prompt_ger += '你喜欢用反问句' if random.random() < 0.01: prompt_ger += '你喜欢用文言文' #额外信息要求 extra_info = '''但是记得回复平淡一些，简短一些,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"{memory_prompt}\n" # prompt += f"{activate_prompt}\n" prompt += f"{prompt_personality}\n" prompt += f"{prompt_ger}\n" prompt += f"{extra_info}\n" return prompt def get_prompt_info(self,message:str,threshold:float): 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,threshold=threshold) else: embedding = get_embedding(message) related_info += self.get_info_from_db(embedding,threshold=threshold) return related_info def get_info_from_db(self, query_embedding: list, limit: int = 1, threshold: float = 0.5) -> str: 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)) # print(f"\033[1;34m[调试]\033[0m获取知识库内容结果: {results}") 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()