-
@ 🔞鸟巢🔞
2025-04-22 16:33:01
### aibot源代码第一部分
```python
import requests
import time
import ssl
import os
import json
import uuid
from pynostr.event import Event, EventKind
from pynostr.relay_manager import RelayManager
from pynostr.message_type import ClientMessageType
from pynostr.key import PrivateKey
from pynostr.filters import FiltersList, Filters
from pynostr.encrypted_dm import EncryptedDirectMessage
from pynostr.utils import get_timestamp
from openai import OpenAI
import gc
from datetime import datetime
import re
def add_relay(relay_manager, relays):
for relay in relays.split(","):
print("Adding relay: " + relay)
relay_manager.add_relay(relay)
print("Added relay: " + relay)
return True
relays = "wss://nostr.data.haus,wss://relay.nostr.net,wss://n.ok0.org,wss://ditto.pub/relay,wss://purplerelay.com,wss://relay.nostr.band,wss://relay.damus.io"
relay_manager_for_pubkey = RelayManager(timeout=5)
add_relay(relay_manager_for_pubkey, relays)
relay_manager_for_ids = RelayManager(timeout=5)
add_relay(relay_manager_for_ids, relays)
openai_client = OpenAI(
# 请用API Key将下行替换为：api_key="sk-xxx", base_url替换为你的大模型服务方url
api_key="sk-xxx",
base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"
)
def extract_referenced_event_ids(event):
referenced_event_ids = []
for tag in event.tags:
if tag[0] == 'e':
referenced_event_ids.append(tag[1])
return referenced_event_ids
def process_event(event):
# 提取引用的事件ID
referenced_event_ids = extract_referenced_event_ids(event)
# 获取引用的事件内容
referenced_events = get_referenced_events(referenced_event_ids, current_event_id=event.id)
# 组合上下文信息
ref_contents = []
if not referenced_events:
return ref_contents
for ref_event in referenced_events:
if ref_event.pubkey == get_keypair()[1]:
ref_contents.append(f"我（大鸟）之前回复的内容: {ref_event.content}")
else:
content = re.sub(r'\b(nostr:)?(nprofile|npub)[0-9a-z]+[\s]*', '', ref_event.content)
ref_contents.append(f"用户之前提问或者交流的内容: {content}")
return ref_contents
def get_referenced_events(event_ids, current_event_id=None):
# 从Nostr网络查询引用的事件
filters = FiltersList([
Filters(ids=event_ids)
])
subscription_id = uuid.uuid1().hex
relay_manager_for_ids.add_subscription_on_all_relays(subscription_id, filters)
relay_manager_for_ids.run_sync()
referenced_events = []
processed_event_ids = set() # 用于存储已处理的事件ID
while relay_manager_for_ids.message_pool.has_events():
event_msg = relay_manager_for_ids.message_pool.get_event()
if event_msg.event.id not in processed_event_ids and event_msg.event.id != current_event_id:
# 过滤掉当前事件ID
# 过滤掉已处理的事件
referenced_events.append(event_msg.event)
processed_event_ids.add(event_msg.event.id)
# 关闭所有连接
relay_manager_for_ids.close_all_relay_connections()
return referenced_events
def system_message():
system_message = f"你叫大鸟，家住鸟巢（鸟巢是你使用的Nostr社交账号名称）。你是一个AI全能助手。你要尽量全面准确的回答用户问题，并且在回答问题时保持与用户提问使用的语种一致。你在回答的问题的时候要适当参考上下文信息（其中包含你之前的回复和用户之前的提问或交流内容），并关注你之前的回答内容避免重复回答（如有用户的问题有重复你可以指出之前已经回答过），如果你不清楚的问题请直接告知用户而不是胡编乱造。"
return(system_message)
def respond(message, ref_contents=None):
messages = [{"role": "user", "content": message}]
sys_message = {"role": "system", "content": system_message()}
if ref_contents and len(ref_contents) > 0:
context_content = "\n".join(ref_contents)
sys_message["content"] += (f"上下文信息:\n{context_content}")
messages.append(sys_message)
print(messages)
completion = openai_client.chat.completions.create(
model="deepseek-r1", # 此处以 deepseek-r1 为例，可按需更换模型名称。
messages=messages
)
thinking = completion.choices[0].message.reasoning_content
answer = completion.choices[0].message.content
response = "### 思考过程：\n\n" + thinking + "\n\n----------\n\n### 正式回答：\n\n" + answer
print(response)
return response
def get_keypair():
nsec_key = "nsec124d4tdzgr7pe0cljzu6wk9xknz0s3mtht8p8cd52kywxj4rh42rs7txxns"
private_key = PrivateKey.from_nsec(nsec_key)
pubkey_hex = private_key.public_key.hex()
return private_key, pubkey_hex
```