mirror of
https://github.com/LiteyukiStudio/nonebot-plugin-marshoai.git
synced 2024-12-02 18:24:59 +08:00
d6d417a784
* 确实，现在可以处理 LaTeX 渲染了，欢迎 PR 新的渲染网址。 * 意外的小问题 * 删掉一个小数点 * 单词拼错了，马上四级，不知道能不能过 * 我是傻逼 * ok，但我肚子痛，等去蹲个坑
415 lines
14 KiB
Python
415 lines
14 KiB
Python
import os
|
||
import json
|
||
import uuid
|
||
import httpx
|
||
import base64
|
||
import mimetypes
|
||
|
||
from typing import Any, Optional
|
||
|
||
from nonebot.log import logger
|
||
|
||
import nonebot_plugin_localstore as store
|
||
|
||
from nonebot_plugin_alconna import (
|
||
Text as TextMsg,
|
||
Image as ImageMsg,
|
||
UniMessage,
|
||
)
|
||
|
||
|
||
# from zhDateTime import DateTime
|
||
from azure.ai.inference.aio import ChatCompletionsClient
|
||
from azure.ai.inference.models import SystemMessage
|
||
|
||
from .config import config
|
||
from .constants import *
|
||
from .deal_latex import ConvertLatex
|
||
|
||
nickname_json = None # 记录昵称
|
||
praises_json = None # 记录夸赞名单
|
||
loaded_target_list = [] # 记录已恢复备份的上下文的列表
|
||
|
||
# noinspection LongLine
|
||
chromium_headers = {
|
||
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
|
||
}
|
||
|
||
|
||
async def get_image_raw_and_type(
|
||
url: str, timeout: int = 10
|
||
) -> Optional[tuple[bytes, str]]:
|
||
"""
|
||
获取图片的二进制数据
|
||
|
||
参数:
|
||
url: str 图片链接
|
||
timeout: int 超时时间 秒
|
||
|
||
return:
|
||
tuple[bytes, str]: 图片二进制数据, 图片MIME格式
|
||
|
||
"""
|
||
|
||
async with httpx.AsyncClient() as client:
|
||
response = await client.get(url, headers=chromium_headers, timeout=timeout)
|
||
if response.status_code == 200:
|
||
# 获取图片数据
|
||
content_type = response.headers.get("Content-Type")
|
||
if not content_type:
|
||
content_type = mimetypes.guess_type(url)[0]
|
||
# image_format = content_type.split("/")[1] if content_type else "jpeg"
|
||
return response.content, str(content_type)
|
||
else:
|
||
return None
|
||
|
||
|
||
async def get_image_b64(url: str, timeout: int = 10) -> Optional[str]:
|
||
"""
|
||
获取图片的base64编码
|
||
|
||
参数:
|
||
url: 图片链接
|
||
timeout: 超时时间 秒
|
||
|
||
return: 图片base64编码
|
||
"""
|
||
|
||
if data_type := await get_image_raw_and_type(url, timeout):
|
||
# image_format = content_type.split("/")[1] if content_type else "jpeg"
|
||
base64_image = base64.b64encode(data_type[0]).decode("utf-8")
|
||
data_url = "data:{};base64,{}".format(data_type[1], base64_image)
|
||
return data_url
|
||
else:
|
||
return None
|
||
|
||
|
||
async def make_chat(
|
||
client: ChatCompletionsClient,
|
||
msg: list,
|
||
model_name: str,
|
||
tools: Optional[list] = None,
|
||
):
|
||
"""调用ai获取回复
|
||
|
||
参数:
|
||
client: 用于与AI模型进行通信
|
||
msg: 消息内容
|
||
model_name: 指定AI模型名"""
|
||
return await client.complete(
|
||
messages=msg,
|
||
model=model_name,
|
||
tools=tools,
|
||
temperature=config.marshoai_temperature,
|
||
max_tokens=config.marshoai_max_tokens,
|
||
top_p=config.marshoai_top_p,
|
||
)
|
||
|
||
|
||
def get_praises():
|
||
global praises_json
|
||
if praises_json is None:
|
||
praises_file = store.get_plugin_data_file(
|
||
"praises.json"
|
||
) # 夸赞名单文件使用localstore存储
|
||
if not os.path.exists(praises_file):
|
||
init_data = {
|
||
"like": [
|
||
{
|
||
"name": "Asankilp",
|
||
"advantages": "赋予了Marsho猫娘人格，使用vim与vscode为Marsho写了许多代码，使Marsho更加可爱",
|
||
}
|
||
]
|
||
}
|
||
with open(praises_file, "w", encoding="utf-8") as f:
|
||
json.dump(init_data, f, ensure_ascii=False, indent=4)
|
||
with open(praises_file, "r", encoding="utf-8") as f:
|
||
data = json.load(f)
|
||
praises_json = data
|
||
return praises_json
|
||
|
||
|
||
async def refresh_praises_json():
|
||
global praises_json
|
||
praises_file = store.get_plugin_data_file("praises.json")
|
||
if not os.path.exists(praises_file):
|
||
init_data = {
|
||
"like": [
|
||
{
|
||
"name": "Asankilp",
|
||
"advantages": "赋予了Marsho猫娘人格，使用vim与vscode为Marsho写了许多代码，使Marsho更加可爱",
|
||
}
|
||
]
|
||
}
|
||
with open(praises_file, "w", encoding="utf-8") as f:
|
||
json.dump(init_data, f, ensure_ascii=False, indent=4)
|
||
with open(praises_file, "r", encoding="utf-8") as f:
|
||
data = json.load(f)
|
||
praises_json = data
|
||
|
||
|
||
def build_praises():
|
||
praises = get_praises()
|
||
result = ["你喜欢以下几个人物，他们有各自的优点："]
|
||
for item in praises["like"]:
|
||
result.append(f"名字：{item['name']}，优点：{item['advantages']}")
|
||
return "\n".join(result)
|
||
|
||
|
||
async def save_context_to_json(name: str, context: Any, path: str):
|
||
context_dir = store.get_plugin_data_dir() / path
|
||
os.makedirs(context_dir, exist_ok=True)
|
||
file_path = os.path.join(context_dir, f"{name}.json")
|
||
with open(file_path, "w", encoding="utf-8") as json_file:
|
||
json.dump(context, json_file, ensure_ascii=False, indent=4)
|
||
|
||
|
||
async def load_context_from_json(name: str, path: str) -> list:
|
||
"""从指定路径加载历史记录"""
|
||
context_dir = store.get_plugin_data_dir() / path
|
||
os.makedirs(context_dir, exist_ok=True)
|
||
file_path = os.path.join(context_dir, f"{name}.json")
|
||
try:
|
||
with open(file_path, "r", encoding="utf-8") as json_file:
|
||
return json.load(json_file)
|
||
except FileNotFoundError:
|
||
return []
|
||
|
||
|
||
async def set_nickname(user_id: str, name: str):
|
||
global nickname_json
|
||
filename = store.get_plugin_data_file("nickname.json")
|
||
if not os.path.exists(filename):
|
||
data = {}
|
||
else:
|
||
with open(filename, "r", encoding="utf-8") as f:
|
||
data = json.load(f)
|
||
data[user_id] = name
|
||
if name == "" and user_id in data:
|
||
del data[user_id]
|
||
with open(filename, "w", encoding="utf-8") as f:
|
||
json.dump(data, f, ensure_ascii=False, indent=4)
|
||
nickname_json = data
|
||
|
||
|
||
# noinspection PyBroadException
|
||
async def get_nicknames():
|
||
"""获取nickname_json, 优先来源于全局变量"""
|
||
global nickname_json
|
||
if nickname_json is None:
|
||
filename = store.get_plugin_data_file("nickname.json")
|
||
try:
|
||
with open(filename, "r", encoding="utf-8") as f:
|
||
nickname_json = json.load(f)
|
||
except Exception:
|
||
nickname_json = {}
|
||
return nickname_json
|
||
|
||
|
||
async def refresh_nickname_json():
|
||
"""强制刷新nickname_json, 刷新全局变量"""
|
||
global nickname_json
|
||
filename = store.get_plugin_data_file("nickname.json")
|
||
# noinspection PyBroadException
|
||
try:
|
||
with open(filename, "r", encoding="utf-8") as f:
|
||
nickname_json = json.load(f)
|
||
except Exception:
|
||
logger.error("Error loading nickname.json")
|
||
|
||
|
||
def get_prompt():
|
||
"""获取系统提示词"""
|
||
prompts = ""
|
||
prompts += config.marshoai_additional_prompt
|
||
if config.marshoai_enable_praises:
|
||
praises_prompt = build_praises()
|
||
prompts += praises_prompt
|
||
marsho_prompt = config.marshoai_prompt
|
||
spell = SystemMessage(content=marsho_prompt + prompts).as_dict()
|
||
return spell
|
||
|
||
|
||
def suggest_solution(errinfo: str) -> str:
|
||
# noinspection LongLine
|
||
suggestions = {
|
||
"content_filter": "消息已被内容过滤器过滤。请调整聊天内容后重试。",
|
||
"RateLimitReached": "模型达到调用速率限制。请稍等一段时间或联系Bot管理员。",
|
||
"tokens_limit_reached": "请求token达到上限。请重置上下文。",
|
||
"content_length_limit": "请求体过大。请重置上下文。",
|
||
"unauthorized": "访问token无效。请联系Bot管理员。",
|
||
"invalid type: parameter messages.content is of type array but should be of type string.": "聊天请求体包含此模型不支持的数据类型。请重置上下文。",
|
||
"At most 1 image(s) may be provided in one request.": "此模型只能在上下文中包含1张图片。如果此前的聊天已经发送过图片，请重置上下文。",
|
||
}
|
||
|
||
for key, suggestion in suggestions.items():
|
||
if key in errinfo:
|
||
return f"\n{suggestion}"
|
||
|
||
return ""
|
||
|
||
|
||
async def get_backup_context(target_id: str, target_private: bool) -> list:
|
||
"""获取历史上下文"""
|
||
global loaded_target_list
|
||
if target_private:
|
||
target_uid = f"private_{target_id}"
|
||
else:
|
||
target_uid = f"group_{target_id}"
|
||
if target_uid not in loaded_target_list:
|
||
loaded_target_list.append(target_uid)
|
||
return await load_context_from_json(
|
||
f"back_up_context_{target_uid}", "contexts/backup"
|
||
)
|
||
return []
|
||
|
||
|
||
"""
|
||
以下函数依照 Mulan PSL v2 协议授权
|
||
|
||
函数: parse_markdown, get_uuid_back2codeblock
|
||
|
||
版权所有 © 2024 金羿ELS
|
||
Copyright (R) 2024 Eilles(EillesWan@outlook.com)
|
||
|
||
Licensed under Mulan PSL v2.
|
||
You can use this software according to the terms and conditions of the Mulan PSL v2.
|
||
You may obtain a copy of Mulan PSL v2 at:
|
||
http://license.coscl.org.cn/MulanPSL2
|
||
THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND,
|
||
EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT,
|
||
MERCHANTABILITY OR FIT FOR A PARTICULAR PURPOSE.
|
||
See the Mulan PSL v2 for more details.
|
||
"""
|
||
|
||
if config.marshoai_enable_richtext_prase:
|
||
|
||
latex_convert = ConvertLatex() # 开启一个转换实例
|
||
|
||
async def get_uuid_back2codeblock(
|
||
msg: str, code_blank_uuid_map: list[tuple[str, str]]
|
||
):
|
||
|
||
for torep, rep in code_blank_uuid_map:
|
||
msg = msg.replace(torep, rep)
|
||
|
||
return msg
|
||
|
||
async def parse_richtext(msg: str) -> UniMessage:
|
||
"""
|
||
人工智能给出的回答一般不会包含 HTML 嵌入其中，但是包含图片或者 LaTeX 公式、代码块，都很正常。
|
||
这个函数会把这些都以图片形式嵌入消息体。
|
||
"""
|
||
|
||
if not IMG_LATEX_PATTERN.search(msg): # 没有图片和LaTeX标签
|
||
return UniMessage(msg)
|
||
|
||
result_msg = UniMessage()
|
||
code_blank_uuid_map = [
|
||
(uuid.uuid4().hex, cbp.group()) for cbp in CODE_BLOCK_PATTERN.finditer(msg)
|
||
]
|
||
|
||
last_tag_index = 0
|
||
|
||
# 代码块渲染麻烦，先不处理
|
||
for rep, torep in code_blank_uuid_map:
|
||
msg = msg.replace(torep, rep)
|
||
|
||
# for to_rep in CODE_SINGLE_PATTERN.finditer(msg):
|
||
# code_blank_uuid_map.append((rep := uuid.uuid4().hex, to_rep.group()))
|
||
# msg = msg.replace(to_rep.group(), rep)
|
||
|
||
# print("#####################\n", msg, "\n\n")
|
||
|
||
# 插入图片
|
||
for each_find_tag in IMG_LATEX_PATTERN.finditer(msg):
|
||
|
||
tag_found = await get_uuid_back2codeblock(
|
||
each_find_tag.group(), code_blank_uuid_map
|
||
)
|
||
result_msg.append(
|
||
TextMsg(
|
||
await get_uuid_back2codeblock(
|
||
msg[last_tag_index : msg.find(tag_found)], code_blank_uuid_map
|
||
)
|
||
)
|
||
)
|
||
|
||
last_tag_index = msg.find(tag_found) + len(tag_found)
|
||
|
||
if each_find_tag.group(1):
|
||
|
||
# 图形一定要优先考虑
|
||
# 别忘了有些图形的地址就是 LaTeX，所以要优先判断
|
||
|
||
image_description = tag_found[2 : tag_found.find("]")]
|
||
image_url = tag_found[tag_found.find("(") + 1 : -1]
|
||
|
||
if image_ := await get_image_raw_and_type(image_url):
|
||
|
||
result_msg.append(
|
||
ImageMsg(
|
||
raw=image_[0],
|
||
mimetype=image_[1],
|
||
name=image_description + ".png",
|
||
)
|
||
)
|
||
result_msg.append(TextMsg("（{}）".format(image_description)))
|
||
|
||
else:
|
||
result_msg.append(TextMsg(tag_found))
|
||
elif each_find_tag.group(2):
|
||
|
||
latex_exp = await get_uuid_back2codeblock(
|
||
each_find_tag.group()
|
||
.replace("$", "")
|
||
.replace("\\(", "")
|
||
.replace("\\)", "")
|
||
.replace("\\[", "")
|
||
.replace("\\]", ""),
|
||
code_blank_uuid_map,
|
||
)
|
||
latex_generate_ok, latex_generate_result = (
|
||
await latex_convert.generate_png(
|
||
latex_exp,
|
||
dpi=300,
|
||
foreground_colour=config.marshoai_main_colour,
|
||
)
|
||
)
|
||
|
||
if latex_generate_ok:
|
||
result_msg.append(
|
||
ImageMsg(
|
||
raw=latex_generate_result,
|
||
mimetype="image/png",
|
||
name="latex.png",
|
||
)
|
||
)
|
||
else:
|
||
result_msg.append(TextMsg(latex_exp + "（公式解析失败）"))
|
||
if isinstance(latex_generate_result, str):
|
||
result_msg.append(TextMsg(latex_generate_result))
|
||
else:
|
||
result_msg.append(
|
||
ImageMsg(
|
||
raw=latex_generate_result,
|
||
mimetype="image/png",
|
||
name="latex_error.png",
|
||
)
|
||
)
|
||
else:
|
||
result_msg.append(TextMsg(tag_found + "（未知内容解析失败）"))
|
||
|
||
result_msg.append(
|
||
TextMsg(
|
||
await get_uuid_back2codeblock(msg[last_tag_index:], code_blank_uuid_map)
|
||
)
|
||
)
|
||
|
||
return result_msg
|
||
|
||
|
||
"""
|
||
Mulan PSL v2 协议授权部分结束
|
||
"""
|