J1147
/
ai-on-wechat
Obserwuj 1
Polub 0
Forkuj 0
Kod Zgłoszenia 0 Oczekujące zmiany 0 Wydania 0 Wiki Aktywność
Nie możesz wybrać więcej, niż 25 tematów Tematy muszą się zaczynać od litery lub cyfry, mogą zawierać myślniki ('-') i mogą mieć do 35 znaków.
1202 Commity
2 Gałęzie
152MB
Drzewo: 36981bf516
Gałęzie Tagi
${ item.name }
Utwórz gałąź ${ searchTerm }
z '36981bf516'
${ noResults }
ai-on-wechat/plugins/coze4upload/coze4upload2.txt

280 lines
12KB
Czysty Wina Historia 

  1. import requests

  2. import json

  3. import plugins

  4. from bridge.reply import Reply, ReplyType

  5. from bridge.context import ContextType

  6. from channel.chat_message import ChatMessage

  7. from plugins import *

  8. from common.log import logger

  9. from common.expired_dict import ExpiredDict

  10. import os

  11. import base64

  12. from pathlib import Path

  13. from PIL import Image

  14. import oss2

  15. 

  16. @plugins.register(

  17.     name="coze4upload",

  18.     desire_priority=-1,

  19.     desc="A plugin for upload",

  20.     version="0.0.01",

  21.     author="",

  22. )

  23. 

  24. class coze4upload(Plugin):

  25.     def __init__(self):

  26.         super().__init__()

  27.         try:

  28.             curdir = os.path.dirname(__file__)

  29.             config_path = os.path.join(curdir, "config.json")

  30.             if os.path.exists(config_path):

  31.                 with open(config_path, "r", encoding="utf-8") as f:

  32.                     self.config = json.load(f)

  33.             else:

  34.                 # 使用父类的方法来加载配置

  35.                 self.config = super().load_config()

  36. 

  37.                 if not self.config:

  38.                     raise Exception("config.json not found")

  39.             # 设置事件处理函数

  40.             self.handlers[Event.ON_HANDLE_CONTEXT] = self.on_handle_context

  41.             self.params_cache = ExpiredDict(300)

  42. 

  43.              # 从配置中提取所需的设置

  44.             self.keys = self.config.get("keys", {})

  45.             self.url_sum = self.config.get("url_sum", {})

  46.             self.search_sum = self.config.get("search_sum", {})

  47.             self.file_sum = self.config.get("file_sum", {})

  48.             self.image_sum = self.config.get("image_sum", {})

  49.             self.note = self.config.get("note", {})

  50. 

  51.             self.sum4all_key = self.keys.get("sum4all_key", "")

  52.             self.search1api_key = self.keys.get("search1api_key", "")

  53.             self.gemini_key = self.keys.get("gemini_key", "")

  54.             self.bibigpt_key = self.keys.get("bibigpt_key", "")

  55.             self.outputLanguage = self.keys.get("outputLanguage", "zh-CN")

  56.             self.opensum_key = self.keys.get("opensum_key", "")

  57.             self.open_ai_api_key = self.keys.get("open_ai_api_key", "")

  58.             self.model = self.keys.get("model", "gpt-3.5-turbo")

  59.             self.open_ai_api_base = self.keys.get("open_ai_api_base", "https://api.openai.com/v1")

  60.             self.xunfei_app_id = self.keys.get("xunfei_app_id", "")

  61.             self.xunfei_api_key = self.keys.get("xunfei_api_key", "")

  62.             self.xunfei_api_secret = self.keys.get("xunfei_api_secret", "")

  63.             self.perplexity_key = self.keys.get("perplexity_key", "")

  64.             self.flomo_key = self.keys.get("flomo_key", "")

  65.             # 之前提示

  66.             self.previous_prompt=''

  67. 

  68.             self.file_sum_enabled = self.file_sum.get("enabled", False)

  69.             self.file_sum_service = self.file_sum.get("service", "")

  70.             self.max_file_size = self.file_sum.get("max_file_size", 15000)

  71.             self.file_sum_group = self.file_sum.get("group", True)

  72.             self.file_sum_qa_prefix = self.file_sum.get("qa_prefix", "问")

  73.             self.file_sum_prompt = self.file_sum.get("prompt", "")

  74. 

  75.             self.image_sum_enabled = self.image_sum.get("enabled", False)

  76.             self.image_sum_service = self.image_sum.get("service", "")

  77.             self.image_sum_group = self.image_sum.get("group", True)

  78.             self.image_sum_qa_prefix = self.image_sum.get("qa_prefix", "问")

  79.             self.image_sum_prompt = self.image_sum.get("prompt", "")

  80. 

  81.             # 初始化成功日志

  82.             logger.info("[file4upload] inited.")

  83.         except Exception as e:

  84.             # 初始化失败日志

  85.             logger.warn(f"file4upload init failed: {e}")

  86. 

  87. 

  88.     def on_handle_context(self, e_context: EventContext):

  89.         context = e_context["context"]

  90.         if context.type not in [ContextType.TEXT, ContextType.SHARING,ContextType.FILE,ContextType.IMAGE]:

  91.             return

  92.         msg: ChatMessage = e_context["context"]["msg"]

  93.         user_id = msg.from_user_id

  94.         content = context.content

  95.         isgroup = e_context["context"].get("isgroup", False)

  96. 

  97.         if isgroup and not self.file_sum_group:

  98.             # 群聊中忽略处理文件

  99.             logger.info("群聊消息，文件处理功能已禁用")

  100.             return

  101.         logger.info("on_handle_context: 处理上下文开始")

  102.         context.get("msg").prepare()

  103. 

  104.         api_key='sk-5z2L4zy9T1w90j6e3T90ANZdyN2zLWClRwFnBzWgzdrG4onx'

  105. 

  106.         

  107.         logger.info(f'当前缓存：self.params_cache:{self.params_cache}') 

  108.         if user_id not in self.params_cache:

  109.             self.params_cache[user_id] = {}

  110.             logger.info(f'初始化缓存:{self.params_cache}') 

  111. 

  112.         if  context.type == ContextType.TEXT and user_id in self.params_cache:

  113.                 self.params_cache[user_id]['previous_prompt']=msg.content

  114. 

  115. 

  116.         # print(f'{msg.__dict__}')

  117.         if context.type == ContextType.IMAGE:

  118.             logger.info('处理图片')

  119.             file_path = context.content

  120.             logger.info(f"on_handle_context: 获取到图片路径 {file_path}，{user_id in self.params_cache}")

  121.             if user_id in self.params_cache:

  122.                 if 'previous_prompt' not in self.params_cache[user_id]:

  123. 

  124.                     reply = Reply()

  125.                     reply.type = ReplyType.TEXT

  126.                     # reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.file_sum_qa_prefix}+问题，可继续追问" 

  127.                     reply.content = f"您刚刚上传了一张图片，请问我有什么可以帮您的呢？"

  128.                     e_context["reply"] = reply

  129.                     e_context.action = EventAction.BREAK

  130.                     

  131.                 # else:

  132.                 print(f'准备抽取文字')

  133.                 file_content=extract_content_by_llm(file_path,api_key)

  134. 

  135.                 if file_content is None:

  136.                     logger.info('不能抽取文字,使用图片oss请求LLM')

  137.                     access_key_id = 'LTAI5tRTG6pLhTpKACJYoPR5'

  138.                     access_key_secret = 'E7dMzeeMxq4VQvLg7Tq7uKf3XWpYfN'

  139.                     # OSS区域对应的Endpoint

  140.                     endpoint = 'http://oss-cn-shanghai.aliyuncs.com'  # 根据你的区域选择

  141.                     # Bucket名称

  142.                     bucket_name = 'cow-agent'

  143.                     local_file_path=file_path

  144.                     oss_file_name=f'cow/{os.path.basename(file_path)}'

  145.                     logger.info(f'oss_file_name:{oss_file_name}\n local_file_path :{local_file_path}')

  146.                     file_content = upload_oss(access_key_id, access_key_secret, endpoint, bucket_name, local_file_path, oss_file_name)

  147.                     logger.info(f'写入图片缓存oss 地址{file_content}') 

  148.                 

  149.                 self.params_cache[user_id]['last_content']=file_content                  

  150.     

  151.             # else:

  152.             #     logger.warn(f'还没有建立会话')

  153.             logger.info('删除图片')

  154.             os.remove(file_path)

  155.         

  156.         if context.type == ContextType.FILE:

  157.             logger.info('处理图片')

  158.             file_path = context.content

  159.             logger.info(f"on_handle_context: 获取到文件路径 {file_path}")

  160.             if user_id in self.params_cache:

  161.                 if 'previous_prompt' not in self.params_cache[user_id]:                  

  162.                     reply = Reply()

  163.                     reply.type = ReplyType.TEXT

  164.                     # reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.file_sum_qa_prefix}+问题，可继续追问" 

  165.                     reply.content = f"您刚刚上传了一份文件，请问我有什么可以帮您的呢？"

  166.                     e_context["reply"] = reply

  167.                     e_context.action = EventAction.BREAK     

  168.                 # else:

  169.                 print(f'准备抽取文字')

  170.                 file_content=extract_content_by_llm(file_path,api_key)

  171.                 if file_content is None:

  172.                     reply = Reply()

  173.                     reply.type = ReplyType.TEXT

  174.                     # reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.file_sum_qa_prefix}+问题，可继续追问" 

  175.                     reply.content = f"不能处理这份文件"

  176.                     e_context["reply"] = reply

  177.                     e_context.action = EventAction.BREAK

  178.                     return

  179.                 else:

  180.                     self.params_cache[user_id]['last_content']=file_content

  181.             

  182.             logger.info('删除图片')

  183.             os.remove(file_path)

  184. 

  185.         # logger.info('previous_prompt' in self.params_cache[user_id])

  186.         # logger.info('last_content' in self.params_cache[user_id])

  187.         is_previous_prompt='previous_prompt' in self.params_cache[user_id]

  188.         is_last_content='last_content' in self.params_cache[user_id]

  189.         logger.info(f"存在提示词 previous_prompt:{is_previous_prompt}")

  190.         logger.info(f'存在内容 last_content:{is_last_content}' )

  191.         if 'previous_prompt' in self.params_cache[user_id] and 'last_content' in self.params_cache[user_id] :

  192.             e_context["context"].type = ContextType.TEXT

  193.             e_context["context"].content = self.params_cache[user_id]['last_content']+'\n\t'+self.params_cache[user_id]['previous_prompt']

  194.             logger.info(f'conze4upload 插件处理上传文件或图片')

  195.             e_context.action = EventAction.CONTINUE

  196.             

  197.             # 清空清空缓存

  198.             self.params_cache.clear()

  199.             logger.info(f'清空缓存后：{self.params_cache}')

  200.        

  201.         # e_context.action = EventAction.BREAK  

  202.        

  203. 

  204. def remove_markdown(text):

  205.     # 替换Markdown的粗体标记

  206.     text = text.replace("**", "")

  207.     # 替换Markdown的标题标记

  208.     text = text.replace("### ", "").replace("## ", "").replace("# ", "")

  209.     return text

  210. 

  211. def extract_content_by_llm(file_path: str, api_key: str) -> str:

  212.     logger.info(f'大模型开始抽取文字')

  213.     try:

  214.         headers = {

  215.             'Authorization': f'Bearer {api_key}'

  216.         }

  217.         data = {

  218.             'purpose': 'file-extract',

  219.         }

  220.         file_name=os.path.basename(file_path)

  221.         files = {

  222.             'file': (file_name, open(Path(file_path), 'rb')),

  223.         }

  224.         # print(files)

  225.         api_url='https://api.moonshot.cn/v1/files'

  226.         response = requests.post(api_url, headers=headers, files=files, data=data)

  227.         response_data = response.json()

  228.         file_id = response_data.get('id')

  229.         response=requests.get(url=f"https://api.moonshot.cn/v1/files/{file_id}/content", headers=headers)

  230.         print(response.text)

  231.         response_data = response.json()

  232.         content = response_data.get('content')

  233.         return content

  234.     except requests.exceptions.RequestException as e:

  235.         logger.error(f"Error calling LLM API: {e}")

  236.         return None

  237. 

  238. def upload_oss(access_key_id, access_key_secret, endpoint, bucket_name, local_file_path, oss_file_name, expiration_days=7):

  239.     """

  240.     上传文件到阿里云OSS并设置生命周期规则，同时返回文件的公共访问地址。

  241.     

  242.     :param access_key_id: 阿里云AccessKey ID

  243.     :param access_key_secret: 阿里云AccessKey Secret

  244.     :param endpoint: OSS区域对应的Endpoint

  245.     :param bucket_name: OSS中的Bucket名称

  246.     :param local_file_path: 本地文件路径

  247.     :param oss_file_name: OSS中的文件存储路径

  248.     :param expiration_days: 文件保存天数，默认7天后删除

  249.     :return: 文件的公共访问地址

  250.     """

  251.     

  252.     # 创建Bucket实例

  253.     auth = oss2.Auth(access_key_id, access_key_secret)

  254.     bucket = oss2.Bucket(auth, endpoint, bucket_name)

  255. 

  256.     ### 1. 设置生命周期规则 ###

  257.     rule_id = f'delete_after_{expiration_days}_days'  # 规则ID

  258.     prefix = oss_file_name.split('/')[0] + '/'  # 设置规则应用的前缀为文件所在目录

  259. 

  260.     # 定义生命周期规则

  261.     rule = oss2.models.LifecycleRule(rule_id, prefix, status=oss2.models.LifecycleRule.ENABLED,

  262.                                      expiration=oss2.models.LifecycleExpiration(days=expiration_days))

  263. 

  264.     # 设置Bucket的生命周期

  265.     lifecycle = oss2.models.BucketLifecycle([rule])

  266.     bucket.put_bucket_lifecycle(lifecycle)

  267. 

  268.     print(f"已设置生命周期规则：文件将在{expiration_days}天后自动删除")

  269. 

  270.     ### 2. 上传文件到OSS ###

  271.     bucket.put_object_from_file(oss_file_name, local_file_path)

  272. 

  273.     ### 3. 构建公共访问URL ###

  274.     file_url = f"http://{bucket_name}.{endpoint.replace('http://', '')}/{oss_file_name}"

  275. 

  276.     print(f"文件上传成功，公共访问地址：{file_url}")

  277.     

  278.     return file_url

  279.