J1147
/
ai-on-wechat


			
							import requests
import json
import plugins
from bridge.reply import Reply, ReplyType
from bridge.context import ContextType
from channel.chat_message import ChatMessage
from plugins import *
from common.log import logger
from common.expired_dict import ExpiredDict
import os
import base64
from pathlib import Path
from PIL import Image
import oss2


EXTENSION_TO_TYPE = {
    'pdf': 'pdf',
    'doc': 'docx', 'docx': 'docx',
    'md': 'md',
    'txt': 'txt',
    'xls': 'excel', 'xlsx': 'excel',
    'csv': 'csv',
    'html': 'html', 'htm': 'html',
    'ppt': 'ppt', 'pptx': 'ppt'
}

@plugins.register(
    name="kimi4upload",
    desire_priority=-1,
    desc="A plugin for upload",
    version="0.0.01",
    author="",
)


class file4upload(Plugin):
    def __init__(self):
        super().__init__()
        try:
            curdir = os.path.dirname(__file__)
            config_path = os.path.join(curdir, "config.json")
            if os.path.exists(config_path):
                with open(config_path, "r", encoding="utf-8") as f:
                    self.config = json.load(f)
            else:
                # 使用父类的方法来加载配置
                self.config = super().load_config()

                if not self.config:
                    raise Exception("config.json not found")
            # 设置事件处理函数
            self.handlers[Event.ON_HANDLE_CONTEXT] = self.on_handle_context
            self.params_cache = ExpiredDict(300)

             # 从配置中提取所需的设置
            self.keys = self.config.get("keys", {})
            self.url_sum = self.config.get("url_sum", {})
            self.search_sum = self.config.get("search_sum", {})
            self.file_sum = self.config.get("file_sum", {})
            self.image_sum = self.config.get("image_sum", {})
            self.note = self.config.get("note", {})

            self.sum4all_key = self.keys.get("sum4all_key", "")
            self.search1api_key = self.keys.get("search1api_key", "")
            self.gemini_key = self.keys.get("gemini_key", "")
            self.bibigpt_key = self.keys.get("bibigpt_key", "")
            self.outputLanguage = self.keys.get("outputLanguage", "zh-CN")
            self.opensum_key = self.keys.get("opensum_key", "")
            self.open_ai_api_key = self.keys.get("open_ai_api_key", "")
            self.model = self.keys.get("model", "gpt-3.5-turbo")
            self.open_ai_api_base = self.keys.get("open_ai_api_base", "https://api.openai.com/v1")
            self.xunfei_app_id = self.keys.get("xunfei_app_id", "")
            self.xunfei_api_key = self.keys.get("xunfei_api_key", "")
            self.xunfei_api_secret = self.keys.get("xunfei_api_secret", "")
            self.perplexity_key = self.keys.get("perplexity_key", "")
            self.flomo_key = self.keys.get("flomo_key", "")
            # 之前提示
            self.previous_prompt=''

            self.file_sum_enabled = self.file_sum.get("enabled", False)
            self.file_sum_service = self.file_sum.get("service", "")
            self.max_file_size = self.file_sum.get("max_file_size", 15000)
            self.file_sum_group = self.file_sum.get("group", True)
            self.file_sum_qa_prefix = self.file_sum.get("qa_prefix", "问")
            self.file_sum_prompt = self.file_sum.get("prompt", "")

            self.image_sum_enabled = self.image_sum.get("enabled", False)
            self.image_sum_service = self.image_sum.get("service", "")
            self.image_sum_group = self.image_sum.get("group", True)
            self.image_sum_qa_prefix = self.image_sum.get("qa_prefix", "问")
            self.image_sum_prompt = self.image_sum.get("prompt", "")

            # 初始化成功日志
            logger.info("[file4upload] inited.")
        except Exception as e:
            # 初始化失败日志
            logger.warn(f"file4upload init failed: {e}")

    def on_handle_context(self, e_context: EventContext):
        context = e_context["context"]
        if context.type not in [ContextType.TEXT, ContextType.SHARING,ContextType.FILE,ContextType.IMAGE]:
            return
        msg: ChatMessage = e_context["context"]["msg"]
        user_id = msg.from_user_id
        content = context.content
        isgroup = e_context["context"].get("isgroup", False)
        # logger.info(f"user_id:{user_id},content:{content},isgroup:{isgroup}")
        # logger.info(f'上下文参数缓存键字典：{self.params_cache.keys}')
        # logger.info(f'user_id in self.params_cache: {user_id in self.params_cache}')
        # 上次提示
        if  context.type == ContextType.TEXT:
            self.previous_prompt=msg.content

        if isgroup and not self.file_sum_group:
            # 群聊中忽略处理文件
            logger.info("群聊消息，文件处理功能已禁用")
            return
        logger.info("on_handle_context: 处理上下文开始")
        context.get("msg").prepare()
        # file_path = context.content
        # logger.info(f"on_handle_context: 获取到文件路径 {file_path}")
        
        api_key='sk-5z2L4zy9T1w90j6e3T90ANZdyN2zLWClRwFnBzWgzdrG4onx'

        if context.type == ContextType.IMAGE:
            file_path = context.content
            logger.info(f"on_handle_context: 获取到文件路径 {file_path}")
            print(f'处理首次上次的图片,准备抽取文字')
            file_content=self.extract_content_by_llm(file_path,api_key)
            self.params_cache[user_id] = {}
            if file_content is not None:
                logger.info('图片中抽取文字,使用使用图片的文字请求LLM')
                
                messages = [{
                                "role": "system",
                                "content": "你是 Kimi，由 Moonshot AI 提供的人工智能助手，你更擅长中文和英文的对话。你会为用户提供安全，有帮助，准确的回答。同时，你会拒绝一切涉及恐怖主义，种族歧视，黄色暴力等问题的回答。Moonshot AI 为专有名词，不可翻译成其他语言。",
                            },{
                                "role": "system",
                                "content": file_content,
                            }]
                self.params_cache[user_id]['last_word_messages']=messages
                self.params_cache[user_id]['last_image_oss']=None
            else:
                logger.info('不能抽取文字,使用图片oss请求LLM')
                # logger.info(f"on_handle_context: 获取到图片路径 {file_path}")
                # base64_image=self.encode_image_to_base64(file_path)
                
                # self.params_cache[user_id]['last_image_oss']=base64_image
                # self.params_cache[user_id]['last_word_messages']=None
                access_key_id = 'LTAI5tRTG6pLhTpKACJYoPR5'
                access_key_secret = 'E7dMzeeMxq4VQvLg7Tq7uKf3XWpYfN'
                # OSS区域对应的Endpoint
                endpoint = 'http://oss-cn-shanghai.aliyuncs.com'  # 根据你的区域选择
                # Bucket名称
                bucket_name = 'cow-agent'
                local_file_path=file_path
                oss_file_name=f'cow/{os.path.basename(file_path)}'
                logger.info(f'oss_file_name:{oss_file_name}\n local_file_path :{local_file_path}')
                file_url = upload_oss(access_key_id, access_key_secret, endpoint, bucket_name, local_file_path, oss_file_name)
                logger.info(f'写入图片缓存oss 地址{file_url}')
                self.params_cache[user_id]['last_image_oss']=file_url
                self.params_cache[user_id]['last_word_messages']=None


            if self.previous_prompt == '':
                reply = Reply()
                reply.type = ReplyType.TEXT
                # 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" 
                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？"
                e_context["reply"] = reply
                e_context.action = EventAction.BREAK
                return
            else:
                e_context.action = EventAction.CONTINUE

        if context.type == ContextType.FILE:
            file_path = context.content
            logger.info(f"on_handle_context: 获取到文件路径 {file_path}")
            print(f'处理首次上次的文件')
            file_content=self.extract_content_by_llm(file_path,api_key)
            if file_content is not None:
                self.params_cache[user_id] = {}
                messages = [{
                                "role": "system",
                                "content": "你是 Kimi，由 Moonshot AI 提供的人工智能助手，你更擅长中文和英文的对话。你会为用户提供安全，有帮助，准确的回答。同时，你会拒绝一切涉及恐怖主义，种族歧视，黄色暴力等问题的回答。Moonshot AI 为专有名词，不可翻译成其他语言。",
                            },{
                                "role": "system",
                                "content": file_content,
                            }]
                self.params_cache[user_id]['last_word_messages']=messages
                self.params_cache[user_id]['last_image_oss']=None

            if self.previous_prompt == '':
                reply = Reply()
                reply.type = ReplyType.TEXT
                # 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" 
                reply.content = f"您刚刚上传了一个文件，请问我有什么可以帮您的呢？如可以问“请总结分析这份报告文件，同时，提供治疗和健康建议。”"
                e_context["reply"] = reply
                e_context.action = EventAction.BREAK
                return
            else:
                e_context.action = EventAction.CONTINUE
                
        # if user_id in self.params_cache and (self.params_cache[user_id]['last_word_messages']!=None):
        if user_id in self.params_cache:
            if 'last_word_messages' in self.params_cache[user_id] and self.params_cache[user_id]['last_word_messages'] is not None: 
                print(f'缓存处理已经上传的文件')
                # last_word_messages=self.params_cache[user_id]['last_word_messages']
                # cache_messages=last_word_messages[:2]
                cache_messages=self.params_cache[user_id]['last_word_messages']
                messages = [
                    *cache_messages, 
                    {
                        "role": "user",
                        "content": self.previous_prompt ,#msg.content,
                    },
                ]
                self.handle_file_upload(messages, e_context)

        # if user_id in self.params_cache and ('last_image_oss' in self.params_cache[user_id]  or self.params_cache[user_id]['last_image_oss']!=None):
        if user_id in self.params_cache:
            if 'last_image_oss' in self.params_cache[user_id] and self.params_cache[user_id]['last_image_oss'] is not None:           
                print(f'缓存处理已经oss图片的文件')
                file_url=self.params_cache[user_id]['last_image_oss']
                messages = [{
                                    "role": "system",
                                    "content": "你是一个能能描述任何图片的智能助手",
                                },
                            {
                                "role": "user",
                                "content": f'{file_url}\n{self.previous_prompt}',
                        }]
                # messages=[
                #     {
                #         "role": "user",
                #         "content": [
                #             {
                #                 "type": "image_url",
                #                 "image_url": {
                #                     "url": f"{file_url}"
                #                 }
                #             },
                #             {
                #                 "type": "text",
                #                 "text": f"{self.previous_prompt}"
                #             }
                #         ]
                #     }
                # ]


                self.handle_images_oos(messages, e_context)           
            
    def handle_file_upload(self, messages, e_context):
        logger.info("handle_file: 向LLM发送内容总结请求")
        
        msg: ChatMessage = e_context["context"]["msg"]
        user_id = msg.from_user_id
        user_params = self.params_cache.get(user_id, {})
        prompt = user_params.get('prompt', self.file_sum_prompt)
        
        self.params_cache[user_id] = {}


        try:
            api_key = "sk-5z2L4zy9T1w90j6e3T90ANZdyN2zLWClRwFnBzWgzdrG4onx"
            # base_url = "https://api.moonshot.cn/v1",
            api_url = "https://api.moonshot.cn/v1/chat/completions"
            headers = {
                'Content-Type': 'application/json',
                'Authorization': f'Bearer {api_key}'
            }

            data={
                "model": "moonshot-v1-128k",
                "messages":messages,
                # "temperature": 0.3
            }
            response = requests.post(url=api_url, headers=headers, data=json.dumps(data))
            logger.info(f'handle_file_upload: 请求文件内容{json.dumps(messages, ensure_ascii=False)}')
            response.raise_for_status()
            response_data = response.json()
            if "choices" in response_data and len(response_data["choices"]) > 0:
                    first_choice = response_data["choices"][0]
                    if "message" in first_choice and "content" in first_choice["message"]:
                        response_content = first_choice["message"]["content"].strip()  # 获取响应内容
                        reply_content = response_content.replace("\\n", "\n")  # 替换 \\n 为 \n
                        # self.params_cache[user_id]['last_word_messages']=messages
                        # if self.params_cache[user_id]['last_word_messages']!=None:
                        #     self.params_cache[user_id]['last_word_messages']=messages

                        self.previous_prompt =''
                    else:
                        logger.error("Content not found in the response")
                        reply_content = "Content not found in the LLM API response"
            else:
                logger.error("No choices available in the response")
                reply_content = "No choices available in the LLM API response"
        except requests.exceptions.RequestException as e:
            logger.error(f"Error calling LLM API: {e}")
            reply_content = f"An error occurred while calling LLM API"

        reply = Reply()
        reply.type = ReplyType.TEXT
        reply.content = f"{remove_markdown(reply_content)}"
        e_context["reply"] = reply
        e_context.action = EventAction.BREAK_PASS
      

    def handle_images_base64(self, messages, e_context):
        logger.info("handle_file: 向LLM发送内容总结请求")
        
        msg: ChatMessage = e_context["context"]["msg"]
        user_id = msg.from_user_id
        user_params = self.params_cache.get(user_id, {})
        prompt = user_params.get('prompt', self.file_sum_prompt)
        
        
        try:
            # api_key = "sk-5z2L4zy9T1w90j6e3T90ANZdyN2zLWClRwFnBzWgzdrG4onx"
            # # base_url = "https://api.moonshot.cn/v1",
            # api_url = "https://api.moonshot.cn/v1/chat/completions"

            
            # api_key = "sk-5dyg7PMUNeoSqHH807453eB06f434c34Ba6fB4764aC8358c"
            # api_url = "http://106.15.182.218:3001/v1/chat/completions"
            # headers = {
            #     'Content-Type': 'application/json',
            #     'Authorization': f'Bearer {api_key}'
            # }

            # data={
            #     "model": "moonshot-v1-128k",
            #     "messages":messages,
            #     # "temperature": 0.3
            # }
            # response = requests.post(url=api_url, headers=headers, json=data)
            
            base64_image=self.encode_image_to_base64('tmp/240926-164856.png')
            api_key = self.open_ai_api_key
            api_base = f"{self.open_ai_api_base}/chat/completions"
            logger.info(api_base)
            payload = {
                "model": "moonshot-v1-128k",
                "messages": [
                    {
                        "role": "user",
                        "content": [
                            {
                                "type": "text",
                                "text": self.previous_prompt
                            },
                            {
                                "type": "image_url",
                                "image_url": {
                                    "url": f"data:image/jpeg;base64,{base64_image}"
                                }
                            }
                        ]
                    }
                ],
                "max_tokens": 3000
            }

            # payload = {
            #     "model": "moonshot-v1-128k",
            #     "messages": messages,
            #     "max_tokens": 3000
            # }
            headers = {
                "Content-Type": "application/json",
                "Authorization": f"Bearer {api_key}"
            }
            logger.info('开始')
            response = requests.post(api_base, headers=headers, json=payload)


            # logger.info(f'handle_file_upload: 请求文件内容{json.dumps(messages, ensure_ascii=False)}')
            response.raise_for_status()
            response_data = response.json()
            if "choices" in response_data and len(response_data["choices"]) > 0:
                    first_choice = response_data["choices"][0]
                    if "message" in first_choice and "content" in first_choice["message"]:
                        response_content = first_choice["message"]["content"].strip()  # 获取响应内容
                        reply_content = response_content.replace("\\n", "\n")  # 替换 \\n 为 \n
                        # self.params_cache[user_id]['last_word_messages']=messages
                        # if self.params_cache[user_id]['last_word_messages']!=None:
                        #     self.params_cache[user_id]['last_word_messages']=messages

                        self.previous_prompt =''
                    else:
                        logger.error("Content not found in the response")
                        reply_content = "Content not found in the LLM API response"
            else:
                logger.error("No choices available in the response")
                reply_content = "No choices available in the LLM API response"
        except requests.exceptions.RequestException as e:
            logger.error(f"Error calling LLM API: {e}")
            reply_content = f"An error occurred while calling LLM API"

        reply = Reply()
        reply.type = ReplyType.TEXT
        reply.content = f"{remove_markdown(reply_content)}"
        e_context["reply"] = reply
        e_context.action = EventAction.BREAK_PASS
      
    def handle_images_oos(self, messages, e_context):
        logger.info("handle_file: 向LLM发送内容总结请求")
        
        msg: ChatMessage = e_context["context"]["msg"]
        user_id = msg.from_user_id
        user_params = self.params_cache.get(user_id, {})
        prompt = user_params.get('prompt', self.file_sum_prompt)
        
        self.params_cache[user_id] = {}


        try:
            api_key = self.open_ai_api_key
            api_base = f"{self.open_ai_api_base}/chat/completions"
            logger.info(api_base)
            payload = {
                "model": "7374349217580056592",
                "messages":messages,
                "max_tokens": 3000
            }

            # payload = {
            #     "model": "moonshot-v1-128k",
            #     "messages": messages,
            #     "max_tokens": 3000
            # }
            headers = {
                "Content-Type": "application/json",
                "Authorization": f"Bearer {api_key}"
            }
            # logger.info('开始')
            response = requests.post(api_base, headers=headers, json=payload)
            logger.info(f'handle_file_upload: 请求文件内容{json.dumps(messages, ensure_ascii=False)}')
            response.raise_for_status()
            response_data = response.json()
            if "choices" in response_data and len(response_data["choices"]) > 0:
                    first_choice = response_data["choices"][0]
                    if "message" in first_choice and "content" in first_choice["message"]:
                        response_content = first_choice["message"]["content"].strip()  # 获取响应内容
                        reply_content = response_content.replace("\\n", "\n")  # 替换 \\n 为 \n
                        # self.params_cache[user_id]['last_word_messages']=messages
                        # if self.params_cache[user_id]['last_word_messages']!=None:
                        #     self.params_cache[user_id]['last_word_messages']=messages

                        self.previous_prompt =''
                    else:
                        logger.error("Content not found in the response")
                        reply_content = "Content not found in the LLM API response"
            else:
                logger.error("No choices available in the response")
                reply_content = "No choices available in the LLM API response"
        except requests.exceptions.RequestException as e:
            logger.error(f"Error calling LLM API: {e}")
            reply_content = f"An error occurred while calling LLM API"

        reply = Reply()
        reply.type = ReplyType.TEXT
        reply.content = f"{remove_markdown(reply_content)}"
        e_context["reply"] = reply
        e_context.action = EventAction.BREAK_PASS
    
    def extract_content_by_llm(self, file_path: str, api_key: str) -> str:
        logger.info(f'大模型开始抽取文字')
        try:
            headers = {
                'Authorization': f'Bearer {api_key}'
            }
            data = {
                'purpose': 'file-extract',
            }
            file_name=os.path.basename(file_path)
            files = {
                'file': (file_name, open(Path(file_path), 'rb')),
            }
            print(files)
            api_url='https://api.moonshot.cn/v1/files'
            response = requests.post(api_url, headers=headers, files=files, data=data)
            # print(response.text)
            response_data = response.json()
            file_id = response_data.get('id')
            # print(f'文件id:{file_id}')
            response=requests.get(url=f"https://api.moonshot.cn/v1/files/{file_id}/content", headers=headers)
            print(response.text)
            response_data = response.json()
            content = response_data.get('content')
            return content
        except requests.exceptions.RequestException as e:
            logger.error(f"Error calling LLM API: {e}")
            return None

    def encode_image_to_base64(self, image_path):
        logger.info(f"开始处理图片: {image_path}")
        try:
            with Image.open(image_path) as img:
                logger.info(f"成功打开图片. 原始大小: {img.size}")
                if img.width > 1024:
                    new_size = (1024, int(img.height*1024/img.width))
                    img = img.resize(new_size)
                    img.save(image_path)  # 保存调整大小后的图片
                    logger.info(f"调整图片大小至: {new_size}")

            with open(image_path, "rb") as image_file:
                img_byte_arr = image_file.read()
                logger.info(f"读取图片完成. 大小: {len(img_byte_arr)} 字节")

            encoded = base64.b64encode(img_byte_arr).decode('ascii')
            logger.info(f"Base64编码完成. 编码后长度: {len(encoded)}")
            return encoded
        except Exception as e:
            logger.error(f"图片编码过程中发生错误: {str(e)}", exc_info=True)
            raise

def remove_markdown(text):
    # 替换Markdown的粗体标记
    text = text.replace("**", "")
    # 替换Markdown的标题标记
    text = text.replace("### ", "").replace("## ", "").replace("# ", "")
    return text


def upload_oss(access_key_id, access_key_secret, endpoint, bucket_name, local_file_path, oss_file_name, expiration_days=7):
    """
    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。
    
    :param access_key_id: 阿里云AccessKey ID
    :param access_key_secret: 阿里云AccessKey Secret
    :param endpoint: OSS区域对应的Endpoint
    :param bucket_name: OSS中的Bucket名称
    :param local_file_path: 本地文件路径
    :param oss_file_name: OSS中的文件存储路径
    :param expiration_days: 文件保存天数，默认7天后删除
    :return: 文件的公共访问地址
    """
    
    # 创建Bucket实例
    auth = oss2.Auth(access_key_id, access_key_secret)
    bucket = oss2.Bucket(auth, endpoint, bucket_name)

    ### 1. 设置生命周期规则 ###
    rule_id = f'delete_after_{expiration_days}_days'  # 规则ID
    prefix = oss_file_name.split('/')[0] + '/'  # 设置规则应用的前缀为文件所在目录

    # 定义生命周期规则
    rule = oss2.models.LifecycleRule(rule_id, prefix, status=oss2.models.LifecycleRule.ENABLED,
                                     expiration=oss2.models.LifecycleExpiration(days=expiration_days))

    # 设置Bucket的生命周期
    lifecycle = oss2.models.BucketLifecycle([rule])
    bucket.put_bucket_lifecycle(lifecycle)

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

    ### 2. 上传文件到OSS ###
    bucket.put_object_from_file(oss_file_name, local_file_path)

    ### 3. 构建公共访问URL ###
    file_url = f"http://{bucket_name}.{endpoint.replace('http://', '')}/{oss_file_name}"

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