|
- import requests
- import json
- import re
- 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
- from docx import Document
- import markdown
- import fitz
- from openpyxl import load_workbook
- import csv
- from bs4 import BeautifulSoup
- from pptx import Presentation
- from PIL import Image
- import base64
- import html
-
-
-
- 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="sum4all",
- desire_priority=2,
- desc="A plugin for summarizing all things",
- version="0.7.10",
- author="fatwang2",
- )
-
- class sum4all(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", "")
-
- # 提取sum服务的配置
- self.url_sum_enabled = self.url_sum.get("enabled", False)
- self.url_sum_service = self.url_sum.get("service", "")
- self.url_sum_group = self.url_sum.get("group", True)
- self.url_sum_qa_enabled = self.url_sum.get("qa_enabled", True)
- self.url_sum_qa_prefix = self.url_sum.get("qa_prefix", "问")
- self.url_sum_prompt = self.url_sum.get("prompt", "")
-
- self.search_sum_enabled = self.search_sum.get("enabled", False)
- self.search_sum_service = self.search_sum.get("service", "")
- self.search_service = self.search_sum.get("search_service", "duckduckgo")
- self.search_sum_group = self.search_sum.get("group", True)
- self.search_sum_search_prefix = self.search_sum.get("search_prefix", "搜")
- self.search_sum_prompt = self.search_sum.get("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", "")
-
- self.note_enabled = self.note.get("enabled", False)
- self.note_service = self.note.get("service", "")
- self.note_prefix = self.note.get("prefix", "记")
-
- # 初始化成功日志
- logger.info("[sum4all] inited.")
- except Exception as e:
- # 初始化失败日志
- logger.warn(f"sum4all 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)
-
- url_match = re.match('https?://(?:[-\w.]|(?:%[\da-fA-F]{2}))+', content)
- unsupported_urls = re.search(r'.*finder\.video\.qq\.com.*|.*support\.weixin\.qq\.com/update.*|.*support\.weixin\.qq\.com/security.*|.*mp\.weixin\.qq\.com/mp/waerrpage.*', content)
-
- # 检查输入是否以"搜索前缀词" 开头
- if content.startswith(self.search_sum_search_prefix) and self.search_sum_enabled:
- # 如果消息来自一个群聊,并且你不希望在群聊中启用搜索功能,直接返回
- if isgroup and not self.search_sum_group:
- return
- # Call new function to handle search operation
- self.call_service(content, e_context, "search")
- return
-
- if user_id in self.params_cache and ('last_file_content' in self.params_cache[user_id] or 'last_image_base64' in self.params_cache[user_id] or 'last_url' in self.params_cache[user_id]):
- # 如果存在最近一次处理的文件路径,触发文件理解函数
- if 'last_file_content' in self.params_cache[user_id] and content.startswith(self.file_sum_qa_prefix):
- logger.info('Content starts with the file_sum_qa_prefix.')
- # 去除关键词和紧随其后的空格
- new_content = content[len(self.file_sum_qa_prefix):]
- self.params_cache[user_id]['prompt'] = new_content
- logger.info('params_cache for user has been successfully updated.')
- self.handle_file(self.params_cache[user_id]['last_file_content'], e_context)
- # 如果存在最近一次处理的图片路径,触发图片理解函数
- elif 'last_image_base64' in self.params_cache[user_id] and content.startswith(self.image_sum_qa_prefix):
- logger.info('Content starts with the image_sum_qa_prefix.')
- # 去除关键词和紧随其后的空格
- new_content = content[len(self.image_sum_qa_prefix):]
- self.params_cache[user_id]['prompt'] = new_content
- logger.info('params_cache for user has been successfully updated.')
- self.handle_image(self.params_cache[user_id]['last_image_base64'], e_context)
-
- # 如果存在最近一次处理的URL,触发URL理解函数
- elif 'last_url' in self.params_cache[user_id] and content.startswith(self.url_sum_qa_prefix):
- logger.info('Content starts with the url_sum_qa_prefix.')
- # 去除关键词和紧随其后的空格
- new_content = content[len(self.url_sum_qa_prefix):]
- self.params_cache[user_id]['prompt'] = new_content
- logger.info('params_cache for user has been successfully updated.')
- self.call_service(self.params_cache[user_id]['last_url'], e_context ,"sum")
- elif 'last_url' in self.params_cache[user_id] and content.startswith(self.note_prefix) and self.note_enabled and not isgroup:
- logger.info('Content starts with the note_prefix.')
- new_content = content[len(self.note_prefix):]
- self.params_cache[user_id]['note'] = new_content
- logger.info('params_cache for user has been successfully updated.')
- self.call_service(self.params_cache[user_id]['last_url'], e_context, "note")
- if context.type == ContextType.FILE:
- 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}")
-
- # 检查是否应该进行文件总结
- if self.file_sum_enabled:
- # 更新params_cache中的last_file_content
- self.params_cache[user_id] = {}
- file_content = self.extract_content(file_path)
- if file_content is None:
- logger.info("文件内容无法提取,跳过处理")
- else:
- self.params_cache[user_id]['last_file_content'] = file_content
- logger.info('Updated last_file_content in params_cache for user.')
- self.handle_file(file_content, e_context)
- else:
- logger.info("文件总结功能已禁用,不对文件内容进行处理")
- # 删除文件
- os.remove(file_path)
- logger.info(f"文件 {file_path} 已删除")
- elif context.type == ContextType.IMAGE:
- if isgroup and not self.image_sum_group:
- # 群聊中忽略处理图片
- logger.info("群聊消息,图片处理功能已禁用")
- return
- logger.info("on_handle_context: 开始处理图片")
- context.get("msg").prepare()
- image_path = context.content
- logger.info(f"on_handle_context: 获取到图片路径 {image_path}")
-
-
- # 检查是否应该进行图片总结
- if self.image_sum_enabled:
- # 将图片路径转换为Base64编码的字符串
- base64_image = self.encode_image_to_base64(image_path)
- # 更新params_cache中的last_image_path
- self.params_cache[user_id] = {}
- self.params_cache[user_id]['last_image_base64'] = base64_image
- logger.info('Updated last_image_base64 in params_cache for user.')
- self.handle_image(base64_image, e_context)
-
- else:
- logger.info("图片总结功能已禁用,不对图片内容进行处理")
- # 删除文件
- os.remove(image_path)
- logger.info(f"文件 {image_path} 已删除")
- elif context.type == ContextType.SHARING and self.url_sum_enabled: #匹配卡片分享
- content = html.unescape(content)
- if unsupported_urls: #匹配不支持总结的卡片
- if isgroup: ##群聊中忽略
- return
- else: ##私聊回复不支持
- logger.info("[sum4all] Unsupported URL : %s", content)
- reply = Reply(type=ReplyType.TEXT, content="不支持总结小程序和视频号")
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- else: #匹配支持总结的卡片
- if isgroup: #处理群聊总结
- if self.url_sum_group: #group_sharing = True进行总结,False则忽略。
- logger.info("[sum4all] Summary URL : %s", content)
- # 更新params_cache中的last_url
- self.params_cache[user_id] = {}
- self.params_cache[user_id]['last_url'] = content
- logger.info('Updated last_url in params_cache for user.')
- self.call_service(content, e_context, "sum")
- return
- else:
- return
- else: #处理私聊总结
- logger.info("[sum4all] Summary URL : %s", content)
- # 更新params_cache中的last_url
- self.params_cache[user_id] = {}
- self.params_cache[user_id]['last_url'] = content
- logger.info('Updated last_url in params_cache for user.')
- self.call_service(content, e_context, "sum")
- return
-
- elif url_match and self.url_sum_enabled: #匹配URL链接
- if unsupported_urls: #匹配不支持总结的网址
- logger.info("[sum4all] Unsupported URL : %s", content)
- reply = Reply(type=ReplyType.TEXT, content="不支持总结小程序和视频号")
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- else:
- logger.info("[sum4all] Summary URL : %s", content)
- # 更新params_cache中的last_url
- self.params_cache[user_id] = {}
- self.params_cache[user_id]['last_url'] = content
- logger.info('Updated last_url in params_cache for user.')
- self.call_service(content, e_context, "sum")
- return
- def call_service(self, content, e_context, service_type):
- if service_type == "search":
- if self.search_sum_service == "openai" or self.search_sum_service == "sum4all" or self.search_sum_service == "gemini":
- self.handle_search(content, e_context)
- elif self.search_sum_service == "perplexity":
- self.handle_perplexity(content, e_context)
- elif service_type == "sum":
- if self.url_sum_service == "bibigpt":
- self.handle_bibigpt(content, e_context)
- elif self.url_sum_service == "openai" or self.url_sum_service == "sum4all" or self.url_sum_service == "gemini":
- self.handle_url(content, e_context)
- elif self.url_sum_service == "opensum":
- self.handle_opensum(content, e_context)
- elif service_type == "note":
- if self.note_service == "flomo":
- self.handle_note(content, e_context)
- def handle_note(self,link,e_context):
- msg: ChatMessage = e_context["context"]["msg"]
- user_id = msg.from_user_id
- title = self.params_cache[user_id].get('title', '')
- content = self.params_cache[user_id].get('content', '')
- note = self.params_cache[user_id].get('note', '')
- # 将这些内容按照一定的格式整合到一起
- note_content = f"#sum4all\n{title}\n📒笔记:{note}\n{content}\n{link}"
- payload = {"content": note_content}
- # 将这个字典转换为JSON格式
- payload_json = json.dumps(payload)
- # 创建一个POST请求
- url = self.flomo_key
- headers = {'Content-Type': 'application/json'}
- # 发送这个POST请求
- response = requests.post(url, headers=headers, data=payload_json)
- reply = Reply()
- reply.type = ReplyType.TEXT
- if response.status_code == 200 and response.json()['code'] == 0:
- reply.content = f"已发送到{self.note_service}"
- else:
- reply.content = "发送失败,错误码:" + str(response.status_code)
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def short_url(self, long_url):
- url = "https://short.fatwang2.com"
- payload = {
- "url": long_url
- }
- headers = {'Content-Type': "application/json"}
- response = requests.request("POST", url, json=payload, headers=headers)
- if response.status_code == 200:
- res_data = response.json()
- # 直接从返回的 JSON 中获取短链接
- short_url = res_data.get('shorturl', None)
-
- if short_url:
- return short_url
- return None
- def handle_url(self, content, e_context):
- logger.info('Handling Sum4All request...')
- # 根据sum_service的值选择API密钥和基础URL
- if self.url_sum_service == "openai":
- api_key = self.open_ai_api_key
- api_base = self.open_ai_api_base
- model = self.model
- elif self.url_sum_service == "sum4all":
- api_key = self.sum4all_key
- api_base = "https://pro.sum4all.site/v1"
- model = "sum4all"
- elif self.url_sum_service == "gemini":
- api_key = self.gemini_key
- model = "gemini"
- api_base = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
- else:
- logger.error(f"未知的sum_service配置: {self.url_sum_service}")
- return
-
- msg: ChatMessage = e_context["context"]["msg"]
- user_id = msg.from_user_id
- user_params = self.params_cache.get(user_id, {})
- isgroup = e_context["context"].get("isgroup", False)
- prompt = user_params.get('prompt', self.url_sum_prompt)
- headers = {
- 'Content-Type': 'application/json',
- 'Authorization': f'Bearer {api_key}'
- }
- payload = json.dumps({
- "link": content,
- "prompt": prompt,
- "model": model,
- "base": api_base
- })
- additional_content = "" # 在 try 块之前初始化 additional_content
-
- try:
- logger.info('Sending request to LLM...')
- api_url = "https://ai.sum4all.site"
- response = requests.post(api_url, headers=headers, data=payload)
- response.raise_for_status()
- logger.info('Received response from LLM.')
- response_data = response.json() # 解析响应的 JSON 数据
- if response_data.get("success"):
- content = response_data["content"].replace("\\n", "\n") # 替换 \\n 为 \n
- self.params_cache[user_id]['content'] = content
-
- # 新增加的部分,用于解析 meta 数据
- meta = response_data.get("meta", {}) # 如果没有 meta 数据,则默认为空字典
- title = meta.get("og:title", "") # 获取 og:title,如果没有则默认为空字符串
- self.params_cache[user_id]['title'] = title
- # 只有当 title 非空时,才加入到回复中
- if title:
- additional_content += f"{title}\n\n"
- reply_content = additional_content + content # 将内容加入回复
-
- else:
- reply_content = "Content not found or error in response"
-
- except requests.exceptions.RequestException as e:
- # 处理可能出现的错误
- logger.error(f"Error calling new combined api: {e}")
- reply_content = f"An error occurred"
-
- reply = Reply()
- reply.type = ReplyType.TEXT
- if not self.url_sum_qa_enabled:
- reply.content = remove_markdown(reply_content)
- elif isgroup or not self.note_enabled:
- # reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.url_sum_qa_prefix}+问题,可继续追问"
- reply.content = f"{remove_markdown(reply_content)}"
- elif self.note_enabled:
- reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.url_sum_qa_prefix}+问题,可继续追问。\n\n📒输入{self.note_prefix}+笔记,可发送当前总结&笔记到{self.note_service}"
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def handle_bibigpt(self, content, e_context):
- headers = {
- 'Content-Type': 'application/json'
- }
- payload_params = {
- "url": content,
- "includeDetail": False,
- "promptConfig": {
- "outputLanguage": self.outputLanguage
- }
- }
-
- payload = json.dumps(payload_params)
- try:
- api_url = f"https://bibigpt.co/api/open/{self.bibigpt_key}"
- response = requests.request("POST",api_url, headers=headers, data=payload)
- response.raise_for_status()
- data = json.loads(response.text)
- summary_original = data.get('summary', 'Summary not available')
- html_url = data.get('htmlUrl', 'HTML URL not available')
- # 获取短链接
- short_url = self.short_url(html_url)
-
- # 如果获取短链接失败,使用 html_url
- if short_url is None:
- short_url = html_url if html_url != 'HTML URL not available' else 'URL not available'
-
- # 移除 "##摘要"、"## 亮点" 和 "-"
- summary = summary_original.split("详细版(支持对话追问)")[0].replace("## 摘要\n", "📌总结:").replace("## 亮点\n", "").replace("- ", "")
- except requests.exceptions.RequestException as e:
- reply = f"An error occurred"
-
- reply = Reply()
- reply.type = ReplyType.TEXT
- reply.content = f"{summary}\n\n详细链接:{short_url}"
-
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def handle_opensum(self, content, e_context):
- headers = {
- 'Content-Type': 'application/json',
- 'Authorization': f'Bearer {self.opensum_key}'
- }
- payload = json.dumps({"link": content})
- try:
- api_url = "https://read.thinkwx.com/api/v1/article/summary"
- response = requests.request("POST",api_url, headers=headers, data=payload)
- response.raise_for_status()
- data = json.loads(response.text)
- summary_data = data.get('data', {}) # 获取data字段
- summary_original = summary_data.get('summary', 'Summary not available')
- # 使用正则表达式提取URL
- url_pattern = r'https:\/\/[^\s]+'
- match = re.search(url_pattern, summary_original)
- html_url = match.group(0) if match else 'HTML URL not available'
- # 获取短链接
- short_url = self.short_url(html_url) if match else html_url
- # 用于移除摘要中的URL及其后的所有内容
- url_pattern_remove = r'https:\/\/[^\s]+[\s\S]*'
- summary = re.sub(url_pattern_remove, '', summary_original).strip()
-
- except requests.exceptions.RequestException as e:
- summary = f"An error occurred"
- short_url = 'URL not available'
-
- reply = Reply()
- reply.type = ReplyType.TEXT
- reply.content = f"{summary}\n\n详细链接:{short_url}"
-
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def handle_search(self, content, e_context):
- # 根据sum_service的值选择API密钥和基础URL
- if self.search_sum_service == "openai":
- api_key = self.open_ai_api_key
- api_base = self.open_ai_api_base
- model = self.model
- elif self.search_sum_service == "sum4all":
- api_key = self.sum4all_key
- api_base = "https://pro.sum4all.site/v1"
- model = "sum4all"
- elif self.search_sum_service == "gemini":
- api_key = self.gemini_key
- model = "gemini"
- api_base = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
-
- else:
- logger.error(f"未知的search_service配置: {self.search_sum_service}")
- return
- headers = {
- 'Content-Type': 'application/json',
- 'Authorization': f'Bearer {api_key}'
- }
- content = content[len(self.search_sum_search_prefix):]
- payload = json.dumps({
- "ur": content,
- "prompt": self.search_sum_prompt,
- "model": model,
- "base": api_base,
- "search1api_key": self.search1api_key,
- "search_service": self.search_service
- })
- try:
- api_url = "https://ai.sum4all.site"
- response = requests.post(api_url, headers=headers, data=payload)
- response.raise_for_status()
- response_data = response.json() # 解析响应的 JSON 数据
- if response_data.get("success"):
- content = response_data["content"].replace("\\n", "\n") # 替换 \\n 为 \n
- reply_content = content # 将内容加入回复
-
- # 解析 meta 数据
- meta = response_data.get("meta", {}) # 如果没有 meta 数据,则默认为空字典
- title = meta.get("og:title", "") # 获取 og:title,如果没有则默认为空字符串
- og_url = meta.get("og:url", "") # 获取 og:url,如果没有则默认为空字符串
- # 打印 title 和 og_url 以调试
- print("Title:", title)
- print("Original URL:", og_url)
- # 只有当 title 和 url 非空时,才加入到回复中
- if title:
- reply_content += f"\n\n参考文章:{title}"
- if og_url:
- short_url = self.short_url(og_url) # 获取短链接
- reply_content += f"\n\n参考链接:{short_url}"
-
- else:
- content = "Content not found or error in response"
-
- except requests.exceptions.RequestException as e:
- # 处理可能出现的错误
- logger.error(f"Error calling new combined api: {e}")
- reply_content = f"An error occurred"
-
- 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_perplexity(self, content, e_context):
-
- headers = {
- 'Content-Type': 'application/json',
- 'Authorization': f'Bearer {self.perplexity_key}'
- }
- data = {
- "model": "sonar-small-online",
- "messages": [
- {"role": "system", "content": self.search_sum_prompt},
- {"role": "user", "content": content}
- ]
- }
- try:
- api_url = "https://api.perplexity.ai/chat/completions"
- response = requests.post(api_url, headers=headers, json=data)
- response.raise_for_status()
- # 处理响应数据
- response_data = response.json()
- # 这里可以根据你的需要处理响应数据
- # 解析 JSON 并获取 content
- 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"]:
- content = first_choice["message"]["content"]
- else:
- print("Content not found in the response")
- else:
- print("No choices available in the response")
- except requests.exceptions.RequestException as e:
- # 处理可能出现的错误
- logger.error(f"Error calling perplexity: {e}")
- reply = Reply()
- reply.type = ReplyType.TEXT
- reply.content = f"{remove_markdown(content)}"
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def get_help_text(self, verbose=False, **kwargs):
- help_text = "Help you summarize all things\n"
- if not verbose:
- return help_text
- help_text += "1.Share me the link and I will summarize it for you\n"
- help_text += f"2.{self.search_sum_search_prefix}+query,I will search online for you\n"
- return help_text
- def handle_file(self, content, e_context):
- logger.info("handle_file: 向LLM发送内容总结请求")
- # 根据sum_service的值选择API密钥和基础URL
- if self.file_sum_service == "openai":
- api_key = self.open_ai_api_key
- api_base = self.open_ai_api_base
- model = self.model
- elif self.file_sum_service == "sum4all":
- api_key = self.sum4all_key
- api_base = "https://pro.sum4all.site/v1"
- model = "sum4all"
- elif self.file_sum_service == "gemini":
- api_key = self.gemini_key
- model = "gemini"
- api_base = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
- else:
- logger.error(f"未知的sum_service配置: {self.file_sum_service}")
- return
- 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)
- if model == "gemini":
- headers = {
- 'Content-Type': 'application/json',
- 'x-goog-api-key': api_key
- }
- data = {
- "contents": [
- {"role": "user", "parts": [{"text": prompt}]},
- {"role": "model", "parts": [{"text": "okay"}]},
- {"role": "user", "parts": [{"text": content}]}
- ],
- "generationConfig": {
- "maxOutputTokens": 800
- }
- }
- api_url = api_base
- else:
- headers = {
- 'Content-Type': 'application/json',
- 'Authorization': f'Bearer {api_key}'
- }
- data = {
- "model": model,
- "messages": [
- {"role": "system", "content": prompt},
- {"role": "user", "content": f'Use the following context as your learned knowledge, inside <context></context> XML tags.\n\t<context>{content}</context>\n\t\n\tWhen answer to user:\n\t- If you don\'t know, just say that you don\'t know.\n\t- If you don\'t know when you are not sure, ask for clarification.\n\tAvoid mentioning that you obtained the information from the context.\n\tAnd answer according to the language of the user\'s question.\n\t\t\t\n\tGiven the context information, answer the query.'}
- ]
- }
- api_url = f"{api_base}/chat/completions"
- try:
- logger.info(f'handle_file 文件请求内容: {json.dumps(data)}')
- response = requests.post(api_url, headers=headers, data=json.dumps(data))
- response.raise_for_status()
- response_data = response.json()
-
- # 解析 JSON 并获取 content
- if model == "gemini":
- if "candidates" in response_data and len(response_data["candidates"]) > 0:
- first_candidate = response_data["candidates"][0]
- if "content" in first_candidate:
- if "parts" in first_candidate["content"] and len(first_candidate["content"]["parts"]) > 0:
- response_content = first_candidate["content"]["parts"][0]["text"].strip() # 获取响应内容
- logger.info(f"Gemini API response content: {response_content}") # 记录响应内容
- reply_content = response_content.replace("\\n", "\n") # 替换 \\n 为 \n
- else:
- logger.error("Parts not found in the Gemini API response content")
- reply_content = "Parts not found in the Gemini API response content"
- else:
- logger.error("Content not found in the Gemini API response candidate")
- reply_content = "Content not found in the Gemini API response candidate"
- else:
- logger.error("No candidates available in the Gemini API response")
- reply_content = "No candidates available in the Gemini API response"
- else:
- 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
- logger.info(f"LLM API response content") # 记录响应内容
- reply_content = response_content.replace("\\n", "\n") # 替换 \\n 为 \n
- 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)}\n\n💬5min内输入{self.file_sum_qa_prefix}+问题,可继续追问"
- reply.content = f"{remove_markdown(reply_content)}"
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
- def read_pdf(self, file_path):
- logger.info(f"开始读取PDF文件:{file_path}")
- doc = fitz.open(file_path)
- content = ' '.join([page.get_text() for page in doc])
- # logger.info(f"PDF文件读取完成:{file_path}")
- logger.info(f"PDF文件读取完成:{file_path},内容:{content}")
-
- return content
- def read_word(self, file_path):
- doc = Document(file_path)
- return ' '.join([p.text for p in doc.paragraphs])
- def read_markdown(self, file_path):
- with open(file_path, 'r', encoding='utf-8') as file:
- md_content = file.read()
- return markdown.markdown(md_content)
- def read_excel(self, file_path):
- workbook = load_workbook(file_path)
- content = ''
- for sheet in workbook:
- for row in sheet.iter_rows():
- content += ' '.join([str(cell.value) for cell in row])
- content += '\n'
- return content
- def read_txt(self, file_path):
- logger.debug(f"开始读取TXT文件: {file_path}")
- try:
- with open(file_path, 'r', encoding='utf-8') as file:
- content = file.read()
- logger.debug(f"TXT文件读取完成: {file_path}")
- logger.debug("TXT文件内容的前50个字符:")
- logger.debug(content[:50]) # 打印文件内容的前50个字符
- return content
- except Exception as e:
- logger.error(f"读取TXT文件时出错: {file_path},错误信息: {str(e)}")
- return ""
- def read_csv(self, file_path):
- content = ''
- with open(file_path, 'r', encoding='utf-8') as csvfile:
- reader = csv.reader(csvfile)
- for row in reader:
- content += ' '.join(row) + '\n'
- return content
- def read_html(self, file_path):
- with open(file_path, 'r', encoding='utf-8') as file:
- soup = BeautifulSoup(file, 'html.parser')
- return soup.get_text()
- def read_ppt(self, file_path):
- presentation = Presentation(file_path)
- content = ''
- for slide in presentation.slides:
- for shape in slide.shapes:
- if hasattr(shape, "text"):
- content += shape.text + '\n'
- return content
- def extract_content(self, file_path):
- logger.info(f"extract_content: 提取文件内容,文件路径: {file_path}")
- file_size = os.path.getsize(file_path) // 1000 # 将文件大小转换为KB
- if file_size > int(self.max_file_size):
- logger.warning(f"文件大小超过限制({self.max_file_size}KB),不进行处理。文件大小: {file_size}KB")
- return None
- file_extension = os.path.splitext(file_path)[1][1:].lower()
- logger.info(f"extract_content: 文件类型为 {file_extension}")
-
- file_type = EXTENSION_TO_TYPE.get(file_extension)
-
- if not file_type:
- logger.error(f"不支持的文件扩展名: {file_extension}")
- return None
-
- read_func = {
- 'pdf': self.read_pdf,
- 'docx': self.read_word,
- 'md': self.read_markdown,
- 'txt': self.read_txt,
- 'excel': self.read_excel,
- 'csv': self.read_csv,
- 'html': self.read_html,
- 'ppt': self.read_ppt
- }.get(file_type)
-
- if not read_func:
- logger.error(f"不支持的文件类型: {file_type}")
- return None
- logger.info("extract_content: 文件内容提取完成")
- return read_func(file_path)
- 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
- # Function to handle OpenAI image processing
- def handle_image(self, base64_image, e_context):
- logger.info("handle_image: 解析图像处理API的响应")
- 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.image_sum_prompt)
-
- if self.image_sum_service == "openai":
- api_key = self.open_ai_api_key
- api_base = f"{self.open_ai_api_base}/chat/completions"
- model = "gpt-4o-mini"
- elif self.image_sum_service == "xunfei":
- api_key = self.xunfei_api_key
- api_base = "https://spark.sum4all.site/v1/chat/completions"
- model = "spark-chat-vision"
- elif self.image_sum_service == "sum4all":
- api_key = self.sum4all_key
- api_base = "https://pro.sum4all.site/v1/chat/completions"
- model = "sum4all-vision"
- elif self.image_sum_service == "gemini":
- api_key = self.gemini_key
- api_base = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
- payload = {
- "contents": [
- {
- "parts": [
- {"text": prompt},
- {
- "inline_data": {
- "mime_type":"image/png",
- "data": base64_image
- }
- }
- ]
- }
- ]
- }
- headers = {
- "Content-Type": "application/json",
- "x-goog-api-key": api_key
- }
- logger.info(f"准备发送请求. Payload大小: {len(json.dumps(payload))} 字节")
-
- else:
- logger.error(f"未知的image_sum_service配置: {self.image_sum_service}")
- return
-
- if self.image_sum_service != "gemini":
- payload = {
- "model": model,
- "messages": [
- {
- "role": "user",
- "content": [
- {
- "type": "text",
- "text": prompt
- },
- {
- "type": "image_url",
- "image_url": {
- "url": f"data:image/jpeg;base64,{base64_image}"
- }
- }
- ]
- }
- ],
- "max_tokens": 3000
- }
- headers = {
- "Content-Type": "application/json",
- "Authorization": f"Bearer {api_key}"
- }
-
- try:
- response = requests.post(api_base, headers=headers, json=payload)
- logger.info(f"API请求已发送. 状态码: {response.status_code}")
- response.raise_for_status()
- logger.info("API响应状态码正常,开始解析JSON")
- response_json = response.json()
- logger.info("JSON解析完成")
-
- if self.image_sum_service == "gemini":
- reply_content = response_json.get('candidates', [{}])[0].get('content', {}).get('parts', [{}])[0].get('text', 'No text found in the response')
- logger.info(f"成功解析Gemini响应. 回复内容长度: {len(reply_content)}")
- else:
- if "choices" in response_json and len(response_json["choices"]) > 0:
- first_choice = response_json["choices"][0]
- if "message" in first_choice and "content" in first_choice["message"]:
- response_content = first_choice["message"]["content"].strip()
- logger.info("LLM API response content")
- reply_content = response_content
- 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 Exception as e:
- logger.error(f"Error processing LLM API response: {e}")
- reply_content = f"An error occurred while processing LLM API response"
-
- reply = Reply()
- reply.type = ReplyType.TEXT
- # reply.content = f"{remove_markdown(reply_content)}\n\n💬5min内输入{self.image_sum_qa_prefix}+问题,可继续追问"
- reply.content = f"{remove_markdown(reply_content)}"
- e_context["reply"] = reply
- e_context.action = EventAction.BREAK_PASS
-
- 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
|