mirror of
https://github.com/ikechan8370/chatgpt-plugin.git
synced 2025-12-16 21:37:11 +00:00
218 lines
7.8 KiB
JavaScript
218 lines
7.8 KiB
JavaScript
import { SendMessageOption } from 'chaite'
|
||
import ChatGPTConfig from '../../config/config.js'
|
||
import { getClientForModel } from '../chaite/vectorizer.js'
|
||
|
||
function collectTextFromResponse (response) {
|
||
if (!response?.contents) {
|
||
return ''
|
||
}
|
||
return response.contents
|
||
.filter(content => content.type === 'text')
|
||
.map(content => content.text || '')
|
||
.join('\n')
|
||
.trim()
|
||
}
|
||
|
||
function parseJSON (text) {
|
||
if (!text) {
|
||
return null
|
||
}
|
||
const trimmed = text.trim()
|
||
const codeBlockMatch = trimmed.match(/^```(?:json)?\s*([\s\S]*?)\s*```$/i)
|
||
const payload = codeBlockMatch ? codeBlockMatch[1] : trimmed
|
||
try {
|
||
return JSON.parse(payload)
|
||
} catch (err) {
|
||
logger.warn('Failed to parse JSON from memory extractor response:', text)
|
||
return null
|
||
}
|
||
}
|
||
|
||
function formatEntry (entry) {
|
||
let str = ''
|
||
try {
|
||
if (typeof entry === 'string') {
|
||
str = entry
|
||
} else {
|
||
str = JSON.stringify(entry)
|
||
}
|
||
} catch (err) {
|
||
str = String(entry)
|
||
}
|
||
const limit = 200
|
||
return str.length > limit ? str.slice(0, limit) + '…' : str
|
||
}
|
||
|
||
function buildGroupSystemPrompt () {
|
||
return `You are a knowledge extraction assistant that specialises in summarising long-term facts from chat transcripts.
|
||
Read the provided group conversation and identify statements that should be stored as long-term knowledge for the group.
|
||
Note that you should only record valuable information, as this will help the LLM assistant use it as objective facts to answer questions in the future.
|
||
Return a JSON array. Each element must contain:
|
||
{
|
||
"fact": 事实内容，必须完整包含事件的各个要素而不能是简单的短语（比如谁参与了事件、做了什么事情、背景时间是什么）（同一件事情尽可能整合为同一条而非拆分，以便利于检索）,
|
||
"topic": 主题关键词，字符串，如 "活动"、"成员信息",
|
||
"importance": 一个介于0和1之间的小数，数值越大表示越重要,
|
||
"source_message_ids": 原始消息ID数组,
|
||
"source_messages": 对应原始消息的简要摘录或合并文本,
|
||
"involved_users": 出现或相关的用户ID数组
|
||
}
|
||
Only include meaningful, verifiable information about group members that is useful for future conversations. Do not record incomplete information. Do not record other common sense which is not specified for the group`
|
||
}
|
||
|
||
function buildGroupUserPrompt (messages) {
|
||
const joined = messages.map(msg => {
|
||
const sender = msg.nickname || msg.user_id || '未知用户'
|
||
return `${sender}: ${msg.text}`
|
||
}).join('\n')
|
||
return `以下是群聊中的一些消息，请根据系统说明提取值得长期记忆的事实，以JSON数组形式返回，不要输出额外说明。
|
||
|
||
${joined}`
|
||
}
|
||
|
||
function buildExistingMemorySection (existingMemories = []) {
|
||
if (!existingMemories || existingMemories.length === 0) {
|
||
return '当前没有任何已知的长期记忆。'
|
||
}
|
||
const lines = existingMemories.map((item, idx) => `${idx + 1}. ${item}`)
|
||
return `以下是关于用户的已知长期记忆，请在提取新记忆时参考，避免重复已有事实，并在信息变更时更新描述：\n${lines.join('\n')}`
|
||
}
|
||
|
||
function buildUserSystemPrompt (existingMemories = []) {
|
||
return `You are an assistant that extracts long-term personal preferences or persona details about a user.
|
||
Given a conversation snippet between the user and the bot, identify durable information such as preferences, nicknames, roles, speaking style, habits, or other facts that remain valid over time.
|
||
Return a JSON array of **strings**, and nothing else. The full response must be a json array!!! Each string must be a short sentence (in the same language as the conversation) describing one piece of long-term memory. Do not include embedded JSON objects, or additional metadata. Ignore temporary topics or uncertain information.
|
||
|
||
${buildExistingMemorySection(existingMemories)}`
|
||
}
|
||
|
||
function buildUserPrompt (messages) {
|
||
const body = messages.map(msg => {
|
||
const prefix = msg.role === 'assistant' ? '机器人' : (msg.nickname || msg.user_id || '用户')
|
||
return `${prefix}: ${msg.text}`
|
||
}).join('\n')
|
||
return `下面是用户与机器人的对话，请根据系统提示提取可长期记忆的个人信息。
|
||
|
||
${body}`
|
||
}
|
||
|
||
async function callModel ({ prompt, systemPrompt, model, maxToken = 4096, temperature = 0.2 }) {
|
||
const { client } = await getClientForModel(model)
|
||
const response = await client.sendMessage({
|
||
role: 'user',
|
||
content: [
|
||
{
|
||
type: 'text',
|
||
text: prompt
|
||
}
|
||
]
|
||
}, SendMessageOption.create({
|
||
model,
|
||
// temperature,
|
||
maxToken,
|
||
systemOverride: systemPrompt,
|
||
disableHistoryRead: true,
|
||
disableHistorySave: true,
|
||
stream: false
|
||
}))
|
||
return collectTextFromResponse(response)
|
||
}
|
||
|
||
function resolveGroupExtractionModel () {
|
||
const config = ChatGPTConfig.memory?.group
|
||
if (config?.extractionModel) {
|
||
return config.extractionModel
|
||
}
|
||
if (ChatGPTConfig.llm?.defaultModel) {
|
||
return ChatGPTConfig.llm.defaultModel
|
||
}
|
||
return ChatGPTConfig.llm?.embeddingModel || ''
|
||
}
|
||
|
||
function resolveUserExtractionModel () {
|
||
const config = ChatGPTConfig.memory?.user
|
||
if (config?.extractionModel) {
|
||
return config.extractionModel
|
||
}
|
||
if (ChatGPTConfig.llm?.defaultModel) {
|
||
return ChatGPTConfig.llm.defaultModel
|
||
}
|
||
return ChatGPTConfig.llm?.embeddingModel || ''
|
||
}
|
||
|
||
export async function extractGroupFacts (messages) {
|
||
if (!messages || messages.length === 0) {
|
||
return []
|
||
}
|
||
const model = resolveGroupExtractionModel()
|
||
if (!model) {
|
||
logger.warn('No model configured for group memory extraction')
|
||
return []
|
||
}
|
||
try {
|
||
logger.debug(`[Memory] start group fact extraction, messages=${messages.length}, model=${model}`)
|
||
const text = await callModel({
|
||
prompt: buildGroupUserPrompt(messages),
|
||
systemPrompt: buildGroupSystemPrompt(),
|
||
model
|
||
})
|
||
const parsed = parseJSON(text)
|
||
if (Array.isArray(parsed)) {
|
||
logger.info(`[Memory] extracted ${parsed.length} group facts`)
|
||
parsed.slice(0, 10).forEach((item, idx) => {
|
||
logger.debug(`[Memory] group fact[${idx}] ${formatEntry(item)}`)
|
||
})
|
||
return parsed
|
||
}
|
||
logger.debug('[Memory] group fact extraction returned non-array content')
|
||
return []
|
||
} catch (err) {
|
||
logger.error('Failed to extract group facts:', err)
|
||
return []
|
||
}
|
||
}
|
||
|
||
export async function extractUserMemories (messages, existingMemories = []) {
|
||
if (!messages || messages.length === 0) {
|
||
return []
|
||
}
|
||
const model = resolveUserExtractionModel()
|
||
if (!model) {
|
||
logger.warn('No model configured for user memory extraction')
|
||
return []
|
||
}
|
||
try {
|
||
logger.debug(`[Memory] start user memory extraction, snippets=${messages.length}, existing=${existingMemories.length}, model=${model}`)
|
||
// logger.debug(`[Memory] memories prompt: ${buildUserPrompt(messages)}`)
|
||
// logger.debug(`[Memory] system prompt: ${buildUserSystemPrompt(existingMemories)}`)
|
||
const text = await callModel({
|
||
prompt: buildUserPrompt(messages),
|
||
systemPrompt: buildUserSystemPrompt(existingMemories),
|
||
model
|
||
})
|
||
const parsed = parseJSON(text)
|
||
if (Array.isArray(parsed)) {
|
||
const sentences = parsed.map(item => {
|
||
if (typeof item === 'string') {
|
||
return item.trim()
|
||
}
|
||
if (item && typeof item === 'object') {
|
||
const possible = item.sentence || item.text || item.value || item.fact
|
||
if (possible) {
|
||
return String(possible).trim()
|
||
}
|
||
}
|
||
return ''
|
||
}).filter(Boolean)
|
||
logger.info(`[Memory] extracted ${sentences.length} user memories`)
|
||
sentences.slice(0, 10).forEach((item, idx) => {
|
||
logger.debug(`[Memory] user memory[${idx}] ${formatEntry(item)}`)
|
||
})
|
||
return sentences
|
||
}
|
||
logger.debug('[Memory] user memory extraction returned non-array content')
|
||
return []
|
||
} catch (err) {
|
||
logger.error('Failed to extract user memories:', err)
|
||
return []
|
||
}
|
||
}
|