281 lines
8.2 KiB
JavaScript
281 lines
8.2 KiB
JavaScript
/**
|
||
* ═══════════════════════════════════════════════════════════
|
||
* AI桥接服务 · AI Bridge Service
|
||
* ═══════════════════════════════════════════════════════════
|
||
*
|
||
* 智库节点 Phase 4 · ZY-SVR-006
|
||
* 守护: 铸渊 · ICE-GL-ZY001
|
||
* 版权: 国作登字-2026-A-00037559
|
||
*
|
||
* 功能:
|
||
* - 接入多模型 AI API（DeepSeek / Kimi / 通义千问）
|
||
* - 自动降级：按优先级尝试，一个失败自动切换下一个
|
||
* - 支持第三方 OpenAI 兼容 API（LLM_BASE_URL）
|
||
* - SSE 流式响应预留
|
||
*
|
||
* 模型优先级:
|
||
* 1. DeepSeek-Chat（成本低·中文强）
|
||
* 2. Kimi / Moonshot（长上下文）
|
||
* 3. 通义千问 Qwen（阿里云原生）
|
||
* 4. 第三方 OpenAI 兼容 API（备用）
|
||
*
|
||
* ═══════════════════════════════════════════════════════════
|
||
*/
|
||
|
||
'use strict';
|
||
|
||
const https = require('https');
|
||
const http = require('http');
|
||
|
||
// ─── 模型配置（从环境变量读取 API Key） ───
|
||
const MODELS = [
|
||
{
|
||
id: 'deepseek',
|
||
name: 'DeepSeek-Chat',
|
||
model: 'deepseek-chat',
|
||
endpoint: 'https://api.deepseek.com/v1/chat/completions',
|
||
envKey: 'DEEPSEEK_API_KEY',
|
||
altEnvKey: 'ZY_DEEPSEEK_API_KEY'
|
||
},
|
||
{
|
||
id: 'kimi',
|
||
name: 'Kimi (Moonshot)',
|
||
model: 'moonshot-v1-8k',
|
||
endpoint: 'https://api.moonshot.cn/v1/chat/completions',
|
||
envKey: 'KIMI_API_KEY',
|
||
altEnvKey: 'ZY_KIMI_API_KEY'
|
||
},
|
||
{
|
||
id: 'qwen',
|
||
name: '通义千问 Qwen',
|
||
model: 'qwen-max',
|
||
endpoint: 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
|
||
envKey: 'DASHSCOPE_API_KEY',
|
||
altEnvKey: 'ZY_QIANWEN_API_KEY'
|
||
},
|
||
{
|
||
id: 'third-party',
|
||
name: '第三方 OpenAI 兼容',
|
||
model: process.env.LLM_MODEL || 'gpt-4o',
|
||
endpoint: (process.env.LLM_BASE_URL || '').replace(/\/$/, '') + '/chat/completions',
|
||
envKey: 'LLM_API_KEY',
|
||
altEnvKey: ''
|
||
}
|
||
];
|
||
|
||
/**
|
||
* 获取模型 API Key
|
||
*/
|
||
function getApiKey(modelCfg) {
|
||
const key = process.env[modelCfg.envKey] || '';
|
||
if (key && key.length > 5) return key;
|
||
if (modelCfg.altEnvKey) {
|
||
const altKey = process.env[modelCfg.altEnvKey] || '';
|
||
if (altKey && altKey.length > 5) return altKey;
|
||
}
|
||
return '';
|
||
}
|
||
|
||
/**
|
||
* 获取可用模型列表
|
||
*/
|
||
function getAvailableModels() {
|
||
return MODELS.filter(m => {
|
||
if (m.id === 'third-party' && !process.env.LLM_BASE_URL) return false;
|
||
return !!getApiKey(m);
|
||
}).map(m => ({ id: m.id, name: m.name, model: m.model }));
|
||
}
|
||
|
||
/**
|
||
* HTTP POST 请求
|
||
*/
|
||
function httpPost(url, body, headers, timeoutMs) {
|
||
return new Promise((resolve, reject) => {
|
||
if (!url) {
|
||
return reject(new Error('无效的 API endpoint'));
|
||
}
|
||
const parsed = new URL(url);
|
||
const transport = parsed.protocol === 'https:' ? https : http;
|
||
|
||
const req = transport.request({
|
||
hostname: parsed.hostname,
|
||
port: parsed.port || (parsed.protocol === 'https:' ? 443 : 80),
|
||
path: parsed.pathname + parsed.search,
|
||
method: 'POST',
|
||
headers: {
|
||
...headers,
|
||
'Content-Length': Buffer.byteLength(body)
|
||
},
|
||
timeout: timeoutMs || 60000
|
||
}, (res) => {
|
||
let data = '';
|
||
res.on('data', chunk => { data += chunk; });
|
||
res.on('end', () => {
|
||
if (res.statusCode >= 400) {
|
||
reject(new Error(`API 错误 (${res.statusCode}): ${data.slice(0, 200)}`));
|
||
return;
|
||
}
|
||
resolve(data);
|
||
});
|
||
});
|
||
|
||
req.on('error', err => reject(new Error(`网络错误: ${err.message}`)));
|
||
req.on('timeout', () => {
|
||
req.destroy();
|
||
reject(new Error('API 请求超时'));
|
||
});
|
||
req.write(body);
|
||
req.end();
|
||
});
|
||
}
|
||
|
||
/**
|
||
* 调用单个模型
|
||
*/
|
||
async function callSingleModel(modelCfg, messages, options) {
|
||
const apiKey = getApiKey(modelCfg);
|
||
if (!apiKey) throw new Error(`${modelCfg.name} API Key 未配置`);
|
||
|
||
const requestBody = {
|
||
model: modelCfg.model,
|
||
messages: messages,
|
||
temperature: options.temperature || 0.7,
|
||
max_tokens: options.maxTokens || 2000,
|
||
stream: false
|
||
};
|
||
|
||
const response = await httpPost(
|
||
modelCfg.endpoint,
|
||
JSON.stringify(requestBody),
|
||
{
|
||
'Content-Type': 'application/json',
|
||
'Authorization': `Bearer ${apiKey}`
|
||
},
|
||
options.timeout || 60000
|
||
);
|
||
|
||
const data = JSON.parse(response);
|
||
|
||
if (!data.choices || !data.choices[0] || !data.choices[0].message) {
|
||
throw new Error(`${modelCfg.name} 返回格式异常`);
|
||
}
|
||
|
||
return {
|
||
content: data.choices[0].message.content,
|
||
model: data.model || modelCfg.model,
|
||
provider: modelCfg.id,
|
||
usage: data.usage || {}
|
||
};
|
||
}
|
||
|
||
/**
|
||
* 智能调用 AI（自动降级）
|
||
*
|
||
* @param {Array} messages OpenAI 格式消息列表
|
||
* @param {object} options { temperature, maxTokens, preferredModel, timeout }
|
||
* @returns {Promise<object>} { content, model, provider, usage }
|
||
*/
|
||
async function callAI(messages, options) {
|
||
options = options || {};
|
||
|
||
// 如果指定了首选模型，优先使用
|
||
let models = MODELS.filter(m => {
|
||
if (m.id === 'third-party' && !process.env.LLM_BASE_URL) return false;
|
||
return !!getApiKey(m);
|
||
});
|
||
|
||
if (options.preferredModel) {
|
||
const preferred = models.find(m => m.id === options.preferredModel);
|
||
if (preferred) {
|
||
models = [preferred, ...models.filter(m => m.id !== options.preferredModel)];
|
||
}
|
||
}
|
||
|
||
if (models.length === 0) {
|
||
// 没有任何 API Key 可用，返回离线回复
|
||
return {
|
||
content: '⚠️ AI服务暂未配置。请在环境变量中设置 DEEPSEEK_API_KEY / KIMI_API_KEY / DASHSCOPE_API_KEY 中的任意一个即可启用AI对话。',
|
||
model: 'offline',
|
||
provider: 'none',
|
||
usage: {}
|
||
};
|
||
}
|
||
|
||
let lastError = null;
|
||
|
||
for (const modelCfg of models) {
|
||
try {
|
||
const result = await callSingleModel(modelCfg, messages, options);
|
||
console.log(`[AI-Bridge] ✅ ${modelCfg.name} 调用成功`);
|
||
return result;
|
||
} catch (err) {
|
||
console.error(`[AI-Bridge] ⚠️ ${modelCfg.name} 调用失败: ${err.message}`);
|
||
lastError = err;
|
||
continue;
|
||
}
|
||
}
|
||
|
||
// 全部失败
|
||
return {
|
||
content: `⚠️ AI服务暂时不可用（已尝试 ${models.length} 个模型）。错误: ${lastError ? lastError.message : '未知'}。请稍后重试。`,
|
||
model: 'fallback',
|
||
provider: 'none',
|
||
usage: {}
|
||
};
|
||
}
|
||
|
||
/**
|
||
* 构建智库 Agent 系统提示词
|
||
*/
|
||
function buildAgentSystemPrompt(agent) {
|
||
const nickname = agent.nickname || '成员';
|
||
const memoryCount = agent.memory ? agent.memory.length : 0;
|
||
const noteCount = agent.notes ? agent.notes.length : 0;
|
||
|
||
// 获取最近的记忆摘要
|
||
const recentMemories = (agent.memory || [])
|
||
.slice(-5)
|
||
.map(m => `- [${m.type}] ${m.content.slice(0, 100)}`)
|
||
.join('\n');
|
||
|
||
return `你是光湖智库的AI阅读助手，服务于成员「${nickname}」。
|
||
|
||
你的核心能力：
|
||
1. 📖 阅读辅助 — 帮助理解小说内容、分析情节、解读人物
|
||
2. ✂️ 拆文整理 — 分析小说结构、提取大纲、梳理人物关系
|
||
3. 📝 摘要生成 — 为章节或整本书生成摘要
|
||
4. 💡 创作建议 — 提供写作灵感、情节走向建议
|
||
5. 🔍 内容分析 — 分析文风、叙事手法、节奏把控
|
||
|
||
成员档案：
|
||
- 昵称：${nickname}
|
||
- 记忆条目：${memoryCount} 条
|
||
- 阅读笔记：${noteCount} 条
|
||
${recentMemories ? '\n最近记忆：\n' + recentMemories : ''}
|
||
|
||
交互规则：
|
||
- 使用中文回复，语气友好且专业
|
||
- 回复简洁有力，避免冗长
|
||
- 涉及小说分析时，给出具体的分析角度
|
||
- 可以使用 emoji 增强表达`;
|
||
}
|
||
|
||
/**
|
||
* 健康检查
|
||
*/
|
||
function healthCheck() {
|
||
const available = getAvailableModels();
|
||
return {
|
||
status: available.length > 0 ? 'ok' : 'no_models',
|
||
available_models: available,
|
||
total_configured: MODELS.length
|
||
};
|
||
}
|
||
|
||
module.exports = {
|
||
callAI,
|
||
getAvailableModels,
|
||
buildAgentSystemPrompt,
|
||
healthCheck
|
||
};
|