316 lines
9.3 KiB
TypeScript
316 lines
9.3 KiB
TypeScript
/* eslint-disable @typescript-eslint/no-explicit-any,no-console */
|
|
|
|
import { NextRequest, NextResponse } from 'next/server';
|
|
|
|
import {
|
|
orchestrateDataSources,
|
|
VideoContext,
|
|
} from '@/lib/ai-orchestrator';
|
|
import { getAuthInfoFromCookie } from '@/lib/auth';
|
|
import { getConfig } from '@/lib/config';
|
|
import { db } from '@/lib/db';
|
|
|
|
export const runtime = 'nodejs';
|
|
|
|
interface ChatMessage {
|
|
role: 'user' | 'assistant';
|
|
content: string;
|
|
}
|
|
|
|
interface ChatRequest {
|
|
message: string;
|
|
context?: VideoContext;
|
|
history?: ChatMessage[];
|
|
}
|
|
|
|
/**
|
|
* OpenAI兼容的流式聊天请求
|
|
*/
|
|
async function streamOpenAIChat(
|
|
messages: ChatMessage[],
|
|
config: {
|
|
apiKey: string;
|
|
baseURL: string;
|
|
model: string;
|
|
temperature: number;
|
|
maxTokens: number;
|
|
},
|
|
enableStreaming = true
|
|
): Promise<ReadableStream | Response> {
|
|
const response = await fetch(`${config.baseURL}/chat/completions`, {
|
|
method: 'POST',
|
|
headers: {
|
|
'Content-Type': 'application/json',
|
|
Authorization: `Bearer ${config.apiKey}`,
|
|
},
|
|
body: JSON.stringify({
|
|
model: config.model,
|
|
messages,
|
|
temperature: config.temperature,
|
|
max_tokens: config.maxTokens,
|
|
stream: enableStreaming,
|
|
}),
|
|
});
|
|
|
|
if (!response.ok) {
|
|
throw new Error(
|
|
`OpenAI API error: ${response.status} ${response.statusText}`
|
|
);
|
|
}
|
|
|
|
return enableStreaming ? response.body! : response;
|
|
}
|
|
|
|
/**
|
|
* 转换流为SSE格式
|
|
*/
|
|
function transformToSSE(
|
|
stream: ReadableStream,
|
|
provider: 'openai' | 'claude' | 'custom'
|
|
): ReadableStream {
|
|
const reader = stream.getReader();
|
|
const decoder = new TextDecoder();
|
|
|
|
return new ReadableStream({
|
|
async start(controller) {
|
|
let buffer = ''; // 缓冲区，用于保存不完整的行
|
|
try {
|
|
while (true) {
|
|
const { done, value } = await reader.read();
|
|
if (done) break;
|
|
|
|
const chunk = decoder.decode(value, { stream: true });
|
|
// 将新chunk与缓冲区拼接
|
|
const text = buffer + chunk;
|
|
// 按换行符分割，最后一个元素可能是不完整的行
|
|
const parts = text.split('\n');
|
|
// 保存最后一个不完整的行到缓冲区
|
|
buffer = parts.pop() || '';
|
|
|
|
// 处理完整的行
|
|
const lines = parts.filter((line) => line.trim() !== '');
|
|
|
|
for (const line of lines) {
|
|
if (line.startsWith('data: ')) {
|
|
const data = line.slice(6).trim();
|
|
|
|
// 跳过空数据
|
|
if (!data) {
|
|
continue;
|
|
}
|
|
|
|
if (data === '[DONE]') {
|
|
controller.enqueue(
|
|
new TextEncoder().encode('data: [DONE]\n\n')
|
|
);
|
|
continue;
|
|
}
|
|
|
|
try {
|
|
const json = JSON.parse(data);
|
|
|
|
// 提取文本内容
|
|
let text = '';
|
|
if (provider === 'claude') {
|
|
// Claude格式
|
|
if (json.type === 'content_block_delta') {
|
|
text = json.delta?.text || '';
|
|
}
|
|
} else {
|
|
// OpenAI格式
|
|
text = json.choices?.[0]?.delta?.content || '';
|
|
}
|
|
|
|
if (text) {
|
|
controller.enqueue(
|
|
new TextEncoder().encode(`data: ${JSON.stringify({ text })}\n\n`)
|
|
);
|
|
}
|
|
} catch (e) {
|
|
// 只在非空数据解析失败时打印错误
|
|
if (data.length > 0) {
|
|
console.error('Parse stream chunk error:', e, 'Data:', data.substring(0, 100));
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// 处理缓冲区中剩余的数据
|
|
if (buffer.trim()) {
|
|
const line = buffer.trim();
|
|
if (line.startsWith('data: ')) {
|
|
const data = line.slice(6).trim();
|
|
if (data && data !== '[DONE]') {
|
|
try {
|
|
const json = JSON.parse(data);
|
|
let text = '';
|
|
if (provider === 'claude') {
|
|
if (json.type === 'content_block_delta') {
|
|
text = json.delta?.text || '';
|
|
}
|
|
} else {
|
|
text = json.choices?.[0]?.delta?.content || '';
|
|
}
|
|
if (text) {
|
|
controller.enqueue(
|
|
new TextEncoder().encode(`data: ${JSON.stringify({ text })}\n\n`)
|
|
);
|
|
}
|
|
} catch (e) {
|
|
console.error('Parse final buffer error:', e);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
} catch (error) {
|
|
console.error('Stream error:', error);
|
|
controller.error(error);
|
|
} finally {
|
|
controller.close();
|
|
}
|
|
},
|
|
});
|
|
}
|
|
|
|
export async function POST(request: NextRequest) {
|
|
try {
|
|
// 1. 验证用户登录
|
|
const authInfo = getAuthInfoFromCookie(request);
|
|
if (!authInfo || !authInfo.username) {
|
|
return NextResponse.json({ error: 'Unauthorized' }, { status: 401 });
|
|
}
|
|
|
|
// 2. 获取AI配置
|
|
const adminConfig = await getConfig();
|
|
const aiConfig = adminConfig.AIConfig;
|
|
|
|
if (!aiConfig || !aiConfig.Enabled) {
|
|
return NextResponse.json(
|
|
{ error: 'AI功能未启用' },
|
|
{ status: 400 }
|
|
);
|
|
}
|
|
|
|
// 3. 权限检查：如果不允许普通用户使用，检查用户角色
|
|
if (!aiConfig.AllowRegularUsers) {
|
|
const username = authInfo.username;
|
|
// 站长始终有权限
|
|
if (username !== process.env.USERNAME) {
|
|
// 检查是否为管理员
|
|
const userInfo = await db.getUserInfoV2(username);
|
|
if (!userInfo || (userInfo.role !== 'admin' && userInfo.role !== 'owner') || userInfo.banned) {
|
|
return NextResponse.json(
|
|
{ error: '该功能仅限站长和管理员使用' },
|
|
{ status: 403 }
|
|
);
|
|
}
|
|
}
|
|
}
|
|
|
|
// 4. 解析请求参数
|
|
const body = (await request.json()) as ChatRequest;
|
|
const { message, context, history = [] } = body;
|
|
|
|
if (!message || typeof message !== 'string') {
|
|
return NextResponse.json(
|
|
{ error: '消息内容不能为空' },
|
|
{ status: 400 }
|
|
);
|
|
}
|
|
|
|
console.log('📨 收到AI聊天请求:', {
|
|
message: message.slice(0, 50),
|
|
context,
|
|
historyLength: history.length,
|
|
});
|
|
|
|
// 4. 使用orchestrator协调数据源
|
|
const orchestrationResult = await orchestrateDataSources(
|
|
message,
|
|
context,
|
|
{
|
|
enableWebSearch: aiConfig.EnableWebSearch,
|
|
webSearchProvider: aiConfig.WebSearchProvider,
|
|
tavilyApiKey: aiConfig.TavilyApiKey,
|
|
serperApiKey: aiConfig.SerperApiKey,
|
|
serpApiKey: aiConfig.SerpApiKey,
|
|
// TMDB 配置
|
|
tmdbApiKey: adminConfig.SiteConfig.TMDBApiKey,
|
|
tmdbProxy: adminConfig.SiteConfig.TMDBProxy,
|
|
tmdbReverseProxy: adminConfig.SiteConfig.TMDBReverseProxy,
|
|
// 决策模型配置（固定使用自定义provider，复用主模型的API配置）
|
|
enableDecisionModel: aiConfig.EnableDecisionModel,
|
|
decisionProvider: 'custom',
|
|
decisionApiKey: aiConfig.CustomApiKey,
|
|
decisionBaseURL: aiConfig.CustomBaseURL,
|
|
decisionModel: aiConfig.DecisionCustomModel,
|
|
}
|
|
);
|
|
|
|
console.log('🎯 数据协调完成, systemPrompt长度:', orchestrationResult.systemPrompt.length);
|
|
|
|
// 5. 构建消息列表
|
|
const systemPrompt = aiConfig.SystemPrompt
|
|
? `${aiConfig.SystemPrompt}\n\n${orchestrationResult.systemPrompt}`
|
|
: orchestrationResult.systemPrompt;
|
|
|
|
const messages: ChatMessage[] = [
|
|
{ role: 'user', content: systemPrompt },
|
|
{ role: 'assistant', content: '明白了，我会按照要求回答用户的问题。' },
|
|
...history,
|
|
{ role: 'user', content: message },
|
|
];
|
|
|
|
// 6. 调用自定义API
|
|
const temperature = aiConfig.Temperature ?? 0.7;
|
|
const maxTokens = aiConfig.MaxTokens ?? 1000;
|
|
const enableStreaming = aiConfig.EnableStreaming !== false; // 默认启用流式响应
|
|
|
|
if (!aiConfig.CustomApiKey || !aiConfig.CustomBaseURL) {
|
|
return NextResponse.json(
|
|
{ error: '自定义API配置不完整' },
|
|
{ status: 400 }
|
|
);
|
|
}
|
|
|
|
const result = await streamOpenAIChat(messages, {
|
|
apiKey: aiConfig.CustomApiKey,
|
|
baseURL: aiConfig.CustomBaseURL,
|
|
model: aiConfig.CustomModel || 'gpt-3.5-turbo',
|
|
temperature,
|
|
maxTokens,
|
|
}, enableStreaming);
|
|
|
|
// 7. 根据是否启用流式响应返回不同格式
|
|
if (enableStreaming) {
|
|
// 流式响应：转换为SSE格式并返回
|
|
const sseStream = transformToSSE(result as ReadableStream, 'openai');
|
|
|
|
return new NextResponse(sseStream, {
|
|
headers: {
|
|
'Content-Type': 'text/event-stream',
|
|
'Cache-Control': 'no-cache',
|
|
Connection: 'keep-alive',
|
|
},
|
|
});
|
|
} else {
|
|
// 非流式响应：等待完整响应后返回JSON
|
|
const response = result as Response;
|
|
const data = await response.json();
|
|
const content = data.choices?.[0]?.message?.content || '';
|
|
|
|
return NextResponse.json({ content });
|
|
}
|
|
} catch (error) {
|
|
console.error('❌ AI聊天API错误:', error);
|
|
return NextResponse.json(
|
|
{
|
|
error: 'AI聊天请求失败',
|
|
details: (error as Error).message,
|
|
},
|
|
{ status: 500 }
|
|
);
|
|
}
|
|
}
|