import { auth } from '@/auth'
import { kv } from '@vercel/kv'
import { OpenAIStream, StreamingTextResponse } from 'ai-connector'
import { Configuration, OpenAIApi } from 'openai-edge'

import { nanoid } from '@/lib/utils'

// export const runtime = 'edge'

const configuration = new Configuration({
  apiKey: process.env.OPENAI_API_KEY
})

const openai = new OpenAIApi(configuration)

if (!process.env.OPENAI_API_KEY) {
  throw new Error('Missing env var from OpenAI')
}

export const POST = auth(async function POST(req: Request) {
  const json = await req.json()
  // @ts-ignore
  console.log(req.auth) // todo fix types
  const messages = json.messages.map((m: any) => ({
    content: m.content,
    role: m.role
  }))
  const res = await openai.createChatCompletion({
    model: 'gpt-3.5-turbo',
    messages,
    temperature: 0.7,
    top_p: 1,
    frequency_penalty: 1,
    max_tokens: 500,
    n: 1,
    stream: true
  })

  const stream = await OpenAIStream(res, {
    async onCompletion(completion) {
      // @ts-ignore
      if (req.auth?.user?.email == null) {
        return
      }
      const title = json.messages[0].content.substring(0, 100)
      const userId = (req as any).auth?.user?.email
      const id = json.id ?? nanoid()
      const createdAt = Date.now()
      const payload = {
        id,
        title,
        userId,
        createdAt,
        messages: [
          ...messages,
          {
            content: completion,
            role: 'assistant'
          }
        ]
      }
      await kv.hmset(`chat:${id}`, payload)
      await kv.zadd(`user:chat:${userId}`, {
        score: createdAt,
        member: `chat:${id}`
      })
    }
  })

  return new StreamingTextResponse(stream)
})