⚡ (openai) Stream chat completion to avoid serverless timeout (#526)
Closes #520
This commit is contained in:
@@ -0,0 +1,103 @@
|
||||
import { parseVariableNumber } from '@/features/variables/parseVariableNumber'
|
||||
import { Connection } from '@planetscale/database'
|
||||
import { decrypt } from '@typebot.io/lib/api/encryption'
|
||||
import {
|
||||
ChatCompletionOpenAIOptions,
|
||||
OpenAICredentials,
|
||||
} from '@typebot.io/schemas/features/blocks/integrations/openai'
|
||||
import { SessionState } from '@typebot.io/schemas/features/chat'
|
||||
import {
|
||||
ParsedEvent,
|
||||
ReconnectInterval,
|
||||
createParser,
|
||||
} from 'eventsource-parser'
|
||||
import type {
|
||||
ChatCompletionRequestMessage,
|
||||
CreateChatCompletionRequest,
|
||||
} from 'openai'
|
||||
|
||||
export const getChatCompletionStream =
|
||||
(conn: Connection) =>
|
||||
async (
|
||||
state: SessionState,
|
||||
options: ChatCompletionOpenAIOptions,
|
||||
messages: ChatCompletionRequestMessage[]
|
||||
) => {
|
||||
if (!options.credentialsId) return
|
||||
const credentials = (
|
||||
await conn.execute('select data, iv from Credentials where id=?', [
|
||||
options.credentialsId,
|
||||
])
|
||||
).rows.at(0) as { data: string; iv: string } | undefined
|
||||
if (!credentials) {
|
||||
console.error('Could not find credentials in database')
|
||||
return
|
||||
}
|
||||
const { apiKey } = (await decrypt(
|
||||
credentials.data,
|
||||
credentials.iv
|
||||
)) as OpenAICredentials['data']
|
||||
|
||||
const temperature = parseVariableNumber(state.typebot.variables)(
|
||||
options.advancedSettings?.temperature
|
||||
)
|
||||
|
||||
const encoder = new TextEncoder()
|
||||
const decoder = new TextDecoder()
|
||||
|
||||
let counter = 0
|
||||
|
||||
const res = await fetch('https://api.openai.com/v1/chat/completions', {
|
||||
headers: {
|
||||
'Content-Type': 'application/json',
|
||||
Authorization: `Bearer ${apiKey}`,
|
||||
},
|
||||
method: 'POST',
|
||||
body: JSON.stringify({
|
||||
messages,
|
||||
model: options.model,
|
||||
temperature,
|
||||
stream: true,
|
||||
} satisfies CreateChatCompletionRequest),
|
||||
})
|
||||
|
||||
const stream = new ReadableStream({
|
||||
async start(controller) {
|
||||
function onParse(event: ParsedEvent | ReconnectInterval) {
|
||||
if (event.type === 'event') {
|
||||
const data = event.data
|
||||
if (data === '[DONE]') {
|
||||
controller.close()
|
||||
return
|
||||
}
|
||||
try {
|
||||
const json = JSON.parse(data) as {
|
||||
choices: { delta: { content: string } }[]
|
||||
}
|
||||
const text = json.choices.at(0)?.delta.content
|
||||
if (counter < 2 && (text?.match(/\n/) || []).length) {
|
||||
return
|
||||
}
|
||||
const queue = encoder.encode(text)
|
||||
controller.enqueue(queue)
|
||||
counter++
|
||||
} catch (e) {
|
||||
controller.error(e)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// stream response (SSE) from OpenAI may be fragmented into multiple chunks
|
||||
// this ensures we properly read chunks & invoke an event for each SSE event stream
|
||||
const parser = createParser(onParse)
|
||||
|
||||
// https://web.dev/streams/#asynchronous-iteration
|
||||
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
||||
for await (const chunk of res.body as any) {
|
||||
parser.feed(decoder.decode(chunk))
|
||||
}
|
||||
},
|
||||
})
|
||||
|
||||
return stream
|
||||
}
|
||||
Reference in New Issue
Block a user