fyelci / serverless-openai-chatgpt-stream

Serverless OpenAI ChatGPT Stream

API Gateway doesn't support streaming, so this is a workaround to stream data from OpenAI ChatGPT.

Lambda offers function urls which allows you to call Lambda function via http. Lambda function urls support streaming, so you can stream data from OpenAI ChatGPT.

Setup and deployment

You need to copy .env.example to .env and fill in the OPENAI_API_KEY environment variable.

• Run npm install to install all the necessary dependencies.

Deploy on AWS, simply run:

$ npm run deploy

# or

$ serverless deploy

This will deploy your function and print your lambda function's url like below:

endpoint: https://XXXXXXXXXXXXXX.lambda-url.eu-west-2.on.aws/

Usage

You can test your endpoint like below

curl --location 'https://XXXXXXXXXXXXXX.lambda-url.eu-west-2.on.aws/' \
--header 'Content-Type: application/json' \
--data '{"chatLog": [{ "role": "user",  "content": "Write me a blog post about London." }]}'

serverless-openai-chatgpt-stream's Issues

Unable to use the project

Hi there,

I've been trying to get this working, but keep running into the issue below.

Issue:

2023-07-30T08:49:26.179Z d6bf5a4e-c18e-4b84-b526-fa1416f1566c ERROR TypeError: The "streams" argument must be specified at new NodeError (node:internal/errors:399:5) at pipelineImpl (node:internal/streams/pipeline:193:11) at node:stream/promises:31:5 at new Promise (<anonymous>) at pipeline (node:stream/promises:20:10) at /var/src/handlers/chat.ts:73:9 at Generator.next (<anonymous>) at fulfilled (/var/task/src/handlers/chat.js:5:58) at processTicksAndRejections (node:internal/process/task_queues:95:5) { code: 'ERR_MISSING_ARGS' }

I run and populate:

cp .env.example .env

Followed by:

npm install
npm run deploy

I added the environment variables into Serverless too, and everything seems to go well and logs the event body correctly up until the issue above.