`LangChainAdapter`

The LangChainAdapter module provides helper functions to transform LangChain output streams into data streams and data stream responses. See the LangChain Adapter documentation for more information.

It supports:

LangChain StringOutputParser streams
LangChain AIMessageChunk streams
LangChain StreamEvents v2 streams

Import

import { LangChainAdapter } from "ai"

API Signature

Methods

toDataStream:

(stream: ReadableStream<LangChainAIMessageChunk> | ReadableStream<string>, AIStreamCallbacksAndOptions) => AIStream

Converts LangChain output streams to data stream.

toDataStreamResponse:

(stream: ReadableStream<LangChainAIMessageChunk> | ReadableStream<string>, options?: {init?: ResponseInit, data?: StreamData, callbacks?: AIStreamCallbacksAndOptions}) => Response

Converts LangChain output streams to data stream response.

mergeIntoDataStream:

(stream: ReadableStream<LangChainStreamEvent> | ReadableStream<LangChainAIMessageChunk> | ReadableStream<string>, options: { dataStream: DataStreamWriter; callbacks?: StreamCallbacks }) => void

Merges LangChain output streams into an existing data stream.

Examples

Convert LangChain Expression Language Stream

app/api/completion/route.ts

import { ChatOpenAI } from '@langchain/openai';
import { LangChainAdapter } from 'ai';

export async function POST(req: Request) {
  const { prompt } = await req.json();

  const model = new ChatOpenAI({
    model: 'gpt-3.5-turbo-0125',
    temperature: 0,
  });

  const stream = await model.stream(prompt);

  return LangChainAdapter.toDataStreamResponse(stream);
}

Convert StringOutputParser Stream

app/api/completion/route.ts

import { ChatOpenAI } from '@langchain/openai';
import { LangChainAdapter } from 'ai';
import { StringOutputParser } from '@langchain/core/output_parsers';

export async function POST(req: Request) {
  const { prompt } = await req.json();

  const model = new ChatOpenAI({
    model: 'gpt-3.5-turbo-0125',
    temperature: 0,
  });

  const parser = new StringOutputParser();
  const stream = await model.pipe(parser).stream(prompt);

  return LangChainAdapter.toDataStreamResponse(stream);
}