langchain-chatbot.js

import { ChatOpenAI } from "langchain/chat_models/openai"
import { PromptTemplate } from "langchain/prompts"
import { SupabaseVectorStore } from 'langchain/vectorstores/supabase'
import { OpenAIEmbeddings } from 'langchain/embeddings/openai'
import { createClient } from '@supabase/supabase-js'
import { StringOutputParser } from 'langchain/schema/output_parser'
import { RunnablePassthrough, RunnableSequence } from "langchain/schema/runnable"

const openAIApiKey = process.env.OPENAI_API_KEY
const supabaseApiKey = process.env.SUPABASE_API_KEY
const supabaseURL = process.env.SUPABASE_URL

function createDatebase() {
  const text = await readFile('aliceInWonderland.txt', 'utf-8');
    
    const splitter = new RecursiveCharacterTextSplitter({
        chunkSize: 500,
        separators: ['\n\n', '\n', ' ', ''], // default setting
        chunkOverlap: 50
    })
    
    const output = await splitter.createDocuments([text])

    const client = createClient(supabaseURL, supabaseApiKey)

    await SupabaseVectorStore.fromDocuments(
        output, 
        new OpenAIEmbeddings({ openAIApiKey }), 
        {
            client,
            tableName: 'documents'
        }
    )
}


const embeddings = new OpenAIEmbeddings({ openAIApiKey })
const client = createClient(supabaseURL, supabaseApiKey)

const vectorStore = new SupabaseVectorStore(embeddings, {
    client,
    tableName: 'documents',
    queryName: 'match_documents'
})

const retriever = vectorStore.asRetriever()

const llm = new ChatOpenAI({ openAIApiKey })

const standaloneQuestionTemplate = `Given a question, convert it to a standalone question: {question} standalone question:`

const standaloneQuestionPrompt = PromptTemplate.fromTemplate(standaloneQuestionTemplate)

const answerTemplate = `You are a helpful and enthusiastic support bot who can answer
    a given question about Alice in Wonderland based on the context provided. Try to find
    the answer in the context. If you really don't know the answer, say "I'm sorry, I don't
    know the answer to that." Don't try to make up an answer. Always speak as if you were
    chatting to a friend.
    context: {context}
    question: {question}
    answer:`

const answerPrompt = PromptTemplate.fromTemplate(answerTemplate)

function combineDocuments(docs){
    return docs.map((doc)=>doc.pageContent).join('\n\n')
}

const standaloneQuestionChain = RunnableSequence.from([
  standaloneQuestionPrompt,
  llm,
  new StringOutputParser()
])

const retrieverChain = RunnableSequence.from([
  prevResult => prevResult.standalone_question,
  retriever,
  combineDocuments
])

const answerChain = RunnableSequence.from([
  answerPrompt,
  llm,
  new StringOutputParser()
])

const chain = RunnableSequence.from([
    {
        standalone_question: standaloneQuestionChain,
        input_variables: new RunnablePassthrough()
    },
    {
        context: retrieverChain,
        question: ({ input_variables }) => input_variables.question
    },
    answerChain
])

const result = await chain.invoke({
    question: question
})