use less langchain

package-lock.json

@@ -19,6 +19,7 @@
                 "eslint": "8.40.0",
                 "eslint-config-next": "13.4.2",
                 "fs-extra": "^11.1.1",
+                "hnswlib-node": "^1.4.2",
                 "langchain": "^0.0.90",
                 "next": "13.4.2",
                 "pdfjs-dist": "^3.7.107",
@@ -911,6 +912,16 @@
                 "node": ">=16.15"
             }
         },
+        "node_modules/@huggingface/inference": {
+            "version": "1.8.0",
+            "resolved": "https://registry.npmjs.org/@huggingface/inference/-/inference-1.8.0.tgz",
+            "integrity": "sha512-Dkh7PiyMf6TINRocQsdceiR5LcqJiUHgWjaBMRpCUOCbs+GZA122VH9q+wodoSptj6rIQf7wIwtDsof+/gd0WA==",
+            "optional": true,
+            "peer": true,
+            "engines": {
+                "node": ">=18"
+            }
+        },
         "node_modules/@humanwhocodes/config-array": {
             "version": "0.11.10",
             "resolved": "https://registry.npmjs.org/@humanwhocodes/config-array/-/config-array-0.11.10.tgz",
@@ -1952,6 +1963,14 @@
             "resolved": "https://registry.npmjs.org/binary-search/-/binary-search-1.3.6.tgz",
             "integrity": "sha512-nbE1WxOTTrUWIfsfZ4aHGYu5DOuNkbxGokjV6Z2kxfJK3uaAb8zNK1muzOeipoLHZjInT4Br88BHpzevc681xA=="
         },
+        "node_modules/bindings": {
+            "version": "1.5.0",
+            "resolved": "https://registry.npmjs.org/bindings/-/bindings-1.5.0.tgz",
+            "integrity": "sha512-p2q/t/mhvuOj/UeLlV6566GD/guowlr0hHxClI0W9m7MWYkL1F0hLo+0Aexs9HSPCtR1SXQ0TD3MMKrXZajbiQ==",
+            "dependencies": {
+                "file-uri-to-path": "1.0.0"
+            }
+        },
         "node_modules/bl": {
             "version": "4.1.0",
             "resolved": "https://registry.npmjs.org/bl/-/bl-4.1.0.tgz",
@@ -2190,6 +2209,13 @@
             "resolved": "https://registry.npmjs.org/chownr/-/chownr-1.1.4.tgz",
             "integrity": "sha512-jJ0bqzaylmJtVnNgzTeSOs8DPavpbYgEr/b0YL8/2GO3xJEhInFmhKMUnEJQjZumK7KXGFhUy89PrsJWlakBVg=="
         },
+        "node_modules/chromadb": {
+            "version": "1.5.2",
+            "resolved": "https://registry.npmjs.org/chromadb/-/chromadb-1.5.2.tgz",
+            "integrity": "sha512-x/rOD7Oo1RiYA+vPK+Ma7CliCHlx26OjUt5J7Z9HZ5Ud1qDrPlvctBycK9Il3zqza96yeUoPQ7gCXHVKNoyvRQ==",
+            "optional": true,
+            "peer": true
+        },
         "node_modules/client-only": {
             "version": "0.0.1",
             "resolved": "https://registry.npmjs.org/client-only/-/client-only-0.0.1.tgz",
@@ -3257,6 +3283,11 @@
                 "node": ">= 12"
             }
         },
+        "node_modules/file-uri-to-path": {
+            "version": "1.0.0",
+            "resolved": "https://registry.npmjs.org/file-uri-to-path/-/file-uri-to-path-1.0.0.tgz",
+            "integrity": "sha512-0Zt+s3L7Vf1biwWZ29aARiVYLx7iMGnEUl9x33fbB/j3jR81u/O2LbqK+Bm1CDSNDKVtJ/YjwY7TUd5SkeLQLw=="
+        },
         "node_modules/filelist": {
             "version": "1.0.4",
             "resolved": "https://registry.npmjs.org/filelist/-/filelist-1.0.4.tgz",
@@ -3767,6 +3798,16 @@
             "integrity": "sha512-8Rf9Y83NBReMnx0gFzA8JImQACstCYWUplepDa9xprwwtmgEZUF0h/i5xSA625zB/I37EtrswSST6OXxwaaIJQ==",
             "optional": true
         },
+        "node_modules/hnswlib-node": {
+            "version": "1.4.2",
+            "resolved": "https://registry.npmjs.org/hnswlib-node/-/hnswlib-node-1.4.2.tgz",
+            "integrity": "sha512-76PIzOaNcX8kOpKwlFPl07uelpctqDMzbiC+Qsk2JWNVkzeU/6iXRk4tfE9z3DoK1RCBrOaFXmQ6RFb1BVF9LA==",
+            "hasInstallScript": true,
+            "dependencies": {
+                "bindings": "^1.5.0",
+                "node-addon-api": "^6.0.0"
+            }
+        },
         "node_modules/hoist-non-react-statics": {
             "version": "3.3.2",
             "resolved": "https://registry.npmjs.org/hoist-non-react-statics/-/hoist-non-react-statics-3.3.2.tgz",

package.json

@@ -19,6 +19,7 @@
         "eslint": "8.40.0",
         "eslint-config-next": "13.4.2",
         "fs-extra": "^11.1.1",
+        "hnswlib-node": "^1.4.2",
         "langchain": "^0.0.90",
         "next": "13.4.2",
         "pdfjs-dist": "^3.7.107",

src/components/ChatWindow.jsx

@@ -3,6 +3,7 @@ import Image from "next/image";
 import { useCallback, useEffect, useState } from "react";
 import MessageList from './MessageList';
 import {FileLoader} from './FileLoader';
+import { db } from '@/utils/db-client';
 import Loader from "./Loader";
 
 function ChatWindow({
@@ -19,16 +20,32 @@ function ChatWindow({
 
   const isReady = loadingStatus.progress === 1;
 
-  const handleSubmit = useCallback(() => {
+  const handleSubmit = useCallback(async () => {
     if (isGenerating || !isReady) {
       return;
     }
     
     if (fileId) {
-      console.log('we have a fileId, so this should be contextual chat')
+      const similarityMatches = 2;
+      const fileContents = await db.docs.get(fileId);
+      const vectorStore = await HNSWLib.fromDocuments(fileContents, new XenovaTransformersEmbeddings());
+      const result = await vectorStore.similaritySearch(userInput, similarityMatches);
+      const qaPrompt = 
+      `You are an AI assistant providing helpful advice. You are given the following extracted parts of a long document and a question. Provide a conversational answer based on the context provided.
+        You should only provide hyperlinks that reference the context below. Do NOT make up hyperlinks.
+        If you can't find the answer in the context below, just say "Hmm, I'm not sure." Don't try to make up an answer.
+        If the question is not related to the context, politely respond that you are tuned to only answer questions that are related to the context.
+        Question: ${userInput}
+        =========
+        ${result}
+        =========
+        Answer:
+        `
+      send(qaPrompt, maxTokens, stopStrings);
+    } else {
+      send(userInput, maxTokens, stopStrings);
     }
-
-    send(userInput, maxTokens, stopStrings);
+    
     setUserInput("");
   }, [
     userInput,

src/pages/api/docChat.ts

@@ -1,58 +0,0 @@
-import type { NextApiRequest, NextApiResponse } from 'next';
-
-import { storesDir } from '@/utils/file-handler';
-
-import { makeChain } from '@/utils/make-chain';
-import XenovaTransformersEmbeddings from '../../embed/hf';
-import { HNSWLib } from 'langchain/vectorstores/hnswlib';
-
-let vectorStore: HNSWLib;
-
-export default async function handler(
-  req: NextApiRequest,
-  res: NextApiResponse,
-) {
-  const { prompt, messages } = req.body;
-
-  if (!prompt) {
-    return res.status(400).json({ message: 'No question in the request' });
-  }
-  // OpenAI recommends replacing newlines with spaces for best results
-  const sanitizedQuestion = prompt.trim().replaceAll('\n', ' ');
-
-  res.writeHead(200, {
-    'Content-Type': 'text/event-stream',
-    'Cache-Control': 'no-cache, no-transform',
-    Connection: 'keep-alive',
-  });
-
-  const sendData = (data: string) => {
-    res.write(`${data}\n\n`);
-  };
-
-  // load store
-  if (!vectorStore) {
-    vectorStore = await HNSWLib.load(storesDir, new XenovaTransformersEmbeddings());
-  }
-  //create chain
-  const chain = makeChain(vectorStore, (token: string) => {
-    // skipping stremaing for now
-    // sendData(JSON.stringify({ data: token }));
-  });
-
-  try {
-    //Ask a question
-    const response = await chain.call({
-      question: sanitizedQuestion,
-      chat_history: messages || [],
-    });
-
-    console.log('response', response);
-    sendData(response.text);
-  } catch (error) {
-    console.log('error', error);
-  } finally {
-    // sendData('[DONE]');
-    res.end();
-  }
-}

src/pages/api/docHandle.ts

@@ -1,10 +1,7 @@
 import type { NextApiRequest, NextApiResponse } from 'next';
 import {
-  readHNSWLibModelFromLocal,
-  storesDir,
   vectorStoreToHNSWLibModel,
 } from '@/utils/file-handler';
-import fs from 'fs-extra';
 import { RecursiveCharacterTextSplitter } from 'langchain/text_splitter';
 import { HNSWLib } from 'langchain/vectorstores/hnswlib';
 import XenovaTransformersEmbeddings from '../../embed/hf'
@@ -27,20 +24,6 @@ export default async function handler(
     return res.status(400).json({ message: 'No question in the request' });
   }
 
-  /*
-  use dexie instead to get contents 
-  const exists = await fs.exists(storesDir);
-  console.log(exists);
-
-  if (exists) {
-    console.log('read from ' + storesDir);
-    const model = await readHNSWLibModelFromLocal();
-    return res.status(200).send({
-      ...model,
-    });
-  }
-  */
-
   const vectorStore = await handleDocs(text);
   const model = await vectorStoreToHNSWLibModel(vectorStore);
   res.status(200).send({

src/utils/make-chain.ts

@@ -1,57 +0,0 @@
-import { CallbackManager } from 'langchain/callbacks';
-import { ConversationalRetrievalQAChain } from 'langchain/chains';
-import { OpenAIChat } from 'langchain/llms';
-import { PromptTemplate } from 'langchain/prompts';
-import { BufferMemory } from "langchain/memory";
-import { HNSWLib } from 'langchain/vectorstores/hnswlib';
-
-export const defaultPrompts = {
-  CONDENSE_PROMPT: `Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question.
-
-  Chat History:
-  {chat_history}
-  Follow Up Input: {question}
-  Standalone question:`,
-  QA_PROMPT: `You are an AI assistant providing helpful advice. You are given the following extracted parts of a long document and a question. Provide a conversational answer based on the context provided.
-  You should only provide hyperlinks that reference the context below. Do NOT make up hyperlinks.
-  If you can't find the answer in the context below, just say "Hmm, I'm not sure." Don't try to make up an answer.
-  If the question is not related to the context, politely respond that you are tuned to only answer questions that are related to the context.
-
-  Question: {question}
-  =========
-  {context}
-  =========
-  Answer:`,
-};
-
-const CONDENSE_PROMPT = PromptTemplate.fromTemplate(
-  defaultPrompts.CONDENSE_PROMPT,
-);
-
-const QA_PROMPT = PromptTemplate.fromTemplate(defaultPrompts.QA_PROMPT);
-
-export const makeChain = (
-  vectorstore: HNSWLib,
-  onTokenStream?: (token: string) => void,
-) => {
-
-  const model = new OpenAIChat({
-    temperature: 0.8,
-    modelName: "OpenAIModelID.GPT_3_5",
-    streaming: false,
-    callbackManager: onTokenStream
-      ? CallbackManager.fromHandlers({
-          async handleLLMNewToken(token) {
-            onTokenStream(token);
-          },
-        })
-      : undefined,
-  })
-
-  return ConversationalRetrievalQAChain.fromLLM(
-    model, vectorstore.asRetriever(),
-    { 
-      memory: new BufferMemory({
-        memoryKey: "chat_history", // Must be set to "chat_history"
-      }), qaTemplate: defaultPrompts.QA_PROMPT, questionGeneratorTemplate: defaultPrompts.CONDENSE_PROMPT })
-};