v1.7.6

app/ui/package.json

@@ -1,7 +1,7 @@
 {
   "name": "app",
   "private": true,
-  "version": "1.7.5",
+  "version": "1.7.6",
   "type": "module",
   "scripts": {
     "dev": "vite",

app/ui/src/components/Common/BotForm.tsx

@@ -26,6 +26,8 @@ import { YoutubeIcon } from "../Icons/YoutubeIcon";
 import { ApiIcon } from "../Icons/ApiIcon";
 import { SitemapIcon } from "../Icons/SitemapIcon";
 import { useCreateConfig } from "../../hooks/useCreateConfig";
+import { useQuery } from "@tanstack/react-query";
+import api from "../../services/api";
 
 type Props = {
   createBot: (values: any) => void;
@@ -48,6 +50,9 @@ export const BotForm = ({
 }: Props) => {
   const { data: botConfig, status: botConfigStatus } = useCreateConfig();
 
+  const youtubeMode = Form.useWatch(["options", "youtube_mode"], form);
+  const url = Form.useWatch(["content"], form);
+
   const [availableSources] = React.useState([
     {
       id: 1,
@@ -331,21 +336,36 @@ export const BotForm = ({
               },
             ]}
           >
-            <Input type="url" placeholder="Enter the youtube URL" />
+            <Input
+              type="url"
+              size="large"
+              placeholder="Enter the youtube URL"
+            />
           </Form.Item>
 
-          <p className="text-sm text-gray-500">
-            If you find any issues, please report them on{" "}
-            <a
-              href="https://github.com/n4ze3m/dialoqbase/issues/new?title=Github%20issue&labels=bug"
-              target="_blank"
-              rel="noreferrer"
-              className="font-medium text-indigo-600 hover:text-indigo-500"
-            >
-              GitHub
-            </a>
-            .
-          </p>
+          <Form.Item
+            name={["options", "youtube_mode"]}
+            label="Youtube mode"
+            rules={[
+              {
+                required: true,
+                message: "Please select a youtube mode",
+              },
+            ]}
+          >
+            <Select
+              options={[
+                {
+                  label: "Transcribe using Whisper",
+                  value: "whisper",
+                },
+                {
+                  label: "Youtube Transcript",
+                  value: "transcript",
+                },
+              ]}
+            />
+          </Form.Item>
         </>
       ),
     },
@@ -426,6 +446,31 @@ export const BotForm = ({
   const [selectedSource, _setSelectedSource] = React.useState<any>(
     showEmbeddingAndModels ? null : availableSources[0]
   );
+  const { data: transcripts, isLoading: isFetchingTranscript } = useQuery({
+    queryKey: [
+      "fetchYoutubeTranscript",
+      url,
+      youtubeMode,
+      selectedSource?.value,
+    ],
+    queryFn: async () => {
+      if (Boolean(url) && youtubeMode === "transcript") {
+        const res = await api.get("/yt/transcript?url=" + url);
+        return res.data["data"] as {
+          name: {
+            simpleText: string;
+          };
+          languageCode: string;
+        }[];
+      }
+
+      return [];
+    },
+    enabled:
+      Boolean(url) &&
+      youtubeMode === "transcript" &&
+      selectedSource?.value === "youtube",
+  });
 
   return (
     <>
@@ -446,6 +491,7 @@ export const BotForm = ({
               method: "get",
               headers: "{}",
               body: "{}",
+              youtube_mode: "whisper",
             },
           }}
         >
@@ -533,6 +579,45 @@ export const BotForm = ({
             </Row>
           )}
 
+          {selectedSource &&
+            selectedSource.value === "youtube" &&
+            youtubeMode === "transcript" && (
+              <>
+                <Form.Item
+                  name={["options", "language_code"]}
+                  label="Select language"
+                  rules={[
+                    {
+                      required: true,
+                      message: "Please select a language",
+                    },
+                  ]}
+                >
+                  <Select
+                    loading={isFetchingTranscript}
+                    placeholder="Select language"
+                    options={transcripts?.map((transcript) => ({
+                      label: transcript.name.simpleText,
+                      value: transcript.languageCode,
+                    }))}
+                  />
+                </Form.Item>
+
+                <p className="text-sm text-gray-500">
+                  If you find any issues, please report them on{" "}
+                  <a
+                    href="https://github.com/n4ze3m/dialoqbase/issues/new?title=Github%20issue&labels=bug"
+                    target="_blank"
+                    rel="noreferrer"
+                    className="font-medium text-indigo-600 hover:text-indigo-500"
+                  >
+                    GitHub
+                  </a>
+                  .
+                </p>
+              </>
+            )}
+
           <Form.Item hidden={!showEmbeddingAndModels} noStyle>
             <Divider />
           </Form.Item>

package.json

@@ -1,6 +1,6 @@
 {
   "name": "dialoqbase",
-  "version": "1.7.5",
+  "version": "1.7.6",
   "description": "Create chatbots with ease",
   "scripts": {
     "ui:dev": "pnpm run --filter ui dev",

server/package.json

@@ -22,6 +22,9 @@
   "keywords": [],
   "author": "",
   "license": "MIT",
+  "engines": {
+    "node": ">=18"
+  },
   "dependencies": {
     "@fastify/autoload": "^5.0.0",
     "@fastify/cookie": "^9.1.0",
@@ -38,7 +41,11 @@
     "@google-ai/generativelanguage": "^2.0.0",
     "@grammyjs/files": "^1.0.4",
     "@huggingface/inference": "1",
-    "@langchain/google-genai": "^0.0.2",
+    "@langchain/anthropic": "^0.1.4",
+    "@langchain/cohere": "^0.0.5",
+    "@langchain/community": "^0.0.35",
+    "@langchain/google-genai": "^0.0.10",
+    "@langchain/openai": "^0.0.18",
     "@prisma/client": "^5.9.1",
     "@slack/bolt": "^3.13.2",
     "@supabase/supabase-js": "^2.24.0",
@@ -64,7 +71,7 @@
     "html-to-text": "^9.0.5",
     "ignore": "^5.2.4",
     "ioredis": "^5.3.2",
-    "langchain": "^0.0.206",
+    "langchain": "^0.1.25",
     "mammoth": "^1.6.0",
     "pdf-parse": "^1.1.1",
     "pdfjs-dist": "^3.7.107",
@@ -74,6 +81,7 @@
     "ts-node": "^10.9.1",
     "unique-names-generator": "^4.7.1",
     "wavefile": "^11.0.0",
+    "yt-transcript": "^0.0.2",
     "ytdl-core": "^4.11.5"
   },
   "devDependencies": {
@@ -86,6 +94,9 @@
     "tap": "^16.1.0",
     "typescript": "^4.5.4"
   },
+  "resolutions": {
+    "@langchain/core": "0.1.43"
+  },
   "prisma": {
     "seed": "ts-node prisma/seed.ts"
   }

server/prisma/seed.ts

@@ -248,7 +248,7 @@ const LLMS: {
       model_id: "gemini-pro",
       name: "Gemini Pro (Google)",
       model_type: "chat",
-      stream_available: false,
+      stream_available: true,
       model_provider: "Google",
       local_model: false,
       config: "{}",
@@ -271,24 +271,24 @@ const LLMS: {
       local_model: false,
       config: "{}",
     },
-    // {
-    //   name: "Claude 3 Opus 20240229 (Anthropic)",
-    //   model_id: "claude-3-opus-20240229-dbase",
-    //   model_type: "chat",
-    //   model_provider: "Anthropic",
-    //   stream_available: true,
-    //   local_model: false,
-    //   config: "{}",
-    // },
-    // {
-    //   name: "Claude 3 Sonnet 20240229 (Anthropic)",
-    //   model_id: "claude-3-sonnet-20240229-dbase",
-    //   model_type: "chat",
-    //   model_provider: "Anthropic",
-    //   stream_available: true,
-    //   local_model: false,
-    //   config: "{}",
-    // },
+    {
+      name: "Claude 3 Opus 20240229 (Anthropic)",
+      model_id: "claude-3-opus-20240229-dbase",
+      model_type: "chat",
+      model_provider: "Anthropic",
+      stream_available: true,
+      local_model: false,
+      config: "{}",
+    },
+    {
+      name: "Claude 3 Sonnet 20240229 (Anthropic)",
+      model_id: "claude-3-sonnet-20240229-dbase",
+      model_type: "chat",
+      model_provider: "Anthropic",
+      stream_available: true,
+      local_model: false,
+      config: "{}",
+    },
   ];
 
 const EMBEDDING_MODELS: {
@@ -469,10 +469,23 @@ const replaceOldEmbeddings = async () => {
 
 }
 
+
+const updateGeminiStreamingToTrue = async () => {
+  await prisma.dialoqbaseModels.update({
+    where: {
+      model_id: "gemini-pro",
+    },
+    data: {
+      stream_available: true,
+    },
+  });
+}
+
 const main = async () => {
   await newModels();
   await removeTensorflowSupport();
   await replaceOldEmbeddings();
+  await updateGeminiStreamingToTrue()
 };
 
 main()

server/src/chain/index.ts

@@ -1,19 +1,19 @@
-import { BaseLanguageModel } from "langchain/base_language";
-import { Document } from "langchain/document";
+import { BaseLanguageModel } from "@langchain/core/language_models/base";
+import { Document } from "@langchain/core/documents";
 import {
   ChatPromptTemplate,
   MessagesPlaceholder,
   PromptTemplate,
-} from "langchain/prompts";
-import { AIMessage, BaseMessage, HumanMessage } from "langchain/schema";
-import { StringOutputParser } from "langchain/schema/output_parser";
+} from "@langchain/core/prompts";
+import { AIMessage, BaseMessage, HumanMessage } from "@langchain/core/messages";
+import { StringOutputParser } from "@langchain/core/output_parsers";
 import {
   Runnable,
   RunnableBranch,
   RunnableLambda,
   RunnableMap,
   RunnableSequence,
-} from "langchain/schema/runnable";
+} from "@langchain/core/runnables";
 type RetrievalChainInput = {
   chat_history: string;
   question: string;

server/src/embeddings/transformer-embedding.ts

@@ -1,4 +1,4 @@
-import { Embeddings, EmbeddingsParams } from "langchain/embeddings/base";
+import { Embeddings, EmbeddingsParams } from "@langchain/core/embeddings";
 import {
   piplelineTransformer,
   //@ts-ignore

server/src/handlers/api/v1/bot/playground/chat.handler.ts

@@ -4,7 +4,7 @@ import { DialoqbaseVectorStore } from "../../../../../utils/store";
 import { embeddings } from "../../../../../utils/embeddings";
 import { chatModelProvider } from "../../../../../utils/models";
 import { DialoqbaseHybridRetrival } from "../../../../../utils/hybrid";
-import { BaseRetriever } from "langchain/schema/retriever";
+import { BaseRetriever } from "@langchain/core/retrievers";
 import { Document } from "langchain/document";
 import {
   createChain,

server/src/handlers/api/v1/yt/get.handler.ts

@@ -0,0 +1,22 @@
+import { FastifyRequest, FastifyReply } from "fastify";
+import { GetYoutubeTranscript } from "./types";
+import { YtTranscript } from "yt-transcript"
+
+const getYoutubeTranscript = async (url: string) => {
+    const transcript = new YtTranscript({ url });
+    const data = await transcript.listAllTranscripts();
+    return data;
+}
+
+export const getYoutubeTranscriptHandler = async (
+    request: FastifyRequest<GetYoutubeTranscript>,
+    reply: FastifyReply
+) => {
+    const { url } = request.query;
+
+    const available = await getYoutubeTranscript(url);
+
+    return {
+        data: available as any
+    }
+}

server/src/handlers/api/v1/yt/index.ts

@@ -0,0 +1 @@
+export * from "./get.handler"

server/src/handlers/api/v1/yt/types.ts

@@ -0,0 +1,5 @@
+export type GetYoutubeTranscript = {
+    Querystring: {
+        url: string;
+    }
+}

server/src/handlers/bot/api.handler.ts

@@ -5,7 +5,7 @@ import { embeddings } from "../../utils/embeddings";
 import { chatModelProvider } from "../../utils/models";
 import { nextTick } from "./post.handler";
 import { Document } from "langchain/document";
-import { BaseRetriever } from "langchain/schema/retriever";
+import { BaseRetriever } from "@langchain/core/retrievers";
 import { DialoqbaseHybridRetrival } from "../../utils/hybrid";
 import { createChain, groupMessagesByConversation } from "../../chain";
 

server/src/handlers/bot/post.handler.ts

@@ -3,7 +3,7 @@ import { ChatRequestBody } from "./types";
 import { DialoqbaseVectorStore } from "../../utils/store";
 import { embeddings } from "../../utils/embeddings";
 import { chatModelProvider } from "../../utils/models";
-import { BaseRetriever } from "langchain/schema/retriever";
+import { BaseRetriever } from "@langchain/core/retrievers";
 import { DialoqbaseHybridRetrival } from "../../utils/hybrid";
 import { Document } from "langchain/document";
 import { createChain, groupMessagesByConversation } from "../../chain";

server/src/integration/handlers/discord.handler.ts

@@ -4,7 +4,7 @@ import { DialoqbaseVectorStore } from "../../utils/store";
 import { chatModelProvider } from "../../utils/models";
 import { DialoqbaseHybridRetrival } from "../../utils/hybrid";
 import { Document } from "langchain/document";
-import { BaseRetriever } from "langchain/schema/retriever";
+import { BaseRetriever } from "@langchain/core/retrievers";
 import { createChain } from "../../chain";
 const prisma = new PrismaClient();