feat: Add support for Mammoth library for docx file uploads

package.json

@@ -36,6 +36,7 @@
     "i18next-browser-languagedetector": "^7.2.0",
     "langchain": "^0.1.28",
     "lucide-react": "^0.350.0",
+    "mammoth": "^1.7.2",
     "ml-distance": "^4.0.1",
     "pdfjs-dist": "4.0.379",
     "property-information": "^6.4.1",

src/assets/locale/en/knowledge.json

@@ -32,7 +32,7 @@
         "uploadFile": {
             "label": "Upload File",
             "uploadText": "Drag and drop a file here or click to upload",
-            "uploadHint": "Supported file types: .pdf, .csv, .txt, .md",
+            "uploadHint": "Supported file types: .pdf, .csv, .txt, .md, .docx",
             "required": "File is required"
         },
         "submit": "Submit",

src/assets/locale/ml/knowledge.json

@@ -32,7 +32,7 @@
         "uploadFile": {
             "label": "ഫയല്‍ അപ്‌ലോഡ് ചെയ്യുക",
             "uploadText": "ഇവിടെ ഒരു ഫയല്‍ എടുത്തിടുക അല്ലെങ്കില്‍ അപ്‌ലോഡ് ചെയ്യാന്‍ ക്ലിക്ക് ചെയ്യുക",
-            "uploadHint": "പിന്തുണയുള്ള ഫയല്‍ തരങ്ങള്‍: .pdf, .csv, .txt, .md",
+            "uploadHint": "പിന്തുണയുള്ള ഫയല്‍ തരങ്ങള്‍: .pdf, .csv, .txt, .md,.docx",
             "required": "ഫയല്‍ ആവശ്യമാണ്"
         },
         "submit": "സമര്‍പ്പിക്കുക",

src/assets/locale/ru/knowledge.json

@@ -32,7 +32,7 @@
         "uploadFile": {
             "label": "Загрузить файл",
             "uploadText": "Перетащите файл сюда или нажмите, чтобы загрузить",
-            "uploadHint": "Поддерживаемые типы файлов: .pdf, .csv, .txt, .md",
+            "uploadHint": "Поддерживаемые типы файлов: .pdf, .csv, .txt, .md,.docx",
             "required": "Файл обязателен"
         },
         "submit": "Отправить",

src/components/Option/Knowledge/AddKnowledge.tsx

@@ -90,15 +90,16 @@ export const AddKnowledge = ({ open, setOpen }: Props) => {
             return e?.fileList
           }}>
           <Upload.Dragger
-            accept={".pdf, .csv, .txt, .md"}
+            accept={".pdf, .csv, .txt, .md, .docx"}
             multiple={true}
             maxCount={10}
             beforeUpload={(file) => {
               const allowedTypes = [
                 "application/pdf",
-                // "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
                 "text/csv",
-                "text/plain"
+                "text/plain",
+                "text/markdown",
+                "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
               ]
                 .map((type) => type.toLowerCase())
                 .join(", ")

src/db/index.ts

@@ -7,6 +7,7 @@ type HistoryInfo = {
   id: string
   title: string
   is_rag: boolean
+  message_source?: "copilot" | "web-ui"
   createdAt: number
 }
 
@@ -224,10 +225,10 @@ export const generateID = () => {
   })
 }
 
-export const saveHistory = async (title: string, is_rag?: boolean) => {
+export const saveHistory = async (title: string, is_rag?: boolean, message_source?: "copilot" | "web-ui") => {
   const id = generateID()
   const createdAt = Date.now()
-  const history = { id, title, createdAt, is_rag }
+  const history = { id, title, createdAt, is_rag, message_source }
   const db = new PageAssitDatabase()
   await db.addChatHistory(history)
   return history
@@ -465,3 +466,17 @@ export const importPrompts = async (prompts: Prompts) => {
     await db.addPrompt(prompt)
   }
 }
+
+export const getRecentChatFromCopilot = async () => {
+  const db = new PageAssitDatabase()
+  const chatHistories = await db.getChatHistories()
+  if (chatHistories.length === 0) return null
+  const history = chatHistories.find(
+    (history) => history.message_source === "copilot"
+  )
+  if (!history) return null
+
+  const messages = await db.getChatHistory(history.id)
+
+  return { history, messages }
+}

src/hooks/chat-helper/index.ts

@@ -11,7 +11,8 @@ export const saveMessageOnError = async ({
   historyId,
   selectedModel,
   setHistoryId,
-  isRegenerating
+  isRegenerating,
+  message_source = "web-ui"
 }: {
   e: any
   setHistory: (history: ChatHistory) => void
@@ -22,7 +23,8 @@ export const saveMessageOnError = async ({
   historyId: string | null
   selectedModel: string
   setHistoryId: (historyId: string) => void
-  isRegenerating: boolean
+  isRegenerating: boolean,
+  message_source?: "copilot" | "web-ui"
 }) => {
   if (
     e?.name === "AbortError" ||
@@ -65,7 +67,7 @@ export const saveMessageOnError = async ({
         2
       )
     } else {
-      const newHistoryId = await saveHistory(userMessage)
+      const newHistoryId = await saveHistory(userMessage, false, message_source)
       if (!isRegenerating) {
         await saveMessage(
           newHistoryId.id,
@@ -103,7 +105,8 @@ export const saveMessageOnSuccess = async ({
   message,
   image,
   fullText,
-  source
+  source,
+  message_source = "web-ui"
 }: {
   historyId: string | null
   setHistoryId: (historyId: string) => void
@@ -112,7 +115,8 @@ export const saveMessageOnSuccess = async ({
   message: string
   image: string
   fullText: string
-  source: any[]
+  source: any[],
+  message_source?: "copilot" | "web-ui"
 }) => {
   if (historyId) {
     if (!isRegenerate) {
@@ -136,7 +140,7 @@ export const saveMessageOnSuccess = async ({
       2
     )
   } else {
-    const newHistoryId = await saveHistory(message)
+    const newHistoryId = await saveHistory(message, false, message_source)
     await saveMessage(
       newHistoryId.id,
       selectedModel,

src/hooks/useMessage.tsx

@@ -329,7 +329,8 @@ export const useMessage = () => {
         message,
         image,
         fullText,
-        source
+        source,
+        message_source: "copilot"
       })
 
       setIsProcessing(false)
@@ -345,7 +346,8 @@ export const useMessage = () => {
         setHistory,
         setHistoryId,
         userMessage: message,
-        isRegenerating: isRegenerate
+        isRegenerating: isRegenerate,
+        message_source: "copilot"
       })
 
       if (!errorSave) {
@@ -535,7 +537,8 @@ export const useMessage = () => {
         message,
         image,
         fullText,
-        source: []
+        source: [],
+        message_source: "copilot"
       })
 
       setIsProcessing(false)
@@ -551,7 +554,8 @@ export const useMessage = () => {
         setHistory,
         setHistoryId,
         userMessage: message,
-        isRegenerating: isRegenerate
+        isRegenerating: isRegenerate,
+        message_source: "copilot"
       })
 
       if (!errorSave) {

src/libs/process-knowledge.ts

@@ -9,6 +9,18 @@ import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { PageAssistVectorStore } from "./PageAssistVectorStore"
 import { PageAssisCSVUrlLoader } from "@/loader/csv"
 import { PageAssisTXTUrlLoader } from "@/loader/txt"
+import { PageAssistDocxLoader } from "@/loader/docx"
+
+const readAsArrayBuffer = (file: File): Promise<ArrayBuffer> => {
+  return new Promise((resolve, reject) => {
+    const reader = new FileReader()
+    reader.onload = () => {
+      resolve(reader.result as ArrayBuffer)
+    }
+    reader.onerror = reject
+    reader.readAsArrayBuffer(file)
+  })
+}
 
 export const processKnowledge = async (msg: any, id: string): Promise<void> => {
   console.log(`Processing knowledge with id: ${id}`)
@@ -58,6 +70,26 @@ export const processKnowledge = async (msg: any, id: string): Promise<void> => {
           knownledge_id: knowledge.id,
           file_id: doc.source_id
         })
+      } else if (doc.type === "docx" || doc.type === "application/vnd.openxmlformats-officedocument.wordprocessingml.document") {
+        try {
+          const loader = new PageAssistDocxLoader({
+            fileName: doc.filename,
+            buffer: await toArrayBufferFromBase64(
+              doc.content
+            )
+          })
+
+          let docs = await loader.load()
+
+          const chunks = await textSplitter.splitDocuments(docs)
+
+          await PageAssistVectorStore.fromDocuments(chunks, ollamaEmbedding, {
+            knownledge_id: knowledge.id,
+            file_id: doc.source_id
+          })
+        } catch (error) {
+          console.error(`Error processing knowledge with id: ${id}`, error)
+        }
       } else {
         const loader = new PageAssisTXTUrlLoader({
           name: doc.filename,

src/loader/docx.ts

@@ -0,0 +1,33 @@
+import { BaseDocumentLoader } from "langchain/document_loaders/base"
+import { Document } from "@langchain/core/documents"
+import * as mammoth from "mammoth"
+
+export interface WebLoaderParams {
+    fileName: string
+    buffer: ArrayBuffer
+}
+
+export class PageAssistDocxLoader
+    extends BaseDocumentLoader
+    implements WebLoaderParams {
+    fileName: string
+    buffer: ArrayBuffer
+
+    constructor({ fileName, buffer }: WebLoaderParams) {
+        super()
+        this.fileName = fileName
+        this.buffer = buffer
+    }
+
+    public async load(): Promise<Document[]> {
+        const data = await mammoth.extractRawText({
+            arrayBuffer: this.buffer
+        })
+        const text = data.value
+        const meta = { source: this.fileName }
+        if (text) {
+            return [new Document({ pageContent: text, metadata: meta })]
+        }
+        return []
+    }
+}

src/routes/sidepanel-chat.tsx

@@ -1,16 +1,34 @@
+import {
+  formatToChatHistory,
+  formatToMessage,
+  getRecentChatFromCopilot
+} from "@/db"
 import React from "react"
 import { SidePanelBody } from "~/components/Sidepanel/Chat/body"
 import { SidepanelForm } from "~/components/Sidepanel/Chat/form"
 import { SidepanelHeader } from "~/components/Sidepanel/Chat/header"
 import { useMessage } from "~/hooks/useMessage"
 
- const SidepanelChat = () => {
+const SidepanelChat = () => {
   const drop = React.useRef<HTMLDivElement>(null)
   const [dropedFile, setDropedFile] = React.useState<File | undefined>()
   const [dropState, setDropState] = React.useState<
     "idle" | "dragging" | "error"
   >("idle")
-  const {chatMode} = useMessage()
+  const { chatMode, messages, setHistory, setHistoryId, setMessages } =
+    useMessage()
+
+  const setRecentMessagesOnLoad = async () => {
+    if (messages.length === 0) {
+      const recentChat = await getRecentChatFromCopilot()
+      if (recentChat) {
+        setHistoryId(recentChat.history.id)
+        setHistory(formatToChatHistory(recentChat.messages))
+        setMessages(formatToMessage(recentChat.messages))
+      }
+    }
+  }
+
   React.useEffect(() => {
     if (!drop.current) {
       return
@@ -67,6 +85,7 @@ import { useMessage } from "~/hooks/useMessage"
       }
     }
   }, [])
+
   return (
     <div
       ref={drop}

src/utils/to-source.ts

@@ -10,6 +10,7 @@ export const toBase64 = (file: File | Blob): Promise<string> => {
   })
 }
 
+
 export const toArrayBufferFromBase64 = async (base64: string) => {
   const res = await fetch(base64)
   const blob = await res.blob()