Update localization messages for Chinese, English, and Japanese languages

src/assets/locale/en/settings.json

@@ -51,6 +51,23 @@
                 "success": "Import Success",
                 "error": "Import Error"
             }
+        },
+        "tts": {
+            "heading": "Text-to-Speech Settings",
+            "ttsEnabled": {
+                "label": "Enable Text-to-Speech"
+            },
+            "ttsProvider": {
+                "label": "Text-to-Speech Provider",
+                "placeholder": "Select a provider"
+            },
+            "ttsVoice": {
+                "label": "Text-to-Speech Voice",
+                "placeholder": "Select a voice"
+            },
+            "ssmlEnabled": {
+                "label": "Enable SSML (Speech Synthesis Markup Language)"
+            }
         }
     },
     "manageModels": {

src/assets/locale/ja-JP/settings.json

@@ -54,6 +54,23 @@
                 "success": "インポート成功",
                 "error": "インポートエラー"
             }
+        },
+        "tts": {
+            "heading": "テキスト読み上げ設定",
+            "ttsEnabled": {
+                "label": "テキスト読み上げを有効にする"
+            },
+            "ttsProvider": {
+                "label": "テキスト読み上げプロバイダー",
+                "placeholder": "プロバイダーを選択"
+            },
+            "ttsVoice": {
+                "label": "テキスト読み上げの音声",
+                "placeholder": "音声を選択"
+            },
+            "ssmlEnabled": {
+                "label": "SSML (Speech Synthesis Markup Language) を有効にする"
+            }
         }
     },
     "manageModels": {

src/assets/locale/ml/settings.json

@@ -54,6 +54,23 @@
                 "success": "ഇമ്പോർട്ട് വിജയകരമായി",
                 "error": "ഇമ്പോർട്ട് പരാജയപ്പെട്ടു"
             }
+        },
+        "tts": {
+            "heading": "ടെക്സ്റ്റ്-ടു-സ്പീച്ച് ക്രമീകരണങ്ങൾ",
+            "ttsEnabled": {
+                "label": "ടെക്സ്റ്റ്-ടു-സ്പീച്ച് പ്രവർത്തനക്ഷമമാക്കുക"
+            },
+            "ttsProvider": {
+                "label": "ടെക്സ്റ്റ്-ടു-സ്പീച്ച് പ്രോവൈഡർ",
+                "placeholder": "ഒരു പ്രോവൈഡർ തിരഞ്ഞെടുക്കുക"
+            },
+            "ttsVoice": {
+                "label": "ടെക്സ്റ്റ്-ടു-സ്പീച്ച് വോയ്സ്",
+                "placeholder": "ഒരു വോയ്സ് തിരഞ്ഞെടുക്കുക"
+            },
+            "ssmlEnabled": {
+                "label": "SSML (സ്പീച്ച് സിന്തസിസ് മാർക്കപ്പ് ലാംഗ്വേജ്) പ്രവർത്തനക്ഷമമാക്കുക"
+            }
         }
     },
     "manageModels": {

src/assets/locale/zh/settings.json

@@ -54,6 +54,23 @@
                 "success": "导入成功",
                 "error": "导入错误"
             }
+        },
+        "tts": {
+            "heading": "文本转语音设置",
+            "ttsEnabled": {
+                "label": "启用文本转语音"
+            },
+            "ttsProvider": {
+                "label": "文本转语音提供商",
+                "placeholder": "选择一个提供商"
+            },
+            "ttsVoice": {
+                "label": "文本转语音语音",
+                "placeholder": "选择一种语音"
+            },
+            "ssmlEnabled": {
+                "label": "启用SSML(语音合成标记语言)"
+            }
         }
     },
     "manageModels": {

src/components/Common/PageAssistProvider.tsx

@@ -11,6 +11,8 @@ export const PageAssistProvider = ({
   const [controller, setController] = React.useState<AbortController | null>(
     null
   )
+  const [embeddingController, setEmbeddingController] =
+    React.useState<AbortController | null>(null)
 
   return (
     <PageAssistContext.Provider
@@ -19,7 +21,10 @@ export const PageAssistProvider = ({
         setMessages,
 
         controller,
-        setController
+        setController,
+
+        embeddingController,
+        setEmbeddingController
       }}>
       {children}
     </PageAssistContext.Provider>

src/components/Sidepanel/Chat/form.tsx

@@ -8,7 +8,7 @@ import { Checkbox, Dropdown, Image, Tooltip } from "antd"
 import { useSpeechRecognition } from "~/hooks/useSpeechRecognition"
 import { useWebUI } from "~/store/webui"
 import { defaultEmbeddingModelForRag } from "~/services/ollama"
-import { ImageIcon, MicIcon, X } from "lucide-react"
+import { ImageIcon, MicIcon, StopCircleIcon, X } from "lucide-react"
 import { useTranslation } from "react-i18next"
 
 type Props = {
@@ -56,8 +56,13 @@ export const SidepanelForm = ({ dropedFile }: Props) => {
 
   useDynamicTextareaSize(textareaRef, form.values.message, 120)
 
-  const { onSubmit, selectedModel, chatMode, speechToTextLanguage } =
+  const {
-    useMessage()
+    onSubmit,
+    selectedModel,
+    chatMode,
+    speechToTextLanguage,
+    stopStreamingRequest
+  } = useMessage()
   const { isListening, start, stop, transcript } = useSpeechRecognition()
 
   React.useEffect(() => {
@@ -217,6 +222,7 @@ export const SidepanelForm = ({ dropedFile }: Props) => {
                     <ImageIcon className="h-5 w-5" />
                   </button>
                 </Tooltip>
+                {!isSending ? (
                   <Dropdown.Button
                     htmlType="submit"
                     disabled={isSending}
@@ -270,6 +276,16 @@ export const SidepanelForm = ({ dropedFile }: Props) => {
                       {t("common:submit")}
                     </div>
                   </Dropdown.Button>
+                ) : (
+                  <Tooltip title={t("tooltip.stopStreaming")}>
+                    <button
+                      type="button"
+                      onClick={stopStreamingRequest}
+                      className="text-gray-800 dark:text-gray-300">
+                      <StopCircleIcon className="h-6 w-6" />
+                    </button>
+                  </Tooltip>
+                )}
               </div>
             </div>
           </form>

src/context/index.tsx

@@ -7,6 +7,9 @@ interface PageAssistContext {
 
   controller: AbortController | null
   setController: Dispatch<SetStateAction<AbortController>>
+
+  embeddingController: AbortController | null
+  setEmbeddingController: Dispatch<SetStateAction<AbortController>>
 }
 
 export const PageAssistContext = createContext<PageAssistContext>({
@@ -14,7 +17,10 @@ export const PageAssistContext = createContext<PageAssistContext>({
   setMessages: () => {},
 
   controller: null,
-  setController: () => {}
+  setController: () => {},
+
+  embeddingController: null,
+  setEmbeddingController: () => {}
 })
 
 export const usePageAssist = () => {

src/hooks/useMessage.tsx

@@ -6,24 +6,36 @@ import {
   promptForRag,
   systemPromptForNonRag
 } from "~/services/ollama"
-import { useStoreMessage, type Message } from "~/store"
+import { type Message } from "~/store/option"
+import { useStoreMessage } from "~/store"
 import { ChatOllama } from "@langchain/community/chat_models/ollama"
 import { HumanMessage, SystemMessage } from "@langchain/core/messages"
 import { getDataFromCurrentTab } from "~/libs/get-html"
 import { OllamaEmbeddings } from "@langchain/community/embeddings/ollama"
-import {
-  createChatWithWebsiteChain,
-  groupMessagesByConversation
-} from "~/chain/chat-with-website"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 import { memoryEmbedding } from "@/utils/memory-embeddings"
+import { ChatHistory } from "@/store/option"
+import { generateID } from "@/db"
+import { saveMessageOnError, saveMessageOnSuccess } from "./chat-helper"
+import { notification } from "antd"
+import { useTranslation } from "react-i18next"
+import { usePageAssist } from "@/context"
+import { formatDocs } from "@/chain/chat-with-x"
+import { OllamaEmbeddingsPageAssist } from "@/models/OllamaEmbedding"
 
 export const useMessage = () => {
   const {
-    history,
+    controller: abortController,
+    setController: setAbortController,
     messages,
-    setHistory,
     setMessages,
+    embeddingController,
+    setEmbeddingController
+  } = usePageAssist()
+  const { t } = useTranslation("option")
+  const {
+    history,
+    setHistory,
     setStreaming,
     streaming,
     setIsFirstMessage,
@@ -45,8 +57,6 @@ export const useMessage = () => {
     setCurrentURL
   } = useStoreMessage()
 
-  const abortControllerRef = React.useRef<AbortController | null>(null)
-
   const [keepTrackOfEmbedding, setKeepTrackOfEmbedding] = React.useState<{
     [key: string]: MemoryVectorStore
   }>({})
@@ -62,8 +72,58 @@ export const useMessage = () => {
     setStreaming(false)
   }
 
-  const chatWithWebsiteMode = async (message: string) => {
+  const chatWithWebsiteMode = async (
-    try {
+    message: string,
+    image: string,
+    isRegenerate: boolean,
+    messages: Message[],
+    history: ChatHistory,
+    signal: AbortSignal,
+    embeddingSignal: AbortSignal
+  ) => {
+    const url = await getOllamaURL()
+
+    const ollama = new ChatOllama({
+      model: selectedModel!,
+      baseUrl: cleanUrl(url)
+    })
+
+    let newMessage: Message[] = []
+    let generateMessageId = generateID()
+
+    if (!isRegenerate) {
+      newMessage = [
+        ...messages,
+        {
+          isBot: false,
+          name: "You",
+          message,
+          sources: [],
+          images: []
+        },
+        {
+          isBot: true,
+          name: selectedModel,
+          message: "▋",
+          sources: [],
+          id: generateMessageId
+        }
+      ]
+    } else {
+      newMessage = [
+        ...messages,
+        {
+          isBot: true,
+          name: selectedModel,
+          message: "▋",
+          sources: [],
+          id: generateMessageId
+        }
+      ]
+    }
+    setMessages(newMessage)
+    let fullText = ""
+    let contentToSave = ""
     let isAlreadyExistEmbedding: MemoryVectorStore
     let embedURL: string, embedHTML: string, embedType: string
     let embedPDF: { content: string; page: number }[] = []
@@ -80,39 +140,19 @@ export const useMessage = () => {
       isAlreadyExistEmbedding = keepTrackOfEmbedding[currentURL]
       embedURL = currentURL
     }
-      let newMessage: Message[] = [
-        ...messages,
-        {
-          isBot: false,
-          name: "You",
-          message,
-          sources: []
-        },
-        {
-          isBot: true,
-          name: selectedModel,
-          message: "▋",
-          sources: []
-        }
-      ]
 
-      const appendingIndex = newMessage.length - 1
     setMessages(newMessage)
     const ollamaUrl = await getOllamaURL()
     const embeddingModle = await defaultEmbeddingModelForRag()
 
-      const ollamaEmbedding = new OllamaEmbeddings({
+    const ollamaEmbedding = new OllamaEmbeddingsPageAssist({
       model: embeddingModle || selectedModel,
-        baseUrl: cleanUrl(ollamaUrl)
+      baseUrl: cleanUrl(ollamaUrl),
+      signal: embeddingSignal
     })
-
-      const ollamaChat = new ChatOllama({
-        model: selectedModel,
-        baseUrl: cleanUrl(ollamaUrl)
-      })
-
     let vectorstore: MemoryVectorStore
 
+    try {
       if (isAlreadyExistEmbedding) {
         vectorstore = isAlreadyExistEmbedding
       } else {
@@ -127,91 +167,175 @@ export const useMessage = () => {
           url: embedURL
         })
       }
-
+      let query = message
       const { ragPrompt: systemPrompt, ragQuestionPrompt: questionPrompt } =
         await promptForRag()
+      if (newMessage.length > 2) {
+        const lastTenMessages = newMessage.slice(-10)
+        lastTenMessages.pop()
+        const chat_history = lastTenMessages
+          .map((message) => {
+            return `${message.isBot ? "Assistant: " : "Human: "}${message.message}`
+          })
+          .join("\n")
+        const promptForQuestion = questionPrompt
+          .replaceAll("{chat_history}", chat_history)
+          .replaceAll("{question}", message)
+        const questionOllama = new ChatOllama({
+          model: selectedModel!,
+          baseUrl: cleanUrl(url)
+        })
+        const response = await questionOllama.invoke(promptForQuestion)
+        query = response.content.toString()
+      }
 
-      const sanitizedQuestion = message.trim().replaceAll("\n", " ")
+      const docs = await vectorstore.similaritySearch(query, 4)
+      const context = formatDocs(docs)
+      const source = docs.map((doc) => {
+        return {
+          ...doc,
+          name: doc?.metadata?.source || "untitled",
+          type: doc?.metadata?.type || "unknown",
+          mode: "chat",
+          url: ""
+        }
+      })
+      message = message.trim().replaceAll("\n", " ")
 
-      const chain = createChatWithWebsiteChain({
+      let humanMessage = new HumanMessage({
-        llm: ollamaChat,
+        content: [
-        question_llm: ollamaChat,
+          {
-        question_template: questionPrompt,
+            text: systemPrompt
-        response_template: systemPrompt,
+              .replace("{context}", context)
-        retriever: vectorstore.asRetriever()
+              .replace("{question}", message),
+            type: "text"
+          }
+        ]
       })
 
-      const chunks = await chain.stream({
+      const applicationChatHistory = generateHistory(history)
-        question: sanitizedQuestion,
+
-        chat_history: groupMessagesByConversation(history)
+      const chunks = await ollama.stream(
-      })
+        [...applicationChatHistory, humanMessage],
+        {
+          signal: signal
+        }
+      )
       let count = 0
       for await (const chunk of chunks) {
+        contentToSave += chunk.content
+        fullText += chunk.content
         if (count === 0) {
           setIsProcessing(true)
-          newMessage[appendingIndex].message = chunk + "▋"
-          setMessages(newMessage)
-        } else {
-          newMessage[appendingIndex].message =
-            newMessage[appendingIndex].message.slice(0, -1) + chunk + "▋"
-          setMessages(newMessage)
         }
-
+        setMessages((prev) => {
+          return prev.map((message) => {
+            if (message.id === generateMessageId) {
+              return {
+                ...message,
+                message: fullText.slice(0, -1) + "▋"
+              }
+            }
+            return message
+          })
+        })
         count++
       }
-
+      // update the message with the full text
-      newMessage[appendingIndex].message = newMessage[
+      setMessages((prev) => {
-        appendingIndex
+        return prev.map((message) => {
-      ].message.slice(0, -1)
+          if (message.id === generateMessageId) {
+            return {
+              ...message,
+              message: fullText,
+              sources: source
+            }
+          }
+          return message
+        })
+      })
 
       setHistory([
         ...history,
         {
           role: "user",
-          content: message
+          content: message,
+          image
         },
         {
           role: "assistant",
-          content: newMessage[appendingIndex].message
+          content: fullText
         }
       ])
 
-      setIsProcessing(false)
+      await saveMessageOnSuccess({
-    } catch (e) {
+        historyId,
+        setHistoryId,
+        isRegenerate,
+        selectedModel: selectedModel,
+        message,
+        image,
+        fullText,
+        source
+      })
+
       setIsProcessing(false)
       setStreaming(false)
+    } catch (e) {
+      const errorSave = await saveMessageOnError({
+        e,
+        botMessage: fullText,
+        history,
+        historyId,
+        image,
+        selectedModel,
+        setHistory,
+        setHistoryId,
+        userMessage: message,
+        isRegenerating: isRegenerate
+      })
 
-      setMessages([
+      if (!errorSave) {
-        ...messages,
+        notification.error({
-        {
+          message: t("error"),
-          isBot: true,
+          description: e?.message || t("somethingWentWrong")
-          name: selectedModel,
+        })
-          message: `Error in chat with website mode. Check out the following logs:
-          
-~~~
-${e?.message}
- ~~~
-        `,
-          sources: []
       }
-      ])
+      setIsProcessing(false)
+      setStreaming(false)
+      setIsProcessing(false)
+      setStreaming(false)
+      setIsEmbedding(false)
+    } finally {
+      setAbortController(null)
+      setEmbeddingController(null)
     }
   }
 
-  const normalChatMode = async (message: string, image: string) => {
+  const normalChatMode = async (
+    message: string,
+    image: string,
+    isRegenerate: boolean,
+    messages: Message[],
+    history: ChatHistory,
+    signal: AbortSignal
+  ) => {
     const url = await getOllamaURL()
 
     if (image.length > 0) {
       image = `data:image/jpeg;base64,${image.split(",")[1]}`
     }
-    abortControllerRef.current = new AbortController()
 
     const ollama = new ChatOllama({
-      model: selectedModel,
+      model: selectedModel!,
       baseUrl: cleanUrl(url)
     })
 
-    let newMessage: Message[] = [
+    let newMessage: Message[] = []
+    let generateMessageId = generateID()
+
+    if (!isRegenerate) {
+      newMessage = [
         ...messages,
         {
           isBot: false,
@@ -224,12 +348,25 @@ ${e?.message}
           isBot: true,
           name: selectedModel,
           message: "▋",
-        sources: []
+          sources: [],
+          id: generateMessageId
         }
       ]
-
+    } else {
-    const appendingIndex = newMessage.length - 1
+      newMessage = [
+        ...messages,
+        {
+          isBot: true,
+          name: selectedModel,
+          message: "▋",
+          sources: [],
+          id: generateMessageId
+        }
+      ]
+    }
     setMessages(newMessage)
+    let fullText = ""
+    let contentToSave = ""
 
     try {
       const prompt = await systemPromptForNonRag()
@@ -277,29 +414,41 @@ ${e?.message}
       const chunks = await ollama.stream(
         [...applicationChatHistory, humanMessage],
         {
-          signal: abortControllerRef.current.signal
+          signal: signal
         }
       )
       let count = 0
       for await (const chunk of chunks) {
+        contentToSave += chunk.content
+        fullText += chunk.content
         if (count === 0) {
           setIsProcessing(true)
-          newMessage[appendingIndex].message = chunk.content + "▋"
-          setMessages(newMessage)
-        } else {
-          newMessage[appendingIndex].message =
-            newMessage[appendingIndex].message.slice(0, -1) +
-            chunk.content +
-            "▋"
-          setMessages(newMessage)
         }
-
+        setMessages((prev) => {
+          return prev.map((message) => {
+            if (message.id === generateMessageId) {
+              return {
+                ...message,
+                message: fullText.slice(0, -1) + "▋"
+              }
+            }
+            return message
+          })
+        })
         count++
       }
 
-      newMessage[appendingIndex].message = newMessage[
+      setMessages((prev) => {
-        appendingIndex
+        return prev.map((message) => {
-      ].message.slice(0, -1)
+          if (message.id === generateMessageId) {
+            return {
+              ...message,
+              message: fullText.slice(0, -1)
+            }
+          }
+          return message
+        })
+      })
 
       setHistory([
         ...history,
@@ -310,28 +459,49 @@ ${e?.message}
         },
         {
           role: "assistant",
-          content: newMessage[appendingIndex].message
+          content: fullText
         }
       ])
 
-      setIsProcessing(false)
+      await saveMessageOnSuccess({
-    } catch (e) {
+        historyId,
+        setHistoryId,
+        isRegenerate,
+        selectedModel: selectedModel,
+        message,
+        image,
+        fullText,
+        source: []
+      })
+
       setIsProcessing(false)
       setStreaming(false)
+      setIsProcessing(false)
+      setStreaming(false)
+    } catch (e) {
+      const errorSave = await saveMessageOnError({
+        e,
+        botMessage: fullText,
+        history,
+        historyId,
+        image,
+        selectedModel,
+        setHistory,
+        setHistoryId,
+        userMessage: message,
+        isRegenerating: isRegenerate
+      })
 
-      setMessages([
+      if (!errorSave) {
-        ...messages,
+        notification.error({
-        {
+          message: t("error"),
-          isBot: true,
+          description: e?.message || t("somethingWentWrong")
-          name: selectedModel,
+        })
-          message: `Something went wrong. Check out the following logs:
-        \`\`\`
-        ${e?.message}
-        \`\`\`
-        `,
-          sources: []
       }
-      ])
+      setIsProcessing(false)
+      setStreaming(false)
+    } finally {
+      setAbortController(null)
     }
   }
 
@@ -342,20 +512,40 @@ ${e?.message}
     message: string
     image: string
   }) => {
+    const newController = new AbortController()
+    let signal = newController.signal
+    setAbortController(newController)
+
     if (chatMode === "normal") {
-      await normalChatMode(message, image)
+      await normalChatMode(message, image, false, messages, history, signal)
     } else {
-      await chatWithWebsiteMode(message)
+      const newEmbeddingController = new AbortController()
+      let embeddingSignal = newEmbeddingController.signal
+      setEmbeddingController(newEmbeddingController)
+      await chatWithWebsiteMode(
+        message,
+        image,
+        false,
+        messages,
+        history,
+        signal,
+        embeddingSignal
+      )
     }
   }
 
   const stopStreamingRequest = () => {
-    if (abortControllerRef.current) {
+    if (isEmbedding) {
-      abortControllerRef.current.abort()
+      if (embeddingController) {
-      abortControllerRef.current = null
+        embeddingController.abort()
+        setEmbeddingController(null)
+      }
+    }
+    if (abortController) {
+      abortController.abort()
+      setAbortController(null)
     }
   }
-
   return {
     messages,
     setMessages,

src/models/OllamaEmbedding.ts

@@ -0,0 +1,255 @@
+import { Embeddings, EmbeddingsParams } from "@langchain/core/embeddings"
+import type { StringWithAutocomplete } from "@langchain/core/utils/types"
+
+export interface OllamaInput {
+  embeddingOnly?: boolean
+  f16KV?: boolean
+  frequencyPenalty?: number
+  headers?: Record<string, string>
+  keepAlive?: string
+  logitsAll?: boolean
+  lowVram?: boolean
+  mainGpu?: number
+  model?: string
+  baseUrl?: string
+  mirostat?: number
+  mirostatEta?: number
+  mirostatTau?: number
+  numBatch?: number
+  numCtx?: number
+  numGpu?: number
+  numGqa?: number
+  numKeep?: number
+  numPredict?: number
+  numThread?: number
+  penalizeNewline?: boolean
+  presencePenalty?: number
+  repeatLastN?: number
+  repeatPenalty?: number
+  ropeFrequencyBase?: number
+  ropeFrequencyScale?: number
+  temperature?: number
+  stop?: string[]
+  tfsZ?: number
+  topK?: number
+  topP?: number
+  typicalP?: number
+  useMLock?: boolean
+  useMMap?: boolean
+  vocabOnly?: boolean
+  format?: StringWithAutocomplete<"json">
+}
+export interface OllamaRequestParams {
+  model: string
+  format?: StringWithAutocomplete<"json">
+  images?: string[]
+  options: {
+    embedding_only?: boolean
+    f16_kv?: boolean
+    frequency_penalty?: number
+    logits_all?: boolean
+    low_vram?: boolean
+    main_gpu?: number
+    mirostat?: number
+    mirostat_eta?: number
+    mirostat_tau?: number
+    num_batch?: number
+    num_ctx?: number
+    num_gpu?: number
+    num_gqa?: number
+    num_keep?: number
+    num_thread?: number
+    num_predict?: number
+    penalize_newline?: boolean
+    presence_penalty?: number
+    repeat_last_n?: number
+    repeat_penalty?: number
+    rope_frequency_base?: number
+    rope_frequency_scale?: number
+    temperature?: number
+    stop?: string[]
+    tfs_z?: number
+    top_k?: number
+    top_p?: number
+    typical_p?: number
+    use_mlock?: boolean
+    use_mmap?: boolean
+    vocab_only?: boolean
+  }
+}
+
+type CamelCasedRequestOptions = Omit<
+  OllamaInput,
+  "baseUrl" | "model" | "format" | "headers"
+>
+
+/**
+ * Interface for OllamaEmbeddings parameters. Extends EmbeddingsParams and
+ * defines additional parameters specific to the OllamaEmbeddings class.
+ */
+interface OllamaEmbeddingsParams extends EmbeddingsParams {
+  /** The Ollama model to use, e.g: "llama2:13b" */
+  model?: string
+
+  /** Base URL of the Ollama server, defaults to "http://localhost:11434" */
+  baseUrl?: string
+
+  /** Extra headers to include in the Ollama API request */
+  headers?: Record<string, string>
+
+  /** Defaults to "5m" */
+  keepAlive?: string
+
+  /** Advanced Ollama API request parameters in camelCase, see
+   * https://github.com/jmorganca/ollama/blob/main/docs/modelfile.md#valid-parameters-and-values
+   * for details of the available parameters.
+   */
+  requestOptions?: CamelCasedRequestOptions
+
+  signal?: AbortSignal
+}
+
+export class OllamaEmbeddingsPageAssist extends Embeddings {
+  model = "llama2"
+
+  baseUrl = "http://localhost:11434"
+
+  headers?: Record<string, string>
+
+  keepAlive = "5m"
+
+  requestOptions?: OllamaRequestParams["options"]
+
+  signal?: AbortSignal
+
+  constructor(params?: OllamaEmbeddingsParams) {
+    super({ maxConcurrency: 1, ...params })
+
+    if (params?.model) {
+      this.model = params.model
+    }
+
+    if (params?.baseUrl) {
+      this.baseUrl = params.baseUrl
+    }
+
+    if (params?.headers) {
+      this.headers = params.headers
+    }
+
+    if (params?.keepAlive) {
+      this.keepAlive = params.keepAlive
+    }
+
+    if (params?.requestOptions) {
+      this.requestOptions = this._convertOptions(params.requestOptions)
+    }
+
+    if (params?.signal) {
+      this.signal = params.signal
+    }
+  }
+
+  /** convert camelCased Ollama request options like "useMMap" to
+   * the snake_cased equivalent which the ollama API actually uses.
+   * Used only for consistency with the llms/Ollama and chatModels/Ollama classes
+   */
+  _convertOptions(requestOptions: CamelCasedRequestOptions) {
+    const snakeCasedOptions: Record<string, unknown> = {}
+    const mapping: Record<keyof CamelCasedRequestOptions, string> = {
+      embeddingOnly: "embedding_only",
+      f16KV: "f16_kv",
+      frequencyPenalty: "frequency_penalty",
+      keepAlive: "keep_alive",
+      logitsAll: "logits_all",
+      lowVram: "low_vram",
+      mainGpu: "main_gpu",
+      mirostat: "mirostat",
+      mirostatEta: "mirostat_eta",
+      mirostatTau: "mirostat_tau",
+      numBatch: "num_batch",
+      numCtx: "num_ctx",
+      numGpu: "num_gpu",
+      numGqa: "num_gqa",
+      numKeep: "num_keep",
+      numPredict: "num_predict",
+      numThread: "num_thread",
+      penalizeNewline: "penalize_newline",
+      presencePenalty: "presence_penalty",
+      repeatLastN: "repeat_last_n",
+      repeatPenalty: "repeat_penalty",
+      ropeFrequencyBase: "rope_frequency_base",
+      ropeFrequencyScale: "rope_frequency_scale",
+      temperature: "temperature",
+      stop: "stop",
+      tfsZ: "tfs_z",
+      topK: "top_k",
+      topP: "top_p",
+      typicalP: "typical_p",
+      useMLock: "use_mlock",
+      useMMap: "use_mmap",
+      vocabOnly: "vocab_only"
+    }
+
+    for (const [key, value] of Object.entries(requestOptions)) {
+      const snakeCasedOption = mapping[key as keyof CamelCasedRequestOptions]
+      if (snakeCasedOption) {
+        snakeCasedOptions[snakeCasedOption] = value
+      }
+    }
+    return snakeCasedOptions
+  }
+
+  async _request(prompt: string): Promise<number[]> {
+    const { model, baseUrl, keepAlive, requestOptions } = this
+
+    let formattedBaseUrl = baseUrl
+    if (formattedBaseUrl.startsWith("http://localhost:")) {
+      // Node 18 has issues with resolving "localhost"
+      // See https://github.com/node-fetch/node-fetch/issues/1624
+      formattedBaseUrl = formattedBaseUrl.replace(
+        "http://localhost:",
+        "http://127.0.0.1:"
+      )
+    }
+
+    const response = await fetch(`${formattedBaseUrl}/api/embeddings`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        ...this.headers
+      },
+      body: JSON.stringify({
+        prompt,
+        model,
+        keep_alive: keepAlive,
+        options: requestOptions
+      }),
+      signal: this.signal
+    })
+    if (!response.ok) {
+      throw new Error(
+        `Request to Ollama server failed: ${response.status} ${response.statusText}`
+      )
+    }
+
+    const json = await response.json()
+    return json.embedding
+  }
+
+  async _embed(texts: string[]): Promise<number[][]> {
+    const embeddings: number[][] = await Promise.all(
+      texts.map((text) => this.caller.call(() => this._request(text)))
+    )
+
+    return embeddings
+  }
+
+  async embedDocuments(documents: string[]) {
+    return this._embed(documents)
+  }
+
+  async embedQuery(document: string) {
+    return (await this.embedDocuments([document]))[0]
+  }
+}

src/public/_locales/en/messages.json

@@ -6,6 +6,6 @@
         "message": "Use your locally running AI models to assist you in your web browsing."
     },
     "openSidePanelToChat": {
-        "message": "Open Side Panel to Chat"
+        "message": "Open Copilot to Chat"
     }
 }

src/public/_locales/ja/messages.json

@@ -6,6 +6,6 @@
     "message": "ローカルで実行中のAIモデルを使って、Webブラウジングをアシストします。"
   },
   "openSidePanelToChat": {
-      "message": "サイドパネルを開いてチャット"
+    "message": "チャットするためにCopilotを開く"
   }
 }

src/public/_locales/zh_CN/messages.json

@@ -6,6 +6,6 @@
         "message": "使用本地运行的 AI 模型来辅助您的网络浏览。"
     },
     "openSidePanelToChat": {
-        "message": "打开侧边栏进行聊天"
+        "message": "打开Copilot进行聊天"
     }
 }

src/utils/memory-embeddings.ts

@@ -2,15 +2,22 @@ import { PageAssistHtmlLoader } from "~/loader/html"
 import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 import { OllamaEmbeddings } from "@langchain/community/embeddings/ollama"
-import { defaultEmbeddingChunkOverlap, defaultEmbeddingChunkSize } from "@/services/ollama"
+import {
+  defaultEmbeddingChunkOverlap,
+  defaultEmbeddingChunkSize
+} from "@/services/ollama"
 import { PageAssistPDFLoader } from "@/loader/pdf"
 
-
+export const getLoader = ({
-export const getLoader = ({ html, pdf, type, url }: {
+  html,
-    url: string,
+  pdf,
-    html: string,
+  type,
-    type: string,
+  url
-    pdf: { content: string, page: number }[]
+}: {
+  url: string
+  html: string
+  type: string
+  pdf: { content: string; page: number }[]
 }) => {
   if (type === "pdf") {
     return new PageAssistPDFLoader({
@@ -25,19 +32,25 @@ export const getLoader = ({ html, pdf, type, url }: {
   }
 }
 
-export const memoryEmbedding = async (
+export const memoryEmbedding = async ({
-    { html,
+  html,
-        keepTrackOfEmbedding, ollamaEmbedding, pdf, setIsEmbedding, setKeepTrackOfEmbedding, type, url }: {
+  keepTrackOfEmbedding,
-            url: string,
+  ollamaEmbedding,
-            html: string,
+  pdf,
-            type: string,
+  setIsEmbedding,
-            pdf: { content: string, page: number }[],
+  setKeepTrackOfEmbedding,
-            keepTrackOfEmbedding: Record<string, MemoryVectorStore>,
+  type,
-            ollamaEmbedding: OllamaEmbeddings,
+  url
-            setIsEmbedding: (value: boolean) => void,
+}: {
+  url: string
+  html: string
+  type: string
+  pdf: { content: string; page: number }[]
+  keepTrackOfEmbedding: Record<string, MemoryVectorStore>
+  ollamaEmbedding: OllamaEmbeddings
+  setIsEmbedding: (value: boolean) => void
   setKeepTrackOfEmbedding: (value: Record<string, MemoryVectorStore>) => void
-        }
+}) => {
-) => {
   setIsEmbedding(true)
 
   const loader = getLoader({ html, pdf, type, url })