feat: Add text splitting configuration options

src/assets/locale/ar/settings.json

@@ -334,6 +334,14 @@
         "label": "عدد المستندات المسترجعة",
         "placeholder": "أدخل عدد المستندات المسترجعة",
         "required": "الرجاء إدخال عدد المستندات المسترجعة"
+      },
+      "splittingSeparator": {
+        "label": "الفاصل",
+        "placeholder": "أدخل الفاصل (مثال: \\n\\n)",
+        "required": "الرجاء إدخال الفاصل"
+      },
+      "splittingStrategy": {
+        "label": "مقسم النص"
       }
     },
     "prompt": {
@@ -355,4 +363,5 @@
   },
   "chromeAiSettings": {
     "title": "إعدادات Chrome AI"
-  }}
+  }
+}

src/assets/locale/da/settings.json

@@ -331,6 +331,14 @@
         "label": "Antal Hentede Dokumenter",
         "placeholder": "Indtast Number of Retrieved Documents",
         "required": "Venligst indtast the number of retrieved documents"
+      },
+      "splittingSeparator": {
+        "label": "Separator",
+        "placeholder": "Indtast Separator (f.eks. \\n\\n)",
+        "required": "Indtast venligst en separator"
+      },
+      "splittingStrategy": {
+        "label": "Tekst Splitter"
       }
     },
     "prompt": {

src/assets/locale/de/settings.json

@@ -331,6 +331,14 @@
         "label": "Anzahl der abgerufenen Dokumente",
         "placeholder": "Anzahl der abgerufenen Dokumente eingeben",
         "required": "Bitte geben Sie die Anzahl der abgerufenen Dokumente ein"
+      },
+      "splittingSeparator": {
+        "label": "Separator",
+        "placeholder": "Separator eingeben (z.B. \\n\\n)",
+        "required": "Bitte geben Sie einen Separator ein"
+      },
+      "splittingStrategy": {
+        "label": "Text-Splitter"
       }
     },
     "prompt": {

src/assets/locale/en/settings.json

@@ -72,7 +72,7 @@
         }
       },
       "braveApi": {
-        "label":  "Brave API Key",
+        "label": "Brave API Key",
         "placeholder": "Enter your Brave API key"
       },
       "googleDomain": {
@@ -337,6 +337,14 @@
         "label": "Number of Retrieved Documents",
         "placeholder": "Enter Number of Retrieved Documents",
         "required": "Please enter the number of retrieved documents"
+      },
+      "splittingSeparator": {
+        "label": "Separator",
+        "placeholder": "Enter Separator (e.g., \\n\\n)",
+        "required": "Please enter a separator"
+      },
+      "splittingStrategy": {
+        "label": "Text Splitter"
       }
     },
     "prompt": {

src/assets/locale/es/settings.json

@@ -331,6 +331,14 @@
         "label": "Número de Documentos Recuperados",
         "placeholder": "Ingrese el Número de Documentos Recuperados",
         "required": "Por favor, ingrese el número de documentos recuperados"
+      },
+      "splittingSeparator": {
+        "label": "Separador",
+        "placeholder": "Ingrese el separador (ej., \\n\\n)",
+        "required": "Por favor, ingrese un separador"
+      },
+      "splittingStrategy": {
+        "label": "Divisor de Texto"
       }
     },
     "prompt": {

src/assets/locale/fa/settings.json

@@ -327,6 +327,14 @@
         "label": "تعداد اسناد بازیابی شده",
         "placeholder": "تعداد اسناد بازیابی شده را وارد کنید",
         "required": "لطفاً تعداد اسناد بازیابی شده را وارد کنید"
+      },
+      "splittingSeparator": {
+        "label": "جداکننده",
+        "placeholder": "جداکننده را وارد کنید (مثلاً \\n\\n)",
+        "required": "لطفاً یک جداکننده وارد کنید"
+      },
+      "splittingStrategy": {
+        "label": "تقسیم‌کننده متن"
       }
     },
     "prompt": {

src/assets/locale/fr/settings.json

@@ -331,6 +331,14 @@
         "label": "Nombre de documents récupérés",
         "placeholder": "Entrez le nombre de documents récupérés",
         "required": "Veuillez saisir le nombre de documents récupérés"
+      },
+      "splittingSeparator": {
+        "label": "Séparateur",
+        "placeholder": "Entrez le séparateur (par exemple, \\n\\n)",
+        "required": "Veuillez saisir un séparateur"
+      },
+      "splittingStrategy": {
+        "label": "Diviseur de texte"
       }
     },
     "prompt": {

src/assets/locale/it/settings.json

@@ -331,6 +331,14 @@
         "label": "Numero di Documenti Recuperati",
         "placeholder": "Inserisci il Numero di Documenti Recuperati",
         "required": "Inserisci il numero di documenti recuperati"
+      },
+      "splittingSeparator": {
+        "label": "Separatore",
+        "placeholder": "Inserisci il Separatore (es. \\n\\n)",
+        "required": "Inserisci un separatore"
+      },
+      "splittingStrategy": {
+        "label": "Divisore di Testo"
       }
     },
     "prompt": {

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

@@ -334,6 +334,14 @@
         "label": "取得ドキュメント数",
         "placeholder": "取得ドキュメント数を入力",
         "required": "取得ドキュメント数を入力してください"
+      },
+      "splittingSeparator": {
+        "label": "セパレーター",
+        "placeholder": "セパレーターを入力（例：\\n\\n）",
+        "required": "セパレーターを入力してください"
+      },
+      "splittingStrategy": {
+        "label": "テキスト分割方式"
       }
     },
     "prompt": {

src/assets/locale/ko/settings.json

@@ -334,6 +334,14 @@
         "label": "검색 문서 수",
         "placeholder": "검색 문서 수 입력",
         "required": "검색 문서 수를 입력해주세요"
+      },
+      "splittingSeparator": {
+        "label": "구분자",
+        "placeholder": "구분자 입력 (예: \\n\\n)",
+        "required": "구분자를 입력해주세요"
+      },
+      "splittingStrategy": {
+        "label": "텍스트 분할기"
       }
     },
     "prompt": {

src/assets/locale/ml/settings.json

@@ -334,6 +334,14 @@
         "label": "വീണ്ടെടുത്ത രേഖകളുടെ എണ്ണം",
         "placeholder": "വീണ്ടെടുത്ത രേഖകളുടെ എണ്ണം നൽകുക",
         "required": "ദയവായി വീണ്ടെടുത്ത രേഖകളുടെ എണ്ണം നൽകുക"
+      },
+      "splittingSeparator": {
+        "label": "വിഭജന ചിഹ്നം",
+        "placeholder": "വിഭജന ചിഹ്നം നൽകുക (ഉദാ: \\n\\n)",
+        "required": "ദയവായി ഒരു വിഭജന ചിഹ്നം നൽകുക"
+      },
+      "splittingStrategy": {
+        "label": "ടെക്സ്റ്റ് സ്പ്ലിറ്റർ"
       }
     },
     "prompt": {

src/assets/locale/no/settings.json

@@ -331,6 +331,14 @@
         "label": "Antall hentede dokumenter",
         "placeholder": "Skriv inn antall hentede dokumenter",
         "required": "Vennligst skriv inn antall hentede dokumenter"
+      },
+      "splittingSeparator": {
+        "label": "Separator",
+        "placeholder": "Skriv inn separator (f.eks. \\n\\n)",
+        "required": "Vennligst skriv inn en separator"
+      },
+      "splittingStrategy": {
+        "label": "Tekstdeler"
       }
     },
     "prompt": {

src/assets/locale/pt-BR/settings.json

@@ -331,6 +331,14 @@
         "label": "Número de Documentos Recuperados",
         "placeholder": "Digite o Número de Documentos Recuperados",
         "required": "Por favor, insira o número de documentos recuperados"
+      },
+      "splittingSeparator": {
+        "label": "Separador",
+        "placeholder": "Digite o Separador (ex: \\n\\n)",
+        "required": "Por favor, insira um separador"
+      },
+      "splittingStrategy": {
+        "label": "Divisor de Texto"
       }
     },
     "prompt": {

src/assets/locale/ru/settings.json

@@ -333,6 +333,14 @@
         "label": "Количество извлеченных документов",
         "placeholder": "Введите количество извлеченных документов",
         "required": "Пожалуйста, введите количество извлеченных документов"
+      },
+      "splittingSeparator": {
+        "label": "Разделитель",
+        "placeholder": "Введите разделитель (например, \\n\\n)",
+        "required": "Пожалуйста, введите разделитель"
+      },
+      "splittingStrategy": {
+        "label": "Разделитель текста"
       }
     },
     "prompt": {

src/assets/locale/sv/settings.json

@@ -331,6 +331,14 @@
         "label": "Antal hämtade dokument",
         "placeholder": "Ange antal hämtade dokument",
         "required": "Vänligen ange antal hämtade dokument"
+      },
+      "splittingSeparator": {
+        "label": "Separator",
+        "placeholder": "Ange separator (t.ex. \\n\\n)",
+        "required": "Vänligen ange en separator"
+      },
+      "splittingStrategy": {
+        "label": "Textdelare"
       }
     },
     "prompt": {

src/assets/locale/uk/settings.json

@@ -331,6 +331,14 @@
         "label": "Кількість отриманих документів",
         "placeholder": "Ввести кількість отриманих документів",
         "required": "Будь ласка, введіть кількість документів"
+      },
+      "splittingSeparator": {
+        "label": "Роздільник",
+        "placeholder": "Введіть роздільник (напр., \\n\\n)",
+        "required": "Будь ласка, введіть роздільник"
+      },
+      "splittingStrategy": {
+        "label": "Розділювач тексту"
       }
     },
     "prompt": {

src/assets/locale/zh/settings.json

@@ -336,6 +336,14 @@
         "label": "检索文档数量",
         "placeholder": "输入检索文档数量",
         "required": "请输入检索文档数量"
+      },
+      "splittingSeparator": {
+        "label": "分隔符",
+        "placeholder": "输入分隔符（例如：\\n\\n）",
+        "required": "请输入分隔符"
+      },
+      "splittingStrategy": {
+        "label": "文本分割器"
       }
     },
     "prompt": {

src/components/Option/Settings/rag.tsx

@@ -1,10 +1,12 @@
 import { useMutation, useQuery, useQueryClient } from "@tanstack/react-query"
-import { Form, InputNumber, Select, Skeleton } from "antd"
+import { Form, Input, InputNumber, Select, Skeleton } from "antd"
 import { SaveButton } from "~/components/Common/SaveButton"
 import {
   defaultEmbeddingChunkOverlap,
   defaultEmbeddingChunkSize,
   defaultEmbeddingModelForRag,
+  defaultSplittingStrategy,
+  defaultSsplttingSeparator,
   getEmbeddingModels,
   saveForRag
 } from "~/services/ollama"
@@ -16,7 +18,8 @@ import { ProviderIcons } from "@/components/Common/ProviderIcon"
 
 export const RagSettings = () => {
   const { t } = useTranslation("settings")
-
+  const [form] = Form.useForm()
+  const splittingStrategy = Form.useWatch("splittingStrategy", form)
   const queryClient = useQueryClient()
 
   const { data: ollamaInfo, status } = useQuery({
@@ -28,14 +31,18 @@ export const RagSettings = () => {
         chunkSize,
         defaultEM,
         totalFilePerKB,
-        noOfRetrievedDocs
+        noOfRetrievedDocs,
+        splittingStrategy,
+        splittingSeparator
       ] = await Promise.all([
         getEmbeddingModels({ returnEmpty: true }),
         defaultEmbeddingChunkOverlap(),
         defaultEmbeddingChunkSize(),
         defaultEmbeddingModelForRag(),
         getTotalFilePerKB(),
-        getNoOfRetrievedDocs()
+        getNoOfRetrievedDocs(),
+        defaultSplittingStrategy(),
+        defaultSsplttingSeparator()
       ])
       return {
         models: allModels,
@@ -43,7 +50,9 @@ export const RagSettings = () => {
         chunkSize,
         defaultEM,
         totalFilePerKB,
-        noOfRetrievedDocs
+        noOfRetrievedDocs,
+        splittingStrategy,
+        splittingSeparator
       }
     }
   })
@@ -55,13 +64,17 @@ export const RagSettings = () => {
       overlap: number
       totalFilePerKB: number
       noOfRetrievedDocs: number
+      strategy: string
+      separator: string
     }) => {
       await saveForRag(
         data.model,
         data.chunkSize,
         data.overlap,
         data.totalFilePerKB,
-        data.noOfRetrievedDocs
+        data.noOfRetrievedDocs,
+        data.strategy,
+        data.separator
       )
       return true
     },
@@ -85,6 +98,7 @@ export const RagSettings = () => {
               <div className="border border-b border-gray-200 dark:border-gray-600 mt-3 mb-6"></div>
             </div>
             <Form
+              form={form}
               layout="vertical"
               onFinish={(data) => {
                 saveRAG({
@@ -92,7 +106,9 @@ export const RagSettings = () => {
                   chunkSize: data.chunkSize,
                   overlap: data.chunkOverlap,
                   totalFilePerKB: data.totalFilePerKB,
-                  noOfRetrievedDocs: data.noOfRetrievedDocs
+                  noOfRetrievedDocs: data.noOfRetrievedDocs,
+                  separator: data.splittingSeparator,
+                  strategy: data.splittingStrategy
                 })
               }}
               initialValues={{
@@ -100,7 +116,9 @@ export const RagSettings = () => {
                 chunkOverlap: ollamaInfo?.chunkOverlap,
                 defaultEM: ollamaInfo?.defaultEM,
                 totalFilePerKB: ollamaInfo?.totalFilePerKB,
-                noOfRetrievedDocs: ollamaInfo?.noOfRetrievedDocs
+                noOfRetrievedDocs: ollamaInfo?.noOfRetrievedDocs,
+                splittingStrategy: ollamaInfo?.splittingStrategy,
+                splittingSeparator: ollamaInfo?.splittingSeparator
               }}>
               <Form.Item
                 name="defaultEM"
@@ -140,6 +158,50 @@ export const RagSettings = () => {
                 />
               </Form.Item>
 
+              <Form.Item
+                name="splittingStrategy"
+                label={t("rag.ragSettings.splittingStrategy.label")}
+                rules={[
+                  {
+                    required: true,
+                    message: t("rag.ragSettings.model.required")
+                  }
+                ]}>
+                <Select
+                  size="large"
+                  showSearch
+                  style={{ width: "100%" }}
+                  className="mt-4"
+                  options={[
+                    "RecursiveCharacterTextSplitter",
+                    "CharacterTextSplitter"
+                  ].map((e) => ({
+                    label: e,
+                    value: e
+                  }))}
+                />
+              </Form.Item>
+
+              {splittingStrategy !== "RecursiveCharacterTextSplitter" && (
+                <Form.Item
+                  name="splittingSeparator"
+                  label={t("rag.ragSettings.splittingSeparator.label")}
+                  rules={[
+                    {
+                      required: true,
+                      message: t("rag.ragSettings.splittingSeparator.required")
+                    }
+                  ]}>
+                  <Input
+                    size="large"
+                    style={{ width: "100%" }}
+                    placeholder={t(
+                      "rag.ragSettings.splittingSeparator.placeholder"
+                    )}
+                  />
+                </Form.Item>
+              )}
+
               <Form.Item
                 name="chunkSize"
                 label={t("rag.ragSettings.chunkSize.label")}

src/libs/process-knowledge.ts

@@ -1,11 +1,6 @@
 import { getKnowledgeById, updateKnowledgeStatus } from "@/db/knowledge"
 import { PageAssistPDFUrlLoader } from "@/loader/pdf-url"
-import {
+import { getOllamaURL } from "@/services/ollama"
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize,
-  getOllamaURL
-} from "@/services/ollama"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { PageAssistVectorStore } from "./PageAssistVectorStore"
 import { PageAssisCSVUrlLoader } from "@/loader/csv"
 import { PageAssisTXTUrlLoader } from "@/loader/txt"
@@ -13,7 +8,7 @@ import { PageAssistDocxLoader } from "@/loader/docx"
 import { cleanUrl } from "./clean-url"
 import { sendEmbeddingCompleteNotification } from "./send-notification"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
-
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 
 export const processKnowledge = async (msg: any, id: string): Promise<void> => {
   console.log(`Processing knowledge with id: ${id}`)
@@ -32,12 +27,8 @@ export const processKnowledge = async (msg: any, id: string): Promise<void> => {
       baseUrl: cleanUrl(ollamaUrl),
       model: knowledge.embedding_model
     })
-    const chunkSize = await defaultEmbeddingChunkSize()
+
-    const chunkOverlap = await defaultEmbeddingChunkOverlap()
+    const textSplitter = await getPageAssistTextSplitter()
-    const textSplitter = new RecursiveCharacterTextSplitter({
-      chunkSize,
-      chunkOverlap
-    })
 
     for (const doc of knowledge.source) {
       if (doc.type === "pdf" || doc.type === "application/pdf") {
@@ -65,13 +56,15 @@ 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") {
+      } else if (
+        doc.type === "docx" ||
+        doc.type ===
+          "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+      ) {
         try {
           const loader = new PageAssistDocxLoader({
             fileName: doc.filename,
-            buffer: await toArrayBufferFromBase64(
+            buffer: await toArrayBufferFromBase64(doc.content)
-              doc.content
-            )
           })
 
           let docs = await loader.load()

src/services/ollama.ts

@@ -8,6 +8,9 @@ import { ollamaFormatAllCustomModels } from "@/db/models"
 
 
 const storage = new Storage()
+const storage2 = new Storage({
+  area: "local"
+})
 
 const DEFAULT_OLLAMA_URL = "http://127.0.0.1:11434"
 const DEFAULT_ASK_FOR_MODEL_SELECTION_EVERY_TIME = true
@@ -310,6 +313,22 @@ export const defaultEmbeddingChunkSize = async () => {
   return parseInt(embeddingChunkSize)
 }
 
+export const defaultSplittingStrategy = async () => {
+  const splittingStrategy = await storage.get("defaultSplittingStrategy")
+  if (!splittingStrategy || splittingStrategy.length === 0) {
+    return "RecursiveCharacterTextSplitter"
+  }
+  return splittingStrategy
+}
+
+export const defaultSsplttingSeparator = async () => {
+  const splittingSeparator = await storage.get("defaultSplittingSeparator")
+  if (!splittingSeparator || splittingSeparator.length === 0) {
+    return "\\n\\n"
+  }
+  return splittingSeparator
+}
+
 export const defaultEmbeddingChunkOverlap = async () => {
   const embeddingChunkOverlap = await storage.get(
     "defaultEmbeddingChunkOverlap"
@@ -320,6 +339,14 @@ export const defaultEmbeddingChunkOverlap = async () => {
   return parseInt(embeddingChunkOverlap)
 }
 
+export const setDefaultSplittingStrategy = async (strategy: string) => {
+  await storage.set("defaultSplittingStrategy", strategy)
+}
+
+export const setDefaultSplittingSeparator = async (separator: string) => {
+  await storage.set("defaultSplittingSeparator", separator)
+}
+
 export const setDefaultEmbeddingModelForRag = async (model: string) => {
   await storage.set("defaultEmbeddingModel", model)
 }
@@ -337,7 +364,9 @@ export const saveForRag = async (
   chunkSize: number,
   overlap: number,
   totalFilePerKB: number,
-  noOfRetrievedDocs?: number
+  noOfRetrievedDocs?: number,
+  strategy?: string,
+  separator?: string
 ) => {
   await setDefaultEmbeddingModelForRag(model)
   await setDefaultEmbeddingChunkSize(chunkSize)
@@ -346,6 +375,12 @@ export const saveForRag = async (
   if (noOfRetrievedDocs) {
     await setNoOfRetrievedDocs(noOfRetrievedDocs)
   }
+  if (strategy) {
+    await setDefaultSplittingStrategy(strategy)
+  }
+  if (separator) {
+    await setDefaultSplittingSeparator(separator)
+  }
 }
 
 export const getWebSearchPrompt = async () => {

src/utils/memory-embeddings.ts

@@ -1,12 +1,8 @@
 import { PageAssistHtmlLoader } from "~/loader/html"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 
-import {
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize
-} from "@/services/ollama"
 import { PageAssistPDFLoader } from "@/loader/pdf"
 import { PAMemoryVectorStore } from "@/libs/PAMemoryVectorStore"
+import { getPageAssistTextSplitter } from "./text-splitter"
 
 export const getLoader = ({
   html,
@@ -54,12 +50,7 @@ export const memoryEmbedding = async ({
   setIsEmbedding(true)
   const loader = getLoader({ html, pdf, type, url })
   const docs = await loader.load()
-  const chunkSize = await defaultEmbeddingChunkSize()
+  const textSplitter = await getPageAssistTextSplitter()
-  const chunkOverlap = await defaultEmbeddingChunkOverlap()
-  const textSplitter = new RecursiveCharacterTextSplitter({
-    chunkSize,
-    chunkOverlap
-  })
 
   const chunks = await textSplitter.splitDocuments(docs)
 

src/utils/text-splitter.ts

@@ -0,0 +1,37 @@
+import {
+  RecursiveCharacterTextSplitter,
+  CharacterTextSplitter
+} from "langchain/text_splitter"
+
+import {
+  defaultEmbeddingChunkOverlap,
+  defaultEmbeddingChunkSize,
+  defaultSsplttingSeparator,
+  defaultSplittingStrategy
+} from "@/services/ollama"
+
+export const getPageAssistTextSplitter = async () => {
+  const chunkSize = await defaultEmbeddingChunkSize()
+  const chunkOverlap = await defaultEmbeddingChunkOverlap()
+  const splittingStrategy = await defaultSplittingStrategy()
+
+  switch (splittingStrategy) {
+    case "CharacterTextSplitter":
+      console.log("Using CharacterTextSplitter")
+      const splittingSeparator = await defaultSsplttingSeparator()
+      const processedSeparator = splittingSeparator
+        .replace(/\\n/g, "\n")
+        .replace(/\\t/g, "\t")
+        .replace(/\\r/g, "\r")
+      return new CharacterTextSplitter({
+        chunkSize,
+        chunkOverlap,
+        separator: processedSeparator
+      })
+    default:
+      return new RecursiveCharacterTextSplitter({
+        chunkSize,
+        chunkOverlap
+      })
+  }
+}

src/web/search-engines/brave-api.ts

@@ -2,15 +2,13 @@ import { cleanUrl } from "~/libs/clean-url"
 import { getIsSimpleInternetSearch, totalSearchResults, getBraveApiKey } from "@/services/search"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
 import type { Document } from "@langchain/core/documents"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 import { PageAssistHtmlLoader } from "~/loader/html"
 import {
-    defaultEmbeddingChunkOverlap,
-    defaultEmbeddingChunkSize,
     defaultEmbeddingModelForRag,
     getOllamaURL
 } from "~/services/ollama"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 
 interface BraveAPIResult {
     title: string
@@ -70,12 +68,7 @@ export const braveAPISearch = async (query: string) => {
         baseUrl: cleanUrl(ollamaUrl)
     })
 
-    const chunkSize = await defaultEmbeddingChunkSize()
+    const textSplitter = await getPageAssistTextSplitter()
-    const chunkOverlap = await defaultEmbeddingChunkOverlap()
-    const textSplitter = new RecursiveCharacterTextSplitter({
-        chunkSize,
-        chunkOverlap
-    })
 
     const chunks = await textSplitter.splitDocuments(docs)
     const store = new MemoryVectorStore(ollamaEmbedding)

src/web/search-engines/brave.ts

@@ -3,8 +3,6 @@ import { urlRewriteRuntime } from "@/libs/runtime"
 import { PageAssistHtmlLoader } from "@/loader/html"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
 import {
-    defaultEmbeddingChunkOverlap,
-    defaultEmbeddingChunkSize,
     defaultEmbeddingModelForRag,
     getOllamaURL
 } from "@/services/ollama"
@@ -12,10 +10,10 @@ import {
     getIsSimpleInternetSearch,
     totalSearchResults
 } from "@/services/search"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 
 import type { Document } from "@langchain/core/documents"
 import * as cheerio from "cheerio"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 
 export const localBraveSearch = async (query: string) => {
@@ -87,12 +85,8 @@ export const webBraveSearch = async (query: string) => {
         baseUrl: cleanUrl(ollamaUrl)
     })
 
-    const chunkSize = await defaultEmbeddingChunkSize()
+ 
-    const chunkOverlap = await defaultEmbeddingChunkOverlap()
+    const textSplitter = await getPageAssistTextSplitter();
-    const textSplitter = new RecursiveCharacterTextSplitter({
-        chunkSize,
-        chunkOverlap
-    })
 
     const chunks = await textSplitter.splitDocuments(docs)
 

src/web/search-engines/duckduckgo.ts

@@ -3,8 +3,6 @@ import { urlRewriteRuntime } from "@/libs/runtime"
 import { PageAssistHtmlLoader } from "@/loader/html"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
 import {
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize,
   defaultEmbeddingModelForRag,
   getOllamaURL
 } from "@/services/ollama"
@@ -12,9 +10,9 @@ import {
   getIsSimpleInternetSearch,
   totalSearchResults
 } from "@/services/search"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 import type { Document } from "@langchain/core/documents"
 import * as cheerio from "cheerio"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 
 export const localDuckDuckGoSearch = async (query: string) => {
@@ -90,12 +88,7 @@ export const webDuckDuckGoSearch = async (query: string) => {
     baseUrl: cleanUrl(ollamaUrl)
   })
 
-  const chunkSize = await defaultEmbeddingChunkSize()
+  const textSplitter = await getPageAssistTextSplitter()
-  const chunkOverlap = await defaultEmbeddingChunkOverlap()
-  const textSplitter = new RecursiveCharacterTextSplitter({
-    chunkSize,
-    chunkOverlap
-  })
 
   const chunks = await textSplitter.splitDocuments(docs)
 

src/web/search-engines/google.ts

@@ -4,15 +4,13 @@ import {
   getIsSimpleInternetSearch,
   totalSearchResults
 } from "@/services/search"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 import type { Document } from "@langchain/core/documents"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 import { cleanUrl } from "~/libs/clean-url"
 import { urlRewriteRuntime } from "~/libs/runtime"
 import { PageAssistHtmlLoader } from "~/loader/html"
 import {
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize,
   defaultEmbeddingModelForRag,
   getOllamaURL
 } from "~/services/ollama"
@@ -91,13 +89,9 @@ export const webGoogleSearch = async (query: string) => {
     baseUrl: cleanUrl(ollamaUrl)
   })
 
-  const chunkSize = await defaultEmbeddingChunkSize()
+  
-  const chunkOverlap = await defaultEmbeddingChunkOverlap()
+  const textSplitter = await getPageAssistTextSplitter()
-  const textSplitter = new RecursiveCharacterTextSplitter({
+  
-    chunkSize,
-    chunkOverlap
-  })
-
   const chunks = await textSplitter.splitDocuments(docs)
 
   const store = new MemoryVectorStore(ollamaEmbedding)

src/web/search-engines/searxng.ts

@@ -3,15 +3,13 @@ import { cleanUrl } from "~/libs/clean-url"
 import { getSearxngURL, isSearxngJSONMode, getIsSimpleInternetSearch, totalSearchResults } from "@/services/search"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
 import type { Document } from "@langchain/core/documents"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 import { PageAssistHtmlLoader } from "~/loader/html"
 import {
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize,
   defaultEmbeddingModelForRag,
   getOllamaURL
 } from "~/services/ollama"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 
 interface SearxNGJSONResult {
   title: string
@@ -73,13 +71,9 @@ export const searxngSearch = async (query: string) => {
     baseUrl: cleanUrl(ollamaUrl)
   })
 
-  const chunkSize = await defaultEmbeddingChunkSize()
-  const chunkOverlap = await defaultEmbeddingChunkOverlap()
-  const textSplitter = new RecursiveCharacterTextSplitter({
-    chunkSize,
-    chunkOverlap
-  })
 
+  const textSplitter = await getPageAssistTextSplitter();
+  
   const chunks = await textSplitter.splitDocuments(docs)
   const store = new MemoryVectorStore(ollamaEmbedding)
   await store.addDocuments(chunks)

src/web/search-engines/sogou.ts

@@ -3,8 +3,6 @@ import { urlRewriteRuntime } from "@/libs/runtime"
 import { PageAssistHtmlLoader } from "@/loader/html"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
 import {
-  defaultEmbeddingChunkOverlap,
-  defaultEmbeddingChunkSize,
   defaultEmbeddingModelForRag,
   getOllamaURL
 } from "@/services/ollama"
@@ -12,9 +10,9 @@ import {
   getIsSimpleInternetSearch,
   totalSearchResults
 } from "@/services/search"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
 import type { Document } from "@langchain/core/documents"
 import * as cheerio from "cheerio"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 const getCorrectTargeUrl = async (url: string) => {
   if (!url) return ""
@@ -104,12 +102,7 @@ export const webSogouSearch = async (query: string) => {
     baseUrl: cleanUrl(ollamaUrl)
   })
 
-  const chunkSize = await defaultEmbeddingChunkSize()
+  const textSplitter = await getPageAssistTextSplitter()
-  const chunkOverlap = await defaultEmbeddingChunkOverlap()
-  const textSplitter = new RecursiveCharacterTextSplitter({
-    chunkSize,
-    chunkOverlap
-  })
 
   const chunks = await textSplitter.splitDocuments(docs)
 

src/web/website/index.ts

@@ -1,8 +1,9 @@
 import { cleanUrl } from "@/libs/clean-url"
 import { PageAssistHtmlLoader } from "@/loader/html"
 import { pageAssistEmbeddingModel } from "@/models/embedding"
-import { defaultEmbeddingChunkOverlap, defaultEmbeddingChunkSize, defaultEmbeddingModelForRag, getOllamaURL } from "@/services/ollama"
+import { defaultEmbeddingModelForRag, getOllamaURL } from "@/services/ollama"
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
+import { getPageAssistTextSplitter } from "@/utils/text-splitter"
+
 import { MemoryVectorStore } from "langchain/vectorstores/memory"
 
 export const processSingleWebsite = async (url: string, query: string) => {
@@ -20,12 +21,8 @@ export const processSingleWebsite = async (url: string, query: string) => {
         baseUrl: cleanUrl(ollamaUrl)
     })
 
-    const chunkSize = await defaultEmbeddingChunkSize()
+
-    const chunkOverlap = await defaultEmbeddingChunkOverlap()
+    const textSplitter = await getPageAssistTextSplitter()
-    const textSplitter = new RecursiveCharacterTextSplitter({
-        chunkSize,
-        chunkOverlap
-    })
 
     const chunks = await textSplitter.splitDocuments(docs)