fix: update type definitions for formRef and onSubmit in StepTwoContent component

- Changed formRef type to allow for null values.
- Updated onSubmit parameter type from Record<string, any> to Record<string, unknown> for better type safety.

web/app/components/datasets/documents/create-from-pipeline/hooks/index.ts

@@ -0,0 +1,5 @@
+export { useAddDocumentsSteps } from './use-add-documents-steps'
+export { useDatasourceActions } from './use-datasource-actions'
+export { useDatasourceOptions } from './use-datasource-options'
+export { useLocalFile, useOnlineDocument, useOnlineDrive, useWebsiteCrawl } from './use-datasource-store'
+export { useDatasourceUIState } from './use-datasource-ui-state'

web/app/components/datasets/documents/create-from-pipeline/hooks/use-add-documents-steps.ts

@@ -0,0 +1,41 @@
+import { useCallback, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import { AddDocumentsStep } from '../types'
+
+/**
+ * Hook for managing add documents wizard steps
+ */
+export const useAddDocumentsSteps = () => {
+  const { t } = useTranslation()
+  const [currentStep, setCurrentStep] = useState(1)
+
+  const handleNextStep = useCallback(() => {
+    setCurrentStep(preStep => preStep + 1)
+  }, [])
+
+  const handleBackStep = useCallback(() => {
+    setCurrentStep(preStep => preStep - 1)
+  }, [])
+
+  const steps = [
+    {
+      label: t('addDocuments.steps.chooseDatasource', { ns: 'datasetPipeline' }),
+      value: AddDocumentsStep.dataSource,
+    },
+    {
+      label: t('addDocuments.steps.processDocuments', { ns: 'datasetPipeline' }),
+      value: AddDocumentsStep.processDocuments,
+    },
+    {
+      label: t('addDocuments.steps.processingDocuments', { ns: 'datasetPipeline' }),
+      value: AddDocumentsStep.processingDocuments,
+    },
+  ]
+
+  return {
+    steps,
+    currentStep,
+    handleNextStep,
+    handleBackStep,
+  }
+}

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-actions.ts

@@ -0,0 +1,321 @@
+import type { StoreApi } from 'zustand'
+import type { DataSourceShape } from '@/app/components/datasets/documents/create-from-pipeline/data-source/store'
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { DataSourceNotionPageMap, NotionPage } from '@/models/common'
+import type { CrawlResultItem, DocumentItem, CustomFile as File, FileIndexingEstimateResponse } from '@/models/datasets'
+import type {
+  OnlineDriveFile,
+  PublishedPipelineRunPreviewResponse,
+  PublishedPipelineRunResponse,
+} from '@/models/pipeline'
+import { useCallback, useRef } from 'react'
+import { trackEvent } from '@/app/components/base/amplitude'
+import { DatasourceType } from '@/models/pipeline'
+import { useRunPublishedPipeline } from '@/service/use-pipeline'
+import {
+  buildLocalFileDatasourceInfo,
+  buildOnlineDocumentDatasourceInfo,
+  buildOnlineDriveDatasourceInfo,
+  buildWebsiteCrawlDatasourceInfo,
+} from '../utils/datasource-info-builder'
+
+type DatasourceActionsParams = {
+  datasource: Datasource | undefined
+  datasourceType: string | undefined
+  pipelineId: string | undefined
+  dataSourceStore: StoreApi<DataSourceShape>
+  setEstimateData: (data: FileIndexingEstimateResponse | undefined) => void
+  setBatchId: (id: string) => void
+  setDocuments: (docs: PublishedPipelineRunResponse['documents']) => void
+  handleNextStep: () => void
+  PagesMapAndSelectedPagesId: DataSourceNotionPageMap
+  currentWorkspacePages: { page_id: string }[] | undefined
+  clearOnlineDocumentData: () => void
+  clearWebsiteCrawlData: () => void
+  clearOnlineDriveData: () => void
+  setDatasource: (ds: Datasource) => void
+}
+
+/**
+ * Hook for datasource-related actions (preview, process, etc.)
+ */
+export const useDatasourceActions = ({
+  datasource,
+  datasourceType,
+  pipelineId,
+  dataSourceStore,
+  setEstimateData,
+  setBatchId,
+  setDocuments,
+  handleNextStep,
+  PagesMapAndSelectedPagesId,
+  currentWorkspacePages,
+  clearOnlineDocumentData,
+  clearWebsiteCrawlData,
+  clearOnlineDriveData,
+  setDatasource,
+}: DatasourceActionsParams) => {
+  const isPreview = useRef(false)
+  const formRef = useRef<{ submit: () => void } | null>(null)
+
+  const { mutateAsync: runPublishedPipeline, isIdle, isPending } = useRunPublishedPipeline()
+
+  // Build datasource info for preview (single item)
+  const buildPreviewDatasourceInfo = useCallback(() => {
+    const {
+      previewLocalFileRef,
+      previewOnlineDocumentRef,
+      previewWebsitePageRef,
+      previewOnlineDriveFileRef,
+      currentCredentialId,
+      bucket,
+    } = dataSourceStore.getState()
+
+    const datasourceInfoList: Record<string, unknown>[] = []
+
+    if (datasourceType === DatasourceType.localFile && previewLocalFileRef.current) {
+      datasourceInfoList.push(buildLocalFileDatasourceInfo(
+        previewLocalFileRef.current as File,
+        currentCredentialId,
+      ))
+    }
+
+    if (datasourceType === DatasourceType.onlineDocument && previewOnlineDocumentRef.current) {
+      datasourceInfoList.push(buildOnlineDocumentDatasourceInfo(
+        previewOnlineDocumentRef.current,
+        currentCredentialId,
+      ))
+    }
+
+    if (datasourceType === DatasourceType.websiteCrawl && previewWebsitePageRef.current) {
+      datasourceInfoList.push(buildWebsiteCrawlDatasourceInfo(
+        previewWebsitePageRef.current,
+        currentCredentialId,
+      ))
+    }
+
+    if (datasourceType === DatasourceType.onlineDrive && previewOnlineDriveFileRef.current) {
+      datasourceInfoList.push(buildOnlineDriveDatasourceInfo(
+        previewOnlineDriveFileRef.current,
+        bucket,
+        currentCredentialId,
+      ))
+    }
+
+    return datasourceInfoList
+  }, [dataSourceStore, datasourceType])
+
+  // Build datasource info for processing (all items)
+  const buildProcessDatasourceInfo = useCallback(() => {
+    const {
+      currentCredentialId,
+      localFileList,
+      onlineDocuments,
+      websitePages,
+      bucket,
+      selectedFileIds,
+      onlineDriveFileList,
+    } = dataSourceStore.getState()
+
+    const datasourceInfoList: Record<string, unknown>[] = []
+
+    if (datasourceType === DatasourceType.localFile) {
+      localFileList.forEach((file) => {
+        datasourceInfoList.push(buildLocalFileDatasourceInfo(file.file, currentCredentialId))
+      })
+    }
+
+    if (datasourceType === DatasourceType.onlineDocument) {
+      onlineDocuments.forEach((page) => {
+        datasourceInfoList.push(buildOnlineDocumentDatasourceInfo(page, currentCredentialId))
+      })
+    }
+
+    if (datasourceType === DatasourceType.websiteCrawl) {
+      websitePages.forEach((page) => {
+        datasourceInfoList.push(buildWebsiteCrawlDatasourceInfo(page, currentCredentialId))
+      })
+    }
+
+    if (datasourceType === DatasourceType.onlineDrive) {
+      selectedFileIds.forEach((id) => {
+        const file = onlineDriveFileList.find(f => f.id === id)
+        if (file)
+          datasourceInfoList.push(buildOnlineDriveDatasourceInfo(file, bucket, currentCredentialId))
+      })
+    }
+
+    return datasourceInfoList
+  }, [dataSourceStore, datasourceType])
+
+  // Handle chunk preview
+  const handlePreviewChunks = useCallback(async (data: Record<string, unknown>) => {
+    if (!datasource || !pipelineId)
+      return
+
+    const datasourceInfoList = buildPreviewDatasourceInfo()
+    await runPublishedPipeline({
+      pipeline_id: pipelineId,
+      inputs: data,
+      start_node_id: datasource.nodeId,
+      datasource_type: datasourceType as DatasourceType,
+      datasource_info_list: datasourceInfoList,
+      is_preview: true,
+    }, {
+      onSuccess: (res) => {
+        setEstimateData((res as PublishedPipelineRunPreviewResponse).data.outputs)
+      },
+    })
+  }, [datasource, pipelineId, datasourceType, buildPreviewDatasourceInfo, runPublishedPipeline, setEstimateData])
+
+  // Handle document processing
+  const handleProcess = useCallback(async (data: Record<string, unknown>) => {
+    if (!datasource || !pipelineId)
+      return
+
+    const datasourceInfoList = buildProcessDatasourceInfo()
+    await runPublishedPipeline({
+      pipeline_id: pipelineId,
+      inputs: data,
+      start_node_id: datasource.nodeId,
+      datasource_type: datasourceType as DatasourceType,
+      datasource_info_list: datasourceInfoList,
+      is_preview: false,
+    }, {
+      onSuccess: (res) => {
+        setBatchId((res as PublishedPipelineRunResponse).batch || '')
+        setDocuments((res as PublishedPipelineRunResponse).documents || [])
+        handleNextStep()
+        trackEvent('dataset_document_added', {
+          data_source_type: datasourceType,
+          indexing_technique: 'pipeline',
+        })
+      },
+    })
+  }, [datasource, pipelineId, datasourceType, buildProcessDatasourceInfo, runPublishedPipeline, setBatchId, setDocuments, handleNextStep])
+
+  // Form submission handlers
+  const onClickProcess = useCallback(() => {
+    isPreview.current = false
+    formRef.current?.submit()
+  }, [])
+
+  const onClickPreview = useCallback(() => {
+    isPreview.current = true
+    formRef.current?.submit()
+  }, [])
+
+  const handleSubmit = useCallback((data: Record<string, unknown>) => {
+    if (isPreview.current)
+      handlePreviewChunks(data)
+    else
+      handleProcess(data)
+  }, [handlePreviewChunks, handleProcess])
+
+  // Preview change handlers
+  const handlePreviewFileChange = useCallback((file: DocumentItem) => {
+    const { previewLocalFileRef } = dataSourceStore.getState()
+    previewLocalFileRef.current = file
+    onClickPreview()
+  }, [dataSourceStore, onClickPreview])
+
+  const handlePreviewOnlineDocumentChange = useCallback((page: NotionPage) => {
+    const { previewOnlineDocumentRef } = dataSourceStore.getState()
+    previewOnlineDocumentRef.current = page
+    onClickPreview()
+  }, [dataSourceStore, onClickPreview])
+
+  const handlePreviewWebsiteChange = useCallback((website: CrawlResultItem) => {
+    const { previewWebsitePageRef } = dataSourceStore.getState()
+    previewWebsitePageRef.current = website
+    onClickPreview()
+  }, [dataSourceStore, onClickPreview])
+
+  const handlePreviewOnlineDriveFileChange = useCallback((file: OnlineDriveFile) => {
+    const { previewOnlineDriveFileRef } = dataSourceStore.getState()
+    previewOnlineDriveFileRef.current = file
+    onClickPreview()
+  }, [dataSourceStore, onClickPreview])
+
+  // Select all handler
+  const handleSelectAll = useCallback(() => {
+    const {
+      onlineDocuments,
+      onlineDriveFileList,
+      selectedFileIds,
+      setOnlineDocuments,
+      setSelectedFileIds,
+      setSelectedPagesId,
+    } = dataSourceStore.getState()
+
+    if (datasourceType === DatasourceType.onlineDocument) {
+      const allIds = currentWorkspacePages?.map(page => page.page_id) || []
+      if (onlineDocuments.length < allIds.length) {
+        const selectedPages = Array.from(allIds).map(pageId => PagesMapAndSelectedPagesId[pageId])
+        setOnlineDocuments(selectedPages)
+        setSelectedPagesId(new Set(allIds))
+      }
+      else {
+        setOnlineDocuments([])
+        setSelectedPagesId(new Set())
+      }
+    }
+
+    if (datasourceType === DatasourceType.onlineDrive) {
+      const allKeys = onlineDriveFileList.filter(item => item.type !== 'bucket').map(file => file.id)
+      if (selectedFileIds.length < allKeys.length)
+        setSelectedFileIds(allKeys)
+      else
+        setSelectedFileIds([])
+    }
+  }, [PagesMapAndSelectedPagesId, currentWorkspacePages, dataSourceStore, datasourceType])
+
+  // Clear datasource data based on type
+  const clearDataSourceData = useCallback((dataSource: Datasource) => {
+    const providerType = dataSource.nodeData.provider_type
+    const clearFunctions: Record<string, () => void> = {
+      [DatasourceType.onlineDocument]: clearOnlineDocumentData,
+      [DatasourceType.websiteCrawl]: clearWebsiteCrawlData,
+      [DatasourceType.onlineDrive]: clearOnlineDriveData,
+      [DatasourceType.localFile]: () => {},
+    }
+    clearFunctions[providerType]?.()
+  }, [clearOnlineDocumentData, clearOnlineDriveData, clearWebsiteCrawlData])
+
+  // Switch datasource handler
+  const handleSwitchDataSource = useCallback((dataSource: Datasource) => {
+    const {
+      setCurrentCredentialId,
+      currentNodeIdRef,
+    } = dataSourceStore.getState()
+    clearDataSourceData(dataSource)
+    setCurrentCredentialId('')
+    currentNodeIdRef.current = dataSource.nodeId
+    setDatasource(dataSource)
+  }, [clearDataSourceData, dataSourceStore, setDatasource])
+
+  // Credential change handler
+  const handleCredentialChange = useCallback((credentialId: string) => {
+    const { setCurrentCredentialId } = dataSourceStore.getState()
+    if (datasource)
+      clearDataSourceData(datasource)
+    setCurrentCredentialId(credentialId)
+  }, [clearDataSourceData, dataSourceStore, datasource])
+
+  return {
+    isPreview,
+    formRef,
+    isIdle,
+    isPending,
+    onClickProcess,
+    onClickPreview,
+    handleSubmit,
+    handlePreviewFileChange,
+    handlePreviewOnlineDocumentChange,
+    handlePreviewWebsiteChange,
+    handlePreviewOnlineDriveFileChange,
+    handleSelectAll,
+    handleSwitchDataSource,
+    handleCredentialChange,
+  }
+}

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-options.ts

@@ -0,0 +1,27 @@
+import type { DataSourceOption } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
+import type { Node } from '@/app/components/workflow/types'
+import { useMemo } from 'react'
+import { BlockEnum } from '@/app/components/workflow/types'
+
+/**
+ * Hook for getting datasource options from pipeline nodes
+ */
+export const useDatasourceOptions = (pipelineNodes: Node<DataSourceNodeType>[]) => {
+  const datasourceNodes = pipelineNodes.filter(node => node.data.type === BlockEnum.DataSource)
+
+  const options = useMemo(() => {
+    const options: DataSourceOption[] = []
+    datasourceNodes.forEach((node) => {
+      const label = node.data.title
+      options.push({
+        label,
+        value: node.id,
+        data: node.data,
+      })
+    })
+    return options
+  }, [datasourceNodes])
+
+  return options
+}

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-store.ts

@@ -1,69 +1,12 @@
-import type { DataSourceOption } from '@/app/components/rag-pipeline/components/panel/test-run/types'
-import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
-import type { Node } from '@/app/components/workflow/types'
 import type { DataSourceNotionPageMap, DataSourceNotionWorkspace } from '@/models/common'
-import { useCallback, useMemo, useState } from 'react'
-import { useTranslation } from 'react-i18next'
+import { useCallback, useMemo } from 'react'
 import { useShallow } from 'zustand/react/shallow'
-import { BlockEnum } from '@/app/components/workflow/types'
 import { CrawlStep } from '@/models/datasets'
-import { useDataSourceStore, useDataSourceStoreWithSelector } from './data-source/store'
-import { AddDocumentsStep } from './types'
-
-export const useAddDocumentsSteps = () => {
-  const { t } = useTranslation()
-  const [currentStep, setCurrentStep] = useState(1)
-
-  const handleNextStep = useCallback(() => {
-    setCurrentStep(preStep => preStep + 1)
-  }, [])
-
-  const handleBackStep = useCallback(() => {
-    setCurrentStep(preStep => preStep - 1)
-  }, [])
-
-  const steps = [
-    {
-      label: t('addDocuments.steps.chooseDatasource', { ns: 'datasetPipeline' }),
-      value: AddDocumentsStep.dataSource,
-    },
-    {
-      label: t('addDocuments.steps.processDocuments', { ns: 'datasetPipeline' }),
-      value: AddDocumentsStep.processDocuments,
-    },
-    {
-      label: t('addDocuments.steps.processingDocuments', { ns: 'datasetPipeline' }),
-      value: AddDocumentsStep.processingDocuments,
-    },
-  ]
-
-  return {
-    steps,
-    currentStep,
-    handleNextStep,
-    handleBackStep,
-  }
-}
-
-export const useDatasourceOptions = (pipelineNodes: Node<DataSourceNodeType>[]) => {
-  const datasourceNodes = pipelineNodes.filter(node => node.data.type === BlockEnum.DataSource)
-
-  const options = useMemo(() => {
-    const options: DataSourceOption[] = []
-    datasourceNodes.forEach((node) => {
-      const label = node.data.title
-      options.push({
-        label,
-        value: node.id,
-        data: node.data,
-      })
-    })
-    return options
-  }, [datasourceNodes])
-
-  return options
-}
+import { useDataSourceStore, useDataSourceStoreWithSelector } from '../data-source/store'
 
+/**
+ * Hook for local file datasource store operations
+ */
 export const useLocalFile = () => {
   const {
     localFileList,
@@ -89,6 +32,9 @@ export const useLocalFile = () => {
   }
 }
 
+/**
+ * Hook for online document datasource store operations
+ */
 export const useOnlineDocument = () => {
   const {
     documentsData,
@@ -147,6 +93,9 @@ export const useOnlineDocument = () => {
   }
 }
 
+/**
+ * Hook for website crawl datasource store operations
+ */
 export const useWebsiteCrawl = () => {
   const {
     websitePages,
@@ -186,6 +135,9 @@ export const useWebsiteCrawl = () => {
   }
 }
 
+/**
+ * Hook for online drive datasource store operations
+ */
 export const useOnlineDrive = () => {
   const {
     onlineDriveFileList,

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-ui-state.ts

@@ -0,0 +1,132 @@
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { useMemo } from 'react'
+import { useTranslation } from 'react-i18next'
+import { DatasourceType } from '@/models/pipeline'
+
+type DatasourceUIStateParams = {
+  datasource: Datasource | undefined
+  allFileLoaded: boolean
+  localFileListLength: number
+  onlineDocumentsLength: number
+  websitePagesLength: number
+  selectedFileIdsLength: number
+  onlineDriveFileList: OnlineDriveFile[]
+  isVectorSpaceFull: boolean
+  enableBilling: boolean
+  currentWorkspacePagesLength: number
+  fileUploadConfig: { file_size_limit: number, batch_count_limit: number }
+}
+
+/**
+ * Hook for computing datasource UI state based on datasource type
+ */
+export const useDatasourceUIState = ({
+  datasource,
+  allFileLoaded,
+  localFileListLength,
+  onlineDocumentsLength,
+  websitePagesLength,
+  selectedFileIdsLength,
+  onlineDriveFileList,
+  isVectorSpaceFull,
+  enableBilling,
+  currentWorkspacePagesLength,
+  fileUploadConfig,
+}: DatasourceUIStateParams) => {
+  const { t } = useTranslation()
+  const datasourceType = datasource?.nodeData.provider_type
+
+  const isShowVectorSpaceFull = useMemo(() => {
+    if (!datasource || !datasourceType)
+      return false
+
+    // Lookup table for vector space full condition check
+    const vectorSpaceFullConditions: Record<string, boolean> = {
+      [DatasourceType.localFile]: allFileLoaded,
+      [DatasourceType.onlineDocument]: onlineDocumentsLength > 0,
+      [DatasourceType.websiteCrawl]: websitePagesLength > 0,
+      [DatasourceType.onlineDrive]: onlineDriveFileList.length > 0,
+    }
+
+    const condition = vectorSpaceFullConditions[datasourceType]
+    return condition && isVectorSpaceFull && enableBilling
+  }, [datasource, datasourceType, allFileLoaded, onlineDocumentsLength, websitePagesLength, onlineDriveFileList.length, isVectorSpaceFull, enableBilling])
+
+  // Lookup table for next button disabled conditions
+  const nextBtnDisabled = useMemo(() => {
+    if (!datasource || !datasourceType)
+      return true
+
+    const disabledConditions: Record<string, boolean> = {
+      [DatasourceType.localFile]: isShowVectorSpaceFull || localFileListLength === 0 || !allFileLoaded,
+      [DatasourceType.onlineDocument]: isShowVectorSpaceFull || onlineDocumentsLength === 0,
+      [DatasourceType.websiteCrawl]: isShowVectorSpaceFull || websitePagesLength === 0,
+      [DatasourceType.onlineDrive]: isShowVectorSpaceFull || selectedFileIdsLength === 0,
+    }
+
+    return disabledConditions[datasourceType] ?? true
+  }, [datasource, datasourceType, isShowVectorSpaceFull, localFileListLength, allFileLoaded, onlineDocumentsLength, websitePagesLength, selectedFileIdsLength])
+
+  // Check if select all should be shown
+  const showSelect = useMemo(() => {
+    if (datasourceType === DatasourceType.onlineDocument)
+      return currentWorkspacePagesLength > 0
+
+    if (datasourceType === DatasourceType.onlineDrive) {
+      const nonBucketItems = onlineDriveFileList.filter(item => item.type !== 'bucket')
+      const isBucketList = onlineDriveFileList.some(file => file.type === 'bucket')
+      return !isBucketList && nonBucketItems.length > 0
+    }
+
+    return false
+  }, [currentWorkspacePagesLength, datasourceType, onlineDriveFileList])
+
+  // Total selectable options count
+  const totalOptions = useMemo(() => {
+    if (datasourceType === DatasourceType.onlineDocument)
+      return currentWorkspacePagesLength
+
+    if (datasourceType === DatasourceType.onlineDrive)
+      return onlineDriveFileList.filter(item => item.type !== 'bucket').length
+
+    return undefined
+  }, [currentWorkspacePagesLength, datasourceType, onlineDriveFileList])
+
+  // Selected options count
+  const selectedOptions = useMemo(() => {
+    if (datasourceType === DatasourceType.onlineDocument)
+      return onlineDocumentsLength
+
+    if (datasourceType === DatasourceType.onlineDrive)
+      return selectedFileIdsLength
+
+    return undefined
+  }, [datasourceType, onlineDocumentsLength, selectedFileIdsLength])
+
+  // Tip message for selection
+  const tip = useMemo(() => {
+    if (datasourceType === DatasourceType.onlineDocument)
+      return t('addDocuments.selectOnlineDocumentTip', { ns: 'datasetPipeline', count: 50 })
+
+    if (datasourceType === DatasourceType.onlineDrive) {
+      return t('addDocuments.selectOnlineDriveTip', {
+        ns: 'datasetPipeline',
+        count: fileUploadConfig.batch_count_limit,
+        fileSize: fileUploadConfig.file_size_limit,
+      })
+    }
+
+    return ''
+  }, [datasourceType, fileUploadConfig.batch_count_limit, fileUploadConfig.file_size_limit, t])
+
+  return {
+    datasourceType,
+    isShowVectorSpaceFull,
+    nextBtnDisabled,
+    showSelect,
+    totalOptions,
+    selectedOptions,
+    tip,
+  }
+}

web/app/components/datasets/documents/create-from-pipeline/index.tsx

@@ -2,75 +2,71 @@
 import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
 import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
 import type { Node } from '@/app/components/workflow/types'
-import type { NotionPage } from '@/models/common'
-import type { CrawlResultItem, DocumentItem, CustomFile as File, FileIndexingEstimateResponse } from '@/models/datasets'
-import type {
-  InitialDocumentDetail,
-  OnlineDriveFile,
-  PublishedPipelineRunPreviewResponse,
-  PublishedPipelineRunResponse,
-} from '@/models/pipeline'
+import type { FileIndexingEstimateResponse } from '@/models/datasets'
+import type { InitialDocumentDetail } from '@/models/pipeline'
 import { useBoolean } from 'ahooks'
-import { useCallback, useMemo, useRef, useState } from 'react'
+import { useCallback, useMemo, useState } from 'react'
 import { useTranslation } from 'react-i18next'
-import { trackEvent } from '@/app/components/base/amplitude'
-import Divider from '@/app/components/base/divider'
 import Loading from '@/app/components/base/loading'
 import PlanUpgradeModal from '@/app/components/billing/plan-upgrade-modal'
-import VectorSpaceFull from '@/app/components/billing/vector-space-full'
-import LocalFile from '@/app/components/datasets/documents/create-from-pipeline/data-source/local-file'
-import OnlineDocuments from '@/app/components/datasets/documents/create-from-pipeline/data-source/online-documents'
-import OnlineDrive from '@/app/components/datasets/documents/create-from-pipeline/data-source/online-drive'
-import WebsiteCrawl from '@/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl'
 import { useDatasetDetailContextWithSelector } from '@/context/dataset-detail'
 import { useProviderContextSelector } from '@/context/provider-context'
 import { DatasourceType } from '@/models/pipeline'
 import { useFileUploadConfig } from '@/service/use-common'
-import { usePublishedPipelineInfo, useRunPublishedPipeline } from '@/service/use-pipeline'
-import { TransferMethod } from '@/types/app'
-import UpgradeCard from '../../create/step-one/upgrade-card'
-import Actions from './actions'
-import DataSourceOptions from './data-source-options'
+import { usePublishedPipelineInfo } from '@/service/use-pipeline'
 import { useDataSourceStore } from './data-source/store'
 import DataSourceProvider from './data-source/store/provider'
-import { useAddDocumentsSteps, useLocalFile, useOnlineDocument, useOnlineDrive, useWebsiteCrawl } from './hooks'
+import {
+  useAddDocumentsSteps,
+  useDatasourceActions,
+  useDatasourceUIState,
+  useLocalFile,
+  useOnlineDocument,
+  useOnlineDrive,
+  useWebsiteCrawl,
+} from './hooks'
 import LeftHeader from './left-header'
-import ChunkPreview from './preview/chunk-preview'
-import FilePreview from './preview/file-preview'
-import OnlineDocumentPreview from './preview/online-document-preview'
-import WebsitePreview from './preview/web-preview'
-import ProcessDocuments from './process-documents'
-import Processing from './processing'
+import { StepOneContent, StepThreeContent, StepTwoContent } from './steps'
+import { StepOnePreview, StepTwoPreview } from './steps/preview-panel'
 
 const CreateFormPipeline = () => {
   const { t } = useTranslation()
   const plan = useProviderContextSelector(state => state.plan)
   const enableBilling = useProviderContextSelector(state => state.enableBilling)
   const pipelineId = useDatasetDetailContextWithSelector(s => s.dataset?.pipeline_id)
+  const dataSourceStore = useDataSourceStore()
+
+  // Core state
   const [datasource, setDatasource] = useState<Datasource>()
   const [estimateData, setEstimateData] = useState<FileIndexingEstimateResponse | undefined>(undefined)
   const [batchId, setBatchId] = useState('')
   const [documents, setDocuments] = useState<InitialDocumentDetail[]>([])
-  const dataSourceStore = useDataSourceStore()
-
-  const isPreview = useRef(false)
-  const formRef = useRef<any>(null)
 
+  // Data fetching
   const { data: pipelineInfo, isFetching: isFetchingPipelineInfo } = usePublishedPipelineInfo(pipelineId || '')
   const { data: fileUploadConfigResponse } = useFileUploadConfig()
 
+  const fileUploadConfig = useMemo(() => fileUploadConfigResponse ?? {
+    file_size_limit: 15,
+    batch_count_limit: 5,
+  }, [fileUploadConfigResponse])
+
+  // Steps management
   const {
     steps,
     currentStep,
     handleNextStep: doHandleNextStep,
     handleBackStep,
   } = useAddDocumentsSteps()
+
+  // Datasource-specific hooks
   const {
     localFileList,
     allFileLoaded,
     currentLocalFile,
     hidePreviewLocalFile,
   } = useLocalFile()
+
   const {
     currentWorkspace,
     onlineDocuments,
@@ -79,12 +75,14 @@ const CreateFormPipeline = () => {
     hidePreviewOnlineDocument,
     clearOnlineDocumentData,
   } = useOnlineDocument()
+
   const {
     websitePages,
     currentWebsite,
     hideWebsitePreview,
     clearWebsiteCrawlData,
   } = useWebsiteCrawl()
+
   const {
     onlineDriveFileList,
     selectedFileIds,
@@ -92,43 +90,50 @@ const CreateFormPipeline = () => {
     clearOnlineDriveData,
   } = useOnlineDrive()
 
-  const datasourceType = useMemo(() => datasource?.nodeData.provider_type, [datasource])
+  // Computed values
   const isVectorSpaceFull = plan.usage.vectorSpace >= plan.total.vectorSpace
-  const isShowVectorSpaceFull = useMemo(() => {
-    if (!datasource)
-      return false
-    if (datasourceType === DatasourceType.localFile)
-      return allFileLoaded && isVectorSpaceFull && enableBilling
-    if (datasourceType === DatasourceType.onlineDocument)
-      return onlineDocuments.length > 0 && isVectorSpaceFull && enableBilling
-    if (datasourceType === DatasourceType.websiteCrawl)
-      return websitePages.length > 0 && isVectorSpaceFull && enableBilling
-    if (datasourceType === DatasourceType.onlineDrive)
-      return onlineDriveFileList.length > 0 && isVectorSpaceFull && enableBilling
-    return false
-  }, [allFileLoaded, datasource, datasourceType, enableBilling, isVectorSpaceFull, onlineDocuments.length, onlineDriveFileList.length, websitePages.length])
   const supportBatchUpload = !enableBilling || plan.type !== 'sandbox'
 
+  // UI state
+  const {
+    datasourceType,
+    isShowVectorSpaceFull,
+    nextBtnDisabled,
+    showSelect,
+    totalOptions,
+    selectedOptions,
+    tip,
+  } = useDatasourceUIState({
+    datasource,
+    allFileLoaded,
+    localFileListLength: localFileList.length,
+    onlineDocumentsLength: onlineDocuments.length,
+    websitePagesLength: websitePages.length,
+    selectedFileIdsLength: selectedFileIds.length,
+    onlineDriveFileList,
+    isVectorSpaceFull,
+    enableBilling,
+    currentWorkspacePagesLength: currentWorkspace?.pages.length ?? 0,
+    fileUploadConfig,
+  })
+
+  // Plan upgrade modal
   const [isShowPlanUpgradeModal, {
     setTrue: showPlanUpgradeModal,
     setFalse: hidePlanUpgradeModal,
   }] = useBoolean(false)
+
+  // Next step with batch upload check
   const handleNextStep = useCallback(() => {
     if (!supportBatchUpload) {
-      let isMultiple = false
-      if (datasourceType === DatasourceType.localFile && localFileList.length > 1)
-        isMultiple = true
-
-      if (datasourceType === DatasourceType.onlineDocument && onlineDocuments.length > 1)
-        isMultiple = true
-
-      if (datasourceType === DatasourceType.websiteCrawl && websitePages.length > 1)
-        isMultiple = true
-
-      if (datasourceType === DatasourceType.onlineDrive && selectedFileIds.length > 1)
-        isMultiple = true
-
-      if (isMultiple) {
+      const multipleCheckMap: Record<string, number> = {
+        [DatasourceType.localFile]: localFileList.length,
+        [DatasourceType.onlineDocument]: onlineDocuments.length,
+        [DatasourceType.websiteCrawl]: websitePages.length,
+        [DatasourceType.onlineDrive]: selectedFileIds.length,
+      }
+      const count = datasourceType ? multipleCheckMap[datasourceType] : 0
+      if (count > 1) {
         showPlanUpgradeModal()
         return
       }
@@ -136,334 +141,44 @@ const CreateFormPipeline = () => {
     doHandleNextStep()
   }, [datasourceType, doHandleNextStep, localFileList.length, onlineDocuments.length, selectedFileIds.length, showPlanUpgradeModal, supportBatchUpload, websitePages.length])
 
-  const nextBtnDisabled = useMemo(() => {
-    if (!datasource)
-      return true
-    if (datasourceType === DatasourceType.localFile)
-      return isShowVectorSpaceFull || !localFileList.length || !allFileLoaded
-    if (datasourceType === DatasourceType.onlineDocument)
-      return isShowVectorSpaceFull || !onlineDocuments.length
-    if (datasourceType === DatasourceType.websiteCrawl)
-      return isShowVectorSpaceFull || !websitePages.length
-    if (datasourceType === DatasourceType.onlineDrive)
-      return isShowVectorSpaceFull || !selectedFileIds.length
-    return false
-  }, [datasource, datasourceType, isShowVectorSpaceFull, localFileList.length, allFileLoaded, onlineDocuments.length, websitePages.length, selectedFileIds.length])
+  // Datasource actions
+  const {
+    isPreview,
+    formRef,
+    isIdle,
+    isPending,
+    onClickProcess,
+    onClickPreview,
+    handleSubmit,
+    handlePreviewFileChange,
+    handlePreviewOnlineDocumentChange,
+    handlePreviewWebsiteChange,
+    handlePreviewOnlineDriveFileChange,
+    handleSelectAll,
+    handleSwitchDataSource,
+    handleCredentialChange,
+  } = useDatasourceActions({
+    datasource,
+    datasourceType,
+    pipelineId,
+    dataSourceStore,
+    setEstimateData,
+    setBatchId,
+    setDocuments,
+    handleNextStep,
+    PagesMapAndSelectedPagesId,
+    currentWorkspacePages: currentWorkspace?.pages,
+    clearOnlineDocumentData,
+    clearWebsiteCrawlData,
+    clearOnlineDriveData,
+    setDatasource,
+  })
 
-  const fileUploadConfig = useMemo(() => fileUploadConfigResponse ?? {
-    file_size_limit: 15,
-    batch_count_limit: 5,
-  }, [fileUploadConfigResponse])
-
-  const showSelect = useMemo(() => {
-    if (datasourceType === DatasourceType.onlineDocument) {
-      const pagesCount = currentWorkspace?.pages.length ?? 0
-      return pagesCount > 0
-    }
-    if (datasourceType === DatasourceType.onlineDrive) {
-      const isBucketList = onlineDriveFileList.some(file => file.type === 'bucket')
-      return !isBucketList && onlineDriveFileList.filter((item) => {
-        return item.type !== 'bucket'
-      }).length > 0
-    }
-    return false
-  }, [currentWorkspace?.pages.length, datasourceType, onlineDriveFileList])
-
-  const totalOptions = useMemo(() => {
-    if (datasourceType === DatasourceType.onlineDocument)
-      return currentWorkspace?.pages.length
-    if (datasourceType === DatasourceType.onlineDrive) {
-      return onlineDriveFileList.filter((item) => {
-        return item.type !== 'bucket'
-      }).length
-    }
-  }, [currentWorkspace?.pages.length, datasourceType, onlineDriveFileList])
-
-  const selectedOptions = useMemo(() => {
-    if (datasourceType === DatasourceType.onlineDocument)
-      return onlineDocuments.length
-    if (datasourceType === DatasourceType.onlineDrive)
-      return selectedFileIds.length
-  }, [datasourceType, onlineDocuments.length, selectedFileIds.length])
-
-  const tip = useMemo(() => {
-    if (datasourceType === DatasourceType.onlineDocument)
-      return t('addDocuments.selectOnlineDocumentTip', { ns: 'datasetPipeline', count: 50 })
-    if (datasourceType === DatasourceType.onlineDrive) {
-      return t('addDocuments.selectOnlineDriveTip', {
-        ns: 'datasetPipeline',
-        count: fileUploadConfig.batch_count_limit,
-        fileSize: fileUploadConfig.file_size_limit,
-      })
-    }
-    return ''
-  }, [datasourceType, fileUploadConfig.batch_count_limit, fileUploadConfig.file_size_limit, t])
-
-  const { mutateAsync: runPublishedPipeline, isIdle, isPending } = useRunPublishedPipeline()
-
-  const handlePreviewChunks = useCallback(async (data: Record<string, any>) => {
-    if (!datasource)
-      return
-    const {
-      previewLocalFileRef,
-      previewOnlineDocumentRef,
-      previewWebsitePageRef,
-      previewOnlineDriveFileRef,
-      currentCredentialId,
-    } = dataSourceStore.getState()
-    const datasourceInfoList: Record<string, any>[] = []
-    if (datasourceType === DatasourceType.localFile) {
-      const { id, name, type, size, extension, mime_type } = previewLocalFileRef.current as File
-      const documentInfo = {
-        related_id: id,
-        name,
-        type,
-        size,
-        extension,
-        mime_type,
-        url: '',
-        transfer_method: TransferMethod.local_file,
-        credential_id: currentCredentialId,
-      }
-      datasourceInfoList.push(documentInfo)
-    }
-    if (datasourceType === DatasourceType.onlineDocument) {
-      const { workspace_id, ...rest } = previewOnlineDocumentRef.current!
-      const documentInfo = {
-        workspace_id,
-        page: rest,
-        credential_id: currentCredentialId,
-      }
-      datasourceInfoList.push(documentInfo)
-    }
-    if (datasourceType === DatasourceType.websiteCrawl) {
-      datasourceInfoList.push({
-        ...previewWebsitePageRef.current!,
-        credential_id: currentCredentialId,
-      })
-    }
-    if (datasourceType === DatasourceType.onlineDrive) {
-      const { bucket } = dataSourceStore.getState()
-      const { id, type, name } = previewOnlineDriveFileRef.current!
-      datasourceInfoList.push({
-        bucket,
-        id,
-        name,
-        type,
-        credential_id: currentCredentialId,
-      })
-    }
-    await runPublishedPipeline({
-      pipeline_id: pipelineId!,
-      inputs: data,
-      start_node_id: datasource.nodeId,
-      datasource_type: datasourceType as DatasourceType,
-      datasource_info_list: datasourceInfoList,
-      is_preview: true,
-    }, {
-      onSuccess: (res) => {
-        setEstimateData((res as PublishedPipelineRunPreviewResponse).data.outputs)
-      },
-    })
-  }, [datasource, datasourceType, runPublishedPipeline, pipelineId, dataSourceStore])
-
-  const handleProcess = useCallback(async (data: Record<string, any>) => {
-    if (!datasource)
-      return
-    const { currentCredentialId } = dataSourceStore.getState()
-    const datasourceInfoList: Record<string, any>[] = []
-    if (datasourceType === DatasourceType.localFile) {
-      const {
-        localFileList,
-      } = dataSourceStore.getState()
-      localFileList.forEach((file) => {
-        const { id, name, type, size, extension, mime_type } = file.file
-        const documentInfo = {
-          related_id: id,
-          name,
-          type,
-          size,
-          extension,
-          mime_type,
-          url: '',
-          transfer_method: TransferMethod.local_file,
-          credential_id: currentCredentialId,
-        }
-        datasourceInfoList.push(documentInfo)
-      })
-    }
-    if (datasourceType === DatasourceType.onlineDocument) {
-      const {
-        onlineDocuments,
-      } = dataSourceStore.getState()
-      onlineDocuments.forEach((page) => {
-        const { workspace_id, ...rest } = page
-        const documentInfo = {
-          workspace_id,
-          page: rest,
-          credential_id: currentCredentialId,
-        }
-        datasourceInfoList.push(documentInfo)
-      })
-    }
-    if (datasourceType === DatasourceType.websiteCrawl) {
-      const {
-        websitePages,
-      } = dataSourceStore.getState()
-      websitePages.forEach((websitePage) => {
-        datasourceInfoList.push({
-          ...websitePage,
-          credential_id: currentCredentialId,
-        })
-      })
-    }
-    if (datasourceType === DatasourceType.onlineDrive) {
-      const {
-        bucket,
-        selectedFileIds,
-        onlineDriveFileList,
-      } = dataSourceStore.getState()
-      selectedFileIds.forEach((id) => {
-        const file = onlineDriveFileList.find(file => file.id === id)
-        datasourceInfoList.push({
-          bucket,
-          id: file?.id,
-          name: file?.name,
-          type: file?.type,
-          credential_id: currentCredentialId,
-        })
-      })
-    }
-    await runPublishedPipeline({
-      pipeline_id: pipelineId!,
-      inputs: data,
-      start_node_id: datasource.nodeId,
-      datasource_type: datasourceType as DatasourceType,
-      datasource_info_list: datasourceInfoList,
-      is_preview: false,
-    }, {
-      onSuccess: (res) => {
-        setBatchId((res as PublishedPipelineRunResponse).batch || '')
-        setDocuments((res as PublishedPipelineRunResponse).documents || [])
-        handleNextStep()
-        trackEvent('dataset_document_added', {
-          data_source_type: datasourceType,
-          indexing_technique: 'pipeline',
-        })
-      },
-    })
-  }, [dataSourceStore, datasource, datasourceType, handleNextStep, pipelineId, runPublishedPipeline])
-
-  const onClickProcess = useCallback(() => {
-    isPreview.current = false
-    formRef.current?.submit()
-  }, [])
-
-  const onClickPreview = useCallback(() => {
-    isPreview.current = true
-    formRef.current?.submit()
-  }, [])
-
-  const handleSubmit = useCallback((data: Record<string, any>) => {
-    if (isPreview.current)
-      handlePreviewChunks(data)
-    else
-      handleProcess(data)
-  }, [handlePreviewChunks, handleProcess])
-
-  const handlePreviewFileChange = useCallback((file: DocumentItem) => {
-    const { previewLocalFileRef } = dataSourceStore.getState()
-    previewLocalFileRef.current = file
-    onClickPreview()
-  }, [dataSourceStore, onClickPreview])
-
-  const handlePreviewOnlineDocumentChange = useCallback((page: NotionPage) => {
-    const { previewOnlineDocumentRef } = dataSourceStore.getState()
-    previewOnlineDocumentRef.current = page
-    onClickPreview()
-  }, [dataSourceStore, onClickPreview])
-
-  const handlePreviewWebsiteChange = useCallback((website: CrawlResultItem) => {
-    const { previewWebsitePageRef } = dataSourceStore.getState()
-    previewWebsitePageRef.current = website
-    onClickPreview()
-  }, [dataSourceStore, onClickPreview])
-
-  const handlePreviewOnlineDriveFileChange = useCallback((file: OnlineDriveFile) => {
-    const { previewOnlineDriveFileRef } = dataSourceStore.getState()
-    previewOnlineDriveFileRef.current = file
-    onClickPreview()
-  }, [dataSourceStore, onClickPreview])
-
-  const handleSelectAll = useCallback(() => {
-    const {
-      onlineDocuments,
-      onlineDriveFileList,
-      selectedFileIds,
-      setOnlineDocuments,
-      setSelectedFileIds,
-      setSelectedPagesId,
-    } = dataSourceStore.getState()
-    if (datasourceType === DatasourceType.onlineDocument) {
-      const allIds = currentWorkspace?.pages.map(page => page.page_id) || []
-      if (onlineDocuments.length < allIds.length) {
-        const selectedPages = Array.from(allIds).map(pageId => PagesMapAndSelectedPagesId[pageId])
-        setOnlineDocuments(selectedPages)
-        setSelectedPagesId(new Set(allIds))
-      }
-      else {
-        setOnlineDocuments([])
-        setSelectedPagesId(new Set())
-      }
-    }
-    if (datasourceType === DatasourceType.onlineDrive) {
-      const allKeys = onlineDriveFileList.filter((item) => {
-        return item.type !== 'bucket'
-      }).map(file => file.id)
-      if (selectedFileIds.length < allKeys.length)
-        setSelectedFileIds(allKeys)
-      else
-        setSelectedFileIds([])
-    }
-  }, [PagesMapAndSelectedPagesId, currentWorkspace?.pages, dataSourceStore, datasourceType])
-
-  const clearDataSourceData = useCallback((dataSource: Datasource) => {
-    const providerType = dataSource.nodeData.provider_type
-    if (providerType === DatasourceType.onlineDocument)
-      clearOnlineDocumentData()
-    else if (providerType === DatasourceType.websiteCrawl)
-      clearWebsiteCrawlData()
-    else if (providerType === DatasourceType.onlineDrive)
-      clearOnlineDriveData()
-  }, [clearOnlineDocumentData, clearOnlineDriveData, clearWebsiteCrawlData])
-
-  const handleSwitchDataSource = useCallback((dataSource: Datasource) => {
-    const {
-      setCurrentCredentialId,
-      currentNodeIdRef,
-    } = dataSourceStore.getState()
-    clearDataSourceData(dataSource)
-    setCurrentCredentialId('')
-    currentNodeIdRef.current = dataSource.nodeId
-    setDatasource(dataSource)
-  }, [clearDataSourceData, dataSourceStore])
-
-  const handleCredentialChange = useCallback((credentialId: string) => {
-    const { setCurrentCredentialId } = dataSourceStore.getState()
-    clearDataSourceData(datasource!)
-    setCurrentCredentialId(credentialId)
-  }, [clearDataSourceData, dataSourceStore, datasource])
-
-  if (isFetchingPipelineInfo) {
-    return (
-      <Loading type="app" />
-    )
-  }
+  if (isFetchingPipelineInfo)
+    return <Loading type="app" />
 
   return (
-    <div
-      className="relative flex h-[calc(100vh-56px)] w-full min-w-[1024px] overflow-x-auto rounded-t-2xl border-t border-effects-highlight bg-background-default-subtle"
-    >
+    <div className="relative flex h-[calc(100vh-56px)] w-full min-w-[1024px] overflow-x-auto rounded-t-2xl border-t border-effects-highlight bg-background-default-subtle">
       <div className="h-full min-w-0 flex-1">
         <div className="flex h-full flex-col px-14">
           <LeftHeader
@@ -472,139 +187,77 @@ const CreateFormPipeline = () => {
             currentStep={currentStep}
           />
           <div className="grow overflow-y-auto">
-            {
-              currentStep === 1 && (
-                <div className="flex flex-col gap-y-5 pt-4">
-                  <DataSourceOptions
-                    datasourceNodeId={datasource?.nodeId || ''}
-                    onSelect={handleSwitchDataSource}
-                    pipelineNodes={(pipelineInfo?.graph.nodes || []) as Node<DataSourceNodeType>[]}
-                  />
-                  {datasourceType === DatasourceType.localFile && (
-                    <LocalFile
-                      allowedExtensions={datasource!.nodeData.fileExtensions || []}
-                      supportBatchUpload={supportBatchUpload}
-                    />
-                  )}
-                  {datasourceType === DatasourceType.onlineDocument && (
-                    <OnlineDocuments
-                      nodeId={datasource!.nodeId}
-                      nodeData={datasource!.nodeData}
-                      onCredentialChange={handleCredentialChange}
-                    />
-                  )}
-                  {datasourceType === DatasourceType.websiteCrawl && (
-                    <WebsiteCrawl
-                      nodeId={datasource!.nodeId}
-                      nodeData={datasource!.nodeData}
-                      onCredentialChange={handleCredentialChange}
-                    />
-                  )}
-                  {datasourceType === DatasourceType.onlineDrive && (
-                    <OnlineDrive
-                      nodeId={datasource!.nodeId}
-                      nodeData={datasource!.nodeData}
-                      onCredentialChange={handleCredentialChange}
-                    />
-                  )}
-                  {isShowVectorSpaceFull && (
-                    <VectorSpaceFull />
-                  )}
-                  <Actions
-                    showSelect={showSelect}
-                    totalOptions={totalOptions}
-                    selectedOptions={selectedOptions}
-                    onSelectAll={handleSelectAll}
-                    disabled={nextBtnDisabled}
-                    handleNextStep={handleNextStep}
-                    tip={tip}
-                  />
-                  {
-                    !supportBatchUpload && datasourceType === DatasourceType.localFile && localFileList.length > 0 && (
-                      <>
-                        <Divider type="horizontal" className="my-4 h-px bg-divider-subtle" />
-                        <UpgradeCard />
-                      </>
-                    )
-                  }
-                </div>
-              )
-            }
-            {
-              currentStep === 2 && (
-                <ProcessDocuments
-                  ref={formRef}
-                  dataSourceNodeId={datasource!.nodeId}
-                  isRunning={isPending}
-                  onProcess={onClickProcess}
-                  onPreview={onClickPreview}
-                  onSubmit={handleSubmit}
-                  onBack={handleBackStep}
-                />
-              )
-            }
-            {
-              currentStep === 3 && (
-                <Processing
-                  batchId={batchId}
-                  documents={documents}
-                />
-              )
-            }
+            {currentStep === 1 && (
+              <StepOneContent
+                datasource={datasource}
+                datasourceType={datasourceType}
+                pipelineNodes={(pipelineInfo?.graph.nodes || []) as Node<DataSourceNodeType>[]}
+                supportBatchUpload={supportBatchUpload}
+                localFileListLength={localFileList.length}
+                isShowVectorSpaceFull={isShowVectorSpaceFull}
+                showSelect={showSelect}
+                totalOptions={totalOptions}
+                selectedOptions={selectedOptions}
+                tip={tip}
+                nextBtnDisabled={nextBtnDisabled}
+                onSelectDataSource={handleSwitchDataSource}
+                onCredentialChange={handleCredentialChange}
+                onSelectAll={handleSelectAll}
+                onNextStep={handleNextStep}
+              />
+            )}
+            {currentStep === 2 && (
+              <StepTwoContent
+                formRef={formRef}
+                dataSourceNodeId={datasource!.nodeId}
+                isRunning={isPending}
+                onProcess={onClickProcess}
+                onPreview={onClickPreview}
+                onSubmit={handleSubmit}
+                onBack={handleBackStep}
+              />
+            )}
+            {currentStep === 3 && (
+              <StepThreeContent
+                batchId={batchId}
+                documents={documents}
+              />
+            )}
           </div>
         </div>
       </div>
-      {/* Preview */}
-      {
-        currentStep === 1 && (
-          <div className="h-full min-w-0 flex-1">
-            <div className="flex h-full flex-col pl-2 pt-2">
-              {currentLocalFile && (
-                <FilePreview
-                  file={currentLocalFile}
-                  hidePreview={hidePreviewLocalFile}
-                />
-              )}
-              {currentDocument && (
-                <OnlineDocumentPreview
-                  datasourceNodeId={datasource!.nodeId}
-                  currentPage={currentDocument}
-                  hidePreview={hidePreviewOnlineDocument}
-                />
-              )}
-              {currentWebsite && (
-                <WebsitePreview
-                  currentWebsite={currentWebsite}
-                  hidePreview={hideWebsitePreview}
-                />
-              )}
-            </div>
-          </div>
-        )
-      }
-      {
-        currentStep === 2 && (
-          <div className="h-full min-w-0 flex-1">
-            <div className="flex h-full flex-col pl-2 pt-2">
-              <ChunkPreview
-                dataSourceType={datasourceType as DatasourceType}
-                localFiles={localFileList.map(file => file.file)}
-                onlineDocuments={onlineDocuments}
-                websitePages={websitePages}
-                onlineDriveFiles={selectedOnlineDriveFileList}
-                isIdle={isIdle}
-                isPending={isPending && isPreview.current}
-                estimateData={estimateData}
-                onPreview={onClickPreview}
-                handlePreviewFileChange={handlePreviewFileChange}
-                handlePreviewOnlineDocumentChange={handlePreviewOnlineDocumentChange}
-                handlePreviewWebsitePageChange={handlePreviewWebsiteChange}
-                handlePreviewOnlineDriveFileChange={handlePreviewOnlineDriveFileChange}
-              />
-            </div>
-          </div>
-        )
-      }
+
+      {/* Preview Panel */}
+      {currentStep === 1 && (
+        <StepOnePreview
+          datasource={datasource}
+          currentLocalFile={currentLocalFile}
+          currentDocument={currentDocument}
+          currentWebsite={currentWebsite}
+          hidePreviewLocalFile={hidePreviewLocalFile}
+          hidePreviewOnlineDocument={hidePreviewOnlineDocument}
+          hideWebsitePreview={hideWebsitePreview}
+        />
+      )}
+      {currentStep === 2 && (
+        <StepTwoPreview
+          datasourceType={datasourceType}
+          localFileList={localFileList}
+          onlineDocuments={onlineDocuments}
+          websitePages={websitePages}
+          selectedOnlineDriveFileList={selectedOnlineDriveFileList}
+          isIdle={isIdle}
+          isPendingPreview={isPending && isPreview.current}
+          estimateData={estimateData}
+          onPreview={onClickPreview}
+          handlePreviewFileChange={handlePreviewFileChange}
+          handlePreviewOnlineDocumentChange={handlePreviewOnlineDocumentChange}
+          handlePreviewWebsitePageChange={handlePreviewWebsiteChange}
+          handlePreviewOnlineDriveFileChange={handlePreviewOnlineDriveFileChange}
+        />
+      )}
+
+      {/* Plan Upgrade Modal */}
       {isShowPlanUpgradeModal && (
         <PlanUpgradeModal
           show

web/app/components/datasets/documents/create-from-pipeline/steps/index.ts

@@ -0,0 +1,3 @@
+export { default as StepOneContent } from './step-one-content'
+export { default as StepThreeContent } from './step-three-content'
+export { default as StepTwoContent } from './step-two-content'

web/app/components/datasets/documents/create-from-pipeline/steps/preview-panel.tsx

@@ -0,0 +1,112 @@
+'use client'
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem, CustomFile, DocumentItem, FileIndexingEstimateResponse, FileItem } from '@/models/datasets'
+import type { DatasourceType, OnlineDriveFile } from '@/models/pipeline'
+import { memo } from 'react'
+import ChunkPreview from '../preview/chunk-preview'
+import FilePreview from '../preview/file-preview'
+import OnlineDocumentPreview from '../preview/online-document-preview'
+import WebsitePreview from '../preview/web-preview'
+
+type StepOnePreviewProps = {
+  datasource: Datasource | undefined
+  currentLocalFile: CustomFile | undefined
+  currentDocument: (NotionPage & { workspace_id: string }) | undefined
+  currentWebsite: CrawlResultItem | undefined
+  hidePreviewLocalFile: () => void
+  hidePreviewOnlineDocument: () => void
+  hideWebsitePreview: () => void
+}
+
+export const StepOnePreview = memo(({
+  datasource,
+  currentLocalFile,
+  currentDocument,
+  currentWebsite,
+  hidePreviewLocalFile,
+  hidePreviewOnlineDocument,
+  hideWebsitePreview,
+}: StepOnePreviewProps) => {
+  return (
+    <div className="h-full min-w-0 flex-1">
+      <div className="flex h-full flex-col pl-2 pt-2">
+        {currentLocalFile && (
+          <FilePreview
+            file={currentLocalFile}
+            hidePreview={hidePreviewLocalFile}
+          />
+        )}
+        {currentDocument && (
+          <OnlineDocumentPreview
+            datasourceNodeId={datasource!.nodeId}
+            currentPage={currentDocument}
+            hidePreview={hidePreviewOnlineDocument}
+          />
+        )}
+        {currentWebsite && (
+          <WebsitePreview
+            currentWebsite={currentWebsite}
+            hidePreview={hideWebsitePreview}
+          />
+        )}
+      </div>
+    </div>
+  )
+})
+StepOnePreview.displayName = 'StepOnePreview'
+
+type StepTwoPreviewProps = {
+  datasourceType: string | undefined
+  localFileList: FileItem[]
+  onlineDocuments: (NotionPage & { workspace_id: string })[]
+  websitePages: CrawlResultItem[]
+  selectedOnlineDriveFileList: OnlineDriveFile[]
+  isIdle: boolean
+  isPendingPreview: boolean
+  estimateData: FileIndexingEstimateResponse | undefined
+  onPreview: () => void
+  handlePreviewFileChange: (file: DocumentItem) => void
+  handlePreviewOnlineDocumentChange: (page: NotionPage) => void
+  handlePreviewWebsitePageChange: (website: CrawlResultItem) => void
+  handlePreviewOnlineDriveFileChange: (file: OnlineDriveFile) => void
+}
+
+export const StepTwoPreview = memo(({
+  datasourceType,
+  localFileList,
+  onlineDocuments,
+  websitePages,
+  selectedOnlineDriveFileList,
+  isIdle,
+  isPendingPreview,
+  estimateData,
+  onPreview,
+  handlePreviewFileChange,
+  handlePreviewOnlineDocumentChange,
+  handlePreviewWebsitePageChange,
+  handlePreviewOnlineDriveFileChange,
+}: StepTwoPreviewProps) => {
+  return (
+    <div className="h-full min-w-0 flex-1">
+      <div className="flex h-full flex-col pl-2 pt-2">
+        <ChunkPreview
+          dataSourceType={datasourceType as DatasourceType}
+          localFiles={localFileList.map(file => file.file)}
+          onlineDocuments={onlineDocuments}
+          websitePages={websitePages}
+          onlineDriveFiles={selectedOnlineDriveFileList}
+          isIdle={isIdle}
+          isPending={isPendingPreview}
+          estimateData={estimateData}
+          onPreview={onPreview}
+          handlePreviewFileChange={handlePreviewFileChange}
+          handlePreviewOnlineDocumentChange={handlePreviewOnlineDocumentChange}
+          handlePreviewWebsitePageChange={handlePreviewWebsitePageChange}
+          handlePreviewOnlineDriveFileChange={handlePreviewOnlineDriveFileChange}
+        />
+      </div>
+    </div>
+  )
+})
+StepTwoPreview.displayName = 'StepTwoPreview'

web/app/components/datasets/documents/create-from-pipeline/steps/step-one-content.tsx

@@ -0,0 +1,110 @@
+'use client'
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
+import type { Node } from '@/app/components/workflow/types'
+import { memo } from 'react'
+import Divider from '@/app/components/base/divider'
+import VectorSpaceFull from '@/app/components/billing/vector-space-full'
+import LocalFile from '@/app/components/datasets/documents/create-from-pipeline/data-source/local-file'
+import OnlineDocuments from '@/app/components/datasets/documents/create-from-pipeline/data-source/online-documents'
+import OnlineDrive from '@/app/components/datasets/documents/create-from-pipeline/data-source/online-drive'
+import WebsiteCrawl from '@/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl'
+import { DatasourceType } from '@/models/pipeline'
+import UpgradeCard from '../../../create/step-one/upgrade-card'
+import Actions from '../actions'
+import DataSourceOptions from '../data-source-options'
+
+type StepOneContentProps = {
+  datasource: Datasource | undefined
+  datasourceType: string | undefined
+  pipelineNodes: Node<DataSourceNodeType>[]
+  supportBatchUpload: boolean
+  localFileListLength: number
+  isShowVectorSpaceFull: boolean
+  showSelect: boolean
+  totalOptions: number | undefined
+  selectedOptions: number | undefined
+  tip: string
+  nextBtnDisabled: boolean
+  onSelectDataSource: (dataSource: Datasource) => void
+  onCredentialChange: (credentialId: string) => void
+  onSelectAll: () => void
+  onNextStep: () => void
+}
+
+const StepOneContent = ({
+  datasource,
+  datasourceType,
+  pipelineNodes,
+  supportBatchUpload,
+  localFileListLength,
+  isShowVectorSpaceFull,
+  showSelect,
+  totalOptions,
+  selectedOptions,
+  tip,
+  nextBtnDisabled,
+  onSelectDataSource,
+  onCredentialChange,
+  onSelectAll,
+  onNextStep,
+}: StepOneContentProps) => {
+  const showUpgradeCard = !supportBatchUpload
+    && datasourceType === DatasourceType.localFile
+    && localFileListLength > 0
+
+  return (
+    <div className="flex flex-col gap-y-5 pt-4">
+      <DataSourceOptions
+        datasourceNodeId={datasource?.nodeId || ''}
+        onSelect={onSelectDataSource}
+        pipelineNodes={pipelineNodes}
+      />
+      {datasourceType === DatasourceType.localFile && (
+        <LocalFile
+          allowedExtensions={datasource!.nodeData.fileExtensions || []}
+          supportBatchUpload={supportBatchUpload}
+        />
+      )}
+      {datasourceType === DatasourceType.onlineDocument && (
+        <OnlineDocuments
+          nodeId={datasource!.nodeId}
+          nodeData={datasource!.nodeData}
+          onCredentialChange={onCredentialChange}
+        />
+      )}
+      {datasourceType === DatasourceType.websiteCrawl && (
+        <WebsiteCrawl
+          nodeId={datasource!.nodeId}
+          nodeData={datasource!.nodeData}
+          onCredentialChange={onCredentialChange}
+        />
+      )}
+      {datasourceType === DatasourceType.onlineDrive && (
+        <OnlineDrive
+          nodeId={datasource!.nodeId}
+          nodeData={datasource!.nodeData}
+          onCredentialChange={onCredentialChange}
+        />
+      )}
+      {isShowVectorSpaceFull && <VectorSpaceFull />}
+      <Actions
+        showSelect={showSelect}
+        totalOptions={totalOptions}
+        selectedOptions={selectedOptions}
+        onSelectAll={onSelectAll}
+        disabled={nextBtnDisabled}
+        handleNextStep={onNextStep}
+        tip={tip}
+      />
+      {showUpgradeCard && (
+        <>
+          <Divider type="horizontal" className="my-4 h-px bg-divider-subtle" />
+          <UpgradeCard />
+        </>
+      )}
+    </div>
+  )
+}
+
+export default memo(StepOneContent)

web/app/components/datasets/documents/create-from-pipeline/steps/step-three-content.tsx

@@ -0,0 +1,23 @@
+'use client'
+import type { InitialDocumentDetail } from '@/models/pipeline'
+import { memo } from 'react'
+import Processing from '../processing'
+
+type StepThreeContentProps = {
+  batchId: string
+  documents: InitialDocumentDetail[]
+}
+
+const StepThreeContent = ({
+  batchId,
+  documents,
+}: StepThreeContentProps) => {
+  return (
+    <Processing
+      batchId={batchId}
+      documents={documents}
+    />
+  )
+}
+
+export default memo(StepThreeContent)

web/app/components/datasets/documents/create-from-pipeline/steps/step-two-content.tsx

@@ -0,0 +1,38 @@
+'use client'
+import type { RefObject } from 'react'
+import { memo } from 'react'
+import ProcessDocuments from '../process-documents'
+
+type StepTwoContentProps = {
+  formRef: RefObject<{ submit: () => void } | null>
+  dataSourceNodeId: string
+  isRunning: boolean
+  onProcess: () => void
+  onPreview: () => void
+  onSubmit: (data: Record<string, unknown>) => void
+  onBack: () => void
+}
+
+const StepTwoContent = ({
+  formRef,
+  dataSourceNodeId,
+  isRunning,
+  onProcess,
+  onPreview,
+  onSubmit,
+  onBack,
+}: StepTwoContentProps) => {
+  return (
+    <ProcessDocuments
+      ref={formRef}
+      dataSourceNodeId={dataSourceNodeId}
+      isRunning={isRunning}
+      onProcess={onProcess}
+      onPreview={onPreview}
+      onSubmit={onSubmit}
+      onBack={onBack}
+    />
+  )
+}
+
+export default memo(StepTwoContent)

web/app/components/datasets/documents/create-from-pipeline/utils/datasource-info-builder.ts

@@ -0,0 +1,63 @@
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem, CustomFile as File } from '@/models/datasets'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { TransferMethod } from '@/types/app'
+
+/**
+ * Build datasource info for local files
+ */
+export const buildLocalFileDatasourceInfo = (
+  file: File,
+  credentialId: string,
+): Record<string, unknown> => ({
+  related_id: file.id,
+  name: file.name,
+  type: file.type,
+  size: file.size,
+  extension: file.extension,
+  mime_type: file.mime_type,
+  url: '',
+  transfer_method: TransferMethod.local_file,
+  credential_id: credentialId,
+})
+
+/**
+ * Build datasource info for online documents
+ */
+export const buildOnlineDocumentDatasourceInfo = (
+  page: NotionPage & { workspace_id: string },
+  credentialId: string,
+): Record<string, unknown> => {
+  const { workspace_id, ...rest } = page
+  return {
+    workspace_id,
+    page: rest,
+    credential_id: credentialId,
+  }
+}
+
+/**
+ * Build datasource info for website crawl
+ */
+export const buildWebsiteCrawlDatasourceInfo = (
+  page: CrawlResultItem,
+  credentialId: string,
+): Record<string, unknown> => ({
+  ...page,
+  credential_id: credentialId,
+})
+
+/**
+ * Build datasource info for online drive
+ */
+export const buildOnlineDriveDatasourceInfo = (
+  file: OnlineDriveFile,
+  bucket: string,
+  credentialId: string,
+): Record<string, unknown> => ({
+  bucket,
+  id: file.id,
+  name: file.name,
+  type: file.type,
+  credential_id: credentialId,
+})