Merge remote-tracking branch 'origin/main' into feat/support-agent-sandbox

web/app/components/datasets/documents/components/list.tsx

@@ -30,9 +30,10 @@ import { useDatasetDetailContextWithSelector as useDatasetDetailContext } from '
 import useTimestamp from '@/hooks/use-timestamp'
 import { ChunkingMode, DataSourceType, DocumentActionType } from '@/models/datasets'
 import { DatasourceType } from '@/models/pipeline'
-import { useDocumentArchive, useDocumentBatchRetryIndex, useDocumentDelete, useDocumentDisable, useDocumentEnable } from '@/service/knowledge/use-document'
+import { useDocumentArchive, useDocumentBatchRetryIndex, useDocumentDelete, useDocumentDisable, useDocumentDownloadZip, useDocumentEnable } from '@/service/knowledge/use-document'
 import { asyncRunSafe } from '@/utils'
 import { cn } from '@/utils/classnames'
+import { downloadBlob } from '@/utils/download'
 import { formatNumber } from '@/utils/format'
 import BatchAction from '../detail/completed/common/batch-action'
 import StatusItem from '../status-item'
@@ -222,6 +223,7 @@ const DocumentList: FC<IDocumentListProps> = ({
   const { mutateAsync: disableDocument } = useDocumentDisable()
   const { mutateAsync: deleteDocument } = useDocumentDelete()
   const { mutateAsync: retryIndexDocument } = useDocumentBatchRetryIndex()
+  const { mutateAsync: requestDocumentsZip, isPending: isDownloadingZip } = useDocumentDownloadZip()
 
   const handleAction = (actionName: DocumentActionType) => {
     return async () => {
@@ -300,6 +302,39 @@ const DocumentList: FC<IDocumentListProps> = ({
     return dataSourceType === DatasourceType.onlineDrive
   }, [])
 
+  const downloadableSelectedIds = useMemo(() => {
+    const selectedSet = new Set(selectedIds)
+    return localDocs
+      .filter(doc => selectedSet.has(doc.id) && doc.data_source_type === DataSourceType.FILE)
+      .map(doc => doc.id)
+  }, [localDocs, selectedIds])
+
+  /**
+   * Generate a random ZIP filename for bulk document downloads.
+   * We intentionally avoid leaking dataset info in the exported archive name.
+   */
+  const generateDocsZipFileName = useCallback((): string => {
+    // Prefer UUID for uniqueness; fall back to time+random when unavailable.
+    const randomPart = (typeof crypto !== 'undefined' && typeof crypto.randomUUID === 'function')
+      ? crypto.randomUUID()
+      : `${Date.now().toString(36)}${Math.random().toString(36).slice(2, 10)}`
+    return `${randomPart}-docs.zip`
+  }, [])
+
+  const handleBatchDownload = useCallback(async () => {
+    if (isDownloadingZip)
+      return
+
+    // Download as a single ZIP to avoid browser caps on multiple automatic downloads.
+    const [e, blob] = await asyncRunSafe(requestDocumentsZip({ datasetId, documentIds: downloadableSelectedIds }))
+    if (e || !blob) {
+      Toast.notify({ type: 'error', message: t('actionMsg.downloadUnsuccessfully', { ns: 'common' }) })
+      return
+    }
+
+    downloadBlob({ data: blob, fileName: generateDocsZipFileName() })
+  }, [datasetId, downloadableSelectedIds, generateDocsZipFileName, isDownloadingZip, requestDocumentsZip, t])
+
   return (
     <div className="relative mt-3 flex h-full w-full flex-col">
       <div className="relative h-0 grow overflow-x-auto">
@@ -463,6 +498,7 @@ const DocumentList: FC<IDocumentListProps> = ({
           onArchive={handleAction(DocumentActionType.archive)}
           onBatchEnable={handleAction(DocumentActionType.enable)}
           onBatchDisable={handleAction(DocumentActionType.disable)}
+          onBatchDownload={downloadableSelectedIds.length > 0 ? handleBatchDownload : undefined}
           onBatchDelete={handleAction(DocumentActionType.delete)}
           onEditMetadata={showEditModal}
           onBatchReIndex={hasErrorDocumentsSelected ? handleBatchReIndex : undefined}

web/app/components/datasets/documents/components/operations.tsx

@@ -1,8 +1,10 @@
 import type { OperationName } from '../types'
 import type { CommonResponse } from '@/models/common'
+import type { DocumentDownloadResponse } from '@/service/datasets'
 import {
   RiArchive2Line,
   RiDeleteBinLine,
+  RiDownload2Line,
   RiEditLine,
   RiEqualizer2Line,
   RiLoopLeftLine,
@@ -28,6 +30,7 @@ import {
   useDocumentArchive,
   useDocumentDelete,
   useDocumentDisable,
+  useDocumentDownload,
   useDocumentEnable,
   useDocumentPause,
   useDocumentResume,
@@ -37,6 +40,7 @@ import {
 } from '@/service/knowledge/use-document'
 import { asyncRunSafe } from '@/utils'
 import { cn } from '@/utils/classnames'
+import { downloadUrl } from '@/utils/download'
 import s from '../style.module.css'
 import RenameModal from './rename-modal'
 
@@ -69,7 +73,7 @@ const Operations = ({
   scene = 'list',
   className = '',
 }: OperationsProps) => {
-  const { id, enabled = false, archived = false, data_source_type, display_status } = detail || {}
+  const { id, name, enabled = false, archived = false, data_source_type, display_status } = detail || {}
   const [showModal, setShowModal] = useState(false)
   const [deleting, setDeleting] = useState(false)
   const { notify } = useContext(ToastContext)
@@ -80,6 +84,7 @@ const Operations = ({
   const { mutateAsync: enableDocument } = useDocumentEnable()
   const { mutateAsync: disableDocument } = useDocumentDisable()
   const { mutateAsync: deleteDocument } = useDocumentDelete()
+  const { mutateAsync: downloadDocument, isPending: isDownloading } = useDocumentDownload()
   const { mutateAsync: syncDocument } = useSyncDocument()
   const { mutateAsync: syncWebsite } = useSyncWebsite()
   const { mutateAsync: pauseDocument } = useDocumentPause()
@@ -158,6 +163,24 @@ const Operations = ({
     onUpdate()
   }, [onUpdate])
 
+  const handleDownload = useCallback(async () => {
+    // Avoid repeated clicks while the signed URL request is in-flight.
+    if (isDownloading)
+      return
+
+    // Request a signed URL first (it points to `/files/<id>/file-preview?...&as_attachment=true`).
+    const [e, res] = await asyncRunSafe<DocumentDownloadResponse>(
+      downloadDocument({ datasetId, documentId: id }) as Promise<DocumentDownloadResponse>,
+    )
+    if (e || !res?.url) {
+      notify({ type: 'error', message: t('actionMsg.downloadUnsuccessfully', { ns: 'common' }) })
+      return
+    }
+
+    // Trigger download without navigating away (helps avoid duplicate downloads in some browsers).
+    downloadUrl({ url: res.url, fileName: name })
+  }, [datasetId, downloadDocument, id, isDownloading, name, notify, t])
+
   return (
     <div className="flex items-center" onClick={e => e.stopPropagation()}>
       {isListScene && !embeddingAvailable && (
@@ -214,6 +237,20 @@ const Operations = ({
                       <RiEditLine className="h-4 w-4 text-text-tertiary" />
                       <span className={s.actionName}>{t('list.table.rename', { ns: 'datasetDocuments' })}</span>
                     </div>
+                    {data_source_type === DataSourceType.FILE && (
+                      <div
+                        className={s.actionItem}
+                        onClick={(evt) => {
+                          evt.preventDefault()
+                          evt.stopPropagation()
+                          evt.nativeEvent.stopImmediatePropagation?.()
+                          handleDownload()
+                        }}
+                      >
+                        <RiDownload2Line className="h-4 w-4 text-text-tertiary" />
+                        <span className={s.actionName}>{t('list.action.download', { ns: 'datasetDocuments' })}</span>
+                      </div>
+                    )}
                     {['notion_import', DataSourceType.WEB].includes(data_source_type) && (
                       <div className={s.actionItem} onClick={() => onOperate('sync')}>
                         <RiLoopLeftLine className="h-4 w-4 text-text-tertiary" />
@@ -223,6 +260,23 @@ const Operations = ({
                     <Divider className="my-1" />
                   </>
                 )}
+                {archived && data_source_type === DataSourceType.FILE && (
+                  <>
+                    <div
+                      className={s.actionItem}
+                      onClick={(evt) => {
+                        evt.preventDefault()
+                        evt.stopPropagation()
+                        evt.nativeEvent.stopImmediatePropagation?.()
+                        handleDownload()
+                      }}
+                    >
+                      <RiDownload2Line className="h-4 w-4 text-text-tertiary" />
+                      <span className={s.actionName}>{t('list.action.download', { ns: 'datasetDocuments' })}</span>
+                    </div>
+                    <Divider className="my-1" />
+                  </>
+                )}
                 {!archived && display_status?.toLowerCase() === 'indexing' && (
                   <div className={s.actionItem} onClick={() => onOperate('pause')}>
                     <RiPauseCircleLine className="h-4 w-4 text-text-tertiary" />

web/app/components/datasets/documents/detail/completed/common/batch-action.tsx

@@ -1,5 +1,5 @@
 import type { FC } from 'react'
-import { RiArchive2Line, RiCheckboxCircleLine, RiCloseCircleLine, RiDeleteBinLine, RiDraftLine, RiRefreshLine } from '@remixicon/react'
+import { RiArchive2Line, RiCheckboxCircleLine, RiCloseCircleLine, RiDeleteBinLine, RiDownload2Line, RiDraftLine, RiRefreshLine } from '@remixicon/react'
 import { useBoolean } from 'ahooks'
 import * as React from 'react'
 import { useTranslation } from 'react-i18next'
@@ -14,6 +14,7 @@ type IBatchActionProps = {
   selectedIds: string[]
   onBatchEnable: () => void
   onBatchDisable: () => void
+  onBatchDownload?: () => void
   onBatchDelete: () => Promise<void>
   onArchive?: () => void
   onEditMetadata?: () => void
@@ -26,6 +27,7 @@ const BatchAction: FC<IBatchActionProps> = ({
   selectedIds,
   onBatchEnable,
   onBatchDisable,
+  onBatchDownload,
   onArchive,
   onBatchDelete,
   onEditMetadata,
@@ -103,6 +105,16 @@ const BatchAction: FC<IBatchActionProps> = ({
             <span className="px-0.5">{t(`${i18nPrefix}.reIndex`, { ns: 'dataset' })}</span>
           </Button>
         )}
+        {onBatchDownload && (
+          <Button
+            variant="ghost"
+            className="gap-x-0.5 px-3"
+            onClick={onBatchDownload}
+          >
+            <RiDownload2Line className="size-4" />
+            <span className="px-0.5">{t(`${i18nPrefix}.download`, { ns: 'dataset' })}</span>
+          </Button>
+        )}
         <Button
           variant="ghost"
           destructive