Merge branch 'feat/parent-child-retrieval' of https://github.com/langgenius/dify into feat/parent-child-retrieval

web/app/components/datasets/documents/detail/completed/child-segment-detail.tsx

@@ -1,4 +1,4 @@
-import React, { type FC, useState } from 'react'
+import React, { type FC, useMemo, useState } from 'react'
 import { useTranslation } from 'react-i18next'
 import {
   RiCloseLine,
@@ -56,18 +56,33 @@ const ChildSegmentDetail: FC<IChildSegmentDetailProps> = ({
     onUpdate(chunkId, childChunkInfo?.id || '', content)
   }
 
+  const wordCountText = useMemo(() => {
+    const count = content.length
+    return `${formatNumber(count)} ${t('datasetDocuments.segment.characters', { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [content.length])
+
+  const EditTimeText = useMemo(() => {
+    const timeText = formatTime({
+      date: (childChunkInfo?.updated_at ?? 0) * 1000,
+      dateFormat: 'MM/DD/YYYY h:mm:ss',
+    })
+    return `${t('datasetDocuments.segment.editedAt')} ${timeText}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [childChunkInfo?.updated_at])
+
   return (
     <div className={'flex flex-col h-full'}>
       <div className={classNames('flex items-center justify-between', fullScreen ? 'py-3 pr-4 pl-6 border border-divider-subtle' : 'pt-3 pr-3 pl-4')}>
         <div className='flex flex-col'>
-          <div className='text-text-primary system-xl-semibold'>{'Edit Child Chunk'}</div>
+          <div className='text-text-primary system-xl-semibold'>{t('datasetDocuments.segment.editChildChunk')}</div>
           <div className='flex items-center gap-x-2'>
-            <SegmentIndexTag positionId={childChunkInfo?.position || ''} labelPrefix='Child-Chunk' />
+            <SegmentIndexTag positionId={childChunkInfo?.position || ''} labelPrefix={t('datasetDocuments.segment.childChunk') as string} />
             <Dot />
-            <span className='text-text-tertiary system-xs-medium'>{formatNumber(content.length)} {t('datasetDocuments.segment.characters')}</span>
+            <span className='text-text-tertiary system-xs-medium'>{wordCountText}</span>
             <Dot />
             <span className='text-text-tertiary system-xs-medium'>
-              {`Edited at ${formatTime({ date: (childChunkInfo?.created_at ?? 0) * 1000, dateFormat: 'MM/DD/YYYY h:mm:ss' })}`}
+              {EditTimeText}
             </span>
           </div>
         </div>

web/app/components/datasets/documents/detail/completed/child-segment-list.tsx

@@ -1,11 +1,13 @@
 import { type FC, useMemo, useState } from 'react'
 import { RiArrowDownSLine, RiArrowRightSLine } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
 import { EditSlice } from '../../../formatted-text/flavours/edit-slice'
 import { useDocumentContext } from '../index'
 import type { ChildChunkDetail } from '@/models/datasets'
 import Input from '@/app/components/base/input'
 import classNames from '@/utils/classnames'
 import Divider from '@/app/components/base/divider'
+import { formatNumber } from '@/utils/format'
 
 type IChildSegmentCardProps = {
   childChunks: ChildChunkDetail[]
@@ -30,6 +32,7 @@ const ChildSegmentList: FC<IChildSegmentCardProps> = ({
   total,
   inputValue,
 }) => {
+  const { t } = useTranslation()
   const parentMode = useDocumentContext(s => s.parentMode)
 
   const [collapsed, setCollapsed] = useState(true)
@@ -50,10 +53,25 @@ const ChildSegmentList: FC<IChildSegmentCardProps> = ({
     return enabled ? '' : 'opacity-50 group-hover/card:opacity-100'
   }, [enabled])
 
+  const totalText = useMemo(() => {
+    const text = isFullDocMode
+      ? !total
+        ? '--'
+        : formatNumber(total)
+      : formatNumber(childChunks.length)
+    const count = isFullDocMode
+      ? text === '--'
+        ? 0
+        : total
+      : childChunks.length
+    return `${isFullDocMode ? count : childChunks.length} ${t('datasetDocuments.segment.childChunks', { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [isFullDocMode, total, childChunks.length])
+
   return (
-    <div className={classNames('flex flex-col', contentOpacity, isParagraphMode ? 'p-1 pb-2' : 'px-3 grow overflow-y-hidden')}>
+    <div className={classNames('flex flex-col', contentOpacity, isParagraphMode ? 'p-1 pb-2' : 'px-3 grow')}>
       {isFullDocMode ? <Divider type='horizontal' className='h-[1px] bg-divider-subtle my-1' /> : null}
-      <div className={classNames('flex items-center justify-between', isFullDocMode ? 'pt-2 pb-3' : '')}>
+      <div className={classNames('flex items-center justify-between', isFullDocMode ? 'pt-2 pb-3 sticky top-0 left-0 bg-components-panel-bg' : '')}>
         <div className={classNames('h-7 flex items-center pl-1 pr-3 rounded-lg', (isParagraphMode && collapsed) ? 'bg-dataset-child-chunk-expand-btn-bg' : '')} onClick={(event) => {
           event.stopPropagation()
           toggleCollapse()
@@ -67,16 +85,16 @@ const ChildSegmentList: FC<IChildSegmentCardProps> = ({
                 : (<RiArrowDownSLine className='w-4 h-4 text-text-secondary mr-0.5' />)
               : null
           }
-          <span className='text-text-secondary system-sm-semibold-uppercase'>{`${total} CHILD CHUNKS`}</span>
+          <span className='text-text-secondary system-sm-semibold-uppercase'>{totalText}</span>
           <span className={classNames('text-text-quaternary text-xs font-medium pl-1.5', isParagraphMode ? 'hidden group-hover/card:inline-block' : '')}>·</span>
           <button
-            className={classNames('px-1.5 py-1 text-components-button-secondary-accent-text system-xs-semibold', isParagraphMode ? 'hidden group-hover/card:inline-block' : '')}
+            className={classNames('px-1.5 py-1 text-components-button-secondary-accent-text system-xs-semibold-uppercase', isParagraphMode ? 'hidden group-hover/card:inline-block' : '')}
             onClick={(event) => {
               event.stopPropagation()
               handleAddNewChildChunk?.(parentChunkId)
             }}
           >
-            ADD
+            {t('common.operation.add')}
           </button>
         </div>
         {isFullDocMode
@@ -91,14 +109,14 @@ const ChildSegmentList: FC<IChildSegmentCardProps> = ({
           : null}
       </div>
       {(isFullDocMode || !collapsed)
-        ? <div className={classNames('flex gap-x-0.5', isFullDocMode ? 'grow overflow-y-auto' : '')}>
+        ? <div className={classNames('flex gap-x-0.5', isFullDocMode ? 'grow' : '')}>
           {isParagraphMode && <Divider type='vertical' className='h-auto w-[2px] mx-[7px] bg-text-accent-secondary' />}
           <div className={classNames('w-full !leading-5 flex flex-col', isParagraphMode ? 'gap-y-2' : 'gap-y-3')}>
             {childChunks.map((childChunk) => {
               const edited = childChunk.updated_at !== childChunk.created_at
               return <EditSlice
                 key={childChunk.id}
-                label={`C-${childChunk.position}${edited ? ' · EDITED' : ''}`}
+                label={`C-${childChunk.position}${edited ? ` · ${t('datasetDocuments.segment.edited')}` : ''}`}
                 text={childChunk.content}
                 onDelete={() => onDelete?.(childChunk.segment_id, childChunk.id)}
                 className='line-clamp-3'

web/app/components/datasets/documents/detail/completed/display-toggle.tsx

@@ -1,4 +1,5 @@
 import React, { type FC } from 'react'
+import { useTranslation } from 'react-i18next'
 import { RiLineHeight } from '@remixicon/react'
 import Tooltip from '@/app/components/base/tooltip'
 import { Collapse } from '@/app/components/base/icons/src/public/knowledge'
@@ -12,9 +13,11 @@ const DisplayToggle: FC<DisplayToggleProps> = ({
   isCollapsed,
   toggleCollapsed,
 }) => {
+  const { t } = useTranslation()
+
   return (
     <Tooltip
-      popupContent={isCollapsed ? 'Expand chunks' : 'Collapse chunks'}
+      popupContent={isCollapsed ? t('datasetDocuments.segment.expandChunks') : t('datasetDocuments.segment.collapseChunks')}
       popupClassName='text-text-secondary system-xs-medium border-[0.5px] border-components-panel-border'
     >
       <button

web/app/components/datasets/documents/detail/completed/index.tsx

@@ -328,8 +328,14 @@ const Completed: FC<ICompletedProps> = ({
   }, [segments, isAllSelected, selectedSegmentIds])
 
   const totalText = useMemo(() => {
-    return segmentListData?.total ? formatNumber(segmentListData.total) : '--'
-  }, [segmentListData?.total])
+    const total = segmentListData?.total ? formatNumber(segmentListData.total) : '--'
+    const count = total === '--' ? 0 : segmentListData!.total
+    const translationKey = (mode === 'hierarchical' && parentMode === 'paragraph')
+      ? 'datasetDocuments.segment.parentChunks'
+      : 'datasetDocuments.segment.chunks'
+    return `${total} ${t(translationKey, { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [segmentListData?.total, mode, parentMode])
 
   const toggleFullScreen = useCallback(() => {
     setFullScreen(!fullScreen)
@@ -480,7 +486,7 @@ const Completed: FC<ICompletedProps> = ({
           mixed={!isAllSelected && isSomeSelected}
           onCheck={onSelectedAll}
         />
-        <div className={cn('system-sm-semibold-uppercase pl-5', s.totalText)}>{totalText} {t('datasetDocuments.segment.chunks')}</div>
+        <div className={cn('system-sm-semibold-uppercase pl-5', s.totalText)}>{totalText}</div>
         <SimpleSelect
           onSelect={onChangeStatus}
           items={[
@@ -505,7 +511,7 @@ const Completed: FC<ICompletedProps> = ({
       {/* Segment list */}
       {
         isFullDocMode
-          ? <>
+          ? <div className='grow relative overflow-x-hidden overflow-y-auto'>
             <SegmentCard
               detail={segments[0]}
               onClick={() => onClickCard(segments[0])}
@@ -522,7 +528,7 @@ const Completed: FC<ICompletedProps> = ({
               total={childChunkListData?.total || 0}
               inputValue={inputValue}
             />
-          </>
+          </div>
           : <SegmentList
             ref={segmentListRef}
             embeddingAvailable={embeddingAvailable}

web/app/components/datasets/documents/detail/completed/new-child-segment.tsx

@@ -102,17 +102,21 @@ const NewChildSegmentModal: FC<NewChildSegmentModalProps> = ({
     }
   }
 
+  const wordCountText = useMemo(() => {
+    const count = content.length
+    return `${formatNumber(count)} ${t('datasetDocuments.segment.characters', { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [content.length])
+
   return (
     <div className={'flex flex-col h-full'}>
       <div className={classNames('flex items-center justify-between', fullScreen ? 'py-3 pr-4 pl-6 border border-divider-subtle' : 'pt-3 pr-3 pl-4')}>
         <div className='flex flex-col'>
-          <div className='text-text-primary system-xl-semibold'>{
-            t('datasetDocuments.segment.addChildChunk')
-          }</div>
+          <div className='text-text-primary system-xl-semibold'>{t('datasetDocuments.segment.addChildChunk')}</div>
           <div className='flex items-center gap-x-2'>
-            <SegmentIndexTag label={'New Child Chunk'} />
+            <SegmentIndexTag label={t('datasetDocuments.segment.newChildChunk') as string} />
             <Dot />
-            <span className='text-text-tertiary system-xs-medium'>{formatNumber(content.length)} {t('datasetDocuments.segment.characters')}</span>
+            <span className='text-text-tertiary system-xs-medium'>{wordCountText}</span>
           </div>
         </div>
         <div className='flex items-center'>

web/app/components/datasets/documents/detail/completed/segment-card.tsx

@@ -111,6 +111,17 @@ const SegmentCard: FC<ISegmentCardProps> = ({
     return content
   }
 
+  const wordCountText = useMemo(() => {
+    const total = formatNumber(word_count)
+    return `${total} ${t('datasetDocuments.segment.characters', { count: word_count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [word_count])
+
+  const labelPrefix = useMemo(() => {
+    return isParentChildMode ? t('datasetDocuments.segment.parentChunk') : t('datasetDocuments.segment.chunk')
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [isParentChildMode])
+
   return (
     <div
       className={cn('w-full px-3 rounded-xl group/card', isFullDocMode ? '' : 'pt-2.5 pb-2 hover:bg-dataset-chunk-detail-card-hover-bg', className)}
@@ -119,15 +130,15 @@ const SegmentCard: FC<ISegmentCardProps> = ({
       <div className='h-5 relative flex items-center justify-between'>
         <>
           <div className='flex items-center gap-x-2'>
-            <SegmentIndexTag positionId={position} className={textOpacity} labelPrefix={`${isParentChildMode ? 'Parent-' : ''}Chunk`} />
+            <SegmentIndexTag positionId={position} className={textOpacity} labelPrefix={labelPrefix} />
             <Dot />
-            <div className={cn('text-text-tertiary system-xs-medium', textOpacity)}>{`${formatNumber(word_count)} Characters`}</div>
+            <div className={cn('text-text-tertiary system-xs-medium', textOpacity)}>{wordCountText}</div>
             <Dot />
-            <div className={cn('text-text-tertiary system-xs-medium', textOpacity)}>{`${formatNumber(hit_count)} Retrieval Count`}</div>
+            <div className={cn('text-text-tertiary system-xs-medium', textOpacity)}>{`${formatNumber(hit_count)} ${t('datasetDocuments.segment.hitCount')}`}</div>
             {chunkEdited && (
               <>
                 <Dot />
-                <Badge text='edited' uppercase className={textOpacity} />
+                <Badge text={t('datasetDocuments.segment.edited') as string} uppercase className={textOpacity} />
               </>
             )}
           </div>

web/app/components/datasets/documents/detail/completed/segment-detail.tsx

@@ -80,15 +80,32 @@ const SegmentDetail: FC<ISegmentDetailProps> = ({
     return mode === 'hierarchical'
   }, [mode])
 
+  const titleText = useMemo(() => {
+    return isEditMode ? t('datasetDocuments.segment.editChunk') : t('datasetDocuments.segment.chunkDetail')
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [isEditMode])
+
+  const wordCountText = useMemo(() => {
+    const total = formatNumber(isEditMode ? question.length : segInfo!.word_count as number)
+    const count = isEditMode ? question.length : segInfo!.word_count as number
+    return `${total} ${t('datasetDocuments.segment.characters', { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [isEditMode, question.length, segInfo?.word_count])
+
+  const labelPrefix = useMemo(() => {
+    return isParentChildMode ? t('datasetDocuments.segment.parentChunk') : t('datasetDocuments.segment.chunk')
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [isParentChildMode])
+
   return (
     <div className={'flex flex-col h-full'}>
       <div className={classNames('flex items-center justify-between', fullScreen ? 'py-3 pr-4 pl-6 border border-divider-subtle' : 'pt-3 pr-3 pl-4')}>
         <div className='flex flex-col'>
-          <div className='text-text-primary system-xl-semibold'>{isEditMode ? 'Edit Chunk' : 'Chunk Detail'}</div>
+          <div className='text-text-primary system-xl-semibold'>{titleText}</div>
           <div className='flex items-center gap-x-2'>
-            <SegmentIndexTag positionId={segInfo?.position || ''} labelPrefix={`${isParentChildMode ? 'Parent-' : ''}Chunk`} />
+            <SegmentIndexTag positionId={segInfo?.position || ''} labelPrefix={labelPrefix} />
             <Dot />
-            <span className='text-text-tertiary system-xs-medium'>{formatNumber(isEditMode ? question.length : segInfo?.word_count as number)} {t('datasetDocuments.segment.characters')}</span>
+            <span className='text-text-tertiary system-xs-medium'>{wordCountText}</span>
           </div>
         </div>
         <div className='flex items-center'>

web/app/components/datasets/documents/detail/new-segment.tsx

@@ -1,4 +1,4 @@
-import { memo, useRef, useState } from 'react'
+import { memo, useMemo, useRef, useState } from 'react'
 import type { FC } from 'react'
 import { useTranslation } from 'react-i18next'
 import { useContext } from 'use-context-selector'
@@ -108,6 +108,12 @@ const NewSegmentModal: FC<NewSegmentModalProps> = ({
     }
   }
 
+  const wordCountText = useMemo(() => {
+    const count = question.length
+    return `${formatNumber(count)} ${t('datasetDocuments.segment.characters', { count })}`
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [question.length])
+
   return (
     <div className={'flex flex-col h-full'}>
       <div className={classNames('flex items-center justify-between', fullScreen ? 'py-3 pr-4 pl-6 border border-divider-subtle' : 'pt-3 pr-3 pl-4')}>
@@ -120,7 +126,7 @@ const NewSegmentModal: FC<NewSegmentModalProps> = ({
           <div className='flex items-center gap-x-2'>
             <SegmentIndexTag label={'New Chunk'} />
             <Dot />
-            <span className='text-text-tertiary system-xs-medium'>{formatNumber(question.length)} {t('datasetDocuments.segment.characters')}</span>
+            <span className='text-text-tertiary system-xs-medium'>{wordCountText}</span>
           </div>
         </div>
         <div className='flex items-center'>

web/app/components/datasets/hit-testing/components/result-item.tsx

@@ -49,7 +49,7 @@ const ResultItem: FC<Props> = ({
             className={cn('w-fit group-hover:opacity-100')}
           />
           <Dot />
-          <div className='system-xs-medium text-text-tertiary'>{word_count} {t('datasetDocuments.segment.characters')}</div>
+          <div className='system-xs-medium text-text-tertiary'>{word_count} {t('datasetDocuments.segment.characters', { count: word_count })}</div>
         </div>
         <Score value={score} />
       </div>

web/app/components/datasets/hit-testing/hit-detail.tsx

@@ -42,7 +42,7 @@ const HitDetail: FC<IHitDetailProps> = ({ segInfo }) => {
           />
           <div className={cn(s.commonIcon, s.typeSquareIcon)} />
           <span className={cn('mr-6', s.numberInfo)}>
-            {segInfo?.word_count} {t('datasetDocuments.segment.characters')}
+            {segInfo?.word_count} {t('datasetDocuments.segment.characters', { count: segInfo?.word_count || 0 })}
           </span>
           <div className={cn(s.commonIcon, s.targetIcon)} />
           <span className={s.numberInfo}>

web/i18n/en-US/dataset-documents.ts

@@ -333,11 +333,21 @@ const translation = {
   },
   segment: {
     paragraphs: 'Paragraphs',
-    chunks: 'CHUNKS',
+    chunks_one: 'CHUNK',
+    chunks_other: 'CHUNKS',
+    parentChunks_one: 'PARENT CHUNK',
+    parentChunks_other: 'PARENT CHUNKS',
+    childChunks_one: 'CHILD CHUNK',
+    childChunks_other: 'CHILD CHUNKS',
+    chunk: 'Chunk',
+    parentChunk: 'Parent-Chunk',
+    childChunk: 'Child-Chunk',
+    newChildChunk: 'New Child Chunk',
     keywords: 'KEYWORDS',
     addKeyWord: 'Add keyword',
     keywordError: 'The maximum length of keyword is 20',
-    characters: 'characters',
+    characters_one: 'character',
+    characters_other: 'characters',
     hitCount: 'Retrieval count',
     vectorHash: 'Vector hash: ',
     questionPlaceholder: 'Add question here',
@@ -354,12 +364,20 @@ const translation = {
     delete: 'Delete this chunk ?',
     chunkAdded: '1 chunk added',
     childChunkAdded: '1 child chunk added',
+    editChunk: 'Edit Chunk',
+    editParentChunk: 'Edit Parent Chunk',
+    editChildChunk: 'Edit Child Chunk',
+    chunkDetail: 'Chunk Detail',
     regenerationConfirmTitle: 'Do you want to regenerate child chunks?',
     regenerationConfirmMessage: 'Regenerating child chunks will overwrite the current child chunks, including edited chunks and newly added chunks. The regeneration cannot be undone.',
     regeneratingTitle: 'Regenerating child chunks',
     regeneratingMessage: 'This may take a moment, please wait...',
     regenerationSuccessTitle: 'Regeneration completed',
     regenerationSuccessMessage: 'You can close this window.',
+    edited: 'EDITED',
+    editedAt: 'Edited at',
+    expandChunks: 'Expand chunks',
+    collapseChunks: 'Collapse chunks',
   },
 }
 

web/i18n/zh-Hans/dataset-documents.ts

@@ -331,11 +331,21 @@ const translation = {
   },
   segment: {
     paragraphs: '段落',
-    chunks: '段落',
+    chunks_one: '分段',
+    chunks_other: '分段',
+    parentChunks_one: '父分段',
+    parentChunks_other: '父分段',
+    childChunks_one: '子分段',
+    childChunks_other: '子分段',
+    chunk: '分段',
+    parentChunk: '父分段',
+    childChunk: '子分段',
+    newChildChunk: '新子分段',
     keywords: '关键词',
     addKeyWord: '添加关键词',
     keywordError: '关键词最大长度为 20',
-    characters: '字符',
+    characters_one: '字符',
+    characters_other: '字符',
     hitCount: '召回次数',
     vectorHash: '向量哈希：',
     questionPlaceholder: '在这里添加问题',
@@ -348,16 +358,24 @@ const translation = {
     newQaSegment: '新问答分段',
     addChunk: '新增分段',
     addChildChunk: '新增子分段',
-    addAnother: '连续新增',
+    addAnother: '继续新增',
     delete: '删除这个分段？',
     chunkAdded: '新增一个分段',
     childChunkAdded: '新增一个子分段',
+    editChunk: '编辑分段',
+    editParentChunk: '编辑父分段',
+    editChildChunk: '编辑子分段',
+    chunkDetail: '分段详情',
     regenerationConfirmTitle: '是否需要重新生成子分段？',
     regenerationConfirmMessage: '重新生成的子分段将会覆盖当前的子分段，包括编辑过的分段和新添加的分段。重新生成操作无法撤销。',
     regeneratingTitle: '正在生成子分段',
     regeneratingMessage: '生成子分段需要一些时间，请耐心等待...',
     regenerationSuccessTitle: '子分段已重新生成',
     regenerationSuccessMessage: '可以关闭窗口',
+    edited: '已编辑',
+    editedAt: '编辑于',
+    expandChunks: '展开分段',
+    collapseChunks: '折叠分段',
   },
 }