import type { FC } from 'react'
import React, { useCallback, useMemo, useState } from 'react'
import useSWR from 'swr'
import { useRouter } from 'next/navigation'
import { useContext } from 'use-context-selector'
import { useTranslation } from 'react-i18next'
import { omit } from 'lodash-es'
import cn from 'classnames'
import { ArrowRightIcon } from '@heroicons/react/24/solid'
import SegmentCard from '../completed/SegmentCard'
import { FieldInfo } from '../metadata'
import style from '../completed/style.module.css'
import { DocumentContext } from '../index'
import s from './style.module.css'
import Button from '@/app/components/base/button'
import Divider from '@/app/components/base/divider'
import { ToastContext } from '@/app/components/base/toast'
import type { FullDocumentDetail, ProcessRuleResponse } from '@/models/datasets'
import type { CommonResponse } from '@/models/common'
import { asyncRunSafe } from '@/utils'
import { formatNumber } from '@/utils/format'
import { fetchIndexingEstimate, fetchProcessRule, pauseDocIndexing, resumeDocIndexing } from '@/service/datasets'
import DatasetDetailContext from '@/context/dataset-detail'
import StopEmbeddingModal from '@/app/components/datasets/create/stop-embedding-modal'
type Props = {
detail?: FullDocumentDetail
stopPosition?: 'top' | 'bottom'
datasetId?: string
documentId?: string
indexingType?: string
}
const StopIcon: FC<{ className?: string }> = ({ className }) => {
return
}
const ResumeIcon: FC<{ className?: string }> = ({ className }) => {
return
}
const RuleDetail: FC<{ sourceData?: ProcessRuleResponse; docName?: string }> = ({ sourceData, docName }) => {
const { t } = useTranslation()
const segmentationRuleMap = {
docName: t('datasetDocuments.embedding.docName'),
mode: t('datasetDocuments.embedding.mode'),
segmentLength: t('datasetDocuments.embedding.segmentLength'),
textCleaning: t('datasetDocuments.embedding.textCleaning'),
}
const getRuleName = (key: string) => {
if (key === 'remove_extra_spaces')
return t('datasetCreation.stepTwo.removeExtraSpaces')
if (key === 'remove_urls_emails')
return t('datasetCreation.stepTwo.removeUrlEmails')
if (key === 'remove_stopwords')
return t('datasetCreation.stepTwo.removeStopwords')
}
const getValue = useCallback((field: string) => {
let value: string | number | undefined = '-'
switch (field) {
case 'docName':
value = docName
break
case 'mode':
value = sourceData?.mode === 'automatic' ? (t('datasetDocuments.embedding.automatic') as string) : (t('datasetDocuments.embedding.custom') as string)
break
case 'segmentLength':
value = sourceData?.rules?.segmentation?.max_tokens
break
default:
value = sourceData?.mode === 'automatic'
? (t('datasetDocuments.embedding.automatic') as string)
// eslint-disable-next-line array-callback-return
: sourceData?.rules?.pre_processing_rules?.map((rule) => {
if (rule.enabled)
return getRuleName(rule.id)
}).filter(Boolean).join(';')
break
}
return value
}, [sourceData, docName])
return
{Object.keys(segmentationRuleMap).map((field) => {
return
})}
}
const EmbeddingDetail: FC = ({ detail, stopPosition = 'top', datasetId: dstId, documentId: docId, indexingType }) => {
const onTop = stopPosition === 'top'
const { t } = useTranslation()
const { notify } = useContext(ToastContext)
const { datasetId = '', documentId = '' } = useContext(DocumentContext)
const { indexingTechnique } = useContext(DatasetDetailContext)
const localDatasetId = dstId ?? datasetId
const localDocumentId = docId ?? documentId
const localIndexingTechnique = indexingType ?? indexingTechnique
// const { data: indexingStatusDetailFromApi, error: indexingStatusErr, mutate: statusMutate } = useSWR({
// action: 'fetchIndexingStatus',
// datasetId: localDatasetId,
// documentId: localDocumentId,
// }, apiParams => fetchIndexingStatus(omit(apiParams, 'action')), {
// refreshInterval: 2500,
// revalidateOnFocus: false,
// })
const [indexingStatusDetail, setIndexingStatusDetail, getIndexingStatusDetail] = useGetState(null)
const fetchIndexingStatus = async () => {
const status = await doFetchIndexingStatus({ datasetId: localDatasetId, documentId: localDocumentId })
setIndexingStatusDetail(status)
}
const [runId, setRunId, getRunId] = useGetState(null)
const startQueryStatus = () => {
const runId = setInterval(() => {
const indexingStatusDetail = getIndexingStatusDetail()
if (indexingStatusDetail?.indexing_status === 'completed') {
// eslint-disable-next-line @typescript-eslint/no-use-before-define
stopQueryStatus()
return
}
fetchIndexingStatus()
}, 2500)
setRunId(runId)
}
const stopQueryStatus = () => {
clearInterval(getRunId())
}
useEffect(() => {
fetchIndexingStatus()
startQueryStatus()
return () => {
stopQueryStatus()
}
}, [])
const { data: indexingEstimateDetail, error: indexingEstimateErr } = useSWR({
action: 'fetchIndexingEstimate',
datasetId: localDatasetId,
documentId: localDocumentId,
}, apiParams => fetchIndexingEstimate(omit(apiParams, 'action')), {
revalidateOnFocus: false,
})
const { data: ruleDetail, error: ruleError } = useSWR({
action: 'fetchProcessRule',
params: { documentId: localDocumentId },
}, apiParams => fetchProcessRule(omit(apiParams, 'action')), {
revalidateOnFocus: false,
})
const [showModal, setShowModal] = useState(false)
const modalShowHandle = () => setShowModal(true)
const modalCloseHandle = () => setShowModal(false)
const router = useRouter()
const navToDocument = () => {
router.push(`/datasets/${localDatasetId}/documents/${localDocumentId}`)
}
const isEmbedding = useMemo(() => ['indexing', 'splitting', 'parsing', 'cleaning'].includes(indexingStatusDetail?.indexing_status || ''), [indexingStatusDetail])
const isEmbeddingCompleted = useMemo(() => ['completed'].includes(indexingStatusDetail?.indexing_status || ''), [indexingStatusDetail])
const isEmbeddingPaused = useMemo(() => ['paused'].includes(indexingStatusDetail?.indexing_status || ''), [indexingStatusDetail])
const isEmbeddingError = useMemo(() => ['error'].includes(indexingStatusDetail?.indexing_status || ''), [indexingStatusDetail])
const percent = useMemo(() => {
const completedCount = indexingStatusDetail?.completed_segments || 0
const totalCount = indexingStatusDetail?.total_segments || 0
if (totalCount === 0)
return 0
const percent = Math.round(completedCount * 100 / totalCount)
return percent > 100 ? 100 : percent
}, [indexingStatusDetail])
const handleSwitch = async () => {
const opApi = isEmbedding ? pauseDocIndexing : resumeDocIndexing
const [e] = await asyncRunSafe(opApi({ datasetId: localDatasetId, documentId: localDocumentId }) as Promise)
if (!e) {
notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
setIndexingStatusDetail(null)
}
else {
notify({ type: 'error', message: t('common.actionMsg.modificationFailed') })
}
}
// if (!ruleDetail && !error)
// return
return (
<>
{isEmbedding && t('datasetDocuments.embedding.processing')}
{isEmbeddingCompleted && t('datasetDocuments.embedding.completed')}
{isEmbeddingPaused && t('datasetDocuments.embedding.paused')}
{isEmbeddingError && t('datasetDocuments.embedding.error')}
{onTop && isEmbedding && (
)}
{onTop && isEmbeddingPaused && (
)}
{/* progress bar */}
{new Array(10).fill('').map((_, idx) =>
)}
{t('datasetDocuments.embedding.segments')} {indexingStatusDetail?.completed_segments}/{indexingStatusDetail?.total_segments} · {percent}%
{localIndexingTechnique === 'high_quaility' && (
{t('datasetDocuments.embedding.highQuality')} · {t('datasetDocuments.embedding.estimate')}
{formatNumber(indexingEstimateDetail?.tokens || 0)}tokens
(
${formatNumber(indexingEstimateDetail?.total_price || 0)})
)}
{localIndexingTechnique === 'economy' && (
{t('datasetDocuments.embedding.economy')} · {t('datasetDocuments.embedding.estimate')}
0tokens
)}
{!onTop && (
{isEmbedding && (
)}
{isEmbeddingPaused && (
)}
)}
{onTop && <>
{t('datasetDocuments.embedding.previewTip')}
{[1, 2, 3].map(v => (
))}
>}
>
)
}
export default EmbeddingDetail