Feat: Support re-segmentation (#114)

Co-authored-by: John Wang <takatost@gmail.com>
Co-authored-by: Jyong <718720800@qq.com>
Co-authored-by: 金伟强 <iamjoel007@gmail.com>
This commit is contained in:
KVOJJJin
2023-06-01 23:19:36 +08:00
committed by GitHub
parent f65a3ad1cc
commit c67f626b66
61 changed files with 1166 additions and 759 deletions

View File

@@ -19,7 +19,7 @@ import type { FullDocumentDetail, ProcessRuleResponse } from '@/models/datasets'
import type { CommonResponse } from '@/models/common'
import { asyncRunSafe } from '@/utils'
import { formatNumber } from '@/utils/format'
import { fetchIndexingEstimate, fetchIndexingStatus, fetchProcessRule, pauseDocIndexing, resumeDocIndexing } from '@/service/datasets'
import { fetchIndexingEstimate, fetchProcessRule, pauseDocIndexing, resumeDocIndexing } from '@/service/datasets'
import DatasetDetailContext from '@/context/dataset-detail'
import StopEmbeddingModal from '@/app/components/datasets/create/stop-embedding-modal'
@@ -118,14 +118,45 @@ const EmbeddingDetail: FC<Props> = ({ detail, stopPosition = 'top', datasetId: d
const localDocumentId = docId ?? documentId
const localIndexingTechnique = indexingType ?? indexingTechnique
const { data: indexingStatusDetail, error: indexingStatusErr, mutate: statusMutate } = useSWR({
action: 'fetchIndexingStatus',
datasetId: localDatasetId,
documentId: localDocumentId,
}, apiParams => fetchIndexingStatus(omit(apiParams, 'action')), {
refreshInterval: 5000,
revalidateOnFocus: false,
})
// 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<any>(null)
const fetchIndexingStatus = async () => {
const status = await doFetchIndexingStatus({ datasetId: localDatasetId, documentId: localDocumentId })
setIndexingStatusDetail(status)
}
const [runId, setRunId, getRunId] = useGetState<any>(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',
@@ -168,7 +199,7 @@ const EmbeddingDetail: FC<Props> = ({ detail, stopPosition = 'top', datasetId: d
const [e] = await asyncRunSafe<CommonResponse>(opApi({ datasetId: localDatasetId, documentId: localDocumentId }) as Promise<CommonResponse>)
if (!e) {
notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
statusMutate()
setIndexingStatusDetail(null)
}
else {
notify({ type: 'error', message: t('common.actionMsg.modificationFailed') })

View File

@@ -0,0 +1,90 @@
'use client'
import React, { useState, useCallback, useEffect } from 'react'
import { useTranslation } from 'react-i18next'
import { useBoolean } from 'ahooks'
import { useContext } from 'use-context-selector'
import { useRouter } from 'next/navigation'
import DatasetDetailContext from '@/context/dataset-detail'
import type { FullDocumentDetail } from '@/models/datasets'
import { fetchTenantInfo } from '@/service/common'
import { fetchDocumentDetail, MetadataType } from '@/service/datasets'
import Loading from '@/app/components/base/loading'
import StepTwo from '@/app/components/datasets/create/step-two'
import AccountSetting from '@/app/components/header/account-setting'
import AppUnavailable from '@/app/components/base/app-unavailable'
type DocumentSettingsProps = {
datasetId: string;
documentId: string;
}
const DocumentSettings = ({ datasetId, documentId }: DocumentSettingsProps) => {
const { t } = useTranslation()
const router = useRouter()
const [hasSetAPIKEY, setHasSetAPIKEY] = useState(true)
const [isShowSetAPIKey, { setTrue: showSetAPIKey, setFalse: hideSetAPIkey }] = useBoolean()
const [hasError, setHasError] = useState(false)
const { indexingTechnique, dataset } = useContext(DatasetDetailContext)
const saveHandler = () => router.push(`/datasets/${datasetId}/documents/${documentId}`)
const cancelHandler = () => router.back()
const checkAPIKey = async () => {
const data = await fetchTenantInfo({ url: '/info' })
const hasSetKey = data.providers.some(({ is_valid }) => is_valid)
setHasSetAPIKEY(hasSetKey)
}
useEffect(() => {
checkAPIKey()
}, [])
const [documentDetail, setDocumentDetail] = useState<FullDocumentDetail | null>(null)
useEffect(() => {
(async () => {
try {
const detail = await fetchDocumentDetail({
datasetId,
documentId,
params: { metadata: 'without' as MetadataType }
})
setDocumentDetail(detail)
} catch (e) {
setHasError(true)
}
})()
}, [datasetId, documentId])
if (hasError) {
return <AppUnavailable code={500} unknownReason={t('datasetCreation.error.unavailable') as string} />
}
return (
<div className='flex' style={{ height: 'calc(100vh - 56px)' }}>
<div className="grow bg-white">
{!documentDetail && <Loading type='app' />}
{dataset && documentDetail && (
<StepTwo
hasSetAPIKEY={hasSetAPIKEY}
onSetting={showSetAPIKey}
datasetId={datasetId}
indexingType={indexingTechnique || ''}
isSetting
documentDetail={documentDetail}
file={documentDetail.data_source_info.upload_file}
onSave={saveHandler}
onCancel={cancelHandler}
/>
)}
</div>
{isShowSetAPIKey && <AccountSetting activeTab="provider" onCancel={async () => {
await checkAPIKey()
hideSetAPIkey()
}} />}
</div>
)
}
export default DocumentSettings