index.tsx 37 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922
  1. 'use client'
  2. import React, { useEffect, useLayoutEffect, useRef, useState } from 'react'
  3. import { useTranslation } from 'react-i18next'
  4. import { useContext } from 'use-context-selector'
  5. import { useBoolean } from 'ahooks'
  6. import { XMarkIcon } from '@heroicons/react/20/solid'
  7. import { RocketLaunchIcon } from '@heroicons/react/24/outline'
  8. import cn from 'classnames'
  9. import Link from 'next/link'
  10. import { groupBy } from 'lodash-es'
  11. import RetrievalMethodInfo from '../../common/retrieval-method-info'
  12. import PreviewItem, { PreviewType } from './preview-item'
  13. import LanguageSelect from './language-select'
  14. import s from './index.module.css'
  15. import type { CreateDocumentReq, CustomFile, FileIndexingEstimateResponse, FullDocumentDetail, IndexingEstimateParams, IndexingEstimateResponse, NotionInfo, PreProcessingRule, ProcessRule, Rules, createDocumentResponse } from '@/models/datasets'
  16. import {
  17. createDocument,
  18. createFirstDocument,
  19. fetchFileIndexingEstimate as didFetchFileIndexingEstimate,
  20. fetchDefaultProcessRule,
  21. } from '@/service/datasets'
  22. import Button from '@/app/components/base/button'
  23. import Loading from '@/app/components/base/loading'
  24. import FloatRightContainer from '@/app/components/base/float-right-container'
  25. import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
  26. import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
  27. import { type RetrievalConfig } from '@/types/app'
  28. import { ensureRerankModelSelected, isReRankModelSelected } from '@/app/components/datasets/common/check-rerank-model'
  29. import Toast from '@/app/components/base/toast'
  30. import { formatNumber } from '@/utils/format'
  31. import type { NotionPage } from '@/models/common'
  32. import { DataSourceType, DocForm } from '@/models/datasets'
  33. import NotionIcon from '@/app/components/base/notion-icon'
  34. import Switch from '@/app/components/base/switch'
  35. import { MessageChatSquare } from '@/app/components/base/icons/src/public/common'
  36. import { HelpCircle, XClose } from '@/app/components/base/icons/src/vender/line/general'
  37. import { useDatasetDetailContext } from '@/context/dataset-detail'
  38. import I18n from '@/context/i18n'
  39. import { IS_CE_EDITION } from '@/config'
  40. import { RETRIEVE_METHOD } from '@/types/app'
  41. import useBreakpoints, { MediaType } from '@/hooks/use-breakpoints'
  42. import Tooltip from '@/app/components/base/tooltip'
  43. import TooltipPlus from '@/app/components/base/tooltip-plus'
  44. import { useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
  45. import { LanguagesSupported } from '@/i18n/language'
  46. type ValueOf<T> = T[keyof T]
  47. type StepTwoProps = {
  48. isSetting?: boolean
  49. documentDetail?: FullDocumentDetail
  50. hasSetAPIKEY: boolean
  51. onSetting: () => void
  52. datasetId?: string
  53. indexingType?: ValueOf<IndexingType>
  54. dataSourceType: DataSourceType
  55. files: CustomFile[]
  56. notionPages?: NotionPage[]
  57. onStepChange?: (delta: number) => void
  58. updateIndexingTypeCache?: (type: string) => void
  59. updateResultCache?: (res: createDocumentResponse) => void
  60. onSave?: () => void
  61. onCancel?: () => void
  62. }
  63. enum SegmentType {
  64. AUTO = 'automatic',
  65. CUSTOM = 'custom',
  66. }
  67. enum IndexingType {
  68. QUALIFIED = 'high_quality',
  69. ECONOMICAL = 'economy',
  70. }
  71. const StepTwo = ({
  72. isSetting,
  73. documentDetail,
  74. hasSetAPIKEY,
  75. onSetting,
  76. datasetId,
  77. indexingType,
  78. dataSourceType,
  79. files,
  80. notionPages = [],
  81. onStepChange,
  82. updateIndexingTypeCache,
  83. updateResultCache,
  84. onSave,
  85. onCancel,
  86. }: StepTwoProps) => {
  87. const { t } = useTranslation()
  88. const { locale } = useContext(I18n)
  89. const media = useBreakpoints()
  90. const isMobile = media === MediaType.mobile
  91. const { dataset: currentDataset, mutateDatasetRes } = useDatasetDetailContext()
  92. const scrollRef = useRef<HTMLDivElement>(null)
  93. const [scrolled, setScrolled] = useState(false)
  94. const previewScrollRef = useRef<HTMLDivElement>(null)
  95. const [previewScrolled, setPreviewScrolled] = useState(false)
  96. const [segmentationType, setSegmentationType] = useState<SegmentType>(SegmentType.AUTO)
  97. const [segmentIdentifier, setSegmentIdentifier] = useState('\\n')
  98. const [max, setMax] = useState(500)
  99. const [overlap, setOverlap] = useState(50)
  100. const [rules, setRules] = useState<PreProcessingRule[]>([])
  101. const [defaultConfig, setDefaultConfig] = useState<Rules>()
  102. const hasSetIndexType = !!indexingType
  103. const [indexType, setIndexType] = useState<ValueOf<IndexingType>>(
  104. (indexingType
  105. || hasSetAPIKEY)
  106. ? IndexingType.QUALIFIED
  107. : IndexingType.ECONOMICAL,
  108. )
  109. const [docForm, setDocForm] = useState<DocForm | string>(
  110. (datasetId && documentDetail) ? documentDetail.doc_form : DocForm.TEXT,
  111. )
  112. const [docLanguage, setDocLanguage] = useState<string>(locale !== LanguagesSupported[1] ? 'English' : 'Chinese')
  113. const [QATipHide, setQATipHide] = useState(false)
  114. const [previewSwitched, setPreviewSwitched] = useState(false)
  115. const [showPreview, { setTrue: setShowPreview, setFalse: hidePreview }] = useBoolean()
  116. const [customFileIndexingEstimate, setCustomFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  117. const [automaticFileIndexingEstimate, setAutomaticFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  118. const [estimateTokes, setEstimateTokes] = useState<Pick<IndexingEstimateResponse, 'tokens' | 'total_price'> | null>(null)
  119. const fileIndexingEstimate = (() => {
  120. return segmentationType === SegmentType.AUTO ? automaticFileIndexingEstimate : customFileIndexingEstimate
  121. })()
  122. const [isCreating, setIsCreating] = useState(false)
  123. const scrollHandle = (e: Event) => {
  124. if ((e.target as HTMLDivElement).scrollTop > 0)
  125. setScrolled(true)
  126. else
  127. setScrolled(false)
  128. }
  129. const previewScrollHandle = (e: Event) => {
  130. if ((e.target as HTMLDivElement).scrollTop > 0)
  131. setPreviewScrolled(true)
  132. else
  133. setPreviewScrolled(false)
  134. }
  135. const getFileName = (name: string) => {
  136. const arr = name.split('.')
  137. return arr.slice(0, -1).join('.')
  138. }
  139. const getRuleName = (key: string) => {
  140. if (key === 'remove_extra_spaces')
  141. return t('datasetCreation.stepTwo.removeExtraSpaces')
  142. if (key === 'remove_urls_emails')
  143. return t('datasetCreation.stepTwo.removeUrlEmails')
  144. if (key === 'remove_stopwords')
  145. return t('datasetCreation.stepTwo.removeStopwords')
  146. }
  147. const ruleChangeHandle = (id: string) => {
  148. const newRules = rules.map((rule) => {
  149. if (rule.id === id) {
  150. return {
  151. id: rule.id,
  152. enabled: !rule.enabled,
  153. }
  154. }
  155. return rule
  156. })
  157. setRules(newRules)
  158. }
  159. const resetRules = () => {
  160. if (defaultConfig) {
  161. setSegmentIdentifier((defaultConfig.segmentation.separator === '\n' ? '\\n' : defaultConfig.segmentation.separator) || '\\n')
  162. setMax(defaultConfig.segmentation.max_tokens)
  163. setOverlap(defaultConfig.segmentation.chunk_overlap)
  164. setRules(defaultConfig.pre_processing_rules)
  165. }
  166. }
  167. const fetchFileIndexingEstimate = async (docForm = DocForm.TEXT) => {
  168. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  169. const res = await didFetchFileIndexingEstimate(getFileIndexingEstimateParams(docForm)!)
  170. if (segmentationType === SegmentType.CUSTOM) {
  171. setCustomFileIndexingEstimate(res)
  172. }
  173. else {
  174. setAutomaticFileIndexingEstimate(res)
  175. indexType === IndexingType.QUALIFIED && setEstimateTokes({ tokens: res.tokens, total_price: res.total_price })
  176. }
  177. }
  178. const confirmChangeCustomConfig = () => {
  179. setCustomFileIndexingEstimate(null)
  180. setShowPreview()
  181. fetchFileIndexingEstimate()
  182. setPreviewSwitched(false)
  183. }
  184. const getIndexing_technique = () => indexingType || indexType
  185. const getProcessRule = () => {
  186. const processRule: ProcessRule = {
  187. rules: {} as any, // api will check this. It will be removed after api refactored.
  188. mode: segmentationType,
  189. }
  190. if (segmentationType === SegmentType.CUSTOM) {
  191. const ruleObj = {
  192. pre_processing_rules: rules,
  193. segmentation: {
  194. separator: segmentIdentifier === '\\n' ? '\n' : segmentIdentifier,
  195. max_tokens: max,
  196. chunk_overlap: overlap,
  197. },
  198. }
  199. processRule.rules = ruleObj
  200. }
  201. return processRule
  202. }
  203. const getNotionInfo = () => {
  204. const workspacesMap = groupBy(notionPages, 'workspace_id')
  205. const workspaces = Object.keys(workspacesMap).map((workspaceId) => {
  206. return {
  207. workspaceId,
  208. pages: workspacesMap[workspaceId],
  209. }
  210. })
  211. return workspaces.map((workspace) => {
  212. return {
  213. workspace_id: workspace.workspaceId,
  214. pages: workspace.pages.map((page) => {
  215. const { page_id, page_name, page_icon, type } = page
  216. return {
  217. page_id,
  218. page_name,
  219. page_icon,
  220. type,
  221. }
  222. }),
  223. }
  224. }) as NotionInfo[]
  225. }
  226. const getFileIndexingEstimateParams = (docForm: DocForm): IndexingEstimateParams | undefined => {
  227. if (dataSourceType === DataSourceType.FILE) {
  228. return {
  229. info_list: {
  230. data_source_type: dataSourceType,
  231. file_info_list: {
  232. file_ids: files.map(file => file.id) as string[],
  233. },
  234. },
  235. indexing_technique: getIndexing_technique() as string,
  236. process_rule: getProcessRule(),
  237. doc_form: docForm,
  238. doc_language: docLanguage,
  239. dataset_id: datasetId as string,
  240. }
  241. }
  242. if (dataSourceType === DataSourceType.NOTION) {
  243. return {
  244. info_list: {
  245. data_source_type: dataSourceType,
  246. notion_info_list: getNotionInfo(),
  247. },
  248. indexing_technique: getIndexing_technique() as string,
  249. process_rule: getProcessRule(),
  250. doc_form: docForm,
  251. doc_language: docLanguage,
  252. dataset_id: datasetId as string,
  253. }
  254. }
  255. }
  256. const {
  257. modelList: rerankModelList,
  258. defaultModel: rerankDefaultModel,
  259. currentModel: isRerankDefaultModelVaild,
  260. } = useModelListAndDefaultModelAndCurrentProviderAndModel(3)
  261. const getCreationParams = () => {
  262. let params
  263. if (segmentationType === SegmentType.CUSTOM && overlap > max) {
  264. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.overlapCheck') })
  265. return
  266. }
  267. if (isSetting) {
  268. params = {
  269. original_document_id: documentDetail?.id,
  270. doc_form: docForm,
  271. doc_language: docLanguage,
  272. process_rule: getProcessRule(),
  273. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  274. retrieval_model: retrievalConfig, // Readonly. If want to changed, just go to settings page.
  275. } as CreateDocumentReq
  276. }
  277. else { // create
  278. const indexMethod = getIndexing_technique()
  279. if (
  280. !isReRankModelSelected({
  281. rerankDefaultModel,
  282. isRerankDefaultModelVaild: !!isRerankDefaultModelVaild,
  283. rerankModelList,
  284. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  285. retrievalConfig,
  286. indexMethod: indexMethod as string,
  287. })
  288. ) {
  289. Toast.notify({ type: 'error', message: t('appDebug.datasetConfig.rerankModelRequired') })
  290. return
  291. }
  292. const postRetrievalConfig = ensureRerankModelSelected({
  293. rerankDefaultModel: rerankDefaultModel!,
  294. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  295. retrievalConfig,
  296. indexMethod: indexMethod as string,
  297. })
  298. params = {
  299. data_source: {
  300. type: dataSourceType,
  301. info_list: {
  302. data_source_type: dataSourceType,
  303. },
  304. },
  305. indexing_technique: getIndexing_technique(),
  306. process_rule: getProcessRule(),
  307. doc_form: docForm,
  308. doc_language: docLanguage,
  309. retrieval_model: postRetrievalConfig,
  310. } as CreateDocumentReq
  311. if (dataSourceType === DataSourceType.FILE) {
  312. params.data_source.info_list.file_info_list = {
  313. file_ids: files.map(file => file.id || '').filter(Boolean),
  314. }
  315. }
  316. if (dataSourceType === DataSourceType.NOTION)
  317. params.data_source.info_list.notion_info_list = getNotionInfo()
  318. }
  319. return params
  320. }
  321. const getRules = async () => {
  322. try {
  323. const res = await fetchDefaultProcessRule({ url: '/datasets/process-rule' })
  324. const separator = res.rules.segmentation.separator
  325. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  326. setMax(res.rules.segmentation.max_tokens)
  327. setOverlap(res.rules.segmentation.chunk_overlap)
  328. setRules(res.rules.pre_processing_rules)
  329. setDefaultConfig(res.rules)
  330. }
  331. catch (err) {
  332. console.log(err)
  333. }
  334. }
  335. const getRulesFromDetail = () => {
  336. if (documentDetail) {
  337. const rules = documentDetail.dataset_process_rule.rules
  338. const separator = rules.segmentation.separator
  339. const max = rules.segmentation.max_tokens
  340. const overlap = rules.segmentation.chunk_overlap
  341. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  342. setMax(max)
  343. setOverlap(overlap)
  344. setRules(rules.pre_processing_rules)
  345. setDefaultConfig(rules)
  346. }
  347. }
  348. const getDefaultMode = () => {
  349. if (documentDetail)
  350. setSegmentationType(documentDetail.dataset_process_rule.mode)
  351. }
  352. const createHandle = async () => {
  353. if (isCreating)
  354. return
  355. setIsCreating(true)
  356. try {
  357. let res
  358. const params = getCreationParams()
  359. if (!params)
  360. return false
  361. setIsCreating(true)
  362. if (!datasetId) {
  363. res = await createFirstDocument({
  364. body: params as CreateDocumentReq,
  365. })
  366. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  367. updateResultCache && updateResultCache(res)
  368. }
  369. else {
  370. res = await createDocument({
  371. datasetId,
  372. body: params as CreateDocumentReq,
  373. })
  374. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  375. updateResultCache && updateResultCache(res)
  376. }
  377. if (mutateDatasetRes)
  378. mutateDatasetRes()
  379. onStepChange && onStepChange(+1)
  380. isSetting && onSave && onSave()
  381. }
  382. catch (err) {
  383. Toast.notify({
  384. type: 'error',
  385. message: `${err}`,
  386. })
  387. }
  388. finally {
  389. setIsCreating(false)
  390. }
  391. }
  392. const handleSwitch = (state: boolean) => {
  393. if (state)
  394. setDocForm(DocForm.QA)
  395. else
  396. setDocForm(DocForm.TEXT)
  397. }
  398. const handleSelect = (language: string) => {
  399. setDocLanguage(language)
  400. }
  401. const changeToEconomicalType = () => {
  402. if (!hasSetIndexType) {
  403. setIndexType(IndexingType.ECONOMICAL)
  404. setDocForm(DocForm.TEXT)
  405. }
  406. }
  407. const previewSwitch = async () => {
  408. setPreviewSwitched(true)
  409. if (segmentationType === SegmentType.AUTO)
  410. setAutomaticFileIndexingEstimate(null)
  411. else
  412. setCustomFileIndexingEstimate(null)
  413. await fetchFileIndexingEstimate(DocForm.QA)
  414. }
  415. useEffect(() => {
  416. // fetch rules
  417. if (!isSetting) {
  418. getRules()
  419. }
  420. else {
  421. getRulesFromDetail()
  422. getDefaultMode()
  423. }
  424. }, [])
  425. useEffect(() => {
  426. scrollRef.current?.addEventListener('scroll', scrollHandle)
  427. return () => {
  428. scrollRef.current?.removeEventListener('scroll', scrollHandle)
  429. }
  430. }, [])
  431. useLayoutEffect(() => {
  432. if (showPreview) {
  433. previewScrollRef.current?.addEventListener('scroll', previewScrollHandle)
  434. return () => {
  435. previewScrollRef.current?.removeEventListener('scroll', previewScrollHandle)
  436. }
  437. }
  438. }, [showPreview])
  439. useEffect(() => {
  440. if (indexingType === IndexingType.ECONOMICAL && docForm === DocForm.QA)
  441. setDocForm(DocForm.TEXT)
  442. }, [indexingType, docForm])
  443. useEffect(() => {
  444. // get indexing type by props
  445. if (indexingType)
  446. setIndexType(indexingType as IndexingType)
  447. else
  448. setIndexType(hasSetAPIKEY ? IndexingType.QUALIFIED : IndexingType.ECONOMICAL)
  449. }, [hasSetAPIKEY, indexingType, datasetId])
  450. useEffect(() => {
  451. if (segmentationType === SegmentType.AUTO) {
  452. setAutomaticFileIndexingEstimate(null)
  453. !isMobile && setShowPreview()
  454. fetchFileIndexingEstimate()
  455. setPreviewSwitched(false)
  456. }
  457. else {
  458. hidePreview()
  459. setCustomFileIndexingEstimate(null)
  460. setPreviewSwitched(false)
  461. }
  462. }, [segmentationType, indexType])
  463. const [retrievalConfig, setRetrievalConfig] = useState(currentDataset?.retrieval_model_dict || {
  464. search_method: RETRIEVE_METHOD.semantic,
  465. reranking_enable: false,
  466. reranking_model: {
  467. reranking_provider_name: rerankDefaultModel?.provider.provider,
  468. reranking_model_name: rerankDefaultModel?.model,
  469. },
  470. top_k: 3,
  471. score_threshold_enabled: false,
  472. score_threshold: 0.5,
  473. } as RetrievalConfig)
  474. return (
  475. <div className='flex w-full h-full'>
  476. <div ref={scrollRef} className='relative h-full w-full overflow-y-scroll'>
  477. <div className={cn(s.pageHeader, scrolled && s.fixed, isMobile && '!px-6')}>
  478. <span>{t('datasetCreation.steps.two')}</span>
  479. {isMobile && (
  480. <Button
  481. className='border-[0.5px] !h-8 hover:outline hover:outline-[0.5px] hover:outline-gray-300 text-gray-700 font-medium bg-white shadow-[0px_1px_2px_0px_rgba(16,24,40,0.05)]'
  482. onClick={setShowPreview}
  483. >
  484. <Tooltip selector='data-preview-toggle'>
  485. <div className="flex flex-row items-center">
  486. <RocketLaunchIcon className="h-4 w-4 mr-1.5 stroke-[1.8px]" />
  487. <span className="text-[13px]">{t('datasetCreation.stepTwo.previewTitleButton')}</span>
  488. </div>
  489. </Tooltip>
  490. </Button>
  491. )}
  492. </div>
  493. <div className={cn(s.form, isMobile && '!px-4')}>
  494. <div className={s.label}>{t('datasetCreation.stepTwo.segmentation')}</div>
  495. <div className='max-w-[640px]'>
  496. <div
  497. className={cn(
  498. s.radioItem,
  499. s.segmentationItem,
  500. segmentationType === SegmentType.AUTO && s.active,
  501. )}
  502. onClick={() => setSegmentationType(SegmentType.AUTO)}
  503. >
  504. <span className={cn(s.typeIcon, s.auto)} />
  505. <span className={cn(s.radio)} />
  506. <div className={s.typeHeader}>
  507. <div className={s.title}>{t('datasetCreation.stepTwo.auto')}</div>
  508. <div className={s.tip}>{t('datasetCreation.stepTwo.autoDescription')}</div>
  509. </div>
  510. </div>
  511. <div
  512. className={cn(
  513. s.radioItem,
  514. s.segmentationItem,
  515. segmentationType === SegmentType.CUSTOM && s.active,
  516. segmentationType === SegmentType.CUSTOM && s.custom,
  517. )}
  518. onClick={() => setSegmentationType(SegmentType.CUSTOM)}
  519. >
  520. <span className={cn(s.typeIcon, s.customize)} />
  521. <span className={cn(s.radio)} />
  522. <div className={s.typeHeader}>
  523. <div className={s.title}>{t('datasetCreation.stepTwo.custom')}</div>
  524. <div className={s.tip}>{t('datasetCreation.stepTwo.customDescription')}</div>
  525. </div>
  526. {segmentationType === SegmentType.CUSTOM && (
  527. <div className={s.typeFormBody}>
  528. <div className={s.formRow}>
  529. <div className='w-full'>
  530. <div className={s.label}>{t('datasetCreation.stepTwo.separator')}</div>
  531. <input
  532. type="text"
  533. className={s.input}
  534. placeholder={t('datasetCreation.stepTwo.separatorPlaceholder') || ''} value={segmentIdentifier}
  535. onChange={e => setSegmentIdentifier(e.target.value)}
  536. />
  537. </div>
  538. </div>
  539. <div className={s.formRow}>
  540. <div className='w-full'>
  541. <div className={s.label}>{t('datasetCreation.stepTwo.maxLength')}</div>
  542. <input
  543. type="number"
  544. className={s.input}
  545. placeholder={t('datasetCreation.stepTwo.maxLength') || ''}
  546. value={max}
  547. min={1}
  548. onChange={e => setMax(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  549. />
  550. </div>
  551. </div>
  552. <div className={s.formRow}>
  553. <div className='w-full'>
  554. <div className={s.label}>
  555. {t('datasetCreation.stepTwo.overlap')}
  556. <TooltipPlus popupContent={
  557. <div className='max-w-[200px]'>
  558. {t('datasetCreation.stepTwo.overlapTip')}
  559. </div>
  560. }>
  561. <HelpCircle className='ml-1 w-3.5 h-3.5 text-gray-400' />
  562. </TooltipPlus>
  563. </div>
  564. <input
  565. type="number"
  566. className={s.input}
  567. placeholder={t('datasetCreation.stepTwo.overlap') || ''}
  568. value={overlap}
  569. min={1}
  570. onChange={e => setOverlap(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  571. />
  572. </div>
  573. </div>
  574. <div className={s.formRow}>
  575. <div className='w-full flex flex-col gap-1'>
  576. <div className={s.label}>{t('datasetCreation.stepTwo.rules')}</div>
  577. {rules.map(rule => (
  578. <div key={rule.id} className={s.ruleItem}>
  579. <input id={rule.id} type="checkbox" checked={rule.enabled} onChange={() => ruleChangeHandle(rule.id)} className="w-4 h-4 rounded border-gray-300 text-blue-700 focus:ring-blue-700" />
  580. <label htmlFor={rule.id} className="ml-2 text-sm font-normal cursor-pointer text-gray-800">{getRuleName(rule.id)}</label>
  581. </div>
  582. ))}
  583. </div>
  584. </div>
  585. <div className={s.formFooter}>
  586. <Button type="primary" className={cn(s.button, '!h-8')} onClick={confirmChangeCustomConfig}>{t('datasetCreation.stepTwo.preview')}</Button>
  587. <Button className={cn(s.button, 'ml-2 !h-8')} onClick={resetRules}>{t('datasetCreation.stepTwo.reset')}</Button>
  588. </div>
  589. </div>
  590. )}
  591. </div>
  592. </div>
  593. <div className={s.label}>{t('datasetCreation.stepTwo.indexMode')}</div>
  594. <div className='max-w-[640px]'>
  595. <div className='flex items-center gap-3 flex-wrap sm:flex-nowrap'>
  596. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.QUALIFIED)) && (
  597. <div
  598. className={cn(
  599. s.radioItem,
  600. s.indexItem,
  601. !hasSetAPIKEY && s.disabled,
  602. !hasSetIndexType && indexType === IndexingType.QUALIFIED && s.active,
  603. hasSetIndexType && s.disabled,
  604. hasSetIndexType && '!w-full',
  605. )}
  606. onClick={() => {
  607. if (hasSetAPIKEY)
  608. setIndexType(IndexingType.QUALIFIED)
  609. }}
  610. >
  611. <span className={cn(s.typeIcon, s.qualified)} />
  612. {!hasSetIndexType && <span className={cn(s.radio)} />}
  613. <div className={s.typeHeader}>
  614. <div className={s.title}>
  615. {t('datasetCreation.stepTwo.qualified')}
  616. {!hasSetIndexType && <span className={s.recommendTag}>{t('datasetCreation.stepTwo.recommend')}</span>}
  617. </div>
  618. <div className={s.tip}>{t('datasetCreation.stepTwo.qualifiedTip')}</div>
  619. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  620. {
  621. estimateTokes
  622. ? (
  623. <div className='text-xs font-medium text-gray-800'>{formatNumber(estimateTokes.tokens)} tokens(<span className='text-yellow-500'>${formatNumber(estimateTokes.total_price)}</span>)</div>
  624. )
  625. : (
  626. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  627. )
  628. }
  629. </div>
  630. {!hasSetAPIKEY && (
  631. <div className={s.warningTip}>
  632. <span>{t('datasetCreation.stepTwo.warning')}&nbsp;</span>
  633. <span className={s.click} onClick={onSetting}>{t('datasetCreation.stepTwo.click')}</span>
  634. </div>
  635. )}
  636. </div>
  637. )}
  638. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.ECONOMICAL)) && (
  639. <div
  640. className={cn(
  641. s.radioItem,
  642. s.indexItem,
  643. !hasSetIndexType && indexType === IndexingType.ECONOMICAL && s.active,
  644. hasSetIndexType && s.disabled,
  645. hasSetIndexType && '!w-full',
  646. )}
  647. onClick={changeToEconomicalType}
  648. >
  649. <span className={cn(s.typeIcon, s.economical)} />
  650. {!hasSetIndexType && <span className={cn(s.radio)} />}
  651. <div className={s.typeHeader}>
  652. <div className={s.title}>{t('datasetCreation.stepTwo.economical')}</div>
  653. <div className={s.tip}>{t('datasetCreation.stepTwo.economicalTip')}</div>
  654. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  655. <div className='text-xs font-medium text-gray-800'>0 tokens</div>
  656. </div>
  657. </div>
  658. )}
  659. </div>
  660. {hasSetIndexType && (
  661. <div className='mt-2 text-xs text-gray-500 font-medium'>
  662. {t('datasetCreation.stepTwo.indexSettedTip')}
  663. <Link className='text-[#155EEF]' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  664. </div>
  665. )}
  666. {IS_CE_EDITION && indexType === IndexingType.QUALIFIED && (
  667. <div className='mt-3 rounded-xl bg-gray-50 border border-gray-100'>
  668. <div className='flex justify-between items-center px-5 py-4'>
  669. <div className='flex justify-center items-center w-8 h-8 rounded-lg bg-indigo-50'>
  670. <MessageChatSquare className='w-4 h-4' />
  671. </div>
  672. <div className='grow mx-3'>
  673. <div className='mb-[2px] text-md font-medium text-gray-900'>{t('datasetCreation.stepTwo.QATitle')}</div>
  674. <div className='inline-flex items-center text-[13px] leading-[18px] text-gray-500'>
  675. <span className='pr-1'>{t('datasetCreation.stepTwo.QALanguage')}</span>
  676. <LanguageSelect currentLanguage={docLanguage} onSelect={handleSelect} />
  677. </div>
  678. </div>
  679. <div className='shrink-0'>
  680. <Switch
  681. defaultValue={docForm === DocForm.QA}
  682. onChange={handleSwitch}
  683. size='md'
  684. />
  685. </div>
  686. </div>
  687. {docForm === DocForm.QA && !QATipHide && (
  688. <div className='flex justify-between items-center px-5 py-2 bg-orange-50 border-t border-amber-100 rounded-b-xl text-[13px] leading-[18px] text-medium text-amber-500'>
  689. {t('datasetCreation.stepTwo.QATip')}
  690. <XClose className='w-4 h-4 text-gray-500 cursor-pointer' onClick={() => setQATipHide(true)} />
  691. </div>
  692. )}
  693. </div>
  694. )}
  695. {/* Retrieval Method Config */}
  696. <div>
  697. {!datasetId
  698. ? (
  699. <div className={s.label}>
  700. {t('datasetSettings.form.retrievalSetting.title')}
  701. <div className='leading-[18px] text-xs font-normal text-gray-500'>
  702. <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/features/retrieval-augment' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
  703. {t('datasetSettings.form.retrievalSetting.longDescription')}
  704. </div>
  705. </div>
  706. )
  707. : (
  708. <div className={cn(s.label, 'flex justify-between items-center')}>
  709. <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
  710. </div>
  711. )}
  712. <div className='max-w-[640px]'>
  713. {!datasetId
  714. ? (<>
  715. {getIndexing_technique() === IndexingType.QUALIFIED
  716. ? (
  717. <RetrievalMethodConfig
  718. value={retrievalConfig}
  719. onChange={setRetrievalConfig}
  720. />
  721. )
  722. : (
  723. <EconomicalRetrievalMethodConfig
  724. value={retrievalConfig}
  725. onChange={setRetrievalConfig}
  726. />
  727. )}
  728. </>)
  729. : (
  730. <div>
  731. <RetrievalMethodInfo
  732. value={retrievalConfig}
  733. />
  734. <div className='mt-2 text-xs text-gray-500 font-medium'>
  735. {t('datasetCreation.stepTwo.retrivalSettedTip')}
  736. <Link className='text-[#155EEF]' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  737. </div>
  738. </div>
  739. )}
  740. </div>
  741. </div>
  742. <div className={s.source}>
  743. <div className={s.sourceContent}>
  744. {dataSourceType === DataSourceType.FILE && (
  745. <>
  746. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.fileSource')}</div>
  747. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  748. <span className={cn(s.fileIcon, files.length && s[files[0].extension || ''])} />
  749. {getFileName(files[0].name || '')}
  750. {files.length > 1 && (
  751. <span className={s.sourceCount}>
  752. <span>{t('datasetCreation.stepTwo.other')}</span>
  753. <span>{files.length - 1}</span>
  754. <span>{t('datasetCreation.stepTwo.fileUnit')}</span>
  755. </span>
  756. )}
  757. </div>
  758. </>
  759. )}
  760. {dataSourceType === DataSourceType.NOTION && (
  761. <>
  762. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.notionSource')}</div>
  763. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  764. <NotionIcon
  765. className='shrink-0 mr-1'
  766. type='page'
  767. src={notionPages[0]?.page_icon}
  768. />
  769. {notionPages[0]?.page_name}
  770. {notionPages.length > 1 && (
  771. <span className={s.sourceCount}>
  772. <span>{t('datasetCreation.stepTwo.other')}</span>
  773. <span>{notionPages.length - 1}</span>
  774. <span>{t('datasetCreation.stepTwo.notionUnit')}</span>
  775. </span>
  776. )}
  777. </div>
  778. </>
  779. )}
  780. </div>
  781. <div className={s.divider} />
  782. <div className={s.segmentCount}>
  783. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateSegment')}</div>
  784. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  785. {
  786. fileIndexingEstimate
  787. ? (
  788. <div className='text-xs font-medium text-gray-800'>{formatNumber(fileIndexingEstimate.total_segments)} </div>
  789. )
  790. : (
  791. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  792. )
  793. }
  794. </div>
  795. </div>
  796. </div>
  797. {!isSetting
  798. ? (
  799. <div className='flex items-center mt-8 py-2'>
  800. <Button onClick={() => onStepChange && onStepChange(-1)}>{t('datasetCreation.stepTwo.lastStep')}</Button>
  801. <div className={s.divider} />
  802. <Button loading={isCreating} type='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.nextStep')}</Button>
  803. </div>
  804. )
  805. : (
  806. <div className='flex items-center mt-8 py-2'>
  807. <Button loading={isCreating} type='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.save')}</Button>
  808. <Button className='ml-2' onClick={onCancel}>{t('datasetCreation.stepTwo.cancel')}</Button>
  809. </div>
  810. )}
  811. </div>
  812. </div>
  813. </div>
  814. <FloatRightContainer isMobile={isMobile} isOpen={showPreview} onClose={hidePreview} footer={null}>
  815. {showPreview && <div ref={previewScrollRef} className={cn(s.previewWrap, isMobile && s.isMobile, 'relative h-full overflow-y-scroll border-l border-[#F2F4F7]')}>
  816. <div className={cn(s.previewHeader, previewScrolled && `${s.fixed} pb-3`)}>
  817. <div className='flex items-center justify-between px-8'>
  818. <div className='grow flex items-center'>
  819. <div>{t('datasetCreation.stepTwo.previewTitle')}</div>
  820. {docForm === DocForm.QA && !previewSwitched && (
  821. <Button className='ml-2 !h-[26px] !py-[3px] !px-2 !text-xs !font-medium !text-primary-600' onClick={previewSwitch}>{t('datasetCreation.stepTwo.previewButton')}</Button>
  822. )}
  823. </div>
  824. <div className='flex items-center justify-center w-6 h-6 cursor-pointer' onClick={hidePreview}>
  825. <XMarkIcon className='h-4 w-4'></XMarkIcon>
  826. </div>
  827. </div>
  828. {docForm === DocForm.QA && !previewSwitched && (
  829. <div className='px-8 pr-12 text-xs text-gray-500'>
  830. <span>{t('datasetCreation.stepTwo.previewSwitchTipStart')}</span>
  831. <span className='text-amber-600'>{t('datasetCreation.stepTwo.previewSwitchTipEnd')}</span>
  832. </div>
  833. )}
  834. </div>
  835. <div className='my-4 px-8 space-y-4'>
  836. {previewSwitched && docForm === DocForm.QA && fileIndexingEstimate?.qa_preview && (
  837. <>
  838. {fileIndexingEstimate?.qa_preview.map((item, index) => (
  839. <PreviewItem type={PreviewType.QA} key={item.question} qa={item} index={index + 1} />
  840. ))}
  841. </>
  842. )}
  843. {(docForm === DocForm.TEXT || !previewSwitched) && fileIndexingEstimate?.preview && (
  844. <>
  845. {fileIndexingEstimate?.preview.map((item, index) => (
  846. <PreviewItem type={PreviewType.TEXT} key={item} content={item} index={index + 1} />
  847. ))}
  848. </>
  849. )}
  850. {previewSwitched && docForm === DocForm.QA && !fileIndexingEstimate?.qa_preview && (
  851. <div className='flex items-center justify-center h-[200px]'>
  852. <Loading type='area' />
  853. </div>
  854. )}
  855. {!previewSwitched && !fileIndexingEstimate?.preview && (
  856. <div className='flex items-center justify-center h-[200px]'>
  857. <Loading type='area' />
  858. </div>
  859. )}
  860. </div>
  861. </div>}
  862. {!showPreview && (
  863. <div className={cn(s.sideTip)}>
  864. <div className={s.tipCard}>
  865. <span className={s.icon} />
  866. <div className={s.title}>{t('datasetCreation.stepTwo.sideTipTitle')}</div>
  867. <div className={s.content}>
  868. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP1')}</p>
  869. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP2')}</p>
  870. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP3')}</p>
  871. <p>{t('datasetCreation.stepTwo.sideTipP4')}</p>
  872. </div>
  873. </div>
  874. </div>
  875. )}
  876. </FloatRightContainer>
  877. </div>
  878. )
  879. }
  880. export default StepTwo