index.tsx 38 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962
  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 {
  9. RiCloseLine,
  10. } from '@remixicon/react'
  11. import Link from 'next/link'
  12. import { groupBy } from 'lodash-es'
  13. import PreviewItem, { PreviewType } from './preview-item'
  14. import LanguageSelect from './language-select'
  15. import s from './index.module.css'
  16. import cn from '@/utils/classnames'
  17. import type { CrawlOptions, CrawlResultItem, CreateDocumentReq, CustomFile, FileIndexingEstimateResponse, FullDocumentDetail, IndexingEstimateParams, IndexingEstimateResponse, NotionInfo, PreProcessingRule, ProcessRule, Rules, createDocumentResponse } from '@/models/datasets'
  18. import {
  19. createDocument,
  20. createFirstDocument,
  21. fetchFileIndexingEstimate as didFetchFileIndexingEstimate,
  22. fetchDefaultProcessRule,
  23. } from '@/service/datasets'
  24. import Button from '@/app/components/base/button'
  25. import Loading from '@/app/components/base/loading'
  26. import FloatRightContainer from '@/app/components/base/float-right-container'
  27. import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
  28. import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
  29. import { type RetrievalConfig } from '@/types/app'
  30. import { ensureRerankModelSelected, isReRankModelSelected } from '@/app/components/datasets/common/check-rerank-model'
  31. import Toast from '@/app/components/base/toast'
  32. import { formatNumber } from '@/utils/format'
  33. import type { NotionPage } from '@/models/common'
  34. import { DataSourceType, DocForm } from '@/models/datasets'
  35. import NotionIcon from '@/app/components/base/notion-icon'
  36. import Switch from '@/app/components/base/switch'
  37. import { MessageChatSquare } from '@/app/components/base/icons/src/public/common'
  38. import { useDatasetDetailContext } from '@/context/dataset-detail'
  39. import I18n from '@/context/i18n'
  40. import { IS_CE_EDITION } from '@/config'
  41. import { RETRIEVE_METHOD } from '@/types/app'
  42. import useBreakpoints, { MediaType } from '@/hooks/use-breakpoints'
  43. import Tooltip from '@/app/components/base/tooltip'
  44. import { useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
  45. import { LanguagesSupported } from '@/i18n/language'
  46. import { ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
  47. import { Globe01 } from '@/app/components/base/icons/src/vender/line/mapsAndTravel'
  48. type ValueOf<T> = T[keyof T]
  49. type StepTwoProps = {
  50. isSetting?: boolean
  51. documentDetail?: FullDocumentDetail
  52. isAPIKeySet: boolean
  53. onSetting: () => void
  54. datasetId?: string
  55. indexingType?: ValueOf<IndexingType>
  56. dataSourceType: DataSourceType
  57. files: CustomFile[]
  58. notionPages?: NotionPage[]
  59. websitePages?: CrawlResultItem[]
  60. crawlOptions?: CrawlOptions
  61. fireCrawlJobId?: string
  62. onStepChange?: (delta: number) => void
  63. updateIndexingTypeCache?: (type: string) => void
  64. updateResultCache?: (res: createDocumentResponse) => void
  65. onSave?: () => void
  66. onCancel?: () => void
  67. }
  68. enum SegmentType {
  69. AUTO = 'automatic',
  70. CUSTOM = 'custom',
  71. }
  72. enum IndexingType {
  73. QUALIFIED = 'high_quality',
  74. ECONOMICAL = 'economy',
  75. }
  76. const StepTwo = ({
  77. isSetting,
  78. documentDetail,
  79. isAPIKeySet,
  80. onSetting,
  81. datasetId,
  82. indexingType,
  83. dataSourceType: inCreatePageDataSourceType,
  84. files,
  85. notionPages = [],
  86. websitePages = [],
  87. crawlOptions,
  88. fireCrawlJobId = '',
  89. onStepChange,
  90. updateIndexingTypeCache,
  91. updateResultCache,
  92. onSave,
  93. onCancel,
  94. }: StepTwoProps) => {
  95. const { t } = useTranslation()
  96. const { locale } = useContext(I18n)
  97. const media = useBreakpoints()
  98. const isMobile = media === MediaType.mobile
  99. const { dataset: currentDataset, mutateDatasetRes } = useDatasetDetailContext()
  100. const isInCreatePage = !datasetId || (datasetId && !currentDataset?.data_source_type)
  101. const dataSourceType = isInCreatePage ? inCreatePageDataSourceType : currentDataset?.data_source_type
  102. const scrollRef = useRef<HTMLDivElement>(null)
  103. const [scrolled, setScrolled] = useState(false)
  104. const previewScrollRef = useRef<HTMLDivElement>(null)
  105. const [previewScrolled, setPreviewScrolled] = useState(false)
  106. const [segmentationType, setSegmentationType] = useState<SegmentType>(SegmentType.AUTO)
  107. const [segmentIdentifier, setSegmentIdentifier] = useState('\\n')
  108. const [max, setMax] = useState(500)
  109. const [overlap, setOverlap] = useState(50)
  110. const [rules, setRules] = useState<PreProcessingRule[]>([])
  111. const [defaultConfig, setDefaultConfig] = useState<Rules>()
  112. const hasSetIndexType = !!indexingType
  113. const [indexType, setIndexType] = useState<ValueOf<IndexingType>>(
  114. (indexingType
  115. || isAPIKeySet)
  116. ? IndexingType.QUALIFIED
  117. : IndexingType.ECONOMICAL,
  118. )
  119. const [docForm, setDocForm] = useState<DocForm | string>(
  120. (datasetId && documentDetail) ? documentDetail.doc_form : DocForm.TEXT,
  121. )
  122. const [docLanguage, setDocLanguage] = useState<string>(
  123. (datasetId && documentDetail) ? documentDetail.doc_language : (locale !== LanguagesSupported[1] ? 'English' : 'Chinese'),
  124. )
  125. const [QATipHide, setQATipHide] = useState(false)
  126. const [previewSwitched, setPreviewSwitched] = useState(false)
  127. const [showPreview, { setTrue: setShowPreview, setFalse: hidePreview }] = useBoolean()
  128. const [customFileIndexingEstimate, setCustomFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  129. const [automaticFileIndexingEstimate, setAutomaticFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  130. const [estimateTokes, setEstimateTokes] = useState<Pick<IndexingEstimateResponse, 'tokens' | 'total_price'> | null>(null)
  131. const fileIndexingEstimate = (() => {
  132. return segmentationType === SegmentType.AUTO ? automaticFileIndexingEstimate : customFileIndexingEstimate
  133. })()
  134. const [isCreating, setIsCreating] = useState(false)
  135. const scrollHandle = (e: Event) => {
  136. if ((e.target as HTMLDivElement).scrollTop > 0)
  137. setScrolled(true)
  138. else
  139. setScrolled(false)
  140. }
  141. const previewScrollHandle = (e: Event) => {
  142. if ((e.target as HTMLDivElement).scrollTop > 0)
  143. setPreviewScrolled(true)
  144. else
  145. setPreviewScrolled(false)
  146. }
  147. const getFileName = (name: string) => {
  148. const arr = name.split('.')
  149. return arr.slice(0, -1).join('.')
  150. }
  151. const getRuleName = (key: string) => {
  152. if (key === 'remove_extra_spaces')
  153. return t('datasetCreation.stepTwo.removeExtraSpaces')
  154. if (key === 'remove_urls_emails')
  155. return t('datasetCreation.stepTwo.removeUrlEmails')
  156. if (key === 'remove_stopwords')
  157. return t('datasetCreation.stepTwo.removeStopwords')
  158. }
  159. const ruleChangeHandle = (id: string) => {
  160. const newRules = rules.map((rule) => {
  161. if (rule.id === id) {
  162. return {
  163. id: rule.id,
  164. enabled: !rule.enabled,
  165. }
  166. }
  167. return rule
  168. })
  169. setRules(newRules)
  170. }
  171. const resetRules = () => {
  172. if (defaultConfig) {
  173. setSegmentIdentifier((defaultConfig.segmentation.separator === '\n' ? '\\n' : defaultConfig.segmentation.separator) || '\\n')
  174. setMax(defaultConfig.segmentation.max_tokens)
  175. setOverlap(defaultConfig.segmentation.chunk_overlap)
  176. setRules(defaultConfig.pre_processing_rules)
  177. }
  178. }
  179. const fetchFileIndexingEstimate = async (docForm = DocForm.TEXT) => {
  180. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  181. const res = await didFetchFileIndexingEstimate(getFileIndexingEstimateParams(docForm)!)
  182. if (segmentationType === SegmentType.CUSTOM) {
  183. setCustomFileIndexingEstimate(res)
  184. }
  185. else {
  186. setAutomaticFileIndexingEstimate(res)
  187. indexType === IndexingType.QUALIFIED && setEstimateTokes({ tokens: res.tokens, total_price: res.total_price })
  188. }
  189. }
  190. const confirmChangeCustomConfig = () => {
  191. setCustomFileIndexingEstimate(null)
  192. setShowPreview()
  193. fetchFileIndexingEstimate()
  194. setPreviewSwitched(false)
  195. }
  196. const getIndexing_technique = () => indexingType || indexType
  197. const getProcessRule = () => {
  198. const processRule: ProcessRule = {
  199. rules: {} as any, // api will check this. It will be removed after api refactored.
  200. mode: segmentationType,
  201. }
  202. if (segmentationType === SegmentType.CUSTOM) {
  203. const ruleObj = {
  204. pre_processing_rules: rules,
  205. segmentation: {
  206. separator: segmentIdentifier === '\\n' ? '\n' : segmentIdentifier,
  207. max_tokens: max,
  208. chunk_overlap: overlap,
  209. },
  210. }
  211. processRule.rules = ruleObj
  212. }
  213. return processRule
  214. }
  215. const getNotionInfo = () => {
  216. const workspacesMap = groupBy(notionPages, 'workspace_id')
  217. const workspaces = Object.keys(workspacesMap).map((workspaceId) => {
  218. return {
  219. workspaceId,
  220. pages: workspacesMap[workspaceId],
  221. }
  222. })
  223. return workspaces.map((workspace) => {
  224. return {
  225. workspace_id: workspace.workspaceId,
  226. pages: workspace.pages.map((page) => {
  227. const { page_id, page_name, page_icon, type } = page
  228. return {
  229. page_id,
  230. page_name,
  231. page_icon,
  232. type,
  233. }
  234. }),
  235. }
  236. }) as NotionInfo[]
  237. }
  238. const getWebsiteInfo = () => {
  239. return {
  240. provider: 'firecrawl',
  241. job_id: fireCrawlJobId,
  242. urls: websitePages.map(page => page.source_url),
  243. only_main_content: crawlOptions?.only_main_content,
  244. }
  245. }
  246. const getFileIndexingEstimateParams = (docForm: DocForm): IndexingEstimateParams | undefined => {
  247. if (dataSourceType === DataSourceType.FILE) {
  248. return {
  249. info_list: {
  250. data_source_type: dataSourceType,
  251. file_info_list: {
  252. file_ids: files.map(file => file.id) as string[],
  253. },
  254. },
  255. indexing_technique: getIndexing_technique() as string,
  256. process_rule: getProcessRule(),
  257. doc_form: docForm,
  258. doc_language: docLanguage,
  259. dataset_id: datasetId as string,
  260. }
  261. }
  262. if (dataSourceType === DataSourceType.NOTION) {
  263. return {
  264. info_list: {
  265. data_source_type: dataSourceType,
  266. notion_info_list: getNotionInfo(),
  267. },
  268. indexing_technique: getIndexing_technique() as string,
  269. process_rule: getProcessRule(),
  270. doc_form: docForm,
  271. doc_language: docLanguage,
  272. dataset_id: datasetId as string,
  273. }
  274. }
  275. if (dataSourceType === DataSourceType.WEB) {
  276. return {
  277. info_list: {
  278. data_source_type: dataSourceType,
  279. website_info_list: getWebsiteInfo(),
  280. },
  281. indexing_technique: getIndexing_technique() as string,
  282. process_rule: getProcessRule(),
  283. doc_form: docForm,
  284. doc_language: docLanguage,
  285. dataset_id: datasetId as string,
  286. }
  287. }
  288. }
  289. const {
  290. modelList: rerankModelList,
  291. defaultModel: rerankDefaultModel,
  292. currentModel: isRerankDefaultModelVaild,
  293. } = useModelListAndDefaultModelAndCurrentProviderAndModel(ModelTypeEnum.rerank)
  294. const getCreationParams = () => {
  295. let params
  296. if (segmentationType === SegmentType.CUSTOM && overlap > max) {
  297. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.overlapCheck') })
  298. return
  299. }
  300. if (isSetting) {
  301. params = {
  302. original_document_id: documentDetail?.id,
  303. doc_form: docForm,
  304. doc_language: docLanguage,
  305. process_rule: getProcessRule(),
  306. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  307. retrieval_model: retrievalConfig, // Readonly. If want to changed, just go to settings page.
  308. } as CreateDocumentReq
  309. }
  310. else { // create
  311. const indexMethod = getIndexing_technique()
  312. if (
  313. !isReRankModelSelected({
  314. rerankDefaultModel,
  315. isRerankDefaultModelVaild: !!isRerankDefaultModelVaild,
  316. rerankModelList,
  317. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  318. retrievalConfig,
  319. indexMethod: indexMethod as string,
  320. })
  321. ) {
  322. Toast.notify({ type: 'error', message: t('appDebug.datasetConfig.rerankModelRequired') })
  323. return
  324. }
  325. const postRetrievalConfig = ensureRerankModelSelected({
  326. rerankDefaultModel: rerankDefaultModel!,
  327. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  328. retrievalConfig,
  329. indexMethod: indexMethod as string,
  330. })
  331. params = {
  332. data_source: {
  333. type: dataSourceType,
  334. info_list: {
  335. data_source_type: dataSourceType,
  336. },
  337. },
  338. indexing_technique: getIndexing_technique(),
  339. process_rule: getProcessRule(),
  340. doc_form: docForm,
  341. doc_language: docLanguage,
  342. retrieval_model: postRetrievalConfig,
  343. } as CreateDocumentReq
  344. if (dataSourceType === DataSourceType.FILE) {
  345. params.data_source.info_list.file_info_list = {
  346. file_ids: files.map(file => file.id || '').filter(Boolean),
  347. }
  348. }
  349. if (dataSourceType === DataSourceType.NOTION)
  350. params.data_source.info_list.notion_info_list = getNotionInfo()
  351. if (dataSourceType === DataSourceType.WEB)
  352. params.data_source.info_list.website_info_list = getWebsiteInfo()
  353. }
  354. return params
  355. }
  356. const getRules = async () => {
  357. try {
  358. const res = await fetchDefaultProcessRule({ url: '/datasets/process-rule' })
  359. const separator = res.rules.segmentation.separator
  360. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  361. setMax(res.rules.segmentation.max_tokens)
  362. setOverlap(res.rules.segmentation.chunk_overlap)
  363. setRules(res.rules.pre_processing_rules)
  364. setDefaultConfig(res.rules)
  365. }
  366. catch (err) {
  367. console.log(err)
  368. }
  369. }
  370. const getRulesFromDetail = () => {
  371. if (documentDetail) {
  372. const rules = documentDetail.dataset_process_rule.rules
  373. const separator = rules.segmentation.separator
  374. const max = rules.segmentation.max_tokens
  375. const overlap = rules.segmentation.chunk_overlap
  376. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  377. setMax(max)
  378. setOverlap(overlap)
  379. setRules(rules.pre_processing_rules)
  380. setDefaultConfig(rules)
  381. }
  382. }
  383. const getDefaultMode = () => {
  384. if (documentDetail)
  385. setSegmentationType(documentDetail.dataset_process_rule.mode)
  386. }
  387. const createHandle = async () => {
  388. if (isCreating)
  389. return
  390. setIsCreating(true)
  391. try {
  392. let res
  393. const params = getCreationParams()
  394. if (!params)
  395. return false
  396. setIsCreating(true)
  397. if (!datasetId) {
  398. res = await createFirstDocument({
  399. body: params as CreateDocumentReq,
  400. })
  401. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  402. updateResultCache && updateResultCache(res)
  403. }
  404. else {
  405. res = await createDocument({
  406. datasetId,
  407. body: params as CreateDocumentReq,
  408. })
  409. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  410. updateResultCache && updateResultCache(res)
  411. }
  412. if (mutateDatasetRes)
  413. mutateDatasetRes()
  414. onStepChange && onStepChange(+1)
  415. isSetting && onSave && onSave()
  416. }
  417. catch (err) {
  418. Toast.notify({
  419. type: 'error',
  420. message: `${err}`,
  421. })
  422. }
  423. finally {
  424. setIsCreating(false)
  425. }
  426. }
  427. const handleSwitch = (state: boolean) => {
  428. if (state)
  429. setDocForm(DocForm.QA)
  430. else
  431. setDocForm(DocForm.TEXT)
  432. }
  433. const handleSelect = (language: string) => {
  434. setDocLanguage(language)
  435. }
  436. const changeToEconomicalType = () => {
  437. if (!hasSetIndexType) {
  438. setIndexType(IndexingType.ECONOMICAL)
  439. setDocForm(DocForm.TEXT)
  440. }
  441. }
  442. const previewSwitch = async () => {
  443. setPreviewSwitched(true)
  444. if (segmentationType === SegmentType.AUTO)
  445. setAutomaticFileIndexingEstimate(null)
  446. else
  447. setCustomFileIndexingEstimate(null)
  448. await fetchFileIndexingEstimate(DocForm.QA)
  449. }
  450. useEffect(() => {
  451. // fetch rules
  452. if (!isSetting) {
  453. getRules()
  454. }
  455. else {
  456. getRulesFromDetail()
  457. getDefaultMode()
  458. }
  459. }, [])
  460. useEffect(() => {
  461. scrollRef.current?.addEventListener('scroll', scrollHandle)
  462. return () => {
  463. scrollRef.current?.removeEventListener('scroll', scrollHandle)
  464. }
  465. }, [])
  466. useLayoutEffect(() => {
  467. if (showPreview) {
  468. previewScrollRef.current?.addEventListener('scroll', previewScrollHandle)
  469. return () => {
  470. previewScrollRef.current?.removeEventListener('scroll', previewScrollHandle)
  471. }
  472. }
  473. }, [showPreview])
  474. useEffect(() => {
  475. if (indexingType === IndexingType.ECONOMICAL && docForm === DocForm.QA)
  476. setDocForm(DocForm.TEXT)
  477. }, [indexingType, docForm])
  478. useEffect(() => {
  479. // get indexing type by props
  480. if (indexingType)
  481. setIndexType(indexingType as IndexingType)
  482. else
  483. setIndexType(isAPIKeySet ? IndexingType.QUALIFIED : IndexingType.ECONOMICAL)
  484. }, [isAPIKeySet, indexingType, datasetId])
  485. useEffect(() => {
  486. if (segmentationType === SegmentType.AUTO) {
  487. setAutomaticFileIndexingEstimate(null)
  488. !isMobile && setShowPreview()
  489. fetchFileIndexingEstimate()
  490. setPreviewSwitched(false)
  491. }
  492. else {
  493. hidePreview()
  494. setCustomFileIndexingEstimate(null)
  495. setPreviewSwitched(false)
  496. }
  497. }, [segmentationType, indexType])
  498. const [retrievalConfig, setRetrievalConfig] = useState(currentDataset?.retrieval_model_dict || {
  499. search_method: RETRIEVE_METHOD.semantic,
  500. reranking_enable: false,
  501. reranking_model: {
  502. reranking_provider_name: rerankDefaultModel?.provider.provider,
  503. reranking_model_name: rerankDefaultModel?.model,
  504. },
  505. top_k: 3,
  506. score_threshold_enabled: false,
  507. score_threshold: 0.5,
  508. } as RetrievalConfig)
  509. return (
  510. <div className='flex w-full h-full'>
  511. <div ref={scrollRef} className='relative h-full w-full overflow-y-scroll'>
  512. <div className={cn(s.pageHeader, scrolled && s.fixed, isMobile && '!px-6')}>
  513. <span>{t('datasetCreation.steps.two')}</span>
  514. {isMobile && (
  515. <Button
  516. 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)]'
  517. onClick={setShowPreview}
  518. >
  519. <Tooltip>
  520. <div className="flex flex-row items-center">
  521. <RocketLaunchIcon className="h-4 w-4 mr-1.5 stroke-[1.8px]" />
  522. <span className="text-[13px]">{t('datasetCreation.stepTwo.previewTitleButton')}</span>
  523. </div>
  524. </Tooltip>
  525. </Button>
  526. )}
  527. </div>
  528. <div className={cn(s.form, isMobile && '!px-4')}>
  529. <div className={s.label}>{t('datasetCreation.stepTwo.segmentation')}</div>
  530. <div className='max-w-[640px]'>
  531. <div
  532. className={cn(
  533. s.radioItem,
  534. s.segmentationItem,
  535. segmentationType === SegmentType.AUTO && s.active,
  536. )}
  537. onClick={() => setSegmentationType(SegmentType.AUTO)}
  538. >
  539. <span className={cn(s.typeIcon, s.auto)} />
  540. <span className={cn(s.radio)} />
  541. <div className={s.typeHeader}>
  542. <div className={s.title}>{t('datasetCreation.stepTwo.auto')}</div>
  543. <div className={s.tip}>{t('datasetCreation.stepTwo.autoDescription')}</div>
  544. </div>
  545. </div>
  546. <div
  547. className={cn(
  548. s.radioItem,
  549. s.segmentationItem,
  550. segmentationType === SegmentType.CUSTOM && s.active,
  551. segmentationType === SegmentType.CUSTOM && s.custom,
  552. )}
  553. onClick={() => setSegmentationType(SegmentType.CUSTOM)}
  554. >
  555. <span className={cn(s.typeIcon, s.customize)} />
  556. <span className={cn(s.radio)} />
  557. <div className={s.typeHeader}>
  558. <div className={s.title}>{t('datasetCreation.stepTwo.custom')}</div>
  559. <div className={s.tip}>{t('datasetCreation.stepTwo.customDescription')}</div>
  560. </div>
  561. {segmentationType === SegmentType.CUSTOM && (
  562. <div className={s.typeFormBody}>
  563. <div className={s.formRow}>
  564. <div className='w-full'>
  565. <div className={s.label}>{t('datasetCreation.stepTwo.separator')}</div>
  566. <input
  567. type="text"
  568. className={s.input}
  569. placeholder={t('datasetCreation.stepTwo.separatorPlaceholder') || ''} value={segmentIdentifier}
  570. onChange={e => setSegmentIdentifier(e.target.value)}
  571. />
  572. </div>
  573. </div>
  574. <div className={s.formRow}>
  575. <div className='w-full'>
  576. <div className={s.label}>{t('datasetCreation.stepTwo.maxLength')}</div>
  577. <input
  578. type="number"
  579. className={s.input}
  580. placeholder={t('datasetCreation.stepTwo.maxLength') || ''}
  581. value={max}
  582. min={1}
  583. onChange={e => setMax(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  584. />
  585. </div>
  586. </div>
  587. <div className={s.formRow}>
  588. <div className='w-full'>
  589. <div className={s.label}>
  590. {t('datasetCreation.stepTwo.overlap')}
  591. <Tooltip
  592. popupContent={
  593. <div className='max-w-[200px]'>
  594. {t('datasetCreation.stepTwo.overlapTip')}
  595. </div>
  596. }
  597. />
  598. </div>
  599. <input
  600. type="number"
  601. className={s.input}
  602. placeholder={t('datasetCreation.stepTwo.overlap') || ''}
  603. value={overlap}
  604. min={1}
  605. onChange={e => setOverlap(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  606. />
  607. </div>
  608. </div>
  609. <div className={s.formRow}>
  610. <div className='w-full flex flex-col gap-1'>
  611. <div className={s.label}>{t('datasetCreation.stepTwo.rules')}</div>
  612. {rules.map(rule => (
  613. <div key={rule.id} className={s.ruleItem}>
  614. <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" />
  615. <label htmlFor={rule.id} className="ml-2 text-sm font-normal cursor-pointer text-gray-800">{getRuleName(rule.id)}</label>
  616. </div>
  617. ))}
  618. </div>
  619. </div>
  620. <div className={s.formFooter}>
  621. <Button variant="primary" className={cn(s.button)} onClick={confirmChangeCustomConfig}>{t('datasetCreation.stepTwo.preview')}</Button>
  622. <Button className={cn(s.button, 'ml-2')} onClick={resetRules}>{t('datasetCreation.stepTwo.reset')}</Button>
  623. </div>
  624. </div>
  625. )}
  626. </div>
  627. </div>
  628. <div className={s.label}>{t('datasetCreation.stepTwo.indexMode')}</div>
  629. <div className='max-w-[640px]'>
  630. <div className='flex items-center gap-3 flex-wrap sm:flex-nowrap'>
  631. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.QUALIFIED)) && (
  632. <div
  633. className={cn(
  634. s.radioItem,
  635. s.indexItem,
  636. !isAPIKeySet && s.disabled,
  637. !hasSetIndexType && indexType === IndexingType.QUALIFIED && s.active,
  638. hasSetIndexType && s.disabled,
  639. hasSetIndexType && '!w-full',
  640. )}
  641. onClick={() => {
  642. if (isAPIKeySet)
  643. setIndexType(IndexingType.QUALIFIED)
  644. }}
  645. >
  646. <span className={cn(s.typeIcon, s.qualified)} />
  647. {!hasSetIndexType && <span className={cn(s.radio)} />}
  648. <div className={s.typeHeader}>
  649. <div className={s.title}>
  650. {t('datasetCreation.stepTwo.qualified')}
  651. {!hasSetIndexType && <span className={s.recommendTag}>{t('datasetCreation.stepTwo.recommend')}</span>}
  652. </div>
  653. <div className={s.tip}>{t('datasetCreation.stepTwo.qualifiedTip')}</div>
  654. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  655. {
  656. estimateTokes
  657. ? (
  658. <div className='text-xs font-medium text-gray-800'>{formatNumber(estimateTokes.tokens)} tokens(<span className='text-yellow-500'>${formatNumber(estimateTokes.total_price)}</span>)</div>
  659. )
  660. : (
  661. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  662. )
  663. }
  664. </div>
  665. {!isAPIKeySet && (
  666. <div className={s.warningTip}>
  667. <span>{t('datasetCreation.stepTwo.warning')}&nbsp;</span>
  668. <span className={s.click} onClick={onSetting}>{t('datasetCreation.stepTwo.click')}</span>
  669. </div>
  670. )}
  671. </div>
  672. )}
  673. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.ECONOMICAL)) && (
  674. <div
  675. className={cn(
  676. s.radioItem,
  677. s.indexItem,
  678. !hasSetIndexType && indexType === IndexingType.ECONOMICAL && s.active,
  679. hasSetIndexType && s.disabled,
  680. hasSetIndexType && '!w-full',
  681. )}
  682. onClick={changeToEconomicalType}
  683. >
  684. <span className={cn(s.typeIcon, s.economical)} />
  685. {!hasSetIndexType && <span className={cn(s.radio)} />}
  686. <div className={s.typeHeader}>
  687. <div className={s.title}>{t('datasetCreation.stepTwo.economical')}</div>
  688. <div className={s.tip}>{t('datasetCreation.stepTwo.economicalTip')}</div>
  689. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  690. <div className='text-xs font-medium text-gray-800'>0 tokens</div>
  691. </div>
  692. </div>
  693. )}
  694. </div>
  695. {hasSetIndexType && (
  696. <div className='mt-2 text-xs text-gray-500 font-medium'>
  697. {t('datasetCreation.stepTwo.indexSettedTip')}
  698. <Link className='text-[#155EEF]' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  699. </div>
  700. )}
  701. {IS_CE_EDITION && indexType === IndexingType.QUALIFIED && (
  702. <div className='mt-3 rounded-xl bg-gray-50 border border-gray-100'>
  703. <div className='flex justify-between items-center px-5 py-4'>
  704. <div className='flex justify-center items-center w-8 h-8 rounded-lg bg-indigo-50'>
  705. <MessageChatSquare className='w-4 h-4' />
  706. </div>
  707. <div className='grow mx-3'>
  708. <div className='mb-[2px] text-md font-medium text-gray-900'>{t('datasetCreation.stepTwo.QATitle')}</div>
  709. <div className='inline-flex items-center text-[13px] leading-[18px] text-gray-500'>
  710. <span className='pr-1'>{t('datasetCreation.stepTwo.QALanguage')}</span>
  711. <LanguageSelect currentLanguage={docLanguage} onSelect={handleSelect} />
  712. </div>
  713. </div>
  714. <div className='shrink-0'>
  715. <Switch
  716. defaultValue={docForm === DocForm.QA}
  717. onChange={handleSwitch}
  718. size='md'
  719. />
  720. </div>
  721. </div>
  722. {docForm === DocForm.QA && !QATipHide && (
  723. <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'>
  724. {t('datasetCreation.stepTwo.QATip')}
  725. <RiCloseLine className='w-4 h-4 text-gray-500 cursor-pointer' onClick={() => setQATipHide(true)} />
  726. </div>
  727. )}
  728. </div>
  729. )}
  730. {/* Retrieval Method Config */}
  731. <div>
  732. {!datasetId
  733. ? (
  734. <div className={s.label}>
  735. {t('datasetSettings.form.retrievalSetting.title')}
  736. <div className='leading-[18px] text-xs font-normal text-gray-500'>
  737. <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-6-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
  738. {t('datasetSettings.form.retrievalSetting.longDescription')}
  739. </div>
  740. </div>
  741. )
  742. : (
  743. <div className={cn(s.label, 'flex justify-between items-center')}>
  744. <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
  745. </div>
  746. )}
  747. <div className='max-w-[640px]'>
  748. {
  749. getIndexing_technique() === IndexingType.QUALIFIED
  750. ? (
  751. <RetrievalMethodConfig
  752. value={retrievalConfig}
  753. onChange={setRetrievalConfig}
  754. />
  755. )
  756. : (
  757. <EconomicalRetrievalMethodConfig
  758. value={retrievalConfig}
  759. onChange={setRetrievalConfig}
  760. />
  761. )
  762. }
  763. </div>
  764. </div>
  765. <div className={s.source}>
  766. <div className={s.sourceContent}>
  767. {dataSourceType === DataSourceType.FILE && (
  768. <>
  769. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.fileSource')}</div>
  770. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  771. <span className={cn(s.fileIcon, files.length && s[files[0].extension || ''])} />
  772. {getFileName(files[0].name || '')}
  773. {files.length > 1 && (
  774. <span className={s.sourceCount}>
  775. <span>{t('datasetCreation.stepTwo.other')}</span>
  776. <span>{files.length - 1}</span>
  777. <span>{t('datasetCreation.stepTwo.fileUnit')}</span>
  778. </span>
  779. )}
  780. </div>
  781. </>
  782. )}
  783. {dataSourceType === DataSourceType.NOTION && (
  784. <>
  785. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.notionSource')}</div>
  786. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  787. <NotionIcon
  788. className='shrink-0 mr-1'
  789. type='page'
  790. src={notionPages[0]?.page_icon}
  791. />
  792. {notionPages[0]?.page_name}
  793. {notionPages.length > 1 && (
  794. <span className={s.sourceCount}>
  795. <span>{t('datasetCreation.stepTwo.other')}</span>
  796. <span>{notionPages.length - 1}</span>
  797. <span>{t('datasetCreation.stepTwo.notionUnit')}</span>
  798. </span>
  799. )}
  800. </div>
  801. </>
  802. )}
  803. {dataSourceType === DataSourceType.WEB && (
  804. <>
  805. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.websiteSource')}</div>
  806. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  807. <Globe01 className='shrink-0 mr-1' />
  808. <span className='grow w-0 truncate'>{websitePages[0].source_url}</span>
  809. {websitePages.length > 1 && (
  810. <span className={s.sourceCount}>
  811. <span>{t('datasetCreation.stepTwo.other')}</span>
  812. <span>{websitePages.length - 1}</span>
  813. <span>{t('datasetCreation.stepTwo.webpageUnit')}</span>
  814. </span>
  815. )}
  816. </div>
  817. </>
  818. )}
  819. </div>
  820. <div className={s.divider} />
  821. <div className={s.segmentCount}>
  822. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateSegment')}</div>
  823. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  824. {
  825. fileIndexingEstimate
  826. ? (
  827. <div className='text-xs font-medium text-gray-800'>{formatNumber(fileIndexingEstimate.total_segments)} </div>
  828. )
  829. : (
  830. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  831. )
  832. }
  833. </div>
  834. </div>
  835. </div>
  836. {!isSetting
  837. ? (
  838. <div className='flex items-center mt-8 py-2'>
  839. <Button onClick={() => onStepChange && onStepChange(-1)}>{t('datasetCreation.stepTwo.previousStep')}</Button>
  840. <div className={s.divider} />
  841. <Button loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.nextStep')}</Button>
  842. </div>
  843. )
  844. : (
  845. <div className='flex items-center mt-8 py-2'>
  846. <Button loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.save')}</Button>
  847. <Button className='ml-2' onClick={onCancel}>{t('datasetCreation.stepTwo.cancel')}</Button>
  848. </div>
  849. )}
  850. </div>
  851. </div>
  852. </div>
  853. <FloatRightContainer isMobile={isMobile} isOpen={showPreview} onClose={hidePreview} footer={null}>
  854. {showPreview && <div ref={previewScrollRef} className={cn(s.previewWrap, isMobile && s.isMobile, 'relative h-full overflow-y-scroll border-l border-[#F2F4F7]')}>
  855. <div className={cn(s.previewHeader, previewScrolled && `${s.fixed} pb-3`)}>
  856. <div className='flex items-center justify-between px-8'>
  857. <div className='grow flex items-center'>
  858. <div>{t('datasetCreation.stepTwo.previewTitle')}</div>
  859. {docForm === DocForm.QA && !previewSwitched && (
  860. <Button className='ml-2' variant='secondary-accent' onClick={previewSwitch}>{t('datasetCreation.stepTwo.previewButton')}</Button>
  861. )}
  862. </div>
  863. <div className='flex items-center justify-center w-6 h-6 cursor-pointer' onClick={hidePreview}>
  864. <XMarkIcon className='h-4 w-4'></XMarkIcon>
  865. </div>
  866. </div>
  867. {docForm === DocForm.QA && !previewSwitched && (
  868. <div className='px-8 pr-12 text-xs text-gray-500'>
  869. <span>{t('datasetCreation.stepTwo.previewSwitchTipStart')}</span>
  870. <span className='text-amber-600'>{t('datasetCreation.stepTwo.previewSwitchTipEnd')}</span>
  871. </div>
  872. )}
  873. </div>
  874. <div className='my-4 px-8 space-y-4'>
  875. {previewSwitched && docForm === DocForm.QA && fileIndexingEstimate?.qa_preview && (
  876. <>
  877. {fileIndexingEstimate?.qa_preview.map((item, index) => (
  878. <PreviewItem type={PreviewType.QA} key={item.question} qa={item} index={index + 1} />
  879. ))}
  880. </>
  881. )}
  882. {(docForm === DocForm.TEXT || !previewSwitched) && fileIndexingEstimate?.preview && (
  883. <>
  884. {fileIndexingEstimate?.preview.map((item, index) => (
  885. <PreviewItem type={PreviewType.TEXT} key={item} content={item} index={index + 1} />
  886. ))}
  887. </>
  888. )}
  889. {previewSwitched && docForm === DocForm.QA && !fileIndexingEstimate?.qa_preview && (
  890. <div className='flex items-center justify-center h-[200px]'>
  891. <Loading type='area' />
  892. </div>
  893. )}
  894. {!previewSwitched && !fileIndexingEstimate?.preview && (
  895. <div className='flex items-center justify-center h-[200px]'>
  896. <Loading type='area' />
  897. </div>
  898. )}
  899. </div>
  900. </div>}
  901. {!showPreview && (
  902. <div className={cn(s.sideTip)}>
  903. <div className={s.tipCard}>
  904. <span className={s.icon} />
  905. <div className={s.title}>{t('datasetCreation.stepTwo.sideTipTitle')}</div>
  906. <div className={s.content}>
  907. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP1')}</p>
  908. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP2')}</p>
  909. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP3')}</p>
  910. <p>{t('datasetCreation.stepTwo.sideTipP4')}</p>
  911. </div>
  912. </div>
  913. </div>
  914. )}
  915. </FloatRightContainer>
  916. </div>
  917. )
  918. }
  919. export default StepTwo