index.tsx 39 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974
  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 RetrievalMethodInfo from '../../common/retrieval-method-info'
  14. import PreviewItem, { PreviewType } from './preview-item'
  15. import LanguageSelect from './language-select'
  16. import s from './index.module.css'
  17. import cn from '@/utils/classnames'
  18. import type { CrawlOptions, CrawlResultItem, CreateDocumentReq, CustomFile, FileIndexingEstimateResponse, FullDocumentDetail, IndexingEstimateParams, IndexingEstimateResponse, NotionInfo, PreProcessingRule, ProcessRule, Rules, createDocumentResponse } from '@/models/datasets'
  19. import {
  20. createDocument,
  21. createFirstDocument,
  22. fetchFileIndexingEstimate as didFetchFileIndexingEstimate,
  23. fetchDefaultProcessRule,
  24. } from '@/service/datasets'
  25. import Button from '@/app/components/base/button'
  26. import Loading from '@/app/components/base/loading'
  27. import FloatRightContainer from '@/app/components/base/float-right-container'
  28. import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
  29. import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
  30. import { type RetrievalConfig } from '@/types/app'
  31. import { ensureRerankModelSelected, isReRankModelSelected } from '@/app/components/datasets/common/check-rerank-model'
  32. import Toast from '@/app/components/base/toast'
  33. import { formatNumber } from '@/utils/format'
  34. import type { NotionPage } from '@/models/common'
  35. import { DataSourceType, DocForm } from '@/models/datasets'
  36. import NotionIcon from '@/app/components/base/notion-icon'
  37. import Switch from '@/app/components/base/switch'
  38. import { MessageChatSquare } from '@/app/components/base/icons/src/public/common'
  39. import { useDatasetDetailContext } from '@/context/dataset-detail'
  40. import I18n from '@/context/i18n'
  41. import { IS_CE_EDITION } from '@/config'
  42. import { RETRIEVE_METHOD } from '@/types/app'
  43. import useBreakpoints, { MediaType } from '@/hooks/use-breakpoints'
  44. import Tooltip from '@/app/components/base/tooltip'
  45. import { useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
  46. import { LanguagesSupported } from '@/i18n/language'
  47. import { ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
  48. import { Globe01 } from '@/app/components/base/icons/src/vender/line/mapsAndTravel'
  49. type ValueOf<T> = T[keyof T]
  50. type StepTwoProps = {
  51. isSetting?: boolean
  52. documentDetail?: FullDocumentDetail
  53. isAPIKeySet: boolean
  54. onSetting: () => void
  55. datasetId?: string
  56. indexingType?: ValueOf<IndexingType>
  57. dataSourceType: DataSourceType
  58. files: CustomFile[]
  59. notionPages?: NotionPage[]
  60. websitePages?: CrawlResultItem[]
  61. crawlOptions?: CrawlOptions
  62. fireCrawlJobId?: string
  63. onStepChange?: (delta: number) => void
  64. updateIndexingTypeCache?: (type: string) => void
  65. updateResultCache?: (res: createDocumentResponse) => void
  66. onSave?: () => void
  67. onCancel?: () => void
  68. }
  69. enum SegmentType {
  70. AUTO = 'automatic',
  71. CUSTOM = 'custom',
  72. }
  73. enum IndexingType {
  74. QUALIFIED = 'high_quality',
  75. ECONOMICAL = 'economy',
  76. }
  77. const StepTwo = ({
  78. isSetting,
  79. documentDetail,
  80. isAPIKeySet,
  81. onSetting,
  82. datasetId,
  83. indexingType,
  84. dataSourceType: inCreatePageDataSourceType,
  85. files,
  86. notionPages = [],
  87. websitePages = [],
  88. crawlOptions,
  89. fireCrawlJobId = '',
  90. onStepChange,
  91. updateIndexingTypeCache,
  92. updateResultCache,
  93. onSave,
  94. onCancel,
  95. }: StepTwoProps) => {
  96. const { t } = useTranslation()
  97. const { locale } = useContext(I18n)
  98. const media = useBreakpoints()
  99. const isMobile = media === MediaType.mobile
  100. const { dataset: currentDataset, mutateDatasetRes } = useDatasetDetailContext()
  101. const isInCreatePage = !datasetId || (datasetId && !currentDataset?.data_source_type)
  102. const dataSourceType = isInCreatePage ? inCreatePageDataSourceType : currentDataset?.data_source_type
  103. const scrollRef = useRef<HTMLDivElement>(null)
  104. const [scrolled, setScrolled] = useState(false)
  105. const previewScrollRef = useRef<HTMLDivElement>(null)
  106. const [previewScrolled, setPreviewScrolled] = useState(false)
  107. const [segmentationType, setSegmentationType] = useState<SegmentType>(SegmentType.AUTO)
  108. const [segmentIdentifier, setSegmentIdentifier] = useState('\\n')
  109. const [max, setMax] = useState(500)
  110. const [overlap, setOverlap] = useState(50)
  111. const [rules, setRules] = useState<PreProcessingRule[]>([])
  112. const [defaultConfig, setDefaultConfig] = useState<Rules>()
  113. const hasSetIndexType = !!indexingType
  114. const [indexType, setIndexType] = useState<ValueOf<IndexingType>>(
  115. (indexingType
  116. || isAPIKeySet)
  117. ? IndexingType.QUALIFIED
  118. : IndexingType.ECONOMICAL,
  119. )
  120. const [docForm, setDocForm] = useState<DocForm | string>(
  121. (datasetId && documentDetail) ? documentDetail.doc_form : DocForm.TEXT,
  122. )
  123. const [docLanguage, setDocLanguage] = useState<string>(locale !== LanguagesSupported[1] ? 'English' : 'Chinese')
  124. const [QATipHide, setQATipHide] = useState(false)
  125. const [previewSwitched, setPreviewSwitched] = useState(false)
  126. const [showPreview, { setTrue: setShowPreview, setFalse: hidePreview }] = useBoolean()
  127. const [customFileIndexingEstimate, setCustomFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  128. const [automaticFileIndexingEstimate, setAutomaticFileIndexingEstimate] = useState<FileIndexingEstimateResponse | null>(null)
  129. const [estimateTokes, setEstimateTokes] = useState<Pick<IndexingEstimateResponse, 'tokens' | 'total_price'> | null>(null)
  130. const fileIndexingEstimate = (() => {
  131. return segmentationType === SegmentType.AUTO ? automaticFileIndexingEstimate : customFileIndexingEstimate
  132. })()
  133. const [isCreating, setIsCreating] = useState(false)
  134. const scrollHandle = (e: Event) => {
  135. if ((e.target as HTMLDivElement).scrollTop > 0)
  136. setScrolled(true)
  137. else
  138. setScrolled(false)
  139. }
  140. const previewScrollHandle = (e: Event) => {
  141. if ((e.target as HTMLDivElement).scrollTop > 0)
  142. setPreviewScrolled(true)
  143. else
  144. setPreviewScrolled(false)
  145. }
  146. const getFileName = (name: string) => {
  147. const arr = name.split('.')
  148. return arr.slice(0, -1).join('.')
  149. }
  150. const getRuleName = (key: string) => {
  151. if (key === 'remove_extra_spaces')
  152. return t('datasetCreation.stepTwo.removeExtraSpaces')
  153. if (key === 'remove_urls_emails')
  154. return t('datasetCreation.stepTwo.removeUrlEmails')
  155. if (key === 'remove_stopwords')
  156. return t('datasetCreation.stepTwo.removeStopwords')
  157. }
  158. const ruleChangeHandle = (id: string) => {
  159. const newRules = rules.map((rule) => {
  160. if (rule.id === id) {
  161. return {
  162. id: rule.id,
  163. enabled: !rule.enabled,
  164. }
  165. }
  166. return rule
  167. })
  168. setRules(newRules)
  169. }
  170. const resetRules = () => {
  171. if (defaultConfig) {
  172. setSegmentIdentifier((defaultConfig.segmentation.separator === '\n' ? '\\n' : defaultConfig.segmentation.separator) || '\\n')
  173. setMax(defaultConfig.segmentation.max_tokens)
  174. setOverlap(defaultConfig.segmentation.chunk_overlap)
  175. setRules(defaultConfig.pre_processing_rules)
  176. }
  177. }
  178. const fetchFileIndexingEstimate = async (docForm = DocForm.TEXT) => {
  179. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  180. const res = await didFetchFileIndexingEstimate(getFileIndexingEstimateParams(docForm)!)
  181. if (segmentationType === SegmentType.CUSTOM) {
  182. setCustomFileIndexingEstimate(res)
  183. }
  184. else {
  185. setAutomaticFileIndexingEstimate(res)
  186. indexType === IndexingType.QUALIFIED && setEstimateTokes({ tokens: res.tokens, total_price: res.total_price })
  187. }
  188. }
  189. const confirmChangeCustomConfig = () => {
  190. setCustomFileIndexingEstimate(null)
  191. setShowPreview()
  192. fetchFileIndexingEstimate()
  193. setPreviewSwitched(false)
  194. }
  195. const getIndexing_technique = () => indexingType || indexType
  196. const getProcessRule = () => {
  197. const processRule: ProcessRule = {
  198. rules: {} as any, // api will check this. It will be removed after api refactored.
  199. mode: segmentationType,
  200. }
  201. if (segmentationType === SegmentType.CUSTOM) {
  202. const ruleObj = {
  203. pre_processing_rules: rules,
  204. segmentation: {
  205. separator: segmentIdentifier === '\\n' ? '\n' : segmentIdentifier,
  206. max_tokens: max,
  207. chunk_overlap: overlap,
  208. },
  209. }
  210. processRule.rules = ruleObj
  211. }
  212. return processRule
  213. }
  214. const getNotionInfo = () => {
  215. const workspacesMap = groupBy(notionPages, 'workspace_id')
  216. const workspaces = Object.keys(workspacesMap).map((workspaceId) => {
  217. return {
  218. workspaceId,
  219. pages: workspacesMap[workspaceId],
  220. }
  221. })
  222. return workspaces.map((workspace) => {
  223. return {
  224. workspace_id: workspace.workspaceId,
  225. pages: workspace.pages.map((page) => {
  226. const { page_id, page_name, page_icon, type } = page
  227. return {
  228. page_id,
  229. page_name,
  230. page_icon,
  231. type,
  232. }
  233. }),
  234. }
  235. }) as NotionInfo[]
  236. }
  237. const getWebsiteInfo = () => {
  238. return {
  239. provider: 'firecrawl',
  240. job_id: fireCrawlJobId,
  241. urls: websitePages.map(page => page.source_url),
  242. only_main_content: crawlOptions?.only_main_content,
  243. }
  244. }
  245. const getFileIndexingEstimateParams = (docForm: DocForm): IndexingEstimateParams | undefined => {
  246. if (dataSourceType === DataSourceType.FILE) {
  247. return {
  248. info_list: {
  249. data_source_type: dataSourceType,
  250. file_info_list: {
  251. file_ids: files.map(file => file.id) as string[],
  252. },
  253. },
  254. indexing_technique: getIndexing_technique() as string,
  255. process_rule: getProcessRule(),
  256. doc_form: docForm,
  257. doc_language: docLanguage,
  258. dataset_id: datasetId as string,
  259. }
  260. }
  261. if (dataSourceType === DataSourceType.NOTION) {
  262. return {
  263. info_list: {
  264. data_source_type: dataSourceType,
  265. notion_info_list: getNotionInfo(),
  266. },
  267. indexing_technique: getIndexing_technique() as string,
  268. process_rule: getProcessRule(),
  269. doc_form: docForm,
  270. doc_language: docLanguage,
  271. dataset_id: datasetId as string,
  272. }
  273. }
  274. if (dataSourceType === DataSourceType.WEB) {
  275. return {
  276. info_list: {
  277. data_source_type: dataSourceType,
  278. website_info_list: getWebsiteInfo(),
  279. },
  280. indexing_technique: getIndexing_technique() as string,
  281. process_rule: getProcessRule(),
  282. doc_form: docForm,
  283. doc_language: docLanguage,
  284. dataset_id: datasetId as string,
  285. }
  286. }
  287. }
  288. const {
  289. modelList: rerankModelList,
  290. defaultModel: rerankDefaultModel,
  291. currentModel: isRerankDefaultModelVaild,
  292. } = useModelListAndDefaultModelAndCurrentProviderAndModel(ModelTypeEnum.rerank)
  293. const getCreationParams = () => {
  294. let params
  295. if (segmentationType === SegmentType.CUSTOM && overlap > max) {
  296. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.overlapCheck') })
  297. return
  298. }
  299. if (isSetting) {
  300. params = {
  301. original_document_id: documentDetail?.id,
  302. doc_form: docForm,
  303. doc_language: docLanguage,
  304. process_rule: getProcessRule(),
  305. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  306. retrieval_model: retrievalConfig, // Readonly. If want to changed, just go to settings page.
  307. } as CreateDocumentReq
  308. }
  309. else { // create
  310. const indexMethod = getIndexing_technique()
  311. if (
  312. !isReRankModelSelected({
  313. rerankDefaultModel,
  314. isRerankDefaultModelVaild: !!isRerankDefaultModelVaild,
  315. rerankModelList,
  316. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  317. retrievalConfig,
  318. indexMethod: indexMethod as string,
  319. })
  320. ) {
  321. Toast.notify({ type: 'error', message: t('appDebug.datasetConfig.rerankModelRequired') })
  322. return
  323. }
  324. const postRetrievalConfig = ensureRerankModelSelected({
  325. rerankDefaultModel: rerankDefaultModel!,
  326. // eslint-disable-next-line @typescript-eslint/no-use-before-define
  327. retrievalConfig,
  328. indexMethod: indexMethod as string,
  329. })
  330. params = {
  331. data_source: {
  332. type: dataSourceType,
  333. info_list: {
  334. data_source_type: dataSourceType,
  335. },
  336. },
  337. indexing_technique: getIndexing_technique(),
  338. process_rule: getProcessRule(),
  339. doc_form: docForm,
  340. doc_language: docLanguage,
  341. retrieval_model: postRetrievalConfig,
  342. } as CreateDocumentReq
  343. if (dataSourceType === DataSourceType.FILE) {
  344. params.data_source.info_list.file_info_list = {
  345. file_ids: files.map(file => file.id || '').filter(Boolean),
  346. }
  347. }
  348. if (dataSourceType === DataSourceType.NOTION)
  349. params.data_source.info_list.notion_info_list = getNotionInfo()
  350. if (dataSourceType === DataSourceType.WEB)
  351. params.data_source.info_list.website_info_list = getWebsiteInfo()
  352. }
  353. return params
  354. }
  355. const getRules = async () => {
  356. try {
  357. const res = await fetchDefaultProcessRule({ url: '/datasets/process-rule' })
  358. const separator = res.rules.segmentation.separator
  359. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  360. setMax(res.rules.segmentation.max_tokens)
  361. setOverlap(res.rules.segmentation.chunk_overlap)
  362. setRules(res.rules.pre_processing_rules)
  363. setDefaultConfig(res.rules)
  364. }
  365. catch (err) {
  366. console.log(err)
  367. }
  368. }
  369. const getRulesFromDetail = () => {
  370. if (documentDetail) {
  371. const rules = documentDetail.dataset_process_rule.rules
  372. const separator = rules.segmentation.separator
  373. const max = rules.segmentation.max_tokens
  374. const overlap = rules.segmentation.chunk_overlap
  375. setSegmentIdentifier((separator === '\n' ? '\\n' : separator) || '\\n')
  376. setMax(max)
  377. setOverlap(overlap)
  378. setRules(rules.pre_processing_rules)
  379. setDefaultConfig(rules)
  380. }
  381. }
  382. const getDefaultMode = () => {
  383. if (documentDetail)
  384. setSegmentationType(documentDetail.dataset_process_rule.mode)
  385. }
  386. const createHandle = async () => {
  387. if (isCreating)
  388. return
  389. setIsCreating(true)
  390. try {
  391. let res
  392. const params = getCreationParams()
  393. if (!params)
  394. return false
  395. setIsCreating(true)
  396. if (!datasetId) {
  397. res = await createFirstDocument({
  398. body: params as CreateDocumentReq,
  399. })
  400. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  401. updateResultCache && updateResultCache(res)
  402. }
  403. else {
  404. res = await createDocument({
  405. datasetId,
  406. body: params as CreateDocumentReq,
  407. })
  408. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  409. updateResultCache && updateResultCache(res)
  410. }
  411. if (mutateDatasetRes)
  412. mutateDatasetRes()
  413. onStepChange && onStepChange(+1)
  414. isSetting && onSave && onSave()
  415. }
  416. catch (err) {
  417. Toast.notify({
  418. type: 'error',
  419. message: `${err}`,
  420. })
  421. }
  422. finally {
  423. setIsCreating(false)
  424. }
  425. }
  426. const handleSwitch = (state: boolean) => {
  427. if (state)
  428. setDocForm(DocForm.QA)
  429. else
  430. setDocForm(DocForm.TEXT)
  431. }
  432. const handleSelect = (language: string) => {
  433. setDocLanguage(language)
  434. }
  435. const changeToEconomicalType = () => {
  436. if (!hasSetIndexType) {
  437. setIndexType(IndexingType.ECONOMICAL)
  438. setDocForm(DocForm.TEXT)
  439. }
  440. }
  441. const previewSwitch = async () => {
  442. setPreviewSwitched(true)
  443. if (segmentationType === SegmentType.AUTO)
  444. setAutomaticFileIndexingEstimate(null)
  445. else
  446. setCustomFileIndexingEstimate(null)
  447. await fetchFileIndexingEstimate(DocForm.QA)
  448. }
  449. useEffect(() => {
  450. // fetch rules
  451. if (!isSetting) {
  452. getRules()
  453. }
  454. else {
  455. getRulesFromDetail()
  456. getDefaultMode()
  457. }
  458. }, [])
  459. useEffect(() => {
  460. scrollRef.current?.addEventListener('scroll', scrollHandle)
  461. return () => {
  462. scrollRef.current?.removeEventListener('scroll', scrollHandle)
  463. }
  464. }, [])
  465. useLayoutEffect(() => {
  466. if (showPreview) {
  467. previewScrollRef.current?.addEventListener('scroll', previewScrollHandle)
  468. return () => {
  469. previewScrollRef.current?.removeEventListener('scroll', previewScrollHandle)
  470. }
  471. }
  472. }, [showPreview])
  473. useEffect(() => {
  474. if (indexingType === IndexingType.ECONOMICAL && docForm === DocForm.QA)
  475. setDocForm(DocForm.TEXT)
  476. }, [indexingType, docForm])
  477. useEffect(() => {
  478. // get indexing type by props
  479. if (indexingType)
  480. setIndexType(indexingType as IndexingType)
  481. else
  482. setIndexType(isAPIKeySet ? IndexingType.QUALIFIED : IndexingType.ECONOMICAL)
  483. }, [isAPIKeySet, indexingType, datasetId])
  484. useEffect(() => {
  485. if (segmentationType === SegmentType.AUTO) {
  486. setAutomaticFileIndexingEstimate(null)
  487. !isMobile && setShowPreview()
  488. fetchFileIndexingEstimate()
  489. setPreviewSwitched(false)
  490. }
  491. else {
  492. hidePreview()
  493. setCustomFileIndexingEstimate(null)
  494. setPreviewSwitched(false)
  495. }
  496. }, [segmentationType, indexType])
  497. const [retrievalConfig, setRetrievalConfig] = useState(currentDataset?.retrieval_model_dict || {
  498. search_method: RETRIEVE_METHOD.semantic,
  499. reranking_enable: false,
  500. reranking_model: {
  501. reranking_provider_name: rerankDefaultModel?.provider.provider,
  502. reranking_model_name: rerankDefaultModel?.model,
  503. },
  504. top_k: 3,
  505. score_threshold_enabled: false,
  506. score_threshold: 0.5,
  507. } as RetrievalConfig)
  508. return (
  509. <div className='flex w-full h-full'>
  510. <div ref={scrollRef} className='relative h-full w-full overflow-y-scroll'>
  511. <div className={cn(s.pageHeader, scrolled && s.fixed, isMobile && '!px-6')}>
  512. <span>{t('datasetCreation.steps.two')}</span>
  513. {isMobile && (
  514. <Button
  515. 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)]'
  516. onClick={setShowPreview}
  517. >
  518. <Tooltip>
  519. <div className="flex flex-row items-center">
  520. <RocketLaunchIcon className="h-4 w-4 mr-1.5 stroke-[1.8px]" />
  521. <span className="text-[13px]">{t('datasetCreation.stepTwo.previewTitleButton')}</span>
  522. </div>
  523. </Tooltip>
  524. </Button>
  525. )}
  526. </div>
  527. <div className={cn(s.form, isMobile && '!px-4')}>
  528. <div className={s.label}>{t('datasetCreation.stepTwo.segmentation')}</div>
  529. <div className='max-w-[640px]'>
  530. <div
  531. className={cn(
  532. s.radioItem,
  533. s.segmentationItem,
  534. segmentationType === SegmentType.AUTO && s.active,
  535. )}
  536. onClick={() => setSegmentationType(SegmentType.AUTO)}
  537. >
  538. <span className={cn(s.typeIcon, s.auto)} />
  539. <span className={cn(s.radio)} />
  540. <div className={s.typeHeader}>
  541. <div className={s.title}>{t('datasetCreation.stepTwo.auto')}</div>
  542. <div className={s.tip}>{t('datasetCreation.stepTwo.autoDescription')}</div>
  543. </div>
  544. </div>
  545. <div
  546. className={cn(
  547. s.radioItem,
  548. s.segmentationItem,
  549. segmentationType === SegmentType.CUSTOM && s.active,
  550. segmentationType === SegmentType.CUSTOM && s.custom,
  551. )}
  552. onClick={() => setSegmentationType(SegmentType.CUSTOM)}
  553. >
  554. <span className={cn(s.typeIcon, s.customize)} />
  555. <span className={cn(s.radio)} />
  556. <div className={s.typeHeader}>
  557. <div className={s.title}>{t('datasetCreation.stepTwo.custom')}</div>
  558. <div className={s.tip}>{t('datasetCreation.stepTwo.customDescription')}</div>
  559. </div>
  560. {segmentationType === SegmentType.CUSTOM && (
  561. <div className={s.typeFormBody}>
  562. <div className={s.formRow}>
  563. <div className='w-full'>
  564. <div className={s.label}>{t('datasetCreation.stepTwo.separator')}</div>
  565. <input
  566. type="text"
  567. className={s.input}
  568. placeholder={t('datasetCreation.stepTwo.separatorPlaceholder') || ''} value={segmentIdentifier}
  569. onChange={e => setSegmentIdentifier(e.target.value)}
  570. />
  571. </div>
  572. </div>
  573. <div className={s.formRow}>
  574. <div className='w-full'>
  575. <div className={s.label}>{t('datasetCreation.stepTwo.maxLength')}</div>
  576. <input
  577. type="number"
  578. className={s.input}
  579. placeholder={t('datasetCreation.stepTwo.maxLength') || ''}
  580. value={max}
  581. min={1}
  582. onChange={e => setMax(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  583. />
  584. </div>
  585. </div>
  586. <div className={s.formRow}>
  587. <div className='w-full'>
  588. <div className={s.label}>
  589. {t('datasetCreation.stepTwo.overlap')}
  590. <Tooltip
  591. popupContent={
  592. <div className='max-w-[200px]'>
  593. {t('datasetCreation.stepTwo.overlapTip')}
  594. </div>
  595. }
  596. />
  597. </div>
  598. <input
  599. type="number"
  600. className={s.input}
  601. placeholder={t('datasetCreation.stepTwo.overlap') || ''}
  602. value={overlap}
  603. min={1}
  604. onChange={e => setOverlap(parseInt(e.target.value.replace(/^0+/, ''), 10))}
  605. />
  606. </div>
  607. </div>
  608. <div className={s.formRow}>
  609. <div className='w-full flex flex-col gap-1'>
  610. <div className={s.label}>{t('datasetCreation.stepTwo.rules')}</div>
  611. {rules.map(rule => (
  612. <div key={rule.id} className={s.ruleItem}>
  613. <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" />
  614. <label htmlFor={rule.id} className="ml-2 text-sm font-normal cursor-pointer text-gray-800">{getRuleName(rule.id)}</label>
  615. </div>
  616. ))}
  617. </div>
  618. </div>
  619. <div className={s.formFooter}>
  620. <Button variant="primary" className={cn(s.button)} onClick={confirmChangeCustomConfig}>{t('datasetCreation.stepTwo.preview')}</Button>
  621. <Button className={cn(s.button, 'ml-2')} onClick={resetRules}>{t('datasetCreation.stepTwo.reset')}</Button>
  622. </div>
  623. </div>
  624. )}
  625. </div>
  626. </div>
  627. <div className={s.label}>{t('datasetCreation.stepTwo.indexMode')}</div>
  628. <div className='max-w-[640px]'>
  629. <div className='flex items-center gap-3 flex-wrap sm:flex-nowrap'>
  630. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.QUALIFIED)) && (
  631. <div
  632. className={cn(
  633. s.radioItem,
  634. s.indexItem,
  635. !isAPIKeySet && s.disabled,
  636. !hasSetIndexType && indexType === IndexingType.QUALIFIED && s.active,
  637. hasSetIndexType && s.disabled,
  638. hasSetIndexType && '!w-full',
  639. )}
  640. onClick={() => {
  641. if (isAPIKeySet)
  642. setIndexType(IndexingType.QUALIFIED)
  643. }}
  644. >
  645. <span className={cn(s.typeIcon, s.qualified)} />
  646. {!hasSetIndexType && <span className={cn(s.radio)} />}
  647. <div className={s.typeHeader}>
  648. <div className={s.title}>
  649. {t('datasetCreation.stepTwo.qualified')}
  650. {!hasSetIndexType && <span className={s.recommendTag}>{t('datasetCreation.stepTwo.recommend')}</span>}
  651. </div>
  652. <div className={s.tip}>{t('datasetCreation.stepTwo.qualifiedTip')}</div>
  653. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  654. {
  655. estimateTokes
  656. ? (
  657. <div className='text-xs font-medium text-gray-800'>{formatNumber(estimateTokes.tokens)} tokens(<span className='text-yellow-500'>${formatNumber(estimateTokes.total_price)}</span>)</div>
  658. )
  659. : (
  660. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  661. )
  662. }
  663. </div>
  664. {!isAPIKeySet && (
  665. <div className={s.warningTip}>
  666. <span>{t('datasetCreation.stepTwo.warning')}&nbsp;</span>
  667. <span className={s.click} onClick={onSetting}>{t('datasetCreation.stepTwo.click')}</span>
  668. </div>
  669. )}
  670. </div>
  671. )}
  672. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.ECONOMICAL)) && (
  673. <div
  674. className={cn(
  675. s.radioItem,
  676. s.indexItem,
  677. !hasSetIndexType && indexType === IndexingType.ECONOMICAL && s.active,
  678. hasSetIndexType && s.disabled,
  679. hasSetIndexType && '!w-full',
  680. )}
  681. onClick={changeToEconomicalType}
  682. >
  683. <span className={cn(s.typeIcon, s.economical)} />
  684. {!hasSetIndexType && <span className={cn(s.radio)} />}
  685. <div className={s.typeHeader}>
  686. <div className={s.title}>{t('datasetCreation.stepTwo.economical')}</div>
  687. <div className={s.tip}>{t('datasetCreation.stepTwo.economicalTip')}</div>
  688. <div className='pb-0.5 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateCost')}</div>
  689. <div className='text-xs font-medium text-gray-800'>0 tokens</div>
  690. </div>
  691. </div>
  692. )}
  693. </div>
  694. {hasSetIndexType && (
  695. <div className='mt-2 text-xs text-gray-500 font-medium'>
  696. {t('datasetCreation.stepTwo.indexSettedTip')}
  697. <Link className='text-[#155EEF]' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  698. </div>
  699. )}
  700. {IS_CE_EDITION && indexType === IndexingType.QUALIFIED && (
  701. <div className='mt-3 rounded-xl bg-gray-50 border border-gray-100'>
  702. <div className='flex justify-between items-center px-5 py-4'>
  703. <div className='flex justify-center items-center w-8 h-8 rounded-lg bg-indigo-50'>
  704. <MessageChatSquare className='w-4 h-4' />
  705. </div>
  706. <div className='grow mx-3'>
  707. <div className='mb-[2px] text-md font-medium text-gray-900'>{t('datasetCreation.stepTwo.QATitle')}</div>
  708. <div className='inline-flex items-center text-[13px] leading-[18px] text-gray-500'>
  709. <span className='pr-1'>{t('datasetCreation.stepTwo.QALanguage')}</span>
  710. <LanguageSelect currentLanguage={docLanguage} onSelect={handleSelect} />
  711. </div>
  712. </div>
  713. <div className='shrink-0'>
  714. <Switch
  715. defaultValue={docForm === DocForm.QA}
  716. onChange={handleSwitch}
  717. size='md'
  718. />
  719. </div>
  720. </div>
  721. {docForm === DocForm.QA && !QATipHide && (
  722. <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'>
  723. {t('datasetCreation.stepTwo.QATip')}
  724. <RiCloseLine className='w-4 h-4 text-gray-500 cursor-pointer' onClick={() => setQATipHide(true)} />
  725. </div>
  726. )}
  727. </div>
  728. )}
  729. {/* Retrieval Method Config */}
  730. <div>
  731. {!datasetId
  732. ? (
  733. <div className={s.label}>
  734. {t('datasetSettings.form.retrievalSetting.title')}
  735. <div className='leading-[18px] text-xs font-normal text-gray-500'>
  736. <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>
  737. {t('datasetSettings.form.retrievalSetting.longDescription')}
  738. </div>
  739. </div>
  740. )
  741. : (
  742. <div className={cn(s.label, 'flex justify-between items-center')}>
  743. <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
  744. </div>
  745. )}
  746. <div className='max-w-[640px]'>
  747. {!datasetId
  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. : (
  764. <div>
  765. <RetrievalMethodInfo
  766. value={retrievalConfig}
  767. />
  768. <div className='mt-2 text-xs text-gray-500 font-medium'>
  769. {t('datasetCreation.stepTwo.retrivalSettedTip')}
  770. <Link className='text-[#155EEF]' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  771. </div>
  772. </div>
  773. )}
  774. </div>
  775. </div>
  776. <div className={s.source}>
  777. <div className={s.sourceContent}>
  778. {dataSourceType === DataSourceType.FILE && (
  779. <>
  780. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.fileSource')}</div>
  781. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  782. <span className={cn(s.fileIcon, files.length && s[files[0].extension || ''])} />
  783. {getFileName(files[0].name || '')}
  784. {files.length > 1 && (
  785. <span className={s.sourceCount}>
  786. <span>{t('datasetCreation.stepTwo.other')}</span>
  787. <span>{files.length - 1}</span>
  788. <span>{t('datasetCreation.stepTwo.fileUnit')}</span>
  789. </span>
  790. )}
  791. </div>
  792. </>
  793. )}
  794. {dataSourceType === DataSourceType.NOTION && (
  795. <>
  796. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.notionSource')}</div>
  797. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  798. <NotionIcon
  799. className='shrink-0 mr-1'
  800. type='page'
  801. src={notionPages[0]?.page_icon}
  802. />
  803. {notionPages[0]?.page_name}
  804. {notionPages.length > 1 && (
  805. <span className={s.sourceCount}>
  806. <span>{t('datasetCreation.stepTwo.other')}</span>
  807. <span>{notionPages.length - 1}</span>
  808. <span>{t('datasetCreation.stepTwo.notionUnit')}</span>
  809. </span>
  810. )}
  811. </div>
  812. </>
  813. )}
  814. {dataSourceType === DataSourceType.WEB && (
  815. <>
  816. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.websiteSource')}</div>
  817. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  818. <Globe01 className='shrink-0 mr-1' />
  819. <span className='grow w-0 truncate'>{websitePages[0].source_url}</span>
  820. {websitePages.length > 1 && (
  821. <span className={s.sourceCount}>
  822. <span>{t('datasetCreation.stepTwo.other')}</span>
  823. <span>{websitePages.length - 1}</span>
  824. <span>{t('datasetCreation.stepTwo.webpageUnit')}</span>
  825. </span>
  826. )}
  827. </div>
  828. </>
  829. )}
  830. </div>
  831. <div className={s.divider} />
  832. <div className={s.segmentCount}>
  833. <div className='mb-2 text-xs font-medium text-gray-500'>{t('datasetCreation.stepTwo.emstimateSegment')}</div>
  834. <div className='flex items-center text-sm leading-6 font-medium text-gray-800'>
  835. {
  836. fileIndexingEstimate
  837. ? (
  838. <div className='text-xs font-medium text-gray-800'>{formatNumber(fileIndexingEstimate.total_segments)} </div>
  839. )
  840. : (
  841. <div className={s.calculating}>{t('datasetCreation.stepTwo.calculating')}</div>
  842. )
  843. }
  844. </div>
  845. </div>
  846. </div>
  847. {!isSetting
  848. ? (
  849. <div className='flex items-center mt-8 py-2'>
  850. <Button onClick={() => onStepChange && onStepChange(-1)}>{t('datasetCreation.stepTwo.previousStep')}</Button>
  851. <div className={s.divider} />
  852. <Button loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.nextStep')}</Button>
  853. </div>
  854. )
  855. : (
  856. <div className='flex items-center mt-8 py-2'>
  857. <Button loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.save')}</Button>
  858. <Button className='ml-2' onClick={onCancel}>{t('datasetCreation.stepTwo.cancel')}</Button>
  859. </div>
  860. )}
  861. </div>
  862. </div>
  863. </div>
  864. <FloatRightContainer isMobile={isMobile} isOpen={showPreview} onClose={hidePreview} footer={null}>
  865. {showPreview && <div ref={previewScrollRef} className={cn(s.previewWrap, isMobile && s.isMobile, 'relative h-full overflow-y-scroll border-l border-[#F2F4F7]')}>
  866. <div className={cn(s.previewHeader, previewScrolled && `${s.fixed} pb-3`)}>
  867. <div className='flex items-center justify-between px-8'>
  868. <div className='grow flex items-center'>
  869. <div>{t('datasetCreation.stepTwo.previewTitle')}</div>
  870. {docForm === DocForm.QA && !previewSwitched && (
  871. <Button className='ml-2' variant='secondary-accent' onClick={previewSwitch}>{t('datasetCreation.stepTwo.previewButton')}</Button>
  872. )}
  873. </div>
  874. <div className='flex items-center justify-center w-6 h-6 cursor-pointer' onClick={hidePreview}>
  875. <XMarkIcon className='h-4 w-4'></XMarkIcon>
  876. </div>
  877. </div>
  878. {docForm === DocForm.QA && !previewSwitched && (
  879. <div className='px-8 pr-12 text-xs text-gray-500'>
  880. <span>{t('datasetCreation.stepTwo.previewSwitchTipStart')}</span>
  881. <span className='text-amber-600'>{t('datasetCreation.stepTwo.previewSwitchTipEnd')}</span>
  882. </div>
  883. )}
  884. </div>
  885. <div className='my-4 px-8 space-y-4'>
  886. {previewSwitched && docForm === DocForm.QA && fileIndexingEstimate?.qa_preview && (
  887. <>
  888. {fileIndexingEstimate?.qa_preview.map((item, index) => (
  889. <PreviewItem type={PreviewType.QA} key={item.question} qa={item} index={index + 1} />
  890. ))}
  891. </>
  892. )}
  893. {(docForm === DocForm.TEXT || !previewSwitched) && fileIndexingEstimate?.preview && (
  894. <>
  895. {fileIndexingEstimate?.preview.map((item, index) => (
  896. <PreviewItem type={PreviewType.TEXT} key={item} content={item} index={index + 1} />
  897. ))}
  898. </>
  899. )}
  900. {previewSwitched && docForm === DocForm.QA && !fileIndexingEstimate?.qa_preview && (
  901. <div className='flex items-center justify-center h-[200px]'>
  902. <Loading type='area' />
  903. </div>
  904. )}
  905. {!previewSwitched && !fileIndexingEstimate?.preview && (
  906. <div className='flex items-center justify-center h-[200px]'>
  907. <Loading type='area' />
  908. </div>
  909. )}
  910. </div>
  911. </div>}
  912. {!showPreview && (
  913. <div className={cn(s.sideTip)}>
  914. <div className={s.tipCard}>
  915. <span className={s.icon} />
  916. <div className={s.title}>{t('datasetCreation.stepTwo.sideTipTitle')}</div>
  917. <div className={s.content}>
  918. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP1')}</p>
  919. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP2')}</p>
  920. <p className='mb-3'>{t('datasetCreation.stepTwo.sideTipP3')}</p>
  921. <p>{t('datasetCreation.stepTwo.sideTipP4')}</p>
  922. </div>
  923. </div>
  924. </div>
  925. )}
  926. </FloatRightContainer>
  927. </div>
  928. )
  929. }
  930. export default StepTwo