index.tsx 47 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172
  1. 'use client'
  2. import type { FC, PropsWithChildren } from 'react'
  3. import React, { useCallback, useEffect, useRef, useState } from 'react'
  4. import { useTranslation } from 'react-i18next'
  5. import { useContext } from 'use-context-selector'
  6. import {
  7. RiAlertFill,
  8. RiArrowLeftLine,
  9. RiSearchEyeLine,
  10. } from '@remixicon/react'
  11. import Link from 'next/link'
  12. import Image from 'next/image'
  13. import { useHover } from 'ahooks'
  14. import SettingCog from '../assets/setting-gear-mod.svg'
  15. import OrangeEffect from '../assets/option-card-effect-orange.svg'
  16. import FamilyMod from '../assets/family-mod.svg'
  17. import Note from '../assets/note-mod.svg'
  18. import FileList from '../assets/file-list-3-fill.svg'
  19. import { indexMethodIcon } from '../icons'
  20. import { PreviewContainer } from '../../preview/container'
  21. import { ChunkContainer, QAPreview } from '../../chunk'
  22. import { PreviewHeader } from '../../preview/header'
  23. import { FormattedText } from '../../formatted-text/formatted'
  24. import { PreviewSlice } from '../../formatted-text/flavours/preview-slice'
  25. import PreviewDocumentPicker from '../../common/document-picker/preview-document-picker'
  26. import s from './index.module.css'
  27. import unescape from './unescape'
  28. import escape from './escape'
  29. import { OptionCard } from './option-card'
  30. import LanguageSelect from './language-select'
  31. import { DelimiterInput, MaxLengthInput, OverlapInput } from './inputs'
  32. import cn from '@/utils/classnames'
  33. import type { CrawlOptions, CrawlResultItem, CreateDocumentReq, CustomFile, DocumentItem, FullDocumentDetail, ParentMode, PreProcessingRule, ProcessRule, Rules, createDocumentResponse } from '@/models/datasets'
  34. import { ChunkingMode, DataSourceType, ProcessMode } from '@/models/datasets'
  35. import Button from '@/app/components/base/button'
  36. import FloatRightContainer from '@/app/components/base/float-right-container'
  37. import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
  38. import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
  39. import type { RetrievalConfig } from '@/types/app'
  40. import { isReRankModelSelected } from '@/app/components/datasets/common/check-rerank-model'
  41. import Toast from '@/app/components/base/toast'
  42. import type { NotionPage } from '@/models/common'
  43. import { DataSourceProvider } from '@/models/common'
  44. import { useDatasetDetailContext } from '@/context/dataset-detail'
  45. import I18n from '@/context/i18n'
  46. import { RETRIEVE_METHOD } from '@/types/app'
  47. import useBreakpoints, { MediaType } from '@/hooks/use-breakpoints'
  48. import { useDefaultModel, useModelList, useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
  49. import { LanguagesSupported } from '@/i18n/language'
  50. import ModelSelector from '@/app/components/header/account-setting/model-provider-page/model-selector'
  51. import type { DefaultModel } from '@/app/components/header/account-setting/model-provider-page/declarations'
  52. import { ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
  53. import Checkbox from '@/app/components/base/checkbox'
  54. import RadioCard from '@/app/components/base/radio-card'
  55. import { FULL_DOC_PREVIEW_LENGTH, IS_CE_EDITION } from '@/config'
  56. import Divider from '@/app/components/base/divider'
  57. import { getNotionInfo, getWebsiteInfo, useCreateDocument, useCreateFirstDocument, useFetchDefaultProcessRule, useFetchFileIndexingEstimateForFile, useFetchFileIndexingEstimateForNotion, useFetchFileIndexingEstimateForWeb } from '@/service/knowledge/use-create-dataset'
  58. import Badge from '@/app/components/base/badge'
  59. import { SkeletonContainer, SkeletonPoint, SkeletonRectangle, SkeletonRow } from '@/app/components/base/skeleton'
  60. import Tooltip from '@/app/components/base/tooltip'
  61. import CustomDialog from '@/app/components/base/dialog'
  62. import { PortalToFollowElem, PortalToFollowElemContent, PortalToFollowElemTrigger } from '@/app/components/base/portal-to-follow-elem'
  63. import { AlertTriangle } from '@/app/components/base/icons/src/vender/solid/alertsAndFeedback'
  64. const TextLabel: FC<PropsWithChildren> = (props) => {
  65. return <label className='text-text-secondary system-sm-semibold'>{props.children}</label>
  66. }
  67. type StepTwoProps = {
  68. isSetting?: boolean
  69. documentDetail?: FullDocumentDetail
  70. isAPIKeySet: boolean
  71. onSetting: () => void
  72. datasetId?: string
  73. indexingType?: IndexingType
  74. retrievalMethod?: string
  75. dataSourceType: DataSourceType
  76. files: CustomFile[]
  77. notionPages?: NotionPage[]
  78. websitePages?: CrawlResultItem[]
  79. crawlOptions?: CrawlOptions
  80. websiteCrawlProvider?: DataSourceProvider
  81. websiteCrawlJobId?: string
  82. onStepChange?: (delta: number) => void
  83. updateIndexingTypeCache?: (type: string) => void
  84. updateRetrievalMethodCache?: (method: string) => void
  85. updateResultCache?: (res: createDocumentResponse) => void
  86. onSave?: () => void
  87. onCancel?: () => void
  88. }
  89. export enum IndexingType {
  90. QUALIFIED = 'high_quality',
  91. ECONOMICAL = 'economy',
  92. }
  93. const DEFAULT_SEGMENT_IDENTIFIER = '\\n\\n'
  94. const DEFAULT_MAXIMUM_CHUNK_LENGTH = 500
  95. const DEFAULT_OVERLAP = 50
  96. const MAXIMUM_CHUNK_TOKEN_LENGTH = Number.parseInt(globalThis.document?.body?.getAttribute('data-public-indexing-max-segmentation-tokens-length') || '4000', 10)
  97. type ParentChildConfig = {
  98. chunkForContext: ParentMode
  99. parent: {
  100. delimiter: string
  101. maxLength: number
  102. }
  103. child: {
  104. delimiter: string
  105. maxLength: number
  106. }
  107. }
  108. const defaultParentChildConfig: ParentChildConfig = {
  109. chunkForContext: 'paragraph',
  110. parent: {
  111. delimiter: '\\n\\n',
  112. maxLength: 500,
  113. },
  114. child: {
  115. delimiter: '\\n',
  116. maxLength: 200,
  117. },
  118. }
  119. const StepTwo = ({
  120. isSetting,
  121. documentDetail,
  122. isAPIKeySet,
  123. datasetId,
  124. indexingType,
  125. dataSourceType: inCreatePageDataSourceType,
  126. files,
  127. notionPages = [],
  128. websitePages = [],
  129. crawlOptions,
  130. websiteCrawlProvider = DataSourceProvider.fireCrawl,
  131. websiteCrawlJobId = '',
  132. onStepChange,
  133. updateIndexingTypeCache,
  134. updateResultCache,
  135. onSave,
  136. onCancel,
  137. updateRetrievalMethodCache,
  138. }: StepTwoProps) => {
  139. const { t } = useTranslation()
  140. const { locale } = useContext(I18n)
  141. const media = useBreakpoints()
  142. const isMobile = media === MediaType.mobile
  143. const { dataset: currentDataset, mutateDatasetRes } = useDatasetDetailContext()
  144. const isInUpload = Boolean(currentDataset)
  145. const isUploadInEmptyDataset = isInUpload && !currentDataset?.doc_form
  146. const isNotUploadInEmptyDataset = !isUploadInEmptyDataset
  147. const isInInit = !isInUpload && !isSetting
  148. const isInCreatePage = !datasetId || (datasetId && !currentDataset?.data_source_type)
  149. const dataSourceType = isInCreatePage ? inCreatePageDataSourceType : currentDataset?.data_source_type
  150. const [segmentationType, setSegmentationType] = useState<ProcessMode>(ProcessMode.general)
  151. const [segmentIdentifier, doSetSegmentIdentifier] = useState(DEFAULT_SEGMENT_IDENTIFIER)
  152. const setSegmentIdentifier = useCallback((value: string, canEmpty?: boolean) => {
  153. doSetSegmentIdentifier(value ? escape(value) : (canEmpty ? '' : DEFAULT_SEGMENT_IDENTIFIER))
  154. }, [])
  155. const [maxChunkLength, setMaxChunkLength] = useState(DEFAULT_MAXIMUM_CHUNK_LENGTH) // default chunk length
  156. const [limitMaxChunkLength, setLimitMaxChunkLength] = useState(MAXIMUM_CHUNK_TOKEN_LENGTH)
  157. const [overlap, setOverlap] = useState(DEFAULT_OVERLAP)
  158. const [rules, setRules] = useState<PreProcessingRule[]>([])
  159. const [defaultConfig, setDefaultConfig] = useState<Rules>()
  160. const hasSetIndexType = !!indexingType
  161. const [indexType, setIndexType] = useState<IndexingType>(
  162. (indexingType
  163. || isAPIKeySet)
  164. ? IndexingType.QUALIFIED
  165. : IndexingType.ECONOMICAL,
  166. )
  167. const [previewFile, setPreviewFile] = useState<DocumentItem>(
  168. (datasetId && documentDetail)
  169. ? documentDetail.file
  170. : files[0],
  171. )
  172. const [previewNotionPage, setPreviewNotionPage] = useState<NotionPage>(
  173. (datasetId && documentDetail)
  174. ? documentDetail.notion_page
  175. : notionPages[0],
  176. )
  177. const [previewWebsitePage, setPreviewWebsitePage] = useState<CrawlResultItem>(
  178. (datasetId && documentDetail)
  179. ? documentDetail.website_page
  180. : websitePages[0],
  181. )
  182. // QA Related
  183. const [isQAConfirmDialogOpen, setIsQAConfirmDialogOpen] = useState(false)
  184. const [docForm, setDocForm] = useState<ChunkingMode>(
  185. (datasetId && documentDetail) ? documentDetail.doc_form as ChunkingMode : ChunkingMode.text,
  186. )
  187. const handleChangeDocform = (value: ChunkingMode) => {
  188. if (value === ChunkingMode.qa && indexType === IndexingType.ECONOMICAL) {
  189. setIsQAConfirmDialogOpen(true)
  190. return
  191. }
  192. if (value === ChunkingMode.parentChild && indexType === IndexingType.ECONOMICAL)
  193. setIndexType(IndexingType.QUALIFIED)
  194. setDocForm(value)
  195. // eslint-disable-next-line ts/no-use-before-define
  196. currentEstimateMutation.reset()
  197. }
  198. const [docLanguage, setDocLanguage] = useState<string>(
  199. (datasetId && documentDetail) ? documentDetail.doc_language : (locale !== LanguagesSupported[1] ? 'English' : 'Chinese'),
  200. )
  201. const [parentChildConfig, setParentChildConfig] = useState<ParentChildConfig>(defaultParentChildConfig)
  202. const getIndexing_technique = () => indexingType || indexType
  203. const currentDocForm = currentDataset?.doc_form || docForm
  204. const getProcessRule = (): ProcessRule => {
  205. if (currentDocForm === ChunkingMode.parentChild) {
  206. return {
  207. rules: {
  208. pre_processing_rules: rules,
  209. segmentation: {
  210. separator: unescape(
  211. parentChildConfig.parent.delimiter,
  212. ),
  213. max_tokens: parentChildConfig.parent.maxLength,
  214. },
  215. parent_mode: parentChildConfig.chunkForContext,
  216. subchunk_segmentation: {
  217. separator: unescape(parentChildConfig.child.delimiter),
  218. max_tokens: parentChildConfig.child.maxLength,
  219. },
  220. },
  221. mode: 'hierarchical',
  222. } as ProcessRule
  223. }
  224. return {
  225. rules: {
  226. pre_processing_rules: rules,
  227. segmentation: {
  228. separator: unescape(segmentIdentifier),
  229. max_tokens: maxChunkLength,
  230. chunk_overlap: overlap,
  231. },
  232. }, // api will check this. It will be removed after api refactored.
  233. mode: segmentationType,
  234. } as ProcessRule
  235. }
  236. const fileIndexingEstimateQuery = useFetchFileIndexingEstimateForFile({
  237. docForm: currentDocForm,
  238. docLanguage,
  239. dataSourceType: DataSourceType.FILE,
  240. files: previewFile
  241. ? [files.find(file => file.name === previewFile.name)!]
  242. : files,
  243. indexingTechnique: getIndexing_technique() as any,
  244. processRule: getProcessRule(),
  245. dataset_id: datasetId!,
  246. })
  247. const notionIndexingEstimateQuery = useFetchFileIndexingEstimateForNotion({
  248. docForm: currentDocForm,
  249. docLanguage,
  250. dataSourceType: DataSourceType.NOTION,
  251. notionPages: [previewNotionPage],
  252. indexingTechnique: getIndexing_technique() as any,
  253. processRule: getProcessRule(),
  254. dataset_id: datasetId || '',
  255. })
  256. const websiteIndexingEstimateQuery = useFetchFileIndexingEstimateForWeb({
  257. docForm: currentDocForm,
  258. docLanguage,
  259. dataSourceType: DataSourceType.WEB,
  260. websitePages: [previewWebsitePage],
  261. crawlOptions,
  262. websiteCrawlProvider,
  263. websiteCrawlJobId,
  264. indexingTechnique: getIndexing_technique() as any,
  265. processRule: getProcessRule(),
  266. dataset_id: datasetId || '',
  267. })
  268. const currentEstimateMutation = dataSourceType === DataSourceType.FILE
  269. ? fileIndexingEstimateQuery
  270. : dataSourceType === DataSourceType.NOTION
  271. ? notionIndexingEstimateQuery
  272. : websiteIndexingEstimateQuery
  273. const fetchEstimate = useCallback(() => {
  274. if (dataSourceType === DataSourceType.FILE)
  275. fileIndexingEstimateQuery.mutate()
  276. if (dataSourceType === DataSourceType.NOTION)
  277. notionIndexingEstimateQuery.mutate()
  278. if (dataSourceType === DataSourceType.WEB)
  279. websiteIndexingEstimateQuery.mutate()
  280. }, [dataSourceType, fileIndexingEstimateQuery, notionIndexingEstimateQuery, websiteIndexingEstimateQuery])
  281. const estimate
  282. = dataSourceType === DataSourceType.FILE
  283. ? fileIndexingEstimateQuery.data
  284. : dataSourceType === DataSourceType.NOTION
  285. ? notionIndexingEstimateQuery.data
  286. : websiteIndexingEstimateQuery.data
  287. const getRuleName = (key: string) => {
  288. if (key === 'remove_extra_spaces')
  289. return t('datasetCreation.stepTwo.removeExtraSpaces')
  290. if (key === 'remove_urls_emails')
  291. return t('datasetCreation.stepTwo.removeUrlEmails')
  292. if (key === 'remove_stopwords')
  293. return t('datasetCreation.stepTwo.removeStopwords')
  294. }
  295. const ruleChangeHandle = (id: string) => {
  296. const newRules = rules.map((rule) => {
  297. if (rule.id === id) {
  298. return {
  299. id: rule.id,
  300. enabled: !rule.enabled,
  301. }
  302. }
  303. return rule
  304. })
  305. setRules(newRules)
  306. }
  307. const resetRules = () => {
  308. if (defaultConfig) {
  309. setSegmentIdentifier(defaultConfig.segmentation.separator)
  310. setMaxChunkLength(defaultConfig.segmentation.max_tokens)
  311. setOverlap(defaultConfig.segmentation.chunk_overlap!)
  312. setRules(defaultConfig.pre_processing_rules)
  313. }
  314. setParentChildConfig(defaultParentChildConfig)
  315. }
  316. const updatePreview = () => {
  317. if (segmentationType === ProcessMode.general && maxChunkLength > MAXIMUM_CHUNK_TOKEN_LENGTH) {
  318. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.maxLengthCheck', { limit: MAXIMUM_CHUNK_TOKEN_LENGTH }) })
  319. return
  320. }
  321. fetchEstimate()
  322. }
  323. const {
  324. modelList: rerankModelList,
  325. defaultModel: rerankDefaultModel,
  326. currentModel: isRerankDefaultModelValid,
  327. } = useModelListAndDefaultModelAndCurrentProviderAndModel(ModelTypeEnum.rerank)
  328. const { data: embeddingModelList } = useModelList(ModelTypeEnum.textEmbedding)
  329. const { data: defaultEmbeddingModel } = useDefaultModel(ModelTypeEnum.textEmbedding)
  330. const [embeddingModel, setEmbeddingModel] = useState<DefaultModel>(
  331. currentDataset?.embedding_model
  332. ? {
  333. provider: currentDataset.embedding_model_provider,
  334. model: currentDataset.embedding_model,
  335. }
  336. : {
  337. provider: defaultEmbeddingModel?.provider.provider || '',
  338. model: defaultEmbeddingModel?.model || '',
  339. },
  340. )
  341. const [retrievalConfig, setRetrievalConfig] = useState(currentDataset?.retrieval_model_dict || {
  342. search_method: RETRIEVE_METHOD.semantic,
  343. reranking_enable: false,
  344. reranking_model: {
  345. reranking_provider_name: '',
  346. reranking_model_name: '',
  347. },
  348. top_k: 3,
  349. score_threshold_enabled: false,
  350. score_threshold: 0.5,
  351. } as RetrievalConfig)
  352. useEffect(() => {
  353. if (currentDataset?.retrieval_model_dict)
  354. return
  355. setRetrievalConfig({
  356. search_method: RETRIEVE_METHOD.semantic,
  357. reranking_enable: !!isRerankDefaultModelValid,
  358. reranking_model: {
  359. reranking_provider_name: isRerankDefaultModelValid ? rerankDefaultModel?.provider.provider ?? '' : '',
  360. reranking_model_name: isRerankDefaultModelValid ? rerankDefaultModel?.model ?? '' : '',
  361. },
  362. top_k: 3,
  363. score_threshold_enabled: false,
  364. score_threshold: 0.5,
  365. })
  366. // eslint-disable-next-line react-hooks/exhaustive-deps
  367. }, [rerankDefaultModel, isRerankDefaultModelValid])
  368. const getCreationParams = () => {
  369. let params
  370. if (segmentationType === ProcessMode.general && overlap > maxChunkLength) {
  371. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.overlapCheck') })
  372. return
  373. }
  374. if (segmentationType === ProcessMode.general && maxChunkLength > limitMaxChunkLength) {
  375. Toast.notify({ type: 'error', message: t('datasetCreation.stepTwo.maxLengthCheck', { limit: limitMaxChunkLength }) })
  376. return
  377. }
  378. if (isSetting) {
  379. params = {
  380. original_document_id: documentDetail?.id,
  381. doc_form: currentDocForm,
  382. doc_language: docLanguage,
  383. process_rule: getProcessRule(),
  384. retrieval_model: retrievalConfig, // Readonly. If want to changed, just go to settings page.
  385. embedding_model: embeddingModel.model, // Readonly
  386. embedding_model_provider: embeddingModel.provider, // Readonly
  387. indexing_technique: getIndexing_technique(),
  388. } as CreateDocumentReq
  389. }
  390. else { // create
  391. const indexMethod = getIndexing_technique()
  392. if (
  393. !isReRankModelSelected({
  394. rerankModelList,
  395. retrievalConfig,
  396. indexMethod: indexMethod as string,
  397. })
  398. ) {
  399. Toast.notify({ type: 'error', message: t('appDebug.datasetConfig.rerankModelRequired') })
  400. return
  401. }
  402. params = {
  403. data_source: {
  404. type: dataSourceType,
  405. info_list: {
  406. data_source_type: dataSourceType,
  407. },
  408. },
  409. indexing_technique: getIndexing_technique(),
  410. process_rule: getProcessRule(),
  411. doc_form: currentDocForm,
  412. doc_language: docLanguage,
  413. retrieval_model: retrievalConfig,
  414. embedding_model: embeddingModel.model,
  415. embedding_model_provider: embeddingModel.provider,
  416. } as CreateDocumentReq
  417. if (dataSourceType === DataSourceType.FILE) {
  418. params.data_source.info_list.file_info_list = {
  419. file_ids: files.map(file => file.id || '').filter(Boolean),
  420. }
  421. }
  422. if (dataSourceType === DataSourceType.NOTION)
  423. params.data_source.info_list.notion_info_list = getNotionInfo(notionPages)
  424. if (dataSourceType === DataSourceType.WEB) {
  425. params.data_source.info_list.website_info_list = getWebsiteInfo({
  426. websiteCrawlProvider,
  427. websiteCrawlJobId,
  428. websitePages,
  429. })
  430. }
  431. }
  432. return params
  433. }
  434. const fetchDefaultProcessRuleMutation = useFetchDefaultProcessRule({
  435. onSuccess(data) {
  436. const separator = data.rules.segmentation.separator
  437. setSegmentIdentifier(separator)
  438. setMaxChunkLength(data.rules.segmentation.max_tokens)
  439. setOverlap(data.rules.segmentation.chunk_overlap!)
  440. setRules(data.rules.pre_processing_rules)
  441. setDefaultConfig(data.rules)
  442. setLimitMaxChunkLength(data.limits.indexing_max_segmentation_tokens_length)
  443. },
  444. onError(error) {
  445. Toast.notify({
  446. type: 'error',
  447. message: `${error}`,
  448. })
  449. },
  450. })
  451. const getRulesFromDetail = () => {
  452. if (documentDetail) {
  453. const rules = documentDetail.dataset_process_rule.rules
  454. const separator = rules.segmentation.separator
  455. const max = rules.segmentation.max_tokens
  456. const overlap = rules.segmentation.chunk_overlap
  457. setSegmentIdentifier(separator)
  458. setMaxChunkLength(max)
  459. setOverlap(overlap!)
  460. setRules(rules.pre_processing_rules)
  461. setDefaultConfig(rules)
  462. }
  463. }
  464. const getDefaultMode = () => {
  465. if (documentDetail)
  466. setSegmentationType(documentDetail.dataset_process_rule.mode)
  467. }
  468. const createFirstDocumentMutation = useCreateFirstDocument({
  469. onError(error) {
  470. Toast.notify({
  471. type: 'error',
  472. message: `${error}`,
  473. })
  474. },
  475. })
  476. const createDocumentMutation = useCreateDocument(datasetId!, {
  477. onError(error) {
  478. Toast.notify({
  479. type: 'error',
  480. message: `${error}`,
  481. })
  482. },
  483. })
  484. const isCreating = createFirstDocumentMutation.isPending || createDocumentMutation.isPending
  485. const createHandle = async () => {
  486. const params = getCreationParams()
  487. if (!params)
  488. return false
  489. if (!datasetId) {
  490. await createFirstDocumentMutation.mutateAsync(
  491. params,
  492. {
  493. onSuccess(data) {
  494. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  495. updateResultCache && updateResultCache(data)
  496. updateRetrievalMethodCache && updateRetrievalMethodCache(retrievalConfig.search_method as string)
  497. },
  498. },
  499. )
  500. }
  501. else {
  502. await createDocumentMutation.mutateAsync(params, {
  503. onSuccess(data) {
  504. updateIndexingTypeCache && updateIndexingTypeCache(indexType as string)
  505. updateResultCache && updateResultCache(data)
  506. },
  507. })
  508. }
  509. if (mutateDatasetRes)
  510. mutateDatasetRes()
  511. onStepChange && onStepChange(+1)
  512. isSetting && onSave && onSave()
  513. }
  514. useEffect(() => {
  515. // fetch rules
  516. if (!isSetting) {
  517. fetchDefaultProcessRuleMutation.mutate('/datasets/process-rule')
  518. }
  519. else {
  520. getRulesFromDetail()
  521. getDefaultMode()
  522. }
  523. // eslint-disable-next-line react-hooks/exhaustive-deps
  524. }, [])
  525. useEffect(() => {
  526. // get indexing type by props
  527. if (indexingType)
  528. setIndexType(indexingType as IndexingType)
  529. else
  530. setIndexType(isAPIKeySet ? IndexingType.QUALIFIED : IndexingType.ECONOMICAL)
  531. }, [isAPIKeySet, indexingType, datasetId])
  532. const economyDomRef = useRef<HTMLDivElement>(null)
  533. const isHoveringEconomy = useHover(economyDomRef)
  534. const isModelAndRetrievalConfigDisabled = !!datasetId && !!currentDataset?.data_source_type
  535. return (
  536. <div className='flex w-full h-full'>
  537. <div className={cn('relative h-full w-1/2 py-6 overflow-y-auto', isMobile ? 'px-4' : 'px-12')}>
  538. <div className={'system-md-semibold mb-1'}>{t('datasetCreation.stepTwo.segmentation')}</div>
  539. {((isInUpload && [ChunkingMode.text, ChunkingMode.qa].includes(currentDataset!.doc_form))
  540. || isUploadInEmptyDataset
  541. || isInInit)
  542. && <OptionCard
  543. className='bg-background-section mb-2'
  544. title={t('datasetCreation.stepTwo.general')}
  545. icon={<Image width={20} height={20} src={SettingCog} alt={t('datasetCreation.stepTwo.general')} />}
  546. activeHeaderClassName='bg-dataset-option-card-blue-gradient'
  547. description={t('datasetCreation.stepTwo.generalTip')}
  548. isActive={
  549. [ChunkingMode.text, ChunkingMode.qa].includes(currentDocForm)
  550. }
  551. onSwitched={() =>
  552. handleChangeDocform(ChunkingMode.text)
  553. }
  554. actions={
  555. <>
  556. <Button variant={'secondary-accent'} onClick={() => updatePreview()}>
  557. <RiSearchEyeLine className='h-4 w-4 mr-0.5' />
  558. {t('datasetCreation.stepTwo.previewChunk')}
  559. </Button>
  560. <Button variant={'ghost'} onClick={resetRules}>
  561. {t('datasetCreation.stepTwo.reset')}
  562. </Button>
  563. </>
  564. }
  565. noHighlight={isInUpload && isNotUploadInEmptyDataset}
  566. >
  567. <div className='flex flex-col gap-y-4'>
  568. <div className='flex gap-3'>
  569. <DelimiterInput
  570. value={segmentIdentifier}
  571. onChange={e => setSegmentIdentifier(e.target.value, true)}
  572. />
  573. <MaxLengthInput
  574. unit='tokens'
  575. value={maxChunkLength}
  576. onChange={setMaxChunkLength}
  577. />
  578. <OverlapInput
  579. unit='tokens'
  580. value={overlap}
  581. min={1}
  582. onChange={setOverlap}
  583. />
  584. </div>
  585. <div className='w-full flex flex-col'>
  586. <div className='flex items-center gap-x-2'>
  587. <div className='inline-flex shrink-0'>
  588. <TextLabel>{t('datasetCreation.stepTwo.rules')}</TextLabel>
  589. </div>
  590. <Divider className='grow' bgStyle='gradient' />
  591. </div>
  592. <div className='mt-1'>
  593. {rules.map(rule => (
  594. <div key={rule.id} className={s.ruleItem} onClick={() => {
  595. ruleChangeHandle(rule.id)
  596. }}>
  597. <Checkbox
  598. checked={rule.enabled}
  599. />
  600. <label className="ml-2 system-sm-regular cursor-pointer text-text-secondary">{getRuleName(rule.id)}</label>
  601. </div>
  602. ))}
  603. {IS_CE_EDITION && <>
  604. <Divider type='horizontal' className='my-4 bg-divider-subtle' />
  605. <div className='flex items-center py-0.5'>
  606. <div className='flex items-center' onClick={() => {
  607. if (currentDataset?.doc_form)
  608. return
  609. if (docForm === ChunkingMode.qa)
  610. handleChangeDocform(ChunkingMode.text)
  611. else
  612. handleChangeDocform(ChunkingMode.qa)
  613. }}>
  614. <Checkbox
  615. checked={currentDocForm === ChunkingMode.qa}
  616. disabled={!!currentDataset?.doc_form}
  617. />
  618. <label className="ml-2 system-sm-regular cursor-pointer text-text-secondary">
  619. {t('datasetCreation.stepTwo.useQALanguage')}
  620. </label>
  621. </div>
  622. <LanguageSelect
  623. currentLanguage={docLanguage || locale}
  624. onSelect={setDocLanguage}
  625. disabled={currentDocForm !== ChunkingMode.qa}
  626. />
  627. <Tooltip popupContent={t('datasetCreation.stepTwo.QATip')} />
  628. </div>
  629. {currentDocForm === ChunkingMode.qa && (
  630. <div
  631. style={{
  632. background: 'linear-gradient(92deg, rgba(247, 144, 9, 0.1) 0%, rgba(255, 255, 255, 0.00) 100%)',
  633. }}
  634. className='h-10 mt-2 flex items-center gap-2 rounded-xl backdrop-blur-[5px] border-components-panel-border border shadow-xs px-3 text-xs'
  635. >
  636. <RiAlertFill className='size-4 text-text-warning-secondary' />
  637. <span className='system-xs-medium text-text-primary'>
  638. {t('datasetCreation.stepTwo.QATip')}
  639. </span>
  640. </div>
  641. )}
  642. </>}
  643. </div>
  644. </div>
  645. </div>
  646. </OptionCard>}
  647. {
  648. (
  649. (isInUpload && currentDataset!.doc_form === ChunkingMode.parentChild)
  650. || isUploadInEmptyDataset
  651. || isInInit
  652. )
  653. && <OptionCard
  654. title={t('datasetCreation.stepTwo.parentChild')}
  655. icon={<Image width={20} height={20} src={FamilyMod} alt={t('datasetCreation.stepTwo.parentChild')} />}
  656. effectImg={OrangeEffect.src}
  657. activeHeaderClassName='bg-dataset-option-card-orange-gradient'
  658. description={t('datasetCreation.stepTwo.parentChildTip')}
  659. isActive={currentDocForm === ChunkingMode.parentChild}
  660. onSwitched={() => handleChangeDocform(ChunkingMode.parentChild)}
  661. actions={
  662. <>
  663. <Button variant={'secondary-accent'} onClick={() => updatePreview()}>
  664. <RiSearchEyeLine className='h-4 w-4 mr-0.5' />
  665. {t('datasetCreation.stepTwo.previewChunk')}
  666. </Button>
  667. <Button variant={'ghost'} onClick={resetRules}>
  668. {t('datasetCreation.stepTwo.reset')}
  669. </Button>
  670. </>
  671. }
  672. noHighlight={isInUpload && isNotUploadInEmptyDataset}
  673. >
  674. <div className='flex flex-col gap-4'>
  675. <div>
  676. <div className='flex items-center gap-x-2'>
  677. <div className='inline-flex shrink-0'>
  678. <TextLabel>{t('datasetCreation.stepTwo.parentChunkForContext')}</TextLabel>
  679. </div>
  680. <Divider className='grow' bgStyle='gradient' />
  681. </div>
  682. <RadioCard className='mt-1'
  683. icon={<Image src={Note} alt='' />}
  684. title={t('datasetCreation.stepTwo.paragraph')}
  685. description={t('datasetCreation.stepTwo.paragraphTip')}
  686. isChosen={parentChildConfig.chunkForContext === 'paragraph'}
  687. onChosen={() => setParentChildConfig(
  688. {
  689. ...parentChildConfig,
  690. chunkForContext: 'paragraph',
  691. },
  692. )}
  693. chosenConfig={
  694. <div className='flex gap-3'>
  695. <DelimiterInput
  696. value={parentChildConfig.parent.delimiter}
  697. tooltip={t('datasetCreation.stepTwo.parentChildDelimiterTip')!}
  698. onChange={e => setParentChildConfig({
  699. ...parentChildConfig,
  700. parent: {
  701. ...parentChildConfig.parent,
  702. delimiter: e.target.value ? escape(e.target.value) : '',
  703. },
  704. })}
  705. />
  706. <MaxLengthInput
  707. unit='tokens'
  708. value={parentChildConfig.parent.maxLength}
  709. onChange={value => setParentChildConfig({
  710. ...parentChildConfig,
  711. parent: {
  712. ...parentChildConfig.parent,
  713. maxLength: value,
  714. },
  715. })}
  716. />
  717. </div>
  718. }
  719. />
  720. <RadioCard className='mt-2'
  721. icon={<Image src={FileList} alt='' />}
  722. title={t('datasetCreation.stepTwo.fullDoc')}
  723. description={t('datasetCreation.stepTwo.fullDocTip')}
  724. onChosen={() => setParentChildConfig(
  725. {
  726. ...parentChildConfig,
  727. chunkForContext: 'full-doc',
  728. },
  729. )}
  730. isChosen={parentChildConfig.chunkForContext === 'full-doc'}
  731. />
  732. </div>
  733. <div>
  734. <div className='flex items-center gap-x-2'>
  735. <div className='inline-flex shrink-0'>
  736. <TextLabel>{t('datasetCreation.stepTwo.childChunkForRetrieval')}</TextLabel>
  737. </div>
  738. <Divider className='grow' bgStyle='gradient' />
  739. </div>
  740. <div className='flex gap-3 mt-1'>
  741. <DelimiterInput
  742. value={parentChildConfig.child.delimiter}
  743. tooltip={t('datasetCreation.stepTwo.parentChildChunkDelimiterTip')!}
  744. onChange={e => setParentChildConfig({
  745. ...parentChildConfig,
  746. child: {
  747. ...parentChildConfig.child,
  748. delimiter: e.target.value ? escape(e.target.value) : '',
  749. },
  750. })}
  751. />
  752. <MaxLengthInput
  753. unit='tokens'
  754. value={parentChildConfig.child.maxLength}
  755. onChange={value => setParentChildConfig({
  756. ...parentChildConfig,
  757. child: {
  758. ...parentChildConfig.child,
  759. maxLength: value,
  760. },
  761. })}
  762. />
  763. </div>
  764. </div>
  765. <div>
  766. <div className='flex items-center gap-x-2'>
  767. <div className='inline-flex shrink-0'>
  768. <TextLabel>{t('datasetCreation.stepTwo.rules')}</TextLabel>
  769. </div>
  770. <Divider className='grow' bgStyle='gradient' />
  771. </div>
  772. <div className='mt-1'>
  773. {rules.map(rule => (
  774. <div key={rule.id} className={s.ruleItem} onClick={() => {
  775. ruleChangeHandle(rule.id)
  776. }}>
  777. <Checkbox
  778. checked={rule.enabled}
  779. />
  780. <label className="ml-2 system-sm-regular cursor-pointer text-text-secondary">{getRuleName(rule.id)}</label>
  781. </div>
  782. ))}
  783. </div>
  784. </div>
  785. </div>
  786. </OptionCard>}
  787. <Divider className='my-5' />
  788. <div className={'system-md-semibold mb-1'}>{t('datasetCreation.stepTwo.indexMode')}</div>
  789. <div className='flex items-center gap-2'>
  790. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.QUALIFIED)) && (
  791. <OptionCard className='flex-1'
  792. title={<div className='flex items-center'>
  793. {t('datasetCreation.stepTwo.qualified')}
  794. <Badge className={cn('ml-1 h-[18px]', (!hasSetIndexType && indexType === IndexingType.QUALIFIED) ? 'border-text-accent-secondary text-text-accent-secondary' : '')} uppercase>
  795. {t('datasetCreation.stepTwo.recommend')}
  796. </Badge>
  797. <span className='ml-auto'>
  798. {!hasSetIndexType && <span className={cn(s.radio)} />}
  799. </span>
  800. </div>}
  801. description={t('datasetCreation.stepTwo.qualifiedTip')}
  802. icon={<Image src={indexMethodIcon.high_quality} alt='' />}
  803. isActive={!hasSetIndexType && indexType === IndexingType.QUALIFIED}
  804. disabled={!isAPIKeySet || hasSetIndexType}
  805. onSwitched={() => {
  806. if (isAPIKeySet)
  807. setIndexType(IndexingType.QUALIFIED)
  808. }}
  809. />
  810. )}
  811. {(!hasSetIndexType || (hasSetIndexType && indexingType === IndexingType.ECONOMICAL)) && (
  812. <>
  813. <CustomDialog show={isQAConfirmDialogOpen} onClose={() => setIsQAConfirmDialogOpen(false)} className='w-[432px]'>
  814. <header className='pt-6 mb-4'>
  815. <h2 className='text-lg font-semibold'>
  816. {t('datasetCreation.stepTwo.qaSwitchHighQualityTipTitle')}
  817. </h2>
  818. <p className='font-normal text-sm mt-2'>
  819. {t('datasetCreation.stepTwo.qaSwitchHighQualityTipContent')}
  820. </p>
  821. </header>
  822. <div className='flex gap-2 pb-6'>
  823. <Button className='ml-auto' onClick={() => {
  824. setIsQAConfirmDialogOpen(false)
  825. }}>
  826. {t('datasetCreation.stepTwo.cancel')}
  827. </Button>
  828. <Button variant={'primary'} onClick={() => {
  829. setIsQAConfirmDialogOpen(false)
  830. setIndexType(IndexingType.QUALIFIED)
  831. setDocForm(ChunkingMode.qa)
  832. }}>
  833. {t('datasetCreation.stepTwo.switch')}
  834. </Button>
  835. </div>
  836. </CustomDialog>
  837. <PortalToFollowElem
  838. open={
  839. isHoveringEconomy && docForm !== ChunkingMode.text
  840. }
  841. placement={'top'}
  842. >
  843. <PortalToFollowElemTrigger asChild>
  844. <OptionCard className='flex-1'
  845. title={t('datasetCreation.stepTwo.economical')}
  846. description={t('datasetCreation.stepTwo.economicalTip')}
  847. icon={<Image src={indexMethodIcon.economical} alt='' />}
  848. isActive={!hasSetIndexType && indexType === IndexingType.ECONOMICAL}
  849. disabled={!isAPIKeySet || hasSetIndexType || docForm !== ChunkingMode.text}
  850. ref={economyDomRef}
  851. onSwitched={() => {
  852. if (isAPIKeySet && docForm === ChunkingMode.text)
  853. setIndexType(IndexingType.ECONOMICAL)
  854. }}
  855. />
  856. </PortalToFollowElemTrigger>
  857. <PortalToFollowElemContent>
  858. <div className='p-3 bg-components-tooltip-bg border-components-panel-border text-xs font-medium text-text-secondary rounded-lg shadow-lg'>
  859. {
  860. docForm === ChunkingMode.qa
  861. ? t('datasetCreation.stepTwo.notAvailableForQA')
  862. : t('datasetCreation.stepTwo.notAvailableForParentChild')
  863. }
  864. </div>
  865. </PortalToFollowElemContent>
  866. </PortalToFollowElem>
  867. </>)}
  868. </div>
  869. {!hasSetIndexType && indexType === IndexingType.QUALIFIED && (
  870. <div className='mt-2 h-10 p-2 flex items-center gap-x-0.5 rounded-xl border-[0.5px] border-components-panel-border overflow-hidden bg-components-panel-bg-blur backdrop-blur-[5px] shadow-xs'>
  871. <div className='absolute top-0 left-0 right-0 bottom-0 bg-[linear-gradient(92deg,rgba(247,144,9,0.25)_0%,rgba(255,255,255,0.00)_100%)] opacity-40'></div>
  872. <div className='p-1'>
  873. <AlertTriangle className='size-4 text-text-warning-secondary' />
  874. </div>
  875. <span className='system-xs-medium'>{t('datasetCreation.stepTwo.highQualityTip')}</span>
  876. </div>
  877. )}
  878. {hasSetIndexType && indexType === IndexingType.ECONOMICAL && (
  879. <div className='mt-2 system-xs-medium'>
  880. {t('datasetCreation.stepTwo.indexSettingTip')}
  881. <Link className='text-text-accent' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  882. </div>
  883. )}
  884. {/* Embedding model */}
  885. {indexType === IndexingType.QUALIFIED && (
  886. <div className='mt-5'>
  887. <div className={cn('system-md-semibold mb-1', datasetId && 'flex justify-between items-center')}>{t('datasetSettings.form.embeddingModel')}</div>
  888. <ModelSelector
  889. readonly={isModelAndRetrievalConfigDisabled}
  890. triggerClassName={isModelAndRetrievalConfigDisabled ? 'opacity-50' : ''}
  891. defaultModel={embeddingModel}
  892. modelList={embeddingModelList}
  893. onSelect={(model: DefaultModel) => {
  894. setEmbeddingModel(model)
  895. }}
  896. />
  897. {isModelAndRetrievalConfigDisabled && (
  898. <div className='mt-2 system-xs-medium'>
  899. {t('datasetCreation.stepTwo.indexSettingTip')}
  900. <Link className='text-text-accent' href={`/datasets/${datasetId}/settings`}>{t('datasetCreation.stepTwo.datasetSettingLink')}</Link>
  901. </div>
  902. )}
  903. </div>
  904. )}
  905. <Divider className='my-5' />
  906. {/* Retrieval Method Config */}
  907. <div>
  908. {!isModelAndRetrievalConfigDisabled
  909. ? (
  910. <div className={'mb-1'}>
  911. <div className='system-md-semibold mb-0.5'>{t('datasetSettings.form.retrievalSetting.title')}</div>
  912. <div className='body-xs-regular text-text-tertiary'>
  913. <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-text-accent'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
  914. {t('datasetSettings.form.retrievalSetting.longDescription')}
  915. </div>
  916. </div>
  917. )
  918. : (
  919. <div className={cn('system-md-semibold mb-0.5', 'flex justify-between items-center')}>
  920. <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
  921. </div>
  922. )}
  923. <div className=''>
  924. {
  925. getIndexing_technique() === IndexingType.QUALIFIED
  926. ? (
  927. <RetrievalMethodConfig
  928. disabled={isModelAndRetrievalConfigDisabled}
  929. value={retrievalConfig}
  930. onChange={setRetrievalConfig}
  931. />
  932. )
  933. : (
  934. <EconomicalRetrievalMethodConfig
  935. disabled={isModelAndRetrievalConfigDisabled}
  936. value={retrievalConfig}
  937. onChange={setRetrievalConfig}
  938. />
  939. )
  940. }
  941. </div>
  942. </div>
  943. {!isSetting
  944. ? (
  945. <div className='flex items-center mt-8 py-2'>
  946. <Button onClick={() => onStepChange && onStepChange(-1)}>
  947. <RiArrowLeftLine className='w-4 h-4 mr-1' />
  948. {t('datasetCreation.stepTwo.previousStep')}
  949. </Button>
  950. <Button className='ml-auto' loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.nextStep')}</Button>
  951. </div>
  952. )
  953. : (
  954. <div className='flex items-center mt-8 py-2'>
  955. <Button loading={isCreating} variant='primary' onClick={createHandle}>{t('datasetCreation.stepTwo.save')}</Button>
  956. <Button className='ml-2' onClick={onCancel}>{t('datasetCreation.stepTwo.cancel')}</Button>
  957. </div>
  958. )}
  959. </div>
  960. <FloatRightContainer isMobile={isMobile} isOpen={true} onClose={() => { }} footer={null}>
  961. <PreviewContainer
  962. header={<PreviewHeader
  963. title={t('datasetCreation.stepTwo.preview')}
  964. >
  965. <div className='flex items-center gap-1'>
  966. {dataSourceType === DataSourceType.FILE
  967. && <PreviewDocumentPicker
  968. files={files as Array<Required<CustomFile>>}
  969. onChange={(selected) => {
  970. currentEstimateMutation.reset()
  971. setPreviewFile(selected)
  972. currentEstimateMutation.mutate()
  973. }}
  974. // when it is from setting, it just has one file
  975. value={isSetting ? (files[0]! as Required<CustomFile>) : previewFile}
  976. />
  977. }
  978. {dataSourceType === DataSourceType.NOTION
  979. && <PreviewDocumentPicker
  980. files={
  981. notionPages.map(page => ({
  982. id: page.page_id,
  983. name: page.page_name,
  984. extension: 'md',
  985. }))
  986. }
  987. onChange={(selected) => {
  988. currentEstimateMutation.reset()
  989. const selectedPage = notionPages.find(page => page.page_id === selected.id)
  990. setPreviewNotionPage(selectedPage!)
  991. currentEstimateMutation.mutate()
  992. }}
  993. value={{
  994. id: previewNotionPage?.page_id || '',
  995. name: previewNotionPage?.page_name || '',
  996. extension: 'md',
  997. }}
  998. />
  999. }
  1000. {dataSourceType === DataSourceType.WEB
  1001. && <PreviewDocumentPicker
  1002. files={
  1003. websitePages.map(page => ({
  1004. id: page.source_url,
  1005. name: page.title,
  1006. extension: 'md',
  1007. }))
  1008. }
  1009. onChange={(selected) => {
  1010. currentEstimateMutation.reset()
  1011. const selectedPage = websitePages.find(page => page.source_url === selected.id)
  1012. setPreviewWebsitePage(selectedPage!)
  1013. currentEstimateMutation.mutate()
  1014. }}
  1015. value={
  1016. {
  1017. id: previewWebsitePage?.source_url || '',
  1018. name: previewWebsitePage?.title || '',
  1019. extension: 'md',
  1020. }
  1021. }
  1022. />
  1023. }
  1024. {
  1025. currentDocForm !== ChunkingMode.qa
  1026. && <Badge text={t('datasetCreation.stepTwo.previewChunkCount', {
  1027. count: estimate?.total_segments || 0,
  1028. }) as string}
  1029. />
  1030. }
  1031. </div>
  1032. </PreviewHeader>}
  1033. className={cn('flex shrink-0 w-1/2 p-4 pr-0 relative h-full', isMobile && 'w-full max-w-[524px]')}
  1034. mainClassName='space-y-6'
  1035. >
  1036. {currentDocForm === ChunkingMode.qa && estimate?.qa_preview && (
  1037. estimate?.qa_preview.map((item, index) => (
  1038. <ChunkContainer
  1039. key={item.question}
  1040. label={`Chunk-${index + 1}`}
  1041. characterCount={item.question.length + item.answer.length}
  1042. >
  1043. <QAPreview qa={item} />
  1044. </ChunkContainer>
  1045. ))
  1046. )}
  1047. {currentDocForm === ChunkingMode.text && estimate?.preview && (
  1048. estimate?.preview.map((item, index) => (
  1049. <ChunkContainer
  1050. key={item.content}
  1051. label={`Chunk-${index + 1}`}
  1052. characterCount={item.content.length}
  1053. >
  1054. {item.content}
  1055. </ChunkContainer>
  1056. ))
  1057. )}
  1058. {currentDocForm === ChunkingMode.parentChild && currentEstimateMutation.data?.preview && (
  1059. estimate?.preview?.map((item, index) => {
  1060. const indexForLabel = index + 1
  1061. const childChunks = parentChildConfig.chunkForContext === 'full-doc'
  1062. ? item.child_chunks.slice(0, FULL_DOC_PREVIEW_LENGTH)
  1063. : item.child_chunks
  1064. return (
  1065. <ChunkContainer
  1066. key={item.content}
  1067. label={`Chunk-${indexForLabel}`}
  1068. characterCount={item.content.length}
  1069. >
  1070. <FormattedText>
  1071. {childChunks.map((child, index) => {
  1072. const indexForLabel = index + 1
  1073. return (
  1074. <PreviewSlice
  1075. key={child}
  1076. label={`C-${indexForLabel}`}
  1077. text={child}
  1078. tooltip={`Child-chunk-${indexForLabel} · ${child.length} Characters`}
  1079. labelInnerClassName='text-[10px] font-semibold align-bottom leading-7'
  1080. dividerClassName='leading-7'
  1081. />
  1082. )
  1083. })}
  1084. </FormattedText>
  1085. </ChunkContainer>
  1086. )
  1087. })
  1088. )}
  1089. {currentEstimateMutation.isIdle && (
  1090. <div className='h-full w-full flex items-center justify-center'>
  1091. <div className='flex flex-col items-center justify-center gap-3'>
  1092. <RiSearchEyeLine className='size-10 text-text-empty-state-icon' />
  1093. <p className='text-sm text-text-tertiary'>
  1094. {t('datasetCreation.stepTwo.previewChunkTip')}
  1095. </p>
  1096. </div>
  1097. </div>
  1098. )}
  1099. {currentEstimateMutation.isPending && (
  1100. <div className='space-y-6'>
  1101. {Array.from({ length: 10 }, (_, i) => (
  1102. <SkeletonContainer key={i}>
  1103. <SkeletonRow>
  1104. <SkeletonRectangle className="w-20" />
  1105. <SkeletonPoint />
  1106. <SkeletonRectangle className="w-24" />
  1107. </SkeletonRow>
  1108. <SkeletonRectangle className="w-full" />
  1109. <SkeletonRectangle className="w-full" />
  1110. <SkeletonRectangle className="w-[422px]" />
  1111. </SkeletonContainer>
  1112. ))}
  1113. </div>
  1114. )}
  1115. </PreviewContainer>
  1116. </FloatRightContainer>
  1117. </div>
  1118. )
  1119. }
  1120. export default StepTwo