You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

index.tsx 7.3KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203
  1. 'use client'
  2. import React, { useCallback, useRef, useState } from 'react'
  3. import { useTranslation } from 'react-i18next'
  4. import type { CrawlResultItem } from '@/models/datasets'
  5. import { CrawlStep } from '@/models/datasets'
  6. import Header from '../base/header'
  7. import Options from './base/options'
  8. import Crawling from './base/crawling'
  9. import ErrorMessage from './base/error-message'
  10. import CrawledResult from './base/crawled-result'
  11. import {
  12. useDraftPipelinePreProcessingParams,
  13. usePublishedPipelinePreProcessingParams,
  14. } from '@/service/use-pipeline'
  15. import { useDatasetDetailContextWithSelector } from '@/context/dataset-detail'
  16. import { DatasourceType } from '@/models/pipeline'
  17. import { ssePost } from '@/service/base'
  18. import type {
  19. DataSourceNodeCompletedResponse,
  20. DataSourceNodeErrorResponse,
  21. DataSourceNodeProcessingResponse,
  22. } from '@/types/pipeline'
  23. import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
  24. import { useDataSourceStore, useDataSourceStoreWithSelector } from '../store'
  25. import { useShallow } from 'zustand/react/shallow'
  26. import { useModalContextSelector } from '@/context/modal-context'
  27. import { useGetDataSourceAuth } from '@/service/use-datasource'
  28. const I18N_PREFIX = 'datasetCreation.stepOne.website'
  29. export type WebsiteCrawlProps = {
  30. nodeId: string
  31. nodeData: DataSourceNodeType
  32. isInPipeline?: boolean
  33. onCredentialChange: (credentialId: string) => void
  34. }
  35. const WebsiteCrawl = ({
  36. nodeId,
  37. nodeData,
  38. isInPipeline = false,
  39. onCredentialChange,
  40. }: WebsiteCrawlProps) => {
  41. const { t } = useTranslation()
  42. const [totalNum, setTotalNum] = useState(0)
  43. const [crawledNum, setCrawledNum] = useState(0)
  44. const [crawlErrorMessage, setCrawlErrorMessage] = useState('')
  45. const pipelineId = useDatasetDetailContextWithSelector(s => s.dataset?.pipeline_id)
  46. const setShowAccountSettingModal = useModalContextSelector(s => s.setShowAccountSettingModal)
  47. const {
  48. crawlResult,
  49. step,
  50. checkedCrawlResult,
  51. previewIndex,
  52. currentCredentialId,
  53. } = useDataSourceStoreWithSelector(useShallow(state => ({
  54. crawlResult: state.crawlResult,
  55. step: state.step,
  56. checkedCrawlResult: state.websitePages,
  57. previewIndex: state.previewIndex,
  58. currentCredentialId: state.currentCredentialId,
  59. })))
  60. const { data: dataSourceAuth } = useGetDataSourceAuth({
  61. pluginId: nodeData.plugin_id,
  62. provider: nodeData.provider_name,
  63. })
  64. const dataSourceStore = useDataSourceStore()
  65. const usePreProcessingParams = useRef(!isInPipeline ? usePublishedPipelinePreProcessingParams : useDraftPipelinePreProcessingParams)
  66. const { data: paramsConfig, isFetching: isFetchingParams } = usePreProcessingParams.current({
  67. pipeline_id: pipelineId!,
  68. node_id: nodeId,
  69. }, !!pipelineId && !!nodeId)
  70. const isInit = step === CrawlStep.init
  71. const isCrawlFinished = step === CrawlStep.finished
  72. const isRunning = step === CrawlStep.running
  73. const showError = isCrawlFinished && crawlErrorMessage
  74. const datasourceNodeRunURL = !isInPipeline
  75. ? `/rag/pipelines/${pipelineId}/workflows/published/datasource/nodes/${nodeId}/run`
  76. : `/rag/pipelines/${pipelineId}/workflows/draft/datasource/nodes/${nodeId}/run`
  77. const handleCheckedCrawlResultChange = useCallback((checkedCrawlResult: CrawlResultItem[]) => {
  78. const { setWebsitePages } = dataSourceStore.getState()
  79. setWebsitePages(checkedCrawlResult)
  80. }, [dataSourceStore])
  81. const handlePreview = useCallback((website: CrawlResultItem, index: number) => {
  82. const { setCurrentWebsite, setPreviewIndex } = dataSourceStore.getState()
  83. setCurrentWebsite(website)
  84. setPreviewIndex(index)
  85. }, [dataSourceStore])
  86. const handleRun = useCallback(async (value: Record<string, any>) => {
  87. const { setStep, setCrawlResult, currentCredentialId } = dataSourceStore.getState()
  88. setStep(CrawlStep.running)
  89. ssePost(
  90. datasourceNodeRunURL,
  91. {
  92. body: {
  93. inputs: value,
  94. datasource_type: DatasourceType.websiteCrawl,
  95. credential_id: currentCredentialId,
  96. response_mode: 'streaming',
  97. },
  98. },
  99. {
  100. onDataSourceNodeProcessing: (data: DataSourceNodeProcessingResponse) => {
  101. setTotalNum(data.total ?? 0)
  102. setCrawledNum(data.completed ?? 0)
  103. },
  104. onDataSourceNodeCompleted: (data: DataSourceNodeCompletedResponse) => {
  105. const { data: crawlData, time_consuming } = data
  106. const crawlResultData = {
  107. data: crawlData as CrawlResultItem[],
  108. time_consuming: time_consuming ?? 0,
  109. }
  110. setCrawlResult(crawlResultData)
  111. handleCheckedCrawlResultChange(isInPipeline ? [crawlData[0]] : crawlData) // default select the crawl result
  112. setCrawlErrorMessage('')
  113. setStep(CrawlStep.finished)
  114. },
  115. onDataSourceNodeError: (error: DataSourceNodeErrorResponse) => {
  116. setCrawlErrorMessage(error.error || t(`${I18N_PREFIX}.unknownError`))
  117. setStep(CrawlStep.finished)
  118. },
  119. },
  120. )
  121. }, [dataSourceStore, datasourceNodeRunURL, handleCheckedCrawlResultChange, isInPipeline, t])
  122. const handleSubmit = useCallback((value: Record<string, any>) => {
  123. handleRun(value)
  124. }, [handleRun])
  125. const handleSetting = useCallback(() => {
  126. setShowAccountSettingModal({
  127. payload: 'data-source',
  128. })
  129. }, [setShowAccountSettingModal])
  130. const handleCredentialChange = useCallback((credentialId: string) => {
  131. setCrawledNum(0)
  132. setTotalNum(0)
  133. setCrawlErrorMessage('')
  134. onCredentialChange(credentialId)
  135. }, [dataSourceStore, onCredentialChange])
  136. return (
  137. <div className='flex flex-col'>
  138. <Header
  139. docTitle='How to use?'
  140. docLink='https://docs.dify.ai'
  141. onClickConfiguration={handleSetting}
  142. pluginName={nodeData.datasource_label}
  143. currentCredentialId={currentCredentialId}
  144. onCredentialChange={handleCredentialChange}
  145. credentials={dataSourceAuth?.result || []}
  146. />
  147. <div className='mt-2 rounded-xl border border-components-panel-border bg-background-default-subtle'>
  148. <Options
  149. variables={paramsConfig?.variables || []}
  150. step={step}
  151. runDisabled={!currentCredentialId || isFetchingParams}
  152. onSubmit={handleSubmit}
  153. />
  154. </div>
  155. {!isInit && (
  156. <div className='relative flex flex-col'>
  157. {isRunning && (
  158. <Crawling
  159. crawledNum={crawledNum}
  160. totalNum={totalNum}
  161. />
  162. )}
  163. {showError && (
  164. <ErrorMessage
  165. className='mt-2'
  166. title={t(`${I18N_PREFIX}.exceptionErrorTitle`)}
  167. errorMsg={crawlErrorMessage}
  168. />
  169. )}
  170. {isCrawlFinished && !showError && (
  171. <CrawledResult
  172. className='mt-2'
  173. list={crawlResult?.data || []}
  174. checkedList={checkedCrawlResult}
  175. onSelectedChange={handleCheckedCrawlResultChange}
  176. usedTime={Number.parseFloat(crawlResult?.time_consuming as string) || 0}
  177. previewIndex={previewIndex}
  178. onPreview={handlePreview}
  179. showPreview={!isInPipeline}
  180. isMultipleChoice={!isInPipeline} // only support single choice in test run
  181. />
  182. )}
  183. </div>
  184. )}
  185. </div>
  186. )
  187. }
  188. export default React.memo(WebsiteCrawl)