index.tsx 7.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230
  1. 'use client'
  2. import type { FC } from 'react'
  3. import React, { useCallback, useEffect, useState } from 'react'
  4. import { useTranslation } from 'react-i18next'
  5. import UrlInput from '../base/url-input'
  6. import OptionsWrap from '../base/options-wrap'
  7. import CrawledResult from '../base/crawled-result'
  8. import Crawling from '../base/crawling'
  9. import ErrorMessage from '../base/error-message'
  10. import Header from './header'
  11. import Options from './options'
  12. import { useModalContext } from '@/context/modal-context'
  13. import Toast from '@/app/components/base/toast'
  14. import { checkJinaReaderTaskStatus, createJinaReaderTask } from '@/service/datasets'
  15. import { sleep } from '@/utils'
  16. import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
  17. const ERROR_I18N_PREFIX = 'common.errorMsg'
  18. const I18N_PREFIX = 'datasetCreation.stepOne.website'
  19. type Props = {
  20. onPreview: (payload: CrawlResultItem) => void
  21. checkedCrawlResult: CrawlResultItem[]
  22. onCheckedCrawlResultChange: (payload: CrawlResultItem[]) => void
  23. onJobIdChange: (jobId: string) => void
  24. crawlOptions: CrawlOptions
  25. onCrawlOptionsChange: (payload: CrawlOptions) => void
  26. }
  27. enum Step {
  28. init = 'init',
  29. running = 'running',
  30. finished = 'finished',
  31. }
  32. const JinaReader: FC<Props> = ({
  33. onPreview,
  34. checkedCrawlResult,
  35. onCheckedCrawlResultChange,
  36. onJobIdChange,
  37. crawlOptions,
  38. onCrawlOptionsChange,
  39. }) => {
  40. const { t } = useTranslation()
  41. const [step, setStep] = useState<Step>(Step.init)
  42. const [controlFoldOptions, setControlFoldOptions] = useState<number>(0)
  43. useEffect(() => {
  44. if (step !== Step.init)
  45. setControlFoldOptions(Date.now())
  46. }, [step])
  47. const { setShowAccountSettingModal } = useModalContext()
  48. const handleSetting = useCallback(() => {
  49. setShowAccountSettingModal({
  50. payload: 'data-source',
  51. })
  52. }, [setShowAccountSettingModal])
  53. const checkValid = useCallback((url: string) => {
  54. let errorMsg = ''
  55. if (!url) {
  56. errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
  57. field: 'url',
  58. })
  59. }
  60. if (!errorMsg && !((url.startsWith('http://') || url.startsWith('https://'))))
  61. errorMsg = t(`${ERROR_I18N_PREFIX}.urlError`)
  62. if (!errorMsg && (crawlOptions.limit === null || crawlOptions.limit === undefined || crawlOptions.limit === '')) {
  63. errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
  64. field: t(`${I18N_PREFIX}.limit`),
  65. })
  66. }
  67. return {
  68. isValid: !errorMsg,
  69. errorMsg,
  70. }
  71. }, [crawlOptions, t])
  72. const isInit = step === Step.init
  73. const isCrawlFinished = step === Step.finished
  74. const isRunning = step === Step.running
  75. const [crawlResult, setCrawlResult] = useState<{
  76. current: number
  77. total: number
  78. data: CrawlResultItem[]
  79. time_consuming: number | string
  80. } | undefined>(undefined)
  81. const [crawlErrorMessage, setCrawlErrorMessage] = useState('')
  82. const showError = isCrawlFinished && crawlErrorMessage
  83. const waitForCrawlFinished = useCallback(async (jobId: string) => {
  84. try {
  85. const res = await checkJinaReaderTaskStatus(jobId) as any
  86. if (res.status === 'completed') {
  87. return {
  88. isError: false,
  89. data: {
  90. ...res,
  91. total: Math.min(res.total, Number.parseFloat(crawlOptions.limit as string)),
  92. },
  93. }
  94. }
  95. if (res.status === 'failed' || !res.status) {
  96. return {
  97. isError: true,
  98. errorMessage: res.message,
  99. data: {
  100. data: [],
  101. },
  102. }
  103. }
  104. // update the progress
  105. setCrawlResult({
  106. ...res,
  107. total: Math.min(res.total, Number.parseFloat(crawlOptions.limit as string)),
  108. })
  109. onCheckedCrawlResultChange(res.data || []) // default select the crawl result
  110. await sleep(2500)
  111. return await waitForCrawlFinished(jobId)
  112. }
  113. catch (e: any) {
  114. const errorBody = await e.json()
  115. return {
  116. isError: true,
  117. errorMessage: errorBody.message,
  118. data: {
  119. data: [],
  120. },
  121. }
  122. }
  123. }, [crawlOptions.limit])
  124. const handleRun = useCallback(async (url: string) => {
  125. const { isValid, errorMsg } = checkValid(url)
  126. if (!isValid) {
  127. Toast.notify({
  128. message: errorMsg!,
  129. type: 'error',
  130. })
  131. return
  132. }
  133. setStep(Step.running)
  134. try {
  135. const startTime = Date.now()
  136. const res = await createJinaReaderTask({
  137. url,
  138. options: crawlOptions,
  139. }) as any
  140. if (res.data) {
  141. const data = {
  142. current: 1,
  143. total: 1,
  144. data: [{
  145. title: res.data.title,
  146. markdown: res.data.content,
  147. description: res.data.description,
  148. source_url: res.data.url,
  149. }],
  150. time_consuming: (Date.now() - startTime) / 1000,
  151. }
  152. setCrawlResult(data)
  153. onCheckedCrawlResultChange(data.data || [])
  154. setCrawlErrorMessage('')
  155. }
  156. else if (res.job_id) {
  157. const jobId = res.job_id
  158. onJobIdChange(jobId)
  159. const { isError, data, errorMessage } = await waitForCrawlFinished(jobId)
  160. if (isError) {
  161. setCrawlErrorMessage(errorMessage || t(`${I18N_PREFIX}.unknownError`))
  162. }
  163. else {
  164. setCrawlResult(data)
  165. onCheckedCrawlResultChange(data.data || []) // default select the crawl result
  166. setCrawlErrorMessage('')
  167. }
  168. }
  169. }
  170. catch (e) {
  171. setCrawlErrorMessage(t(`${I18N_PREFIX}.unknownError`)!)
  172. console.log(e)
  173. }
  174. finally {
  175. setStep(Step.finished)
  176. }
  177. }, [checkValid, crawlOptions, onJobIdChange, t, waitForCrawlFinished])
  178. return (
  179. <div>
  180. <Header onSetting={handleSetting} />
  181. <div className='mt-2 p-4 pb-0 rounded-xl border border-components-panel-border bg-background-default-subtle'>
  182. <UrlInput onRun={handleRun} isRunning={isRunning} />
  183. <OptionsWrap
  184. className='mt-4'
  185. controlFoldOptions={controlFoldOptions}
  186. >
  187. <Options className='mt-2' payload={crawlOptions} onChange={onCrawlOptionsChange} />
  188. </OptionsWrap>
  189. {!isInit && (
  190. <div className='mt-3 relative left-[-16px] w-[calc(100%_+_32px)] rounded-b-xl'>
  191. {isRunning
  192. && <Crawling
  193. className='mt-2'
  194. crawledNum={crawlResult?.current || 0}
  195. totalNum={crawlResult?.total || Number.parseFloat(crawlOptions.limit as string) || 0}
  196. />}
  197. {showError && (
  198. <ErrorMessage className='rounded-b-xl' title={t(`${I18N_PREFIX}.exceptionErrorTitle`)} errorMsg={crawlErrorMessage} />
  199. )}
  200. {isCrawlFinished && !showError
  201. && <CrawledResult
  202. className='mb-2'
  203. list={crawlResult?.data || []}
  204. checkedList={checkedCrawlResult}
  205. onSelectedChange={onCheckedCrawlResultChange}
  206. onPreview={onPreview}
  207. usedTime={Number.parseFloat(crawlResult?.time_consuming as string) || 0}
  208. />
  209. }
  210. </div>
  211. )}
  212. </div>
  213. </div>
  214. )
  215. }
  216. export default React.memo(JinaReader)