Refactor: dataset creation components and implement Firecrawl functionality

This commit is contained in:
twwu
2025-04-28 13:33:16 +08:00
parent 5b89d36ea1
commit f71b0eccb2
26 changed files with 788 additions and 448 deletions

View File

@@ -6,7 +6,7 @@ import type { DataSourceNotionPageMap, DataSourceNotionWorkspace, NotionPage } f
import { useModalContextSelector } from '@/context/modal-context'
import NotionConnector from '../notion-connector'
import { usePreImportNotionPages } from '@/service/knowledge/use-import'
import Header from './header'
import Header from '../../datasets/create/website/base/header'
type NotionPageSelectorProps = {
value?: string[]
@@ -99,7 +99,11 @@ const NotionPageSelector = ({
<div className='flex flex-col gap-y-2'>
<Header
isInPipeline={isInPipeline}
handleConfigureNotion={handleConfigureNotion}
onClickConfiguration={handleConfigureNotion}
title={'Choose notion pages'}
buttonText={'Configure Notion'}
docTitle={'Notion docs'}
docLink={'https://www.notion.so/docs'}
/>
<div className='rounded-xl border border-components-panel-border bg-background-default-subtle'>
<div className='flex h-12 items-center gap-x-2 rounded-t-xl border-b border-b-divider-regular bg-components-panel-bg p-2'>

View File

@@ -1,17 +1,25 @@
import React from 'react'
import Divider from '../../divider'
import Button from '../../button'
import Divider from '@/app/components/base/divider'
import Button from '@/app/components/base/button'
import cn from '@/utils/classnames'
import { RiBookOpenLine, RiEqualizer2Line } from '@remixicon/react'
type HeaderProps = {
isInPipeline?: boolean
handleConfigureNotion: () => void
onClickConfiguration: () => void
title: string
buttonText: string
docTitle: string
docLink: string
}
const Header = ({
isInPipeline = false,
handleConfigureNotion,
onClickConfiguration,
title,
buttonText,
docTitle,
docLink,
}: HeaderProps) => {
return (
<div className='flex items-center gap-x-2'>
@@ -20,7 +28,7 @@ const Header = ({
'text-text-secondary',
isInPipeline ? 'system-sm-semibold' : 'system-md-semibold',
)}>
Choose notion pages
{title}
</div>
<Divider type='vertical' className='mx-1 h-3.5' />
<Button
@@ -30,23 +38,23 @@ const Header = ({
>
<RiEqualizer2Line
className='h-4 w-4'
onClick={handleConfigureNotion}
onClick={onClickConfiguration}
/>
{!isInPipeline && (
<span className='system-xs-medium'>
Configure Notion
{buttonText}
</span>
)}
</Button>
</div>
<a
className='system-xs-medium flex items-center gap-x-1 text-text-accent'
href='https://www.notion.so/docs'
className='system-xs-medium flex shrink-0 items-center gap-x-1 text-text-accent'
href={docLink}
target='_blank'
rel='noopener noreferrer'
>
<RiBookOpenLine className='size-3.5' />
<span>Notion docs</span>
<span>{docTitle}</span>
</a>
</div>
)

View File

@@ -1,43 +0,0 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import { useTranslation } from 'react-i18next'
import { RiBookOpenLine, RiEqualizer2Line } from '@remixicon/react'
import Button from '@/app/components/base/button'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
onSetting: () => void
}
const Header: FC<Props> = ({
onSetting,
}) => {
const { t } = useTranslation()
return (
<div className='flex h-6 items-center justify-between'>
<div className='flex items-center'>
<div className='text-base font-medium text-text-secondary'>{t(`${I18N_PREFIX}.firecrawlTitle`)}</div>
<div className='ml-2 mr-2 h-3.5 w-px bg-divider-regular' />
<Button className='flex h-6 items-center gap-x-[1px] px-1.5' onClick={onSetting}>
<RiEqualizer2Line className='h-3.5 w-3.5 text-components-button-secondary-text' />
<span className='px-[3px] text-xs font-medium text-components-button-secondary-text'>
{t(`${I18N_PREFIX}.configureFirecrawl`)}
</span>
</Button>
</div>
<a
href='https://docs.firecrawl.dev/introduction'
target='_blank'
rel='noopener noreferrer'
className='inline-flex items-center gap-x-1 text-xs font-medium text-text-accent'
>
<RiBookOpenLine className='h-3.5 w-3.5 text-text-accent' />
<span>{t(`${I18N_PREFIX}.firecrawlDoc`)}</span>
</a>
</div>
)
}
export default React.memo(Header)

View File

@@ -7,13 +7,13 @@ import OptionsWrap from '../base/options-wrap'
import CrawledResult from '../base/crawled-result'
import Crawling from '../base/crawling'
import ErrorMessage from '../base/error-message'
import Header from './header'
import Options from './options'
import { useModalContext } from '@/context/modal-context'
import { useModalContextSelector } from '@/context/modal-context'
import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
import Toast from '@/app/components/base/toast'
import { checkFirecrawlTaskStatus, createFirecrawlTask } from '@/service/datasets'
import { sleep } from '@/utils'
import Header from '../base/header'
const ERROR_I18N_PREFIX = 'common.errorMsg'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
@@ -48,7 +48,7 @@ const FireCrawl: FC<Props> = ({
if (step !== Step.init)
setControlFoldOptions(Date.now())
}, [step])
const { setShowAccountSettingModal } = useModalContext()
const setShowAccountSettingModal = useModalContextSelector(s => s.setShowAccountSettingModal)
const handleSetting = useCallback(() => {
setShowAccountSettingModal({
payload: 'data-source',
@@ -131,7 +131,7 @@ const FireCrawl: FC<Props> = ({
},
}
}
}, [crawlOptions.limit])
}, [crawlOptions.limit, onCheckedCrawlResultChange])
const handleRun = useCallback(async (url: string) => {
const { isValid, errorMsg } = checkValid(url)
@@ -173,11 +173,17 @@ const FireCrawl: FC<Props> = ({
finally {
setStep(Step.finished)
}
}, [checkValid, crawlOptions, onJobIdChange, t, waitForCrawlFinished])
}, [checkValid, crawlOptions, onJobIdChange, t, waitForCrawlFinished, onCheckedCrawlResultChange])
return (
<div>
<Header onSetting={handleSetting} />
<Header
onClickConfiguration={handleSetting}
title={t(`${I18N_PREFIX}.firecrawlTitle`)}
buttonText={t(`${I18N_PREFIX}.configureFirecrawl`)}
docTitle={t(`${I18N_PREFIX}.firecrawlDoc`)}
docLink={'https://docs.firecrawl.dev/introduction'}
/>
<div className='mt-2 rounded-xl border border-components-panel-border bg-background-default-subtle p-4 pb-0'>
<UrlInput onRun={handleRun} isRunning={isRunning} />
<OptionsWrap

View File

@@ -1,30 +0,0 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import cn from '@/utils/classnames'
import { AlertTriangle } from '@/app/components/base/icons/src/vender/solid/alertsAndFeedback'
type Props = {
className?: string
title: string
errorMsg?: string
}
const ErrorMessage: FC<Props> = ({
className,
title,
errorMsg,
}) => {
return (
<div className={cn(className, 'border-t border-gray-200 bg-[#FFFAEB] px-4 py-2')}>
<div className='flex h-5 items-center'>
<AlertTriangle className='mr-2 h-4 w-4 text-[#F79009]' />
<div className='text-sm font-medium text-[#DC6803]'>{title}</div>
</div>
{errorMsg && (
<div className='mt-1 pl-6 text-xs font-normal leading-[18px] text-gray-700'>{errorMsg}</div>
)}
</div>
)
}
export default React.memo(ErrorMessage)

View File

@@ -1,54 +0,0 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import Input from './input'
import cn from '@/utils/classnames'
import Tooltip from '@/app/components/base/tooltip'
type Props = {
className?: string
label: string
labelClassName?: string
value: string | number
onChange: (value: string | number) => void
isRequired?: boolean
placeholder?: string
isNumber?: boolean
tooltip?: string
}
const Field: FC<Props> = ({
className,
label,
labelClassName,
value,
onChange,
isRequired = false,
placeholder = '',
isNumber = false,
tooltip,
}) => {
return (
<div className={cn(className)}>
<div className='flex py-[7px]'>
<div className={cn(labelClassName, 'flex h-[18px] items-center text-[13px] font-medium text-gray-900')}>{label} </div>
{isRequired && <span className='ml-0.5 text-xs font-semibold text-[#D92D20]'>*</span>}
{tooltip && (
<Tooltip
popupContent={
<div className='w-[200px]'>{tooltip}</div>
}
triggerClassName='ml-0.5 w-4 h-4'
/>
)}
</div>
<Input
value={value}
onChange={onChange}
placeholder={placeholder}
isNumber={isNumber}
/>
</div>
)
}
export default React.memo(Field)

View File

@@ -1,58 +0,0 @@
'use client'
import type { FC } from 'react'
import React, { useCallback } from 'react'
type Props = {
value: string | number
onChange: (value: string | number) => void
placeholder?: string
isNumber?: boolean
}
const MIN_VALUE = 0
const Input: FC<Props> = ({
value,
onChange,
placeholder = '',
isNumber = false,
}) => {
const handleChange = useCallback((e: React.ChangeEvent<HTMLInputElement>) => {
const value = e.target.value
if (isNumber) {
let numberValue = Number.parseInt(value, 10) // integer only
if (isNaN(numberValue)) {
onChange('')
return
}
if (numberValue < MIN_VALUE)
numberValue = MIN_VALUE
onChange(numberValue)
return
}
onChange(value)
}, [isNumber, onChange])
const otherOption = (() => {
if (isNumber) {
return {
min: MIN_VALUE,
}
}
return {
}
})()
return (
<input
type={isNumber ? 'number' : 'text'}
{...otherOption}
value={value}
onChange={handleChange}
className='flex h-9 w-full rounded-lg bg-gray-100 px-2 py-1 text-xs leading-normal caret-primary-600 placeholder:text-gray-400 hover:bg-gray-100 focus:bg-gray-50 focus:ring-1 focus:ring-inset focus:ring-gray-200 focus-visible:outline-none'
placeholder={placeholder}
/>
)
}
export default React.memo(Input)

View File

@@ -1,55 +0,0 @@
'use client'
import { useBoolean } from 'ahooks'
import type { FC } from 'react'
import React, { useEffect } from 'react'
import { useTranslation } from 'react-i18next'
import cn from '@/utils/classnames'
import { Settings04 } from '@/app/components/base/icons/src/vender/line/general'
import { ChevronRight } from '@/app/components/base/icons/src/vender/line/arrows'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
className?: string
children: React.ReactNode
controlFoldOptions?: number
}
const OptionsWrap: FC<Props> = ({
className = '',
children,
controlFoldOptions,
}) => {
const { t } = useTranslation()
const [fold, {
toggle: foldToggle,
setTrue: foldHide,
}] = useBoolean(false)
useEffect(() => {
if (controlFoldOptions)
foldHide()
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [controlFoldOptions])
return (
<div className={cn(className, !fold ? 'mb-0' : 'mb-3')}>
<div
className='flex h-[26px] cursor-pointer select-none items-center justify-between py-1'
onClick={foldToggle}
>
<div className='flex items-center text-gray-700'>
<Settings04 className='mr-1 h-4 w-4' />
<div className='text-[13px] font-semibold uppercase text-gray-800'>{t(`${I18N_PREFIX}.options`)}</div>
</div>
<ChevronRight className={cn(!fold && 'rotate-90', 'h-4 w-4 text-gray-500')} />
</div>
{!fold && (
<div className='mb-4'>
{children}
</div>
)}
</div>
)
}
export default React.memo(OptionsWrap)

View File

@@ -1,48 +0,0 @@
'use client'
import type { FC } from 'react'
import React, { useCallback, useState } from 'react'
import { useTranslation } from 'react-i18next'
import Input from './input'
import Button from '@/app/components/base/button'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
isRunning: boolean
onRun: (url: string) => void
}
const UrlInput: FC<Props> = ({
isRunning,
onRun,
}) => {
const { t } = useTranslation()
const [url, setUrl] = useState('')
const handleUrlChange = useCallback((url: string | number) => {
setUrl(url as string)
}, [])
const handleOnRun = useCallback(() => {
if (isRunning)
return
onRun(url)
}, [isRunning, onRun, url])
return (
<div className='flex items-center justify-between'>
<Input
value={url}
onChange={handleUrlChange}
placeholder='https://docs.dify.ai'
/>
<Button
variant='primary'
onClick={handleOnRun}
className='ml-2'
loading={isRunning}
>
{!isRunning ? t(`${I18N_PREFIX}.run`) : ''}
</Button>
</div>
)
}
export default React.memo(UrlInput)

View File

@@ -1,43 +0,0 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import { useTranslation } from 'react-i18next'
import { RiBookOpenLine, RiEqualizer2Line } from '@remixicon/react'
import Button from '@/app/components/base/button'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
onSetting: () => void
}
const Header: FC<Props> = ({
onSetting,
}) => {
const { t } = useTranslation()
return (
<div className='flex h-6 items-center justify-between'>
<div className='flex items-center'>
<div className='text-sm font-semibold text-text-secondary'>{t(`${I18N_PREFIX}.jinaReaderTitle`)}</div>
<div className='ml-2 mr-2 h-3.5 w-px bg-divider-regular' />
<Button className='flex h-6 items-center gap-x-[1px] px-1.5' onClick={onSetting}>
<RiEqualizer2Line className='h-3.5 w-3.5 text-components-button-secondary-text' />
<span className='px-[3px] text-xs font-medium text-components-button-secondary-text'>
{t(`${I18N_PREFIX}.configureJinaReader`)}
</span>
</Button>
</div>
<a
href='https://jina.ai/reader'
target='_blank'
rel='noopener noreferrer'
className='inline-flex items-center gap-x-1 text-xs font-medium text-text-accent'
>
<RiBookOpenLine className='h-3.5 w-3.5 text-text-accent' />
<span>{t(`${I18N_PREFIX}.jinaReaderDoc`)}</span>
</a>
</div>
)
}
export default React.memo(Header)

View File

@@ -7,13 +7,13 @@ import OptionsWrap from '../base/options-wrap'
import CrawledResult from '../base/crawled-result'
import Crawling from '../base/crawling'
import ErrorMessage from '../base/error-message'
import Header from './header'
import Options from './options'
import { useModalContext } from '@/context/modal-context'
import Toast from '@/app/components/base/toast'
import { checkJinaReaderTaskStatus, createJinaReaderTask } from '@/service/datasets'
import { sleep } from '@/utils'
import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
import Header from '../base/header'
const ERROR_I18N_PREFIX = 'common.errorMsg'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
@@ -190,7 +190,13 @@ const JinaReader: FC<Props> = ({
return (
<div>
<Header onSetting={handleSetting} />
<Header
onClickConfiguration={handleSetting}
title={t(`${I18N_PREFIX}.jinaReaderTitle`)}
buttonText={t(`${I18N_PREFIX}.configureJinaReader`)}
docTitle={t(`${I18N_PREFIX}.jinaReaderDoc`)}
docLink={'https://jina.ai/reader'}
/>
<div className='mt-2 rounded-xl border border-components-panel-border bg-background-default-subtle p-4 pb-0'>
<UrlInput onRun={handleRun} isRunning={isRunning} />
<OptionsWrap

View File

@@ -1,43 +0,0 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import { useTranslation } from 'react-i18next'
import { RiBookOpenLine, RiEqualizer2Line } from '@remixicon/react'
import Button from '@/app/components/base/button'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
onSetting: () => void
}
const Header: FC<Props> = ({
onSetting,
}) => {
const { t } = useTranslation()
return (
<div className='flex h-6 items-center justify-between'>
<div className='flex items-center'>
<div className='text-base font-medium text-text-secondary'>{t(`${I18N_PREFIX}.watercrawlTitle`)}</div>
<div className='ml-2 mr-2 h-3.5 w-px bg-divider-regular' />
<Button className='flex h-6 items-center gap-x-[1px] px-1.5' onClick={onSetting}>
<RiEqualizer2Line className='h-3.5 w-3.5 text-components-button-secondary-text' />
<span className='px-[3px] text-xs font-medium text-components-button-secondary-text'>
{t(`${I18N_PREFIX}.configureWatercrawl`)}
</span>
</Button>
</div>
<a
href='https://docs.watercrawl.dev/'
target='_blank'
rel='noopener noreferrer'
className='inline-flex items-center gap-x-1 text-xs font-medium text-text-accent'
>
<RiBookOpenLine className='h-3.5 w-3.5 text-text-accent' />
<span>{t(`${I18N_PREFIX}.watercrawlDoc`)}</span>
</a>
</div>
)
}
export default React.memo(Header)

View File

@@ -7,13 +7,13 @@ import OptionsWrap from '../base/options-wrap'
import CrawledResult from '../base/crawled-result'
import Crawling from '../base/crawling'
import ErrorMessage from '../base/error-message'
import Header from './header'
import Options from './options'
import { useModalContext } from '@/context/modal-context'
import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
import Toast from '@/app/components/base/toast'
import { checkWatercrawlTaskStatus, createWatercrawlTask } from '@/service/datasets'
import { sleep } from '@/utils'
import Header from '../base/header'
const ERROR_I18N_PREFIX = 'common.errorMsg'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
@@ -177,7 +177,13 @@ const WaterCrawl: FC<Props> = ({
return (
<div>
<Header onSetting={handleSetting} />
<Header
onClickConfiguration={handleSetting}
title={t(`${I18N_PREFIX}.watercrawlTitle`)}
buttonText={t(`${I18N_PREFIX}.configureWatercrawl`)}
docTitle={t(`${I18N_PREFIX}.watercrawlDoc`)}
docLink={'https://docs.watercrawl.dev/'}
/>
<div className='mt-2 rounded-xl border border-components-panel-border bg-background-default-subtle p-4 pb-0'>
<UrlInput onRun={handleRun} isRunning={isRunning} />
<OptionsWrap

View File

@@ -35,12 +35,13 @@ const InputFieldForm = ({
const result = schema.safeParse(value)
if (!result.success) {
const issues = result.error.issues
const firstIssue = issues[0].message
const firstIssue = issues[0]
const errorMessage = `"${firstIssue.path.join('.')}" ${firstIssue.message}`
Toast.notify({
type: 'error',
message: firstIssue,
message: errorMessage,
})
return firstIssue
return errorMessage
}
return undefined
},

View File

@@ -0,0 +1,11 @@
import type { CrawlOptions } from '@/models/datasets'
export const DEFAULT_CRAWL_OPTIONS: CrawlOptions = {
crawl_sub_pages: true,
only_main_content: true,
includes: '',
excludes: '',
limit: 10,
max_depth: '',
use_sitemap: true,
}

View File

@@ -1,11 +1,10 @@
'use client'
import type { FC } from 'react'
import React from 'react'
import cn from '@/utils/classnames'
import Checkbox from '@/app/components/base/checkbox'
import Tooltip from '@/app/components/base/tooltip'
type Props = {
type CheckboxWithLabelProps = {
className?: string
isChecked: boolean
onChange: (isChecked: boolean) => void
@@ -14,18 +13,18 @@ type Props = {
tooltip?: string
}
const CheckboxWithLabel: FC<Props> = ({
const CheckboxWithLabel = ({
className = '',
isChecked,
onChange,
label,
labelClassName,
tooltip,
}) => {
}: CheckboxWithLabelProps) => {
return (
<label className={cn(className, 'flex h-7 items-center space-x-2')}>
<label className={cn('flex items-center space-x-2', className)}>
<Checkbox checked={isChecked} onCheck={() => onChange(!isChecked)} />
<div className={cn('text-sm font-normal text-gray-800', labelClassName)}>{label}</div>
<div className={cn('system-sm-medium text-text-secondary', labelClassName)}>{label}</div>
{tooltip && (
<Tooltip
popupContent={

View File

@@ -0,0 +1,45 @@
'use client'
import React, { useCallback } from 'react'
import cn from '@/utils/classnames'
import type { CrawlResultItem as CrawlResultItemType } from '@/models/datasets'
import Checkbox from '@/app/components/base/checkbox'
type CrawledResultItemProps = {
payload: CrawlResultItemType
isChecked: boolean
onCheckChange: (checked: boolean) => void
}
const CrawledResultItem = ({
payload,
isChecked,
onCheckChange,
}: CrawledResultItemProps) => {
const handleCheckChange = useCallback(() => {
onCheckChange(!isChecked)
}, [isChecked, onCheckChange])
return (
<div className={cn('group flex cursor-pointer gap-x-2 rounded-lg p-2 hover:bg-state-base-hover')}>
<Checkbox
className='shrink-0'
checked={isChecked}
onCheck={handleCheckChange}
/>
<div className='flex min-w-0 grow flex-col gap-y-0.5'>
<div
className='system-sm-medium truncate text-text-secondary'
title={payload.title}
>
{payload.title}
</div>
<div
className='system-xs-regular truncate text-text-tertiary'
title={payload.source_url}
>
{payload.source_url}
</div>
</div>
</div>
)
}
export default React.memo(CrawledResultItem)

View File

@@ -0,0 +1,77 @@
'use client'
import React, { useCallback } from 'react'
import { useTranslation } from 'react-i18next'
import cn from '@/utils/classnames'
import type { CrawlResultItem } from '@/models/datasets'
import CheckboxWithLabel from './checkbox-with-label'
import CrawledResultItem from './crawled-result-item'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type CrawledResultProps = {
className?: string
list: CrawlResultItem[]
checkedList: CrawlResultItem[]
onSelectedChange: (selected: CrawlResultItem[]) => void
usedTime: number
}
const CrawledResult = ({
className = '',
list,
checkedList,
onSelectedChange,
usedTime,
}: CrawledResultProps) => {
const { t } = useTranslation()
const isCheckAll = checkedList.length === list.length
const handleCheckedAll = useCallback(() => {
if (!isCheckAll)
onSelectedChange(list)
else
onSelectedChange([])
}, [isCheckAll, list, onSelectedChange])
const handleItemCheckChange = useCallback((item: CrawlResultItem) => {
return (checked: boolean) => {
if (checked)
onSelectedChange([...checkedList, item])
else
onSelectedChange(checkedList.filter(checkedItem => checkedItem.source_url !== item.source_url))
}
}, [checkedList, onSelectedChange])
return (
<div className={cn('flex flex-col gap-y-2', className)}>
<div className='system-sm-medium pt-2 text-text-primary'>
{t(`${I18N_PREFIX}.scrapTimeInfo`, {
total: list.length,
time: usedTime.toFixed(1),
})}
</div>
<div className='rounded-xl border border-components-panel-border bg-components-panel-bg'>
<div className='flex items-center px-4 py-2'>
<CheckboxWithLabel
isChecked={isCheckAll}
onChange={handleCheckedAll} label={isCheckAll ? t(`${I18N_PREFIX}.resetAll`) : t(`${I18N_PREFIX}.selectAll`)}
/>
</div>
<div className='flex flex-col gap-y-px border-t border-divider-subtle bg-background-default-subtle p-2'>
{list.map(item => (
<CrawledResultItem
key={item.source_url}
payload={item}
isChecked={checkedList.some(checkedItem => checkedItem.source_url === item.source_url)}
onCheckChange={handleItemCheckChange(item)}
/>
))}
</div>
</div>
</div>
)
}
export default React.memo(CrawledResult)

View File

@@ -0,0 +1,89 @@
'use client'
import React from 'react'
import { useTranslation } from 'react-i18next'
import cn from '@/utils/classnames'
type CrawlingProps = {
className?: string
crawledNum: number
totalNum: number
}
type BlockProps = {
className?: string
}
type ItemProps = {
firstLineWidth: string
secondLineWidth: string
}
const Block = React.memo(({
className,
}: BlockProps) => {
return <div className={cn('bg-text-quaternary opacity-20', className)} />
})
const Item = React.memo(({
firstLineWidth,
secondLineWidth,
}: ItemProps) => {
return (
<div className='flex gap-x-2 px-2 py-[5px]'>
<div className='py-0.5'>
<Block className='size-4 rounded-[4px]' />
</div>
<div className='flex grow flex-col'>
<div className='flex h-5 w-full items-center'>
<Block className={cn('h-2.5 rounded-sm', firstLineWidth)} />
</div>
<div className='flex h-[18px] w-full items-center'>
<Block className={cn('h-1.5 rounded-sm', secondLineWidth)} />
</div>
</div>
</div>
)
})
const Crawling = ({
className = '',
crawledNum,
totalNum,
}: CrawlingProps) => {
const { t } = useTranslation()
const itemsConfig = [{
firstLineWidth: 'w-[35%]',
secondLineWidth: 'w-[50%]',
}, {
firstLineWidth: 'w-[40%]',
secondLineWidth: 'w-[45%]',
}, {
firstLineWidth: 'w-[30%]',
secondLineWidth: 'w-[36%]',
}]
return (
<div className={cn('mt-2 flex flex-col gap-y-2 pt-2', className)}>
<div className='system-sm-medium text-text-primary'>
{t('datasetCreation.stepOne.website.totalPageScraped')} {crawledNum}/{totalNum}
</div>
<div className='overflow-hidden rounded-xl border border-components-panel-border bg-components-panel-bg'>
<div className='flex items-center gap-x-2 px-4 py-2'>
<Block className='size-4 rounded-[4px]' />
<Block className='h-2.5 w-14 rounded-sm' />
</div>
<div className='flex flex-col gap-px border-t border-divider-subtle bg-background-default-subtle p-2'>
{itemsConfig.map((item, index) => (
<Item
key={index}
firstLineWidth={item.firstLineWidth}
secondLineWidth={item.secondLineWidth}
/>
))}
</div>
</div>
</div>
)
}
export default React.memo(Crawling)

View File

@@ -0,0 +1,34 @@
import React from 'react'
import cn from '@/utils/classnames'
import { RiErrorWarningFill } from '@remixicon/react'
type ErrorMessageProps = {
className?: string
title: string
errorMsg?: string
}
const ErrorMessage = ({
className,
title,
errorMsg,
}: ErrorMessageProps) => {
return (
// eslint-disable-next-line tailwindcss/migration-from-tailwind-2
<div className={cn(
'flex gap-x-0.5 rounded-xl border-[0.5px] border-components-panel-border bg-opacity-40 bg-toast-error-bg p-2 shadow-xs shadow-shadow-shadow-3',
className,
)}>
<div className='flex size-6 items-center justify-center'>
<RiErrorWarningFill className='h-4 w-4 text-text-destructive' />
</div>
<div className='flex flex-col gap-y-0.5 py-1'>
<div className='system-xs-medium text-text-primary'>{title}</div>
{errorMsg && (
<div className='system-xs-regular text-text-secondary'>{errorMsg}</div>
)}
</div>
</div>
)
}
export default React.memo(ErrorMessage)

View File

@@ -0,0 +1,89 @@
import type { BaseConfiguration } from '@/app/components/base/form/form-scenarios/base/types'
import { BaseFieldType } from '@/app/components/base/form/form-scenarios/base/types'
import { useTranslation } from 'react-i18next'
import type { FormData } from './options'
import { z } from 'zod'
const ERROR_I18N_PREFIX = 'common.errorMsg'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
export const useConfigurations = () => {
const { t } = useTranslation()
const configurations: BaseConfiguration<FormData>[] = [
{
type: BaseFieldType.textInput,
variable: 'url',
label: 'URL',
required: true,
showConditions: [],
placeholder: 'https://docs.dify.ai',
},
{
type: BaseFieldType.numberInput,
variable: 'limit',
label: t(`${I18N_PREFIX}.limit`),
required: false,
showConditions: [],
},
{
type: BaseFieldType.numberInput,
variable: 'max_depth',
label: t(`${I18N_PREFIX}.maxDepth`),
required: false,
showConditions: [],
},
{
type: BaseFieldType.textInput,
variable: 'excludes',
label: t(`${I18N_PREFIX}.excludePaths`),
required: false,
showConditions: [],
placeholder: 'blog/*, /about/*',
},
{
type: BaseFieldType.textInput,
variable: 'includes',
label: t(`${I18N_PREFIX}.includeOnlyPaths`),
required: false,
showConditions: [],
placeholder: 'articles/*',
},
{
type: BaseFieldType.checkbox,
variable: 'crawl_sub_pages',
label: t(`${I18N_PREFIX}.crawlSubPage`),
required: false,
showConditions: [],
},
{
type: BaseFieldType.checkbox,
variable: 'only_main_content',
label: t(`${I18N_PREFIX}.extractOnlyMainContent`),
required: false,
showConditions: [],
},
]
return configurations
}
export const useSchema = () => {
const { t } = useTranslation()
const Schema = z.object({
url: z.string().nonempty({
message: t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
field: 'url',
}),
}).regex(/^https?:\/\//, {
message: t(`${ERROR_I18N_PREFIX}.urlError`),
}),
limit: z.number().positive({
message: t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
field: t(`${I18N_PREFIX}.limit`),
}),
}).int(),
}).passthrough()
return Schema
}

View File

@@ -0,0 +1,201 @@
'use client'
import type { FC } from 'react'
import React, { useCallback, useEffect, useState } from 'react'
import { useTranslation } from 'react-i18next'
import { useModalContextSelector } from '@/context/modal-context'
import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
import { checkFirecrawlTaskStatus, createFirecrawlTask } from '@/service/datasets'
import { sleep } from '@/utils'
import Header from '@/app/components/datasets/create/website/base/header'
import type { FormData } from './options'
import Options from './options'
import { useConfigurations } from './hooks'
import Crawling from '../base/crawling'
import ErrorMessage from '../base/error-message'
import CrawledResult from '../base/crawled-result'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
checkedCrawlResult: CrawlResultItem[]
onCheckedCrawlResultChange: (payload: CrawlResultItem[]) => void
onJobIdChange: (jobId: string) => void
crawlOptions: CrawlOptions
onCrawlOptionsChange: (payload: CrawlOptions) => void
}
enum Step {
init = 'init',
running = 'running',
finished = 'finished',
}
const FireCrawl: FC<Props> = ({
checkedCrawlResult,
onCheckedCrawlResultChange,
onJobIdChange,
crawlOptions,
onCrawlOptionsChange,
}) => {
const { t } = useTranslation()
const [step, setStep] = useState<Step>(Step.init)
const [controlFoldOptions, setControlFoldOptions] = useState<number>(0)
const configurations = useConfigurations()
useEffect(() => {
if (step !== Step.init)
setControlFoldOptions(Date.now())
}, [step])
const setShowAccountSettingModal = useModalContextSelector(s => s.setShowAccountSettingModal)
const handleSetting = useCallback(() => {
setShowAccountSettingModal({
payload: 'data-source',
})
}, [setShowAccountSettingModal])
const isInit = step === Step.init
const isCrawlFinished = step === Step.finished
const isRunning = step === Step.running
const [crawlResult, setCrawlResult] = useState<{
current: number
total: number
data: CrawlResultItem[]
time_consuming: number | string
} | undefined>(undefined)
const [crawlErrorMessage, setCrawlErrorMessage] = useState('')
const showError = isCrawlFinished && crawlErrorMessage
const waitForCrawlFinished = useCallback(async (jobId: string) => {
try {
const res = await checkFirecrawlTaskStatus(jobId) as any
if (res.status === 'completed') {
return {
isError: false,
data: {
...res,
total: Math.min(res.total, Number.parseFloat(crawlOptions.limit as string)),
},
}
}
if (res.status === 'error' || !res.status) {
// can't get the error message from the firecrawl api
return {
isError: true,
errorMessage: res.message,
data: {
data: [],
},
}
}
// update the progress
setCrawlResult({
...res,
total: Math.min(res.total, Number.parseFloat(crawlOptions.limit as string)),
})
onCheckedCrawlResultChange(res.data || []) // default select the crawl result
await sleep(2500)
return await waitForCrawlFinished(jobId)
}
catch (e: any) {
const errorBody = await e.json()
return {
isError: true,
errorMessage: errorBody.message,
data: {
data: [],
},
}
}
}, [crawlOptions.limit, onCheckedCrawlResultChange])
const handleRun = useCallback(async (value: FormData) => {
const { url, ...crawlOptions } = value
onCrawlOptionsChange(crawlOptions)
setStep(Step.running)
try {
const passToServerCrawlOptions: any = {
...crawlOptions,
}
if (crawlOptions.max_depth === '')
delete passToServerCrawlOptions.max_depth
const res = await createFirecrawlTask({
url,
options: passToServerCrawlOptions,
}) as any
const jobId = res.job_id
onJobIdChange(jobId)
const { isError, data, errorMessage } = await waitForCrawlFinished(jobId)
if (isError) {
setCrawlErrorMessage(errorMessage || t(`${I18N_PREFIX}.unknownError`))
}
else {
setCrawlResult(data)
onCheckedCrawlResultChange(data.data || []) // default select the crawl result
setCrawlErrorMessage('')
}
}
catch (e) {
setCrawlErrorMessage(t(`${I18N_PREFIX}.unknownError`)!)
console.log(e)
}
finally {
setStep(Step.finished)
}
}, [onCrawlOptionsChange, onJobIdChange, t, waitForCrawlFinished, onCheckedCrawlResultChange])
return (
<div>
<Header
isInPipeline
onClickConfiguration={handleSetting}
title={t(`${I18N_PREFIX}.firecrawlTitle`)}
buttonText={t(`${I18N_PREFIX}.configureFirecrawl`)}
docTitle={t(`${I18N_PREFIX}.firecrawlDoc`)}
docLink={'https://docs.firecrawl.dev/introduction'}
/>
<div className='mt-2 rounded-xl border border-components-panel-border bg-background-default-subtle'>
<Options
initialData={{
...crawlOptions,
url: '',
}}
configurations={configurations}
isRunning={isRunning}
controlFoldOptions={controlFoldOptions}
onSubmit={(value) => {
handleRun(value)
console.log('submit')
}}
/>
</div>
{!isInit && (
<div className='relative'>
{isRunning && (
<Crawling
crawledNum={crawlResult?.current || 0}
totalNum={crawlResult?.total || Number.parseFloat(crawlOptions.limit as string) || 0}
/>
)}
{showError && (
<ErrorMessage
className='mt-2'
title={t(`${I18N_PREFIX}.exceptionErrorTitle`)}
errorMsg={crawlErrorMessage}
/>
)}
{isCrawlFinished && !showError && (
<CrawledResult
className='mt-2'
list={crawlResult?.data || []}
checkedList={checkedCrawlResult}
onSelectedChange={onCheckedCrawlResultChange}
usedTime={Number.parseFloat(crawlResult?.time_consuming as string) || 0}
/>
)}
</div>
)}
</div>
)
}
export default React.memo(FireCrawl)

View File

@@ -0,0 +1,118 @@
import Button from '@/app/components/base/button'
import { useAppForm } from '@/app/components/base/form'
import BaseField from '@/app/components/base/form/form-scenarios/base/field'
import type { BaseConfiguration } from '@/app/components/base/form/form-scenarios/base/types'
import { ArrowDownRoundFill } from '@/app/components/base/icons/src/vender/solid/general'
import type { CrawlOptions } from '@/models/datasets'
import cn from '@/utils/classnames'
import { RiPlayLargeLine } from '@remixicon/react'
import { useBoolean } from 'ahooks'
import { useEffect } from 'react'
import { useTranslation } from 'react-i18next'
import { useSchema } from './hooks'
import Toast from '@/app/components/base/toast'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
export type FormData = {
url: string
} & CrawlOptions
type OptionsProps = {
initialData: FormData
configurations: BaseConfiguration<FormData>[]
isRunning: boolean
controlFoldOptions?: number
onSubmit: (data: FormData) => void
}
const Options = ({
initialData,
configurations,
isRunning,
controlFoldOptions,
onSubmit,
}: OptionsProps) => {
const schema = useSchema()
const form = useAppForm({
defaultValues: initialData,
validators: {
onSubmit: ({ value }) => {
const result = schema.safeParse(value)
if (!result.success) {
const issues = result.error.issues
const firstIssue = issues[0]
const errorMessage = `"${firstIssue.path.join('.')}" ${firstIssue.message}`
Toast.notify({
type: 'error',
message: errorMessage,
})
return errorMessage
}
return undefined
},
},
onSubmit: ({ value }) => {
onSubmit(value)
},
})
const { t } = useTranslation()
const [fold, {
toggle: foldToggle,
setTrue: foldHide,
}] = useBoolean(false)
useEffect(() => {
if (controlFoldOptions !== 0)
foldHide()
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [controlFoldOptions])
return (
<form
className='w-full'
onSubmit={(e) => {
e.preventDefault()
e.stopPropagation()
form.handleSubmit()
}}
>
<div className='flex items-center gap-x-1 px-4 py-2'>
<div
className='flex grow cursor-pointer select-none items-center gap-x-0.5'
onClick={foldToggle}
>
<span className='system-sm-semibold-uppercase text-text-secondary'>
{t(`${I18N_PREFIX}.options`)}
</span>
<ArrowDownRoundFill className={cn('h-4 w-4 shrink-0 text-text-tertiary', fold && '-rotate-90')} />
</div>
<Button
variant='primary'
onClick={form.handleSubmit}
loading={isRunning}
className='shrink-0 gap-x-0.5'
spinnerClassName='!ml-0'
>
<RiPlayLargeLine className='size-4' />
<span className='px-0.5'>{!isRunning ? t(`${I18N_PREFIX}.run`) : t(`${I18N_PREFIX}.running`)}</span>
</Button>
</div>
{!fold && (
<div className='flex flex-col gap-3 border-t border-divider-subtle px-4 py-3'>
{configurations.map((config, index) => {
const FieldComponent = BaseField<FormData>({
initialData,
config,
})
return <FieldComponent key={index} form={form} />
})}
</div>
)}
</form>
)
}
export default Options

View File

@@ -4,30 +4,35 @@ import { useCallback, useMemo, useState } from 'react'
import StepIndicator from './step-indicator'
import { useTestRunSteps } from './hooks'
import DataSourceOptions from './data-source-options'
import type { FileItem } from '@/models/datasets'
import type { CrawlOptions, CrawlResultItem, FileItem } from '@/models/datasets'
import { DataSourceType } from '@/models/datasets'
import LocalFile from './data-source/local-file'
import produce from 'immer'
import Button from '@/app/components/base/button'
import { useTranslation } from 'react-i18next'
import { useProviderContextSelector } from '@/context/provider-context'
import type { NotionPage } from '@/models/common'
import { DataSourceProvider, type NotionPage } from '@/models/common'
import Notion from './data-source/notion'
import VectorSpaceFull from '@/app/components/billing/vector-space-full'
import { DEFAULT_CRAWL_OPTIONS } from './consts'
import Firecrawl from './data-source/website/firecrawl'
const TestRunPanel = () => {
const { t } = useTranslation()
const [currentStep, setCurrentStep] = useState(1)
const [dataSourceType, setDataSourceType] = useState<string>(DataSourceType.FILE)
const [dataSource, setDataSource] = useState<string>(DataSourceProvider.fireCrawl)
const [fileList, setFiles] = useState<FileItem[]>([])
const [notionPages, setNotionPages] = useState<NotionPage[]>([])
const [websitePages, setWebsitePages] = useState<CrawlResultItem[]>([])
const [websiteCrawlJobId, setWebsiteCrawlJobId] = useState('')
const [crawlOptions, setCrawlOptions] = useState<CrawlOptions>(DEFAULT_CRAWL_OPTIONS)
const setShowTestRunPanel = useStore(s => s.setShowTestRunPanel)
const plan = useProviderContextSelector(state => state.plan)
const enableBilling = useProviderContextSelector(state => state.enableBilling)
const steps = useTestRunSteps()
const dataSources = ['upload_file', 'notion_import', 'firecrawl', 'jinareader', 'watercrawl'] // TODO: replace with real data sources
const dataSources = ['upload_file', 'notion_import', 'firecrawl', 'jinareader', 'watercrawl'] // TODO: replace with real data sources from API
const allFileLoaded = (fileList.length > 0 && fileList.every(file => file.file.id))
const isVectorSpaceFull = plan.usage.vectorSpace >= plan.total.vectorSpace
@@ -42,19 +47,19 @@ const TestRunPanel = () => {
}, [fileList, isShowVectorSpaceFull])
const nextBtnDisabled = useMemo(() => {
if (dataSourceType === DataSourceType.FILE)
if (dataSource === DataSourceType.FILE)
return nextDisabled
if (dataSourceType === DataSourceType.NOTION)
if (dataSource === DataSourceType.NOTION)
return isShowVectorSpaceFull || !notionPages.length
return false
}, [dataSourceType, nextDisabled, isShowVectorSpaceFull, notionPages.length])
}, [dataSource, nextDisabled, isShowVectorSpaceFull, notionPages.length])
const handleClose = () => {
setShowTestRunPanel?.(false)
}
const handleDataSourceSelect = useCallback((option: string) => {
setDataSourceType(option)
setDataSource(option)
}, [])
const updateFile = (fileItem: FileItem, progress: number, list: FileItem[]) => {
@@ -81,7 +86,9 @@ const TestRunPanel = () => {
}, [])
return (
<div className='relative flex h-full w-[480px] flex-col rounded-l-2xl border-y-[0.5px] border-l-[0.5px] border-components-panel-border bg-components-panel-bg shadow-xl shadow-shadow-shadow-1'>
<div
className='relative flex h-full w-[480px] flex-col rounded-l-2xl border-y-[0.5px] border-l-[0.5px] border-components-panel-border bg-components-panel-bg shadow-xl shadow-shadow-shadow-1'
>
<button
type='button'
className='absolute right-2.5 top-2.5 flex size-8 items-center justify-center p-1.5'
@@ -95,41 +102,52 @@ const TestRunPanel = () => {
</div>
<StepIndicator steps={steps} currentStep={currentStep} />
</div>
{
currentStep === 1 && (
<>
<div className='flex flex-col gap-y-4 px-4 py-2'>
<DataSourceOptions
dataSources={dataSources}
dataSourceType={dataSourceType}
onSelect={handleDataSourceSelect}
/>
{dataSourceType === DataSourceType.FILE && (
<LocalFile
files={fileList}
updateFile={updateFile}
updateFileList={updateFileList}
notSupportBatchUpload={notSupportBatchUpload}
<div className='grow overflow-y-auto'>
{
currentStep === 1 && (
<>
<div className='flex flex-col gap-y-4 px-4 py-2'>
<DataSourceOptions
dataSources={dataSources}
dataSourceType={dataSource}
onSelect={handleDataSourceSelect}
/>
)}
{dataSourceType === DataSourceType.NOTION && (
<Notion
notionPages={notionPages}
updateNotionPages={updateNotionPages}
/>
)}
{isShowVectorSpaceFull && (
<VectorSpaceFull />
)}
</div>
<div className='flex justify-end p-4 pt-2'>
<Button disabled={nextBtnDisabled} variant='primary' onClick={handleNextStep}>
<span className='px-0.5'>{t('datasetCreation.stepOne.button')}</span>
</Button>
</div>
</>
)
}
{dataSource === DataSourceType.FILE && (
<LocalFile
files={fileList}
updateFile={updateFile}
updateFileList={updateFileList}
notSupportBatchUpload={notSupportBatchUpload}
/>
)}
{dataSource === DataSourceType.NOTION && (
<Notion
notionPages={notionPages}
updateNotionPages={updateNotionPages}
/>
)}
{dataSource === DataSourceProvider.fireCrawl && (
<Firecrawl
checkedCrawlResult={websitePages}
onCheckedCrawlResultChange={setWebsitePages}
onJobIdChange={setWebsiteCrawlJobId}
crawlOptions={crawlOptions}
onCrawlOptionsChange={setCrawlOptions}
/>
)}
{isShowVectorSpaceFull && (
<VectorSpaceFull />
)}
</div>
<div className='flex justify-end p-4 pt-2'>
<Button disabled={nextBtnDisabled} variant='primary' onClick={handleNextStep}>
<span className='px-0.5'>{t('datasetCreation.stepOne.button')}</span>
</Button>
</div>
</>
)
}
</div>
</div>
)
}

View File

@@ -78,6 +78,7 @@ const translation = {
configureWatercrawl: 'Configure Watercrawl',
configureJinaReader: 'Configure Jina Reader',
run: 'Run',
running: 'Running',
firecrawlTitle: 'Extract web content with 🔥Firecrawl',
firecrawlDoc: 'Firecrawl docs',
firecrawlDocLink: 'https://docs.dify.ai/guides/knowledge-base/sync-from-website',

View File

@@ -77,6 +77,7 @@ const translation = {
configureFirecrawl: '配置 Firecrawl',
configureJinaReader: '配置 Jina Reader',
run: '运行',
running: '运行中',
firecrawlTitle: '使用 🔥Firecrawl 提取网页内容',
firecrawlDoc: 'Firecrawl 文档',
firecrawlDocLink: 'https://docs.dify.ai/v/zh-hans/guides/knowledge-base/sync-from-website',