feat: Integrate WaterCrawl.dev as a new knowledge base provider (#16396)

Co-authored-by: crazywoola <427733928@qq.com>
This commit is contained in:
Amir Mohsen Asaran
2025-04-07 06:43:23 +02:00
committed by GitHub
parent 0afad94378
commit f54905e685
24 changed files with 1102 additions and 55 deletions

View File

@@ -0,0 +1,20 @@
<?xml version="1.0" encoding="utf-8"?>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 500 500">
<path style="fill: rgb(0, 23, 87); stroke: rgb(13, 14, 52);" d="M 247.794 213.903 L 246.81 76.976 L 254.345 76.963 L 254.592 213.989 L 247.794 213.903 Z"/>
<ellipse style="fill: rgb(0, 23, 87); stroke: rgb(0, 23, 87);" cx="250.025" cy="43.859" rx="33.966" ry="33.906"/>
<path style="fill: rgb(30, 141, 166); stroke: rgb(30, 141, 166);" d="M 282.472 260.389 L 414.181 330.253 L 410.563 336.234 L 279.38 265.739 L 282.472 260.389 Z"/>
<path style="fill: rgb(15, 17, 57); stroke: rgb(13, 14, 52);" d="M 255.105 281.394 L 254.485 417.656 L 246.156 417.691 L 246.688 280.51 L 255.105 281.394 Z"/>
<path style="paint-order: fill; fill: rgb(30, 141, 166); stroke: rgb(30, 141, 166);" d="M 279.486 229.517 L 410.351 160.07 L 413.923 167.04 L 283.727 235.998 L 279.486 229.517 Z"/>
<path style="fill: rgb(15, 164, 161); stroke: rgb(15, 164, 161);" d="M 88.545 164.884 L 219.797 236.07 L 222.867 229.568 L 90.887 159.47 L 88.545 164.884 Z"/>
<path style="fill: rgb(15, 164, 161); stroke: rgb(15, 164, 161);" d="M 224.76 266.9 L 95.55 334.829 L 92.878 328.37 L 219.955 261.275 L 224.76 266.9 Z"/>
<ellipse style="paint-order: fill; fill: rgb(2, 181, 225); stroke: rgb(2, 181, 225);" cx="251.242" cy="247.466" rx="33.966" ry="33.906"/>
<path style="fill: rgb(13, 14, 52); stroke: rgb(13, 14, 52);" d="M 279.502 433.617 L 408.666 359.443 C 408.666 359.443 412.398 366.965 412.398 366.916 C 412.398 366.867 281.544 440.217 281.544 440.217 L 279.502 433.617 Z"/>
<path style="fill: rgb(13, 14, 52); stroke: rgb(13, 14, 52);" d="M 223.119 431.408 L 96.643 361.068 L 93.265 368.047 L 218.895 438.099 L 223.119 431.408 Z"/>
<ellipse style="fill: rgb(0, 23, 87); stroke: rgb(0, 23, 87);" cx="250.504" cy="451.168" rx="33.966" ry="33.906"/>
<path style="fill: rgb(90, 191, 187); stroke: rgb(90, 191, 187);" d="M 435.665 180.895 L 435.859 316.869 L 443.103 315.579 L 442.56 180.697 L 435.665 180.895 Z"/>
<ellipse style="fill: rgb(0, 23, 87); stroke: rgb(0, 23, 87);" cx="441.06" cy="349.665" rx="33.966" ry="33.906"/>
<ellipse style="fill: rgb(2, 181, 225); stroke: rgb(2, 181, 225);" cx="441.512" cy="147.767" rx="33.966" ry="33.906"/>
<path style="fill: rgb(84, 187, 181); stroke: rgb(84, 187, 181);" d="M 64.755 314.523 L 57.928 315.006 L 58.307 182.961 L 65.169 182.865 L 64.755 314.523 Z"/>
<ellipse style="fill: rgb(0, 23, 87); stroke: rgb(0, 23, 87);" cx="58.177" cy="149.757" rx="33.966" ry="33.906"/>
<ellipse style="fill: rgb(61, 224, 203); stroke: rgb(61, 224, 203);" cx="65.909" cy="348.17" rx="33.966" ry="33.906"/>
</svg>

After

Width:  |  Height:  |  Size: 2.6 KiB

View File

@@ -4,3 +4,10 @@
background-image: url(../assets/jina.png);
background-size: 16px;
}
.watercrawlLogo {
@apply w-5 h-5 bg-center bg-no-repeat inline-block;
/*background-color: #F5FAFF;*/
background-image: url(../assets/watercrawl.svg);
background-size: 16px;
}

View File

@@ -5,6 +5,7 @@ import { useTranslation } from 'react-i18next'
import s from './index.module.css'
import NoData from './no-data'
import Firecrawl from './firecrawl'
import Watercrawl from './watercrawl'
import JinaReader from './jina-reader'
import cn from '@/utils/classnames'
import { useModalContext } from '@/context/modal-context'
@@ -47,7 +48,11 @@ const Website: FC<Props> = ({
// If users have configured one of the providers, select it.
const availableProviders = res.sources.filter((item: DataSourceItem) =>
[DataSourceProvider.jinaReader, DataSourceProvider.fireCrawl].includes(item.provider),
[
DataSourceProvider.jinaReader,
DataSourceProvider.fireCrawl,
DataSourceProvider.waterCrawl,
].includes(item.provider),
)
if (availableProviders.length > 0)
@@ -70,6 +75,8 @@ const Website: FC<Props> = ({
if (!isLoaded)
return null
const source = sources.find(source => source.provider === selectedProvider)
return (
<div>
<div className="mb-4">
@@ -86,7 +93,7 @@ const Website: FC<Props> = ({
)}
onClick={() => setSelectedProvider(DataSourceProvider.jinaReader)}
>
<span className={cn(s.jinaLogo, 'mr-2')} />
<span className={cn(s.jinaLogo, 'mr-2')}/>
<span>Jina Reader</span>
</button>
<button
@@ -100,40 +107,53 @@ const Website: FC<Props> = ({
>
🔥 Firecrawl
</button>
<button
className={cn('flex items-center justify-center rounded-lg px-4 py-2',
selectedProvider === DataSourceProvider.waterCrawl
? 'system-sm-medium border-[1.5px] border-components-option-card-option-selected-border bg-components-option-card-option-selected-bg text-text-primary'
: `system-sm-regular border border-components-option-card-option-border bg-components-option-card-option-bg text-text-secondary
hover:border-components-option-card-option-border-hover hover:bg-components-option-card-option-bg-hover hover:shadow-xs hover:shadow-shadow-shadow-3`,
)}
onClick={() => setSelectedProvider(DataSourceProvider.waterCrawl)}
>
<span className={cn(s.watercrawlLogo, 'mr-2')}/>
<span>WaterCrawl</span>
</button>
</div>
</div>
{
selectedProvider === DataSourceProvider.fireCrawl
? sources.find(source => source.provider === DataSourceProvider.fireCrawl)
? (
<Firecrawl
onPreview={onPreview}
checkedCrawlResult={checkedCrawlResult}
onCheckedCrawlResultChange={onCheckedCrawlResultChange}
onJobIdChange={onJobIdChange}
crawlOptions={crawlOptions}
onCrawlOptionsChange={onCrawlOptionsChange}
/>
)
: (
<NoData onConfig={handleOnConfig} provider={selectedProvider} />
)
: sources.find(source => source.provider === DataSourceProvider.jinaReader)
? (
<JinaReader
onPreview={onPreview}
checkedCrawlResult={checkedCrawlResult}
onCheckedCrawlResultChange={onCheckedCrawlResultChange}
onJobIdChange={onJobIdChange}
crawlOptions={crawlOptions}
onCrawlOptionsChange={onCrawlOptionsChange}
/>
)
: (
<NoData onConfig={handleOnConfig} provider={selectedProvider} />
)
}
{source && selectedProvider === DataSourceProvider.fireCrawl && (
<Firecrawl
onPreview={onPreview}
checkedCrawlResult={checkedCrawlResult}
onCheckedCrawlResultChange={onCheckedCrawlResultChange}
onJobIdChange={onJobIdChange}
crawlOptions={crawlOptions}
onCrawlOptionsChange={onCrawlOptionsChange}
/>
)}
{source && selectedProvider === DataSourceProvider.waterCrawl && (
<Watercrawl
onPreview={onPreview}
checkedCrawlResult={checkedCrawlResult}
onCheckedCrawlResultChange={onCheckedCrawlResultChange}
onJobIdChange={onJobIdChange}
crawlOptions={crawlOptions}
onCrawlOptionsChange={onCrawlOptionsChange}
/>
)}
{source && selectedProvider === DataSourceProvider.jinaReader && (
<JinaReader
onPreview={onPreview}
checkedCrawlResult={checkedCrawlResult}
onCheckedCrawlResultChange={onCheckedCrawlResultChange}
onJobIdChange={onJobIdChange}
crawlOptions={crawlOptions}
onCrawlOptionsChange={onCrawlOptionsChange}
/>
)}
{!source && (
<NoData onConfig={handleOnConfig} provider={selectedProvider}/>
)}
</div>
)
}

View File

@@ -31,6 +31,11 @@ const NoData: FC<Props> = ({
title: t(`${I18N_PREFIX}.fireCrawlNotConfigured`),
description: t(`${I18N_PREFIX}.fireCrawlNotConfiguredDescription`),
},
[DataSourceProvider.waterCrawl]: {
emoji: <span className={s.watercrawlLogo} />,
title: t(`${I18N_PREFIX}.waterCrawlNotConfigured`),
description: t(`${I18N_PREFIX}.waterCrawlNotConfiguredDescription`),
},
}
const currentProvider = providerConfig[provider]

View File

@@ -0,0 +1,43 @@
'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(`${I18N_PREFIX}.watercrawlTitle`)}</div>
<div className='ml-2 mr-2 w-px h-3.5 bg-divider-regular' />
<Button className='flex items-center gap-x-[1px] h-6 px-1.5' onClick={onSetting}>
<RiEqualizer2Line className='w-3.5 h-3.5 text-components-button-secondary-text' />
<span className='text-components-button-secondary-text text-xs font-medium px-[3px]'>
{t(`${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='w-3.5 h-3.5 text-text-accent' />
<span>{t(`${I18N_PREFIX}.watercrawlDoc`)}</span>
</a>
</div>
)
}
export default React.memo(Header)

View File

@@ -0,0 +1,217 @@
'use client'
import type { FC } from 'react'
import React, { useCallback, useEffect, useState } from 'react'
import { useTranslation } from 'react-i18next'
import UrlInput from '../base/url-input'
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'
const ERROR_I18N_PREFIX = 'common.errorMsg'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
onPreview: (payload: CrawlResultItem) => void
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 WaterCrawl: FC<Props> = ({
onPreview,
checkedCrawlResult,
onCheckedCrawlResultChange,
onJobIdChange,
crawlOptions,
onCrawlOptionsChange,
}) => {
const { t } = useTranslation()
const [step, setStep] = useState<Step>(Step.init)
const [controlFoldOptions, setControlFoldOptions] = useState<number>(0)
useEffect(() => {
if (step !== Step.init)
setControlFoldOptions(Date.now())
}, [step])
const { setShowAccountSettingModal } = useModalContext()
const handleSetting = useCallback(() => {
setShowAccountSettingModal({
payload: 'data-source',
})
}, [setShowAccountSettingModal])
const checkValid = useCallback((url: string) => {
let errorMsg = ''
if (!url) {
errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
field: 'url',
})
}
if (!errorMsg && !((url.startsWith('http://') || url.startsWith('https://'))))
errorMsg = t(`${ERROR_I18N_PREFIX}.urlError`)
if (!errorMsg && (crawlOptions.limit === null || crawlOptions.limit === undefined || crawlOptions.limit === '')) {
errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
field: t(`${I18N_PREFIX}.limit`),
})
}
return {
isValid: !errorMsg,
errorMsg,
}
}, [crawlOptions, t])
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): Promise<any> => {
try {
const res = await checkWatercrawlTaskStatus(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 watercrawl 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])
const handleRun = useCallback(async (url: string) => {
const { isValid, errorMsg } = checkValid(url)
if (!isValid) {
Toast.notify({
message: errorMsg!,
type: 'error',
})
return
}
setStep(Step.running)
try {
const passToServerCrawlOptions: any = {
...crawlOptions,
}
if (crawlOptions.max_depth === '')
delete passToServerCrawlOptions.max_depth
const res = await createWatercrawlTask({
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)
}
}, [checkValid, crawlOptions, onJobIdChange, t, waitForCrawlFinished])
return (
<div>
<Header onSetting={handleSetting} />
<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
className='mt-4'
controlFoldOptions={controlFoldOptions}
>
<Options className='mt-2' payload={crawlOptions} onChange={onCrawlOptionsChange} />
</OptionsWrap>
{!isInit && (
<div className='relative left-[-16px] mt-3 w-[calc(100%_+_32px)] rounded-b-xl'>
{isRunning
&& <Crawling
className='mt-2'
crawledNum={crawlResult?.current || 0}
totalNum={crawlResult?.total || Number.parseFloat(crawlOptions.limit as string) || 0}
/>}
{showError && (
<ErrorMessage className='rounded-b-xl' title={t(`${I18N_PREFIX}.exceptionErrorTitle`)} errorMsg={crawlErrorMessage} />
)}
{isCrawlFinished && !showError
&& <CrawledResult
className='mb-2'
list={crawlResult?.data || []}
checkedList={checkedCrawlResult}
onSelectedChange={onCheckedCrawlResultChange}
onPreview={onPreview}
usedTime={Number.parseFloat(crawlResult?.time_consuming as string) || 0}
/>
}
</div>
)}
</div>
</div>
)
}
export default React.memo(WaterCrawl)

View File

@@ -0,0 +1,85 @@
'use client'
import type { FC } from 'react'
import React, { useCallback } from 'react'
import { useTranslation } from 'react-i18next'
import CheckboxWithLabel from '../base/checkbox-with-label'
import Field from '../base/field'
import cn from '@/utils/classnames'
import type { CrawlOptions } from '@/models/datasets'
const I18N_PREFIX = 'datasetCreation.stepOne.website'
type Props = {
className?: string
payload: CrawlOptions
onChange: (payload: CrawlOptions) => void
}
const Options: FC<Props> = ({
className = '',
payload,
onChange,
}) => {
const { t } = useTranslation()
const handleChange = useCallback((key: keyof CrawlOptions) => {
return (value: any) => {
onChange({
...payload,
[key]: value,
})
}
}, [payload, onChange])
return (
<div className={cn(className, ' space-y-2')}>
<CheckboxWithLabel
label={t(`${I18N_PREFIX}.crawlSubPage`)}
isChecked={payload.crawl_sub_pages}
onChange={handleChange('crawl_sub_pages')}
labelClassName='text-[13px] leading-[16px] font-medium text-text-secondary'
/>
<div className='flex justify-between space-x-4'>
<Field
className='shrink-0 grow'
label={t(`${I18N_PREFIX}.limit`)}
value={payload.limit}
onChange={handleChange('limit')}
isNumber
isRequired
/>
<Field
className='shrink-0 grow'
label={t(`${I18N_PREFIX}.maxDepth`)}
value={payload.max_depth}
onChange={handleChange('max_depth')}
isNumber
tooltip={t(`${I18N_PREFIX}.maxDepthTooltip`)!}
/>
</div>
<div className='flex justify-between space-x-4'>
<Field
className='shrink-0 grow'
label={t(`${I18N_PREFIX}.excludePaths`)}
value={payload.excludes}
onChange={handleChange('excludes')}
placeholder='blog/*, /about/*'
/>
<Field
className='shrink-0 grow'
label={t(`${I18N_PREFIX}.includeOnlyPaths`)}
value={payload.includes}
onChange={handleChange('includes')}
placeholder='articles/*'
/>
</div>
<CheckboxWithLabel
label={t(`${I18N_PREFIX}.extractOnlyMainContent`)}
isChecked={payload.only_main_content}
onChange={handleChange('only_main_content')}
labelClassName='text-[13px] leading-[16px] font-medium text-text-secondary'
/>
</div>
)
}
export default React.memo(Options)

View File

@@ -0,0 +1,161 @@
'use client'
import type { FC } from 'react'
import React, { useCallback, useState } from 'react'
import { useTranslation } from 'react-i18next'
import {
PortalToFollowElem,
PortalToFollowElemContent,
} from '@/app/components/base/portal-to-follow-elem'
import { Lock01 } from '@/app/components/base/icons/src/vender/solid/security'
import Button from '@/app/components/base/button'
import type { WatercrawlConfig } from '@/models/common'
import Field from '@/app/components/datasets/create/website/base/field'
import Toast from '@/app/components/base/toast'
import { createDataSourceApiKeyBinding } from '@/service/datasets'
import { LinkExternal02 } from '@/app/components/base/icons/src/vender/line/general'
type Props = {
onCancel: () => void
onSaved: () => void
}
const I18N_PREFIX = 'datasetCreation.watercrawl'
const DEFAULT_BASE_URL = 'https://app.watercrawl.dev'
const ConfigWatercrawlModal: FC<Props> = ({
onCancel,
onSaved,
}) => {
const { t } = useTranslation()
const [isSaving, setIsSaving] = useState(false)
const [config, setConfig] = useState<WatercrawlConfig>({
api_key: '',
base_url: '',
})
const handleConfigChange = useCallback((key: string) => {
return (value: string | number) => {
setConfig(prev => ({ ...prev, [key]: value as string }))
}
}, [])
const handleSave = useCallback(async () => {
if (isSaving)
return
let errorMsg = ''
if (config.base_url && !((config.base_url.startsWith('http://') || config.base_url.startsWith('https://'))))
errorMsg = t('common.errorMsg.urlError')
if (!errorMsg) {
if (!config.api_key) {
errorMsg = t('common.errorMsg.fieldRequired', {
field: 'API Key',
})
}
}
if (errorMsg) {
Toast.notify({
type: 'error',
message: errorMsg,
})
return
}
const postData = {
category: 'website',
provider: 'watercrawl',
credentials: {
auth_type: 'x-api-key',
config: {
api_key: config.api_key,
base_url: config.base_url || DEFAULT_BASE_URL,
},
},
}
try {
setIsSaving(true)
await createDataSourceApiKeyBinding(postData)
Toast.notify({
type: 'success',
message: t('common.api.success'),
})
}
finally {
setIsSaving(false)
}
onSaved()
}, [config.api_key, config.base_url, onSaved, t, isSaving])
return (
<PortalToFollowElem open>
<PortalToFollowElemContent className='w-full h-full z-[60]'>
<div className='fixed inset-0 flex items-center justify-center bg-background-overlay'>
<div className='mx-2 w-[640px] max-h-[calc(100vh-120px)] bg-components-panel-bg shadow-xl rounded-2xl overflow-y-auto'>
<div className='px-8 pt-8'>
<div className='flex justify-between items-center mb-4'>
<div className='system-xl-semibold text-text-primary'>{t(`${I18N_PREFIX}.configWatercrawl`)}</div>
</div>
<div className='space-y-4'>
<Field
label='API Key'
labelClassName='!text-sm'
isRequired
value={config.api_key}
onChange={handleConfigChange('api_key')}
placeholder={t(`${I18N_PREFIX}.apiKeyPlaceholder`)!}
/>
<Field
label='Base URL'
labelClassName='!text-sm'
value={config.base_url}
onChange={handleConfigChange('base_url')}
placeholder={DEFAULT_BASE_URL}
/>
</div>
<div className='my-8 flex justify-between items-center h-8'>
<a className='flex items-center space-x-1 leading-[18px] text-xs font-normal text-text-accent' target='_blank' href='https://app.watercrawl.dev/'>
<span>{t(`${I18N_PREFIX}.getApiKeyLinkText`)}</span>
<LinkExternal02 className='w-3 h-3' />
</a>
<div className='flex'>
<Button
size='large'
className='mr-2'
onClick={onCancel}
>
{t('common.operation.cancel')}
</Button>
<Button
variant='primary'
size='large'
onClick={handleSave}
loading={isSaving}
>
{t('common.operation.save')}
</Button>
</div>
</div>
</div>
<div className='border-t-[0.5px] border-t-divider-regular'>
<div className='flex justify-center items-center py-3 bg-background-section-burn text-xs text-text-tertiary'>
<Lock01 className='mr-1 w-3 h-3 text-text-tertiary' />
{t('common.modelProvider.encrypted.front')}
<a
className='text-text-accent mx-1'
target='_blank' rel='noopener noreferrer'
href='https://pycryptodome.readthedocs.io/en/latest/src/cipher/oaep.html'
>
PKCS1_OAEP
</a>
{t('common.modelProvider.encrypted.back')}
</div>
</div>
</div>
</div>
</PortalToFollowElemContent>
</PortalToFollowElem>
)
}
export default React.memo(ConfigWatercrawlModal)

View File

@@ -5,19 +5,15 @@ import { useTranslation } from 'react-i18next'
import Panel from '../panel'
import { DataSourceType } from '../panel/types'
import ConfigFirecrawlModal from './config-firecrawl-modal'
import ConfigWatercrawlModal from './config-watercrawl-modal'
import ConfigJinaReaderModal from './config-jina-reader-modal'
import cn from '@/utils/classnames'
import s from '@/app/components/datasets/create/website/index.module.css'
import { fetchDataSources, removeDataSourceApiKeyBinding } from '@/service/datasets'
import type {
DataSourceItem,
} from '@/models/common'
import type { DataSourceItem } from '@/models/common'
import { DataSourceProvider } from '@/models/common'
import { useAppContext } from '@/context/app-context'
import {
DataSourceProvider,
} from '@/models/common'
import Toast from '@/app/components/base/toast'
type Props = {
@@ -58,6 +54,16 @@ const DataSourceWebsite: FC<Props> = ({ provider }) => {
return source?.id
}
const getProviderName = (provider: DataSourceProvider): string => {
if (provider === DataSourceProvider.fireCrawl)
return 'Firecrawl'
if (provider === DataSourceProvider.waterCrawl)
return 'WaterCrawl'
return 'Jina Reader'
}
const handleRemove = useCallback((provider: DataSourceProvider) => {
return async () => {
const dataSourceId = getIdByProvider(provider)
@@ -82,27 +88,42 @@ const DataSourceWebsite: FC<Props> = ({ provider }) => {
readOnly={!isCurrentWorkspaceManager}
configuredList={sources.filter(item => item.provider === provider).map(item => ({
id: item.id,
logo: ({ className }: { className: string }) => (
item.provider === DataSourceProvider.fireCrawl
? (
<div className={cn(className, 'ml-3 flex h-5 w-5 items-center justify-center rounded border border-divider-subtle !bg-background-default text-xs font-medium text-text-tertiary')}>🔥</div>
logo: ({ className }: { className: string }) => {
if (item.provider === DataSourceProvider.fireCrawl) {
return (
<div
className={cn(className, 'ml-3 flex h-5 w-5 items-center justify-center rounded border border-divider-subtle !bg-background-default text-xs font-medium text-text-tertiary')}>🔥</div>
)
: (
<div className={cn(className, 'ml-3 flex h-5 w-5 items-center justify-center rounded border border-divider-subtle !bg-background-default text-xs font-medium text-text-tertiary')}>
<span className={s.jinaLogo} />
}
if (item.provider === DataSourceProvider.waterCrawl) {
return (
<div
className={cn(className, 'ml-3 flex h-5 w-5 items-center justify-center rounded border border-divider-subtle !bg-background-default text-xs font-medium text-text-tertiary')}>
<span className={s.watercrawlLogo}/>
</div>
)
),
name: item.provider === DataSourceProvider.fireCrawl ? 'Firecrawl' : 'Jina Reader',
}
return (
<div
className={cn(className, 'ml-3 flex h-5 w-5 items-center justify-center rounded border border-divider-subtle !bg-background-default text-xs font-medium text-text-tertiary')}>
<span className={s.jinaLogo}/>
</div>
)
},
name: getProviderName(item.provider),
isActive: true,
}))}
onRemove={handleRemove(provider)}
/>
{configTarget === DataSourceProvider.fireCrawl && (
<ConfigFirecrawlModal onSaved={handleAdded} onCancel={hideConfig} />
<ConfigFirecrawlModal onSaved={handleAdded} onCancel={hideConfig}/>
)}
{configTarget === DataSourceProvider.waterCrawl && (
<ConfigWatercrawlModal onSaved={handleAdded} onCancel={hideConfig}/>
)}
{configTarget === DataSourceProvider.jinaReader && (
<ConfigJinaReaderModal onSaved={handleAdded} onCancel={hideConfig} />
<ConfigJinaReaderModal onSaved={handleAdded} onCancel={hideConfig}/>
)}
</>

View File

@@ -15,6 +15,7 @@ export default function DataSourcePage() {
<DataSourceNotion workspaces={notionWorkspaces} />
<DataSourceWebsite provider={DataSourceProvider.jinaReader} />
<DataSourceWebsite provider={DataSourceProvider.fireCrawl} />
<DataSourceWebsite provider={DataSourceProvider.waterCrawl} />
</div>
)
}

View File

@@ -41,6 +41,12 @@ const Panel: FC<Props> = ({
const isNotion = type === DataSourceType.notion
const isWebsite = type === DataSourceType.website
const getProviderName = (): string => {
if (provider === DataSourceProvider.fireCrawl) return '🔥 Firecrawl'
if (provider === DataSourceProvider.waterCrawl) return 'WaterCrawl'
return 'Jina Reader'
}
return (
<div className='mb-2 rounded-xl bg-background-section-burn'>
<div className='flex items-center px-3 py-[9px]'>
@@ -50,7 +56,7 @@ const Panel: FC<Props> = ({
<div className='text-sm font-medium text-text-primary'>{t(`common.dataSource.${type}.title`)}</div>
{isWebsite && (
<div className='ml-1 rounded-md bg-components-badge-white-to-dark px-1.5 text-xs font-medium leading-[18px] text-text-secondary'>
<span className='text-text-tertiary'>{t('common.dataSource.website.with')}</span> { provider === DataSourceProvider.fireCrawl ? '🔥 Firecrawl' : 'Jina Reader'}
<span className='text-text-tertiary'>{t('common.dataSource.website.with')}</span> {getProviderName()}
</div>
)}
</div>