feat: [frontend] support vision (#1518)

Co-authored-by: Joel <iamjoel007@gmail.com>
This commit is contained in:
zxhlyh
2023-11-13 22:32:39 +08:00
committed by GitHub
parent 41d0a8b295
commit 6b15827246
74 changed files with 3159 additions and 339 deletions

View File

@@ -1,6 +1,7 @@
'use client'
import type { FC, ReactNode } from 'react'
import React, { useEffect, useLayoutEffect, useRef, useState } from 'react'
import Textarea from 'rc-textarea'
import { useContext } from 'use-context-selector'
import cn from 'classnames'
import Recorder from 'js-audio-recorder'
@@ -10,9 +11,8 @@ import type { DisplayScene, FeedbackFunc, IChatItem, SubmitAnnotationFunc } from
import { TryToAskIcon, stopIcon } from './icon-component'
import Answer from './answer'
import Question from './question'
import Tooltip from '@/app/components/base/tooltip'
import TooltipPlus from '@/app/components/base/tooltip-plus'
import { ToastContext } from '@/app/components/base/toast'
import AutoHeightTextarea from '@/app/components/base/auto-height-textarea'
import Button from '@/app/components/base/button'
import useBreakpoints, { MediaType } from '@/hooks/use-breakpoints'
import VoiceInput from '@/app/components/base/voice-input'
@@ -20,6 +20,10 @@ import { Microphone01 } from '@/app/components/base/icons/src/vender/line/mediaA
import { Microphone01 as Microphone01Solid } from '@/app/components/base/icons/src/vender/solid/mediaAndDevices'
import { XCircle } from '@/app/components/base/icons/src/vender/solid/general'
import type { DataSet } from '@/models/datasets'
import ChatImageUploader from '@/app/components/base/image-uploader/chat-image-uploader'
import ImageList from '@/app/components/base/image-uploader/image-list'
import { TransferMethod, type VisionFile, type VisionSettings } from '@/types/app'
import { useImageFiles } from '@/app/components/base/image-uploader/hooks'
export type IChatProps = {
configElem?: React.ReactNode
@@ -37,7 +41,7 @@ export type IChatProps = {
onFeedback?: FeedbackFunc
onSubmitAnnotation?: SubmitAnnotationFunc
checkCanSend?: () => boolean
onSend?: (message: string) => void
onSend?: (message: string, files: VisionFile[]) => void
displayScene?: DisplayScene
useCurrentUserAvatar?: boolean
isResponsing?: boolean
@@ -54,6 +58,7 @@ export type IChatProps = {
dataSets?: DataSet[]
isShowCitationHitInfo?: boolean
isShowPromptLog?: boolean
visionConfig?: VisionSettings
}
const Chat: FC<IChatProps> = ({
@@ -83,9 +88,19 @@ const Chat: FC<IChatProps> = ({
dataSets,
isShowCitationHitInfo,
isShowPromptLog,
visionConfig,
}) => {
const { t } = useTranslation()
const { notify } = useContext(ToastContext)
const {
files,
onUpload,
onRemove,
onReUpload,
onImageLinkLoadError,
onImageLinkLoadSuccess,
onClear,
} = useImageFiles()
const isUseInputMethod = useRef(false)
const [query, setQuery] = React.useState('')
@@ -114,9 +129,18 @@ const Chat: FC<IChatProps> = ({
const handleSend = () => {
if (!valid() || (checkCanSend && !checkCanSend()))
return
onSend(query)
if (!isResponsing)
setQuery('')
onSend(query, files.filter(file => file.progress !== -1).map(fileItem => ({
type: 'image',
transfer_method: fileItem.type,
url: fileItem.url,
upload_file_id: fileItem.fileId,
})))
if (!files.find(item => item.type === TransferMethod.local_file && !item.fileId)) {
if (files.length)
onClear()
if (!isResponsing)
setQuery('')
}
}
const handleKeyUp = (e: React.KeyboardEvent<HTMLTextAreaElement>) => {
@@ -198,6 +222,8 @@ const Chat: FC<IChatProps> = ({
item={item}
isShowPromptLog={isShowPromptLog}
isResponsing={isResponsing}
// ['https://placekitten.com/360/360', 'https://placekitten.com/360/640']
imgSrcs={(item.message_files && item.message_files?.length > 0) ? item.message_files.map(item => item.url) : []}
/>
)
})}
@@ -246,18 +272,42 @@ const Chat: FC<IChatProps> = ({
</div>
</div>)
}
<div className="relative">
<AutoHeightTextarea
<div className='p-[5.5px] max-h-[150px] bg-white border-[1.5px] border-gray-200 rounded-xl overflow-y-auto'>
{
visionConfig?.enabled && (
<>
<div className='absolute bottom-2 left-2 flex items-center'>
<ChatImageUploader
settings={visionConfig}
onUpload={onUpload}
disabled={files.length >= visionConfig.number_limits}
/>
<div className='mx-1 w-[1px] h-4 bg-black/5' />
</div>
<div className='pl-[52px]'>
<ImageList
list={files}
onRemove={onRemove}
onReUpload={onReUpload}
onImageLinkLoadSuccess={onImageLinkLoadSuccess}
onImageLinkLoadError={onImageLinkLoadError}
/>
</div>
</>
)
}
<Textarea
className={`
block w-full px-2 pr-[118px] py-[7px] leading-5 max-h-none text-sm text-gray-700 outline-none appearance-none resize-none
${visionConfig?.enabled && 'pl-12'}
`}
value={query}
onChange={handleContentChange}
onKeyUp={handleKeyUp}
onKeyDown={handleKeyDown}
minHeight={48}
autoFocus
controlFocus={controlFocus}
className={`${cn(s.textArea)} resize-none block w-full pl-3 bg-gray-50 border border-gray-200 rounded-md focus:outline-none sm:text-sm text-gray-700`}
autoSize
/>
<div className="absolute top-0 right-2 flex items-center h-[48px]">
<div className="absolute bottom-2 right-2 flex items-center h-8">
<div className={`${s.count} mr-4 h-5 leading-5 text-sm bg-gray-50 text-gray-500`}>{query.trim().length}</div>
{
query
@@ -282,9 +332,8 @@ const Chat: FC<IChatProps> = ({
{isMobile
? sendBtn
: (
<Tooltip
selector='send-tip'
htmlContent={
<TooltipPlus
popupContent={
<div>
<div>{t('common.operation.send')} Enter</div>
<div>{t('common.operation.lineBreak')} Shift Enter</div>
@@ -292,7 +341,7 @@ const Chat: FC<IChatProps> = ({
}
>
{sendBtn}
</Tooltip>
</TooltipPlus>
)}
</div>
{

View File

@@ -8,14 +8,16 @@ import Log from '../log'
import MoreInfo from '../more-info'
import AppContext from '@/context/app-context'
import { Markdown } from '@/app/components/base/markdown'
import ImageGallery from '@/app/components/base/image-gallery'
type IQuestionProps = Pick<IChatItem, 'id' | 'content' | 'more' | 'useCurrentUserAvatar'> & {
isShowPromptLog?: boolean
item: IChatItem
imgSrcs?: string[]
isResponsing?: boolean
}
const Question: FC<IQuestionProps> = ({ id, content, more, useCurrentUserAvatar, isShowPromptLog, item, isResponsing }) => {
const Question: FC<IQuestionProps> = ({ id, content, imgSrcs, more, useCurrentUserAvatar, isShowPromptLog, item, isResponsing }) => {
const { userProfile } = useContext(AppContext)
const userName = userProfile?.name
const ref = useRef(null)
@@ -23,6 +25,7 @@ const Question: FC<IQuestionProps> = ({ id, content, more, useCurrentUserAvatar,
return (
<div className={`flex items-start justify-end ${isShowPromptLog && 'first-of-type:pt-[14px]'}`} key={id} ref={ref}>
<div className={s.questionWrapWrap}>
<div className={`${s.question} group relative text-sm text-gray-900`}>
{
isShowPromptLog && !isResponsing && (
@@ -32,6 +35,9 @@ const Question: FC<IQuestionProps> = ({ id, content, more, useCurrentUserAvatar,
<div
className={'mr-2 py-3 px-4 bg-blue-500 rounded-tl-2xl rounded-b-2xl'}
>
{imgSrcs && imgSrcs.length > 0 && (
<ImageGallery srcs={imgSrcs} />
)}
<Markdown content={content} />
</div>
</div>

View File

@@ -1,5 +1,5 @@
import type { Annotation, MessageRating } from '@/models/log'
import type { VisionFile } from '@/types/app'
export type MessageMore = {
time: string
tokens: number
@@ -67,6 +67,7 @@ export type IChatItem = {
useCurrentUserAvatar?: boolean
isOpeningStatement?: boolean
log?: { role: string; text: string }[]
message_files?: VisionFile[]
}
export type MessageEnd = {