feat: add language detection and bidirectional translation utilities

This commit is contained in:
Pleasurecruise 2025-05-27 20:54:57 +08:00
parent 6260ea76ce
commit c7ed9e9551
No known key found for this signature in database
GPG Key ID: E6385136096279B6
4 changed files with 200 additions and 101 deletions

View File

@ -84,6 +84,7 @@
"electron-window-state": "^5.0.3",
"epub": "patch:epub@npm%3A1.3.0#~/.yarn/patches/epub-npm-1.3.0-8325494ffe.patch",
"fast-xml-parser": "^5.2.0",
"franc": "^6.2.0",
"fs-extra": "^11.2.0",
"jsdom": "^26.0.0",
"markdown-it": "^14.1.0",

View File

@ -12,6 +12,13 @@ import { getDefaultTranslateAssistant } from '@renderer/services/AssistantServic
import { getModelUniqId, hasModel } from '@renderer/services/ModelService'
import type { Model, TranslateHistory } from '@renderer/types'
import { runAsyncFunction, uuid } from '@renderer/utils'
import {
createInputScrollHandler,
createOutputScrollHandler,
detectLanguage,
getTargetLanguageForBidirectional,
isLanguageInPair
} from '@renderer/utils/translate'
import { Button, Divider, Empty, Flex, Modal, Popconfirm, Select, Space, Switch, Tooltip } from 'antd'
import TextArea, { TextAreaRef } from 'antd/es/input/TextArea'
import dayjs from 'dayjs'
@ -268,70 +275,6 @@ const TranslatePage: FC = () => {
await db.translate_history.clear()
}
const detectLanguage = async (inputText: string): Promise<string> => {
if (!inputText.trim()) return 'any'
const langPatterns = {
chinese: /[\u4e00-\u9fa5]/,
japanese: /[\u3040-\u30ff\u3400-\u4dbf]/,
korean: /[\uAC00-\uD7AF]/,
russian: /[\u0400-\u04FF]/,
arabic: /[\u0600-\u06FF]/
}
for (const [lang, pattern] of Object.entries(langPatterns)) {
if (pattern.test(inputText)) return lang
}
try {
const prompt = `Identify language: "${inputText.substring(0, 50)}". Reply with one word only: english, chinese, japanese, korean, russian, spanish, french, german, italian, portuguese, arabic.`
let detectedCode = ''
await fetchTranslate({
content: inputText.substring(0, 50),
assistant: {
id: 'lang-detector',
name: 'Language Detector',
prompt,
topics: [],
type: 'translator'
},
onResponse: (response) => {
detectedCode = response.trim().toLowerCase()
}
})
const validCodes = [
'english',
'chinese',
'japanese',
'korean',
'russian',
'spanish',
'french',
'italian',
'portuguese',
'arabic',
'german'
]
return validCodes.find((code) => detectedCode.includes(code)) || 'english'
} catch (error) {
console.error('语言检测错误:', error)
return 'english'
}
}
const getTargetLanguageForBidirectional = (sourceLanguage: string): string => {
if (sourceLanguage === bidirectionalPair[0]) {
return bidirectionalPair[1]
} else if (sourceLanguage === bidirectionalPair[1]) {
return bidirectionalPair[0]
}
return bidirectionalPair[0] === sourceLanguage ? bidirectionalPair[1] : bidirectionalPair[0]
}
const onTranslate = async () => {
if (!text.trim()) return
if (!translateModel) {
@ -350,7 +293,7 @@ const TranslatePage: FC = () => {
let actualTargetLanguage = targetLanguage
if (isBidirectional) {
if (![bidirectionalPair[0], bidirectionalPair[1]].includes(sourceLanguage)) {
if (!isLanguageInPair(sourceLanguage, bidirectionalPair)) {
window.message.warning({
content: t('translate.language.not_pair'),
key: 'translate-message'
@ -358,7 +301,7 @@ const TranslatePage: FC = () => {
setLoading(false)
return
}
actualTargetLanguage = getTargetLanguageForBidirectional(sourceLanguage)
actualTargetLanguage = getTargetLanguageForBidirectional(sourceLanguage, bidirectionalPair)
setTargetLanguage(actualTargetLanguage)
} else {
if (sourceLanguage === targetLanguage) {
@ -453,41 +396,8 @@ const TranslatePage: FC = () => {
}
}
// Handle input area scroll event
const handleInputScroll = (e: React.UIEvent<HTMLTextAreaElement>) => {
if (!isScrollSyncEnabled || !outputTextRef.current || isProgrammaticScroll.current) return
isProgrammaticScroll.current = true
const inputEl = e.currentTarget
const outputEl = outputTextRef.current
// Calculate scroll position by ratio
const inputScrollRatio = inputEl.scrollTop / (inputEl.scrollHeight - inputEl.clientHeight || 1)
outputEl.scrollTop = inputScrollRatio * (outputEl.scrollHeight - outputEl.clientHeight || 1)
requestAnimationFrame(() => {
isProgrammaticScroll.current = false
})
}
// Handle output area scroll event
const handleOutputScroll = (e: React.UIEvent<HTMLDivElement>) => {
const inputEl = textAreaRef.current?.resizableTextArea?.textArea
if (!isScrollSyncEnabled || !inputEl || isProgrammaticScroll.current) return
isProgrammaticScroll.current = true
const outputEl = e.currentTarget
// Calculate scroll position by ratio
const outputScrollRatio = outputEl.scrollTop / (outputEl.scrollHeight - outputEl.clientHeight || 1)
inputEl.scrollTop = outputScrollRatio * (inputEl.scrollHeight - inputEl.clientHeight || 1)
requestAnimationFrame(() => {
isProgrammaticScroll.current = false
})
}
const handleInputScroll = createInputScrollHandler(outputTextRef, isProgrammaticScroll, isScrollSyncEnabled)
const handleOutputScroll = createOutputScrollHandler(textAreaRef, isProgrammaticScroll, isScrollSyncEnabled)
// 获取当前语言状态显示
const getLanguageDisplay = () => {

View File

@ -0,0 +1,154 @@
import { fetchTranslate } from '@renderer/services/ApiService'
import { franc } from 'franc'
import React, { MutableRefObject } from 'react'
/**
*
* @param {string} inputText
* @returns {Promise<string>}
*/
export const detectLanguage = async (inputText: string): Promise<string> => {
if (!inputText.trim()) return 'any'
const text = inputText.trim()
const detectedLangCode = franc(text)
// 映射 ISO 639-3 代码到应用使用的语言代码
const languageMap: Record<string, string> = {
cmn: 'chinese', // 普通话
zho: 'chinese', // 中文
jpn: 'japanese', // 日语
kor: 'korean', // 韩语
rus: 'russian', // 俄语
spa: 'spanish', // 西班牙语
fra: 'french', // 法语
deu: 'german', // 德语
ita: 'italian', // 意大利语
por: 'portuguese', // 葡萄牙语
ara: 'arabic', // 阿拉伯语
eng: 'english' // 英语
}
if (detectedLangCode !== 'und' && languageMap[detectedLangCode]) {
return languageMap[detectedLangCode]
}
try {
const sampleText = text.substring(0, 200)
const prompt = `Identify the primary language in this text: "${sampleText}". Reply with only one word from this list: english, chinese, japanese, korean, russian, spanish, french, german, italian, portuguese, arabic.`
let detectedCode = ''
await fetchTranslate({
content: sampleText,
assistant: {
id: 'lang-detector',
name: 'Language Detector',
prompt,
topics: [],
type: 'translator'
},
onResponse: (response) => {
detectedCode = response.trim().toLowerCase()
}
})
const validCodes = [
'english',
'chinese',
'japanese',
'korean',
'russian',
'spanish',
'french',
'german',
'italian',
'portuguese',
'arabic'
]
return validCodes.find((code) => detectedCode.includes(code)) || 'english'
} catch (error) {
console.error('语言检测错误:', error)
return 'english'
}
}
/**
*
* @param sourceLanguage
* @param languagePair
* @returns
*/
export const getTargetLanguageForBidirectional = (sourceLanguage: string, languagePair: [string, string]): string => {
if (sourceLanguage === languagePair[0]) {
return languagePair[1]
} else if (sourceLanguage === languagePair[1]) {
return languagePair[0]
}
// 默认返回第一个不同于源语言的语言
return languagePair[0] === sourceLanguage ? languagePair[1] : languagePair[0]
}
/**
*
* @param sourceLanguage
* @param languagePair
* @returns
*/
export const isLanguageInPair = (sourceLanguage: string, languagePair: [string, string]): boolean => {
return [languagePair[0], languagePair[1]].includes(sourceLanguage)
}
/**
*
* @param sourceElement
* @param targetElement
* @param isProgrammaticScrollRef
*/
export const handleScrollSync = (
sourceElement: HTMLElement,
targetElement: HTMLElement,
isProgrammaticScrollRef: MutableRefObject<boolean>
): void => {
if (isProgrammaticScrollRef.current) return
isProgrammaticScrollRef.current = true
// 计算滚动位置比例
const scrollRatio = sourceElement.scrollTop / (sourceElement.scrollHeight - sourceElement.clientHeight || 1)
targetElement.scrollTop = scrollRatio * (targetElement.scrollHeight - targetElement.clientHeight || 1)
requestAnimationFrame(() => {
isProgrammaticScrollRef.current = false
})
}
/**
*
*/
export const createInputScrollHandler = (
targetRef: MutableRefObject<HTMLDivElement | null>,
isProgrammaticScrollRef: MutableRefObject<boolean>,
isScrollSyncEnabled: boolean
) => {
return (e: React.UIEvent<HTMLTextAreaElement>) => {
if (!isScrollSyncEnabled || !targetRef.current || isProgrammaticScrollRef.current) return
handleScrollSync(e.currentTarget, targetRef.current, isProgrammaticScrollRef)
}
}
/**
*
*/
export const createOutputScrollHandler = (
textAreaRef: MutableRefObject<any>,
isProgrammaticScrollRef: MutableRefObject<boolean>,
isScrollSyncEnabled: boolean
) => {
return (e: React.UIEvent<HTMLDivElement>) => {
const inputEl = textAreaRef.current?.resizableTextArea?.textArea
if (!isScrollSyncEnabled || !inputEl || isProgrammaticScrollRef.current) return
handleScrollSync(e.currentTarget, inputEl, isProgrammaticScrollRef)
}
}

View File

@ -6010,6 +6010,7 @@ __metadata:
eslint-plugin-unused-imports: "npm:^4.1.4"
fast-diff: "npm:^1.3.0"
fast-xml-parser: "npm:^5.2.0"
franc: "npm:^6.2.0"
fs-extra: "npm:^11.2.0"
html-to-image: "npm:^1.11.13"
husky: "npm:^9.1.7"
@ -7465,6 +7466,13 @@ __metadata:
languageName: node
linkType: hard
"collapse-white-space@npm:^2.0.0":
version: 2.1.0
resolution: "collapse-white-space@npm:2.1.0"
checksum: 10c0/b2e2800f4ab261e62eb27a1fbe853378296e3a726d6695117ed033e82d61fb6abeae4ffc1465d5454499e237005de9cfc52c9562dc7ca4ac759b9a222ef14453
languageName: node
linkType: hard
"color-convert@npm:^1.9.0":
version: 1.9.3
resolution: "color-convert@npm:1.9.3"
@ -10527,6 +10535,15 @@ __metadata:
languageName: node
linkType: hard
"franc@npm:^6.2.0":
version: 6.2.0
resolution: "franc@npm:6.2.0"
dependencies:
trigram-utils: "npm:^2.0.0"
checksum: 10c0/136a08d6e4632f17eae6f0ae93b224b0bf2233dc1d5dbd0b23e479960f6c71c0847bef834d3b6b7c9cefb4f905d5e08fc82b0738bb3ed4a6c83faffcf9fa2a11
languageName: node
linkType: hard
"fresh@npm:^2.0.0":
version: 2.0.0
resolution: "fresh@npm:2.0.0"
@ -14572,6 +14589,13 @@ __metadata:
languageName: node
linkType: hard
"n-gram@npm:^2.0.0":
version: 2.0.2
resolution: "n-gram@npm:2.0.2"
checksum: 10c0/72e2cdc8c37c9253b556a0deb9cd26d5ac59a5d7a38b2d2928927e3959bc7d3cb591d766e30309a4c685dbc51330025cb30c5c6518ee516caf3318aed2635f1b
languageName: node
linkType: hard
"nanoid@npm:^3.3.7, nanoid@npm:^3.3.8":
version: 3.3.11
resolution: "nanoid@npm:3.3.11"
@ -18953,6 +18977,16 @@ __metadata:
languageName: node
linkType: hard
"trigram-utils@npm:^2.0.0":
version: 2.0.1
resolution: "trigram-utils@npm:2.0.1"
dependencies:
collapse-white-space: "npm:^2.0.0"
n-gram: "npm:^2.0.0"
checksum: 10c0/d024dc91a9c0310e75fa68422185e3a32814831971b9e86a2925e74bd1932a30501aa2ac214768f0a545f3db63610ee14b4748ac31532e1bc46c791941d71c6d
languageName: node
linkType: hard
"trim-lines@npm:^3.0.0":
version: 3.0.1
resolution: "trim-lines@npm:3.0.1"