import type { LanguageCode } from './types'
type VideoMetadata = {
title?: string
description?: string
uploader?: string
tags?: string[]
}
const UNICODE_PATTERNS: Record<LanguageCode, RegExp> = {
ko: /[\u3131-\u3163\uac00-\ud7a3]/, // Korean: Hangul Jamo + Syllables
ja: /[\u3040-\u309f\u30a0-\u30ff]/, // Japanese: Hiragana + Katakana
zh: /[\u4e00-\u9fff]/, // Chinese: CJK Unified Ideographs
th: /[\u0e00-\u0e7f]/, // Thai
ar: /[\u0600-\u06ff]/, // Arabic
hi: /[\u0900-\u097f]/, // Hindi (Devanagari)
ru: /[\u0400-\u04ff]/, // Russian (Cyrillic)
el: /[\u0370-\u03ff]/, // Greek
en: /^$/, // English (fallback, no specific pattern)
} as const
const DEFAULT_PRIORITY: LanguageCode[] = ['en', 'ko', 'ja', 'zh']
const extractTextContent = (metadata: VideoMetadata): string => {
const { title = '', description = '', uploader = '', tags = [] } = metadata
return `${title} ${description} ${uploader} ${tags.join(' ')}`
}
const detectLanguagesFromText = (text: string): LanguageCode[] => {
const detectedLanguages: LanguageCode[] = []
for (const [lang, pattern] of Object.entries(UNICODE_PATTERNS)) {
if (pattern.test(text)) {
detectedLanguages.push(lang as LanguageCode)
}
}
return detectedLanguages
}
export const detectVideoLanguage = (metadata: VideoMetadata): LanguageCode[] => {
const content = extractTextContent(metadata)
const detectedLanguages = detectLanguagesFromText(content)
if (detectedLanguages.length > 0) {
const remainingLanguages = DEFAULT_PRIORITY.filter(lang => !detectedLanguages.includes(lang))
return [...detectedLanguages, ...remainingLanguages]
}
return DEFAULT_PRIORITY
}
