BYW-VoiceHub/server/utils/submissionNoteAutoModeration.ts

import type { SongBlacklist } from '~/drizzle/schema'

export type BlacklistType = 'KEYWORD' | 'LANGUAGE' | 'GENRE' | 'SONG'

export interface ModerationContext {
  title?: string | null
  artist?: string | null
  genre?: string | null
}

export interface ModerationResult {
  notePublic: boolean
  publicStatus: 'approved' | 'rejected' | null
  rejectedReasons: string[]
  matchedRules: { type: BlacklistType; value: string }[]
}

function escapeRegex(s: string): string {
  return String(s).replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
}

function isCJK(text: string): boolean {
  return /[\u4e00-\u9fa5]/.test(text)
}

function isLatin(text: string): boolean {
  return /[A-Za-z]/.test(text)
}

function isKana(text: string): boolean {
  return /[\u3040-\u30ff]/.test(text)
}

function isHangul(text: string): boolean {
  return /[\uac00-\ud7af]/.test(text)
}

function isCyrillic(text: string): boolean {
  return /[\u0400-\u04ff]/.test(text)
}

function isArabic(text: string): boolean {
  return /[\u0600-\u06ff]/.test(text)
}

function detectLanguages(text: string): string[] {
  const result: string[] = []
  if (!text) return result
  const normalized = text.toLowerCase()
  if (isCJK(text)) result.push('zh', '中文', 'chinese', 'cjk')
  if (isLatin(text)) result.push('en', '英文', 'english', 'latin')
  if (isKana(text)) result.push('ja', '日文', 'japanese')
  if (isHangul(text)) result.push('ko', '韩文', 'korean')
  if (isCyrillic(text)) result.push('ru', '俄文', 'russian', 'cyrillic')
  if (isArabic(text)) result.push('ar', '阿拉伯文', 'arabic')
  return result
}

export function moderateSubmissionNote(
  note: string | null,
  wantPublic: boolean,
  requiresApproval: boolean,
  song: ModerationContext,
  blacklistRows: SongBlacklist[]
): ModerationResult {
  if (!note || !wantPublic) {
    return {
      notePublic: wantPublic && !!note,
      publicStatus: null,
      rejectedReasons: [],
      matchedRules: []
    }
  }

  const noteLower = note.toLowerCase()
  const noteForKeyword = note.toLowerCase()
  const title = (song.title || '').toLowerCase()
  const artist = (song.artist || '').toLowerCase()
  const genre = (song.genre || '').toLowerCase()

  const rejectedReasons: string[] = []
  const matchedRules: { type: BlacklistType; value: string }[] = []

  const noteLanguages = detectLanguages(note)
  const textLanguages = new Set([
    ...noteLanguages,
    ...detectLanguages(song.title || ''),
    ...detectLanguages(song.artist || ''),
    ...detectLanguages(song.genre || '')
  ])

  for (const row of blacklistRows) {
    const t = row.type as BlacklistType
    const rawValue = row.value
    if (!rawValue) continue
    const value = rawValue.trim()
    if (!value) continue
    const valueLower = value.toLowerCase()

    if (t === 'KEYWORD') {
      // 命中 note / title / artist / genre 的其中一个都算
      const haystacks = [noteForKeyword, title, artist, genre].filter(Boolean) as string[]
      const exact = haystacks.some((h) => h.includes(valueLower))
      const asRegex = (() => {
        try {
          const re = new RegExp(escapeRegex(value), 'i')
          return haystacks.some((h) => re.test(h))
        } catch (_) {
          return false
        }
      })()
      if (exact || asRegex) {
        matchedRules.push({ type: t, value })
        rejectedReasons.push(`命中敏感词：${value}`)
      }
    } else if (t === 'LANGUAGE') {
      const v = valueLower
      const checkList = [
        v,
        ...v.split(/[,，\/|\s]+/).map((s) => s.trim()).filter(Boolean)
      ]
      const hit = checkList.some((lang) => textLanguages.has(lang))
      if (hit) {
        matchedRules.push({ type: t, value })
        rejectedReasons.push(`命中语言限制：${value}`)
      }
    } else if (t === 'GENRE') {
      // 匹配流派：genre 字段或 note/title/artist 中的流派关键词
      const combinedForGenre = `${genre} ${noteForKeyword} ${title} ${artist}`.trim()
      if (!combinedForGenre) continue
      const tokens = valueLower.split(/[,，\/|\s]+/).map((s) => s.trim()).filter(Boolean)
      const hit = tokens.some((tok) => combinedForGenre.includes(tok))
      if (hit) {
        matchedRules.push({ type: t, value })
        rejectedReasons.push(`命中流派限制：${value}`)
      }
    } else {
      // SONG 不参与 note 审查
      continue
    }
  }

  const rejected = matchedRules.length > 0
  return {
    notePublic: rejected ? false : (requiresApproval ? false : wantPublic),
    publicStatus: requiresApproval || matchedRules.length > 0 ? (rejected ? 'rejected' : 'approved') : null,
    rejectedReasons,
    matchedRules
  }
}