koishi-plugin-kkk 1.0.0

package/src/karin/platform/douyin/comments.ts

@@ -0,0 +1,404 @@
+import { DouyinCommentRepliesResponse, DouyinCommentsResponse } from '@ikenxuan/amagi'
+import {
+  createEmojiNode,
+  createLineBreakNode,
+  createMentionNode,
+  createRichTextDocument,
+  createSearchKeywordNode,
+  createTextNode,
+  type RichTextDocument,
+  type RichTextEmojiDefinition,
+  type RichTextNode
+} from '@kkk/richtext'
+import type { DouyinCommentData } from '@template/template/douyin/comment/components/types'
+import decode from 'heic-decode'
+import jpeg from 'jpeg-js'
+
+import { Networks } from '@/module/utils'
+import { douyinFetcher } from '@/module/utils/amagiClient'
+import { Config } from '@/module/utils/Config'
+
+/** 一级评论（`douyin.comments` 的元素） */
+type DyComment = DouyinCommentsResponse['comments'][number]
+/** `text_extra` 项：一级评论那支在生成树里有完整类型 */
+type DyTextExtra = DyComment['text_extra'][number]
+/** 回复评论（`douyin.commentReplies` 的元素） */
+type DyReplyComment = DouyinCommentRepliesResponse['comments'][number]
+
+/**
+ * 回复评论 + 图片列表。
+ *
+ * 生成树这份样本里回复都没带图，`commentReplies` 的 `image_list` 被记成 `null`；
+ * 实际带图时形状与一级评论一致 —— 只在这一处借用 {@link DyComment} 的那一格，
+ * 下游取图就按精确类型写。等 amagi 补上带图的回复样本后删掉本别名。
+ */
+type DyReplyWithImages = Omit<DyReplyComment, 'image_list'> & { image_list: DyComment['image_list'] }
+
+/**
+ * @description 提取评论里的 @ 用户 sec_uid 列表
+ */
+const extractMentionSecUids = (textExtra: DyTextExtra[] | undefined): string[] | null => {
+  if (!Array.isArray(textExtra) || textExtra.length === 0) {
+    return null
+  }
+
+  const secUids = textExtra.map((item) => item.sec_uid).filter((secUid): secUid is string => Boolean(secUid))
+
+  return secUids.length > 0 ? secUids : null
+}
+
+/**
+ * @description 解析评论中的搜索词信息
+ */
+const extractSearchText = (textExtra: DyTextExtra[] | undefined) => {
+  if (!Array.isArray(textExtra) || textExtra.length === 0) {
+    return null
+  }
+
+  const searchItems = textExtra
+    .filter((item) => Boolean(item.search_text))
+    .map((item) => ({
+      search_text: item.search_text ?? '',
+      search_query_id: item.search_query_id ?? ''
+    }))
+
+  return searchItems.length > 0 ? searchItems : null
+}
+
+type DouyinSearchToken = {
+  start: number
+  end: number
+  text: string
+  queryId: string
+}
+
+type DouyinMentionToken = {
+  text: string
+  userId: string
+}
+
+type DouyinCommentItem = DouyinCommentData['CommentsData'][number]
+type DouyinReplyCommentItem = NonNullable<DouyinCommentItem['replyComment']>[number]
+
+/**
+ * 提取评论正文中的搜索词范围。
+ *
+ * 抖音会把高亮搜索词单独放在 `text_extra` 里，但用户实际看到的是“正文里某一段文字高亮”。
+ * 所以这里不能只把它单独透传给 template，而是要把范围信息重新合回正文解析流程里。
+ */
+const extractSearchTokens = (textExtra: DyTextExtra[] | undefined, text: string): DouyinSearchToken[] => {
+  if (!Array.isArray(textExtra) || textExtra.length === 0 || !text) {
+    return []
+  }
+
+  return textExtra
+    .filter(
+      (item) =>
+        typeof item.start === 'number' &&
+        typeof item.end === 'number' &&
+        typeof item.search_text === 'string' &&
+        item.search_text.length > 0 &&
+        item.start >= 0 &&
+        item.end > item.start &&
+        item.end <= text.length
+    )
+    .map((item) => ({
+      start: item.start,
+      end: item.end,
+      text: item.search_text ?? '',
+      queryId: item.search_query_id ?? ''
+    }))
+    .filter((item) => text.slice(item.start, item.end) === item.text)
+    .sort((a, b) => a.start - b.start || b.end - b.start - (a.end - a.start))
+}
+
+/**
+ * 根据抖音 `sec_uid` 反查当前昵称，生成文本中可匹配的 @ 标记。
+ *
+ * 抖音评论正文里通常只保留 `@昵称` 文本，稳定用户 ID 在 `text_extra` 里。
+ * 这里先把 ID 转成 `@昵称 + userId`，后续解析时才能把普通文本切成 mention 节点。
+ */
+const resolveMentionTokens = async (userIds: string[] | null): Promise<DouyinMentionToken[]> => {
+  if (!userIds || userIds.length === 0) {
+    return []
+  }
+
+  const uniqueUserIds = [...new Set(userIds)]
+
+  const mentionTokens = await Promise.all(
+    uniqueUserIds.map(async (secUid) => {
+      try {
+        const userInfo = await douyinFetcher.fetchUserProfile({ sec_uid: secUid })
+        const nickname = userInfo.data.user.nickname?.trim()
+
+        if (!nickname || userInfo.data.user.sec_uid !== secUid) {
+          return null
+        }
+
+        return {
+          text: `@${nickname}`,
+          userId: secUid
+        }
+      } catch {
+        return null
+      }
+    })
+  )
+
+  return mentionTokens.filter((item): item is DouyinMentionToken => Boolean(item)).sort((a, b) => b.text.length - a.text.length)
+}
+
+/**
+ * 把抖音评论原始文本解析成共享富文本 JSON。
+ *
+ * 这里刻意不拼 HTML：
+ * - 普通文本原样放进 text 节点，由 React 渲染时自动转义。
+ * - 换行符转成 lineBreak 节点，保留原来的视觉换行。
+ * - 平台表情转成 emoji 节点，图片 URL 交给 template 侧渲染器再做协议白名单。
+ * - @ 用户转成 mention 节点，template 侧只负责套样式。
+ */
+const buildDouyinRichText = async (
+  text: string,
+  emojiData: RichTextEmojiDefinition[],
+  mentionUserIds: string[] | null,
+  searchTokens: DouyinSearchToken[] = []
+): Promise<RichTextDocument> => {
+  const mentionTokens = await resolveMentionTokens(mentionUserIds)
+  const emojiTokens = emojiData.filter((item) => Boolean(item?.name) && Boolean(item?.url)).sort((a, b) => b.name.length - a.name.length)
+
+  const nodes: RichTextNode[] = []
+  let buffer = ''
+  let index = 0
+
+  const pushBuffer = () => {
+    if (buffer.length > 0) {
+      nodes.push(createTextNode(buffer))
+      buffer = ''
+    }
+  }
+
+  let searchTokenIndex = 0
+
+  while (index < text.length) {
+    while (searchTokens[searchTokenIndex] && searchTokens[searchTokenIndex].start < index) {
+      searchTokenIndex += 1
+    }
+
+    const currentSearchToken = searchTokens[searchTokenIndex]
+    if (currentSearchToken && currentSearchToken.start === index) {
+      pushBuffer()
+      nodes.push(createSearchKeywordNode(currentSearchToken.text, currentSearchToken.queryId))
+      index = currentSearchToken.end
+      searchTokenIndex += 1
+      continue
+    }
+
+    if (text[index] === '\r') {
+      pushBuffer()
+      if (text[index + 1] === '\n') {
+        index += 2
+      } else {
+        index += 1
+      }
+      nodes.push(createLineBreakNode())
+      continue
+    }
+
+    if (text[index] === '\n') {
+      pushBuffer()
+      nodes.push(createLineBreakNode())
+      index += 1
+      continue
+    }
+
+    const matchedMention = mentionTokens.find((item) => text.startsWith(item.text, index))
+    if (matchedMention) {
+      pushBuffer()
+      nodes.push(createMentionNode(matchedMention.text, matchedMention.userId))
+      index += matchedMention.text.length
+      continue
+    }
+
+    const matchedEmoji = emojiTokens.find((item) => text.startsWith(item.name, index))
+    if (matchedEmoji) {
+      pushBuffer()
+      nodes.push(createEmojiNode(matchedEmoji.name, matchedEmoji.url))
+      index += matchedEmoji.name.length
+      continue
+    }
+
+    buffer += text[index]
+    index += 1
+  }
+
+  pushBuffer()
+
+  return createRichTextDocument(nodes, { platform: 'douyin' })
+}
+
+/**
+ * 处理单个评论图片的HEIC转JPG
+ * @param imageUrl 图片URL
+ * @returns 处理后的图片URL
+ */
+const processCommentImage = async (imageUrl: string | null): Promise<string | null> => {
+  if (!imageUrl) return null
+
+  const headers = await new Networks({ url: imageUrl, type: 'arraybuffer' }).getHeaders()
+  if (headers['content-type'] && headers['content-type'] === 'image/heic') {
+    const response = await new Networks({ url: imageUrl, type: 'arraybuffer' }).returnResult()
+
+    // 使用 heic-decode 解码 HEIC 图片
+    const decoded = await decode({ buffer: response.data })
+
+    // 使用 jpeg-js 将 RGBA 数据编码为 JPEG
+    const jpegImageData = {
+      data: Buffer.from(decoded.data),
+      width: decoded.width,
+      height: decoded.height
+    }
+    const jpegBuffer = jpeg.encode(jpegImageData, 90).data
+
+    const base64Image = Buffer.from(jpegBuffer).toString('base64')
+    return `data:image/jpeg;base64,${base64Image}`
+  }
+  return imageUrl
+}
+
+/**
+ *
+ * @param {*} data 评论接口的响应体（`fetchWorkComments` 的 `data.data`）
+ * @param {*} emojidata 处理过后的emoji列表
+ * @returns obj
+ */
+export const douyinComments = async (data: DouyinCommentsResponse, emojidata: RichTextEmojiDefinition[]) => {
+  const commentsData: DouyinCommentItem[] = []
+  let imageUrls: string[] = []
+  if (data.comments === null) return { CommentsData: [], image_url: [] }
+  let id = 1
+  for (const comment of data.comments) {
+    const cid = comment.cid
+    const aweme_id = comment.aweme_id
+    const nickname = comment.user.nickname
+    const userimageurl = comment.user.avatar_thumb.url_list[0]
+    const text = comment.text
+    const ip = comment.ip_label ?? '未知'
+    const label_type = comment.label_type ?? -1
+    const sticker = comment.sticker ? comment.sticker.animate_url.url_list[0] : null
+    const digg_count = comment.digg_count
+    const imageurl =
+      comment.image_list && comment.image_list?.[0] && comment.image_list?.[0].origin_url && comment.image_list?.[0].origin_url.url_list
+        ? comment.image_list?.[0].origin_url.url_list[0]
+        : null
+    const status_label = comment.label_list?.[0]?.text ?? null
+    const userintextlongid = extractMentionSecUids(comment.text_extra)
+    const search_text = extractSearchText(comment.text_extra)
+    const searchTokens = extractSearchTokens(comment.text_extra, text)
+    const richText = await buildDouyinRichText(text, emojidata, userintextlongid, searchTokens)
+
+    // 在循环中处理图片HEIC转JPG
+    const processedImageUrl = await processCommentImage(imageurl)
+
+    // 收集评论图片
+    if (processedImageUrl) {
+      imageUrls.push(
+        processedImageUrl.startsWith('data:image/jpeg;base64,')
+          ? `base64://${processedImageUrl.replace('data:image/jpeg;base64,', '')}`
+          : processedImageUrl
+      )
+    }
+
+    // 收集表情包图片
+    if (sticker) {
+      imageUrls.push(sticker)
+    }
+
+    const replyComment = await douyinFetcher.fetchCommentReplies({
+      aweme_id,
+      comment_id: cid,
+      number: Config.douyin.subCommentLimit
+    })
+
+    const replyComments: DouyinReplyCommentItem[] = []
+
+    if (replyComment.data.comments && replyComment.data.comments.length > 0) {
+      for (const reply of replyComment.data.comments) {
+        // 生成树里回复的 image_list 是 null（样本没录到带图的回复），见 DyReplyWithImages
+        const replyItem = reply as unknown as DyReplyWithImages
+        const replyUserintextlongid = extractMentionSecUids(replyItem.text_extra)
+        const replySearchTokens = extractSearchTokens(replyItem.text_extra, replyItem.text)
+        const replyRichText = await buildDouyinRichText(replyItem.text, emojidata, replyUserintextlongid, replySearchTokens)
+
+        // 处理回复评论的图片列表
+        const replyImageUrl = replyItem.image_list?.[0]?.origin_url?.url_list?.[0]
+        const replyStickerUrl = replyItem.sticker?.animate_url?.url_list?.[0]
+
+        let replyImageList: string[] | null = null
+        if (replyImageUrl) {
+          const processedReplyImage = await processCommentImage(replyImageUrl)
+          if (processedReplyImage) {
+            replyImageList = [processedReplyImage]
+            imageUrls.push(
+              processedReplyImage.startsWith('data:image/jpeg;base64,')
+                ? `base64://${processedReplyImage.replace('data:image/jpeg;base64,', '')}`
+                : processedReplyImage
+            )
+          }
+        } else if (replyStickerUrl) {
+          replyImageList = [replyStickerUrl]
+          imageUrls.push(replyStickerUrl)
+        }
+
+        replyComments.push({
+          create_time: replyItem.create_time,
+          nickname: replyItem.user.nickname,
+          userimageurl: replyItem.user.avatar_thumb.url_list[0],
+          text: replyRichText,
+          digg_count: replyItem.digg_count,
+          ip_label: replyItem.ip_label,
+          text_extra: replyItem.text_extra,
+          label_text: replyItem.label_text,
+          image_list: replyImageList,
+          cid: replyItem.cid,
+          reply_to_reply_id: replyItem.reply_to_reply_id,
+          reply_to_username: replyItem.reply_to_username
+        })
+      }
+    }
+
+    const commentObj: DouyinCommentData['CommentsData'][number] = {
+      id: id++,
+      replyComment: replyComments.length > 0 ? replyComments : undefined,
+      cid,
+      aweme_id,
+      nickname,
+      userimageurl,
+      text: richText,
+      digg_count,
+      ip_label: ip,
+      create_time: comment.create_time,
+      commentimage: processedImageUrl ?? undefined,
+      label_type,
+      sticker: sticker ?? undefined,
+      status_label: status_label ?? undefined,
+      is_At_user_id: userintextlongid,
+      search_text,
+      is_author_digged: comment.is_author_digged ?? false
+    }
+    commentsData.push(commentObj)
+  }
+
+  commentsData.sort((a, b) => b.digg_count - a.digg_count)
+
+  const indexLabelTypeOne = commentsData.findIndex((comment) => comment.label_type === 1)
+
+  if (indexLabelTypeOne !== -1) {
+    const authorComment = commentsData.splice(indexLabelTypeOne, 1)[0]
+    commentsData.unshift(authorComment)
+  }
+
+  return {
+    CommentsData: commentsData,
+    image_url: imageUrls
+  }
+}