NotionNext/lib/db/getSiteData.js

import BLOG from '@/blog.config'
import { getAllCategories } from '@/lib/notion/getAllCategories'
import getAllPageIds from '@/lib/notion/getAllPageIds'
import { getAllTags } from '@/lib/notion/getAllTags'
import { getConfigMapFromConfigPage } from '@/lib/notion/getNotionConfig'
import getPageProperties, {
  adjustPageProperties
} from '@/lib/notion/getPageProperties'
import { fetchInBatches, getPage } from '@/lib/notion/getPostBlocks'
import { compressImage, mapImgUrl } from '@/lib/notion/mapImage'
import { deepClone } from '@/lib/utils'
import { idToUuid } from 'notion-utils'
import { siteConfig } from '../config'
import { extractLangId, extractLangPrefix, getShortId } from '../utils/pageId'

export { getAllTags } from '../notion/getAllTags'
export { getPost } from '../notion/getNotionPost'
export { getPage as getPostBlocks } from '../notion/getPostBlocks'

/**
 * 获取博客数据; 基于Notion实现
 * @param {*} pageId
 * @param {*} from
 * @param {*} locale 语言  zh|en|jp 等等
 * @returns
 *
 */
export async function getGlobalData({
  pageId = BLOG.NOTION_PAGE_ID,
  from,
  locale
}) {
  // 获取站点数据 ， 如果pageId有逗号隔开则分次取数据
  const siteIds = pageId?.split(',') || []
  let data = EmptyData(pageId)

  try {
    for (let index = 0; index < siteIds.length; index++) {
      const siteId = siteIds[index]
      const id = extractLangId(siteId)
      const prefix = extractLangPrefix(siteId)
      // 第一个id站点默认语言
      if (index === 0 || locale === prefix) {
        data = await getSiteDataByPageId({
          pageId: id,
          from
        })
      }
    }
  } catch (error) {
    console.error('异常', error)
  }
  return handleDataBeforeReturn(deepClone(data))
}

/**
 * 获取指定notion的collection数据
 * @param pageId
 * @param from 请求来源
 * @returns {Promise<JSX.Element|*|*[]>}
 */
export async function getSiteDataByPageId({ pageId, from }) {
  // 获取NOTION原始数据，此接支持mem缓存。
  const pageRecordMap = await getPage(pageId, from)
  // 将Notion数据按规则转成站点数据
  const data = await converNotionToSiteDate(pageId, from, deepClone(pageRecordMap))
  return data
}

/**
 * 获取公告
 */
async function getNotice(post) {
  if (!post) {
    return null
  }

  post.blockMap = await getPage(post.id, 'data-notice')
  return post
}

/**
 * 空的默认数据
 * @param {*} pageId
 * @returns
 */
const EmptyData = pageId => {
  const empty = {
    notice: null,
    siteInfo: getSiteInfo({}),
    allPages: [
      {
        id: 1,
        title: `无法获取Notion数据，请检查Notion_ID： \n 当前 ${pageId}`,
        summary:
          '访问文档获取帮助→ https://tangly1024.com/article/vercel-deploy-notion-next',
        status: 'Published',
        type: 'Post',
        slug: '13a171332816461db29d50e9f575b00d',
        publishDay: '2024-11-13',
        pageCoverThumbnail: BLOG.HOME_BANNER_IMAGE,
        date: {
          start_date: '2023-04-24',
          lastEditedDay: '2023-04-24',
          tagItems: []
        }
      }
    ],
    allNavPages: [],
    collection: [],
    collectionQuery: {},
    collectionId: null,
    collectionView: {},
    viewIds: [],
    block: {},
    schema: {},
    tagOptions: [],
    categoryOptions: [],
    rawMetadata: {},
    customNav: [],
    customMenu: [],
    postCount: 1,
    pageIds: [],
    latestPosts: []
  }
  return empty
}

/**
 * 将Notion数据转站点数据
 * 这里统一对数据格式化
 * @returns {Promise<JSX.Element|null|*>}
 */
async function converNotionToSiteDate(pageId, from, pageRecordMap) {
  if (!pageRecordMap) {
    console.error('can`t get Notion Data ; Which id is: ', pageId)
    return {}
  }
  pageId = idToUuid(pageId)
  let block = pageRecordMap.block || {}
  const rawMetadata = block[pageId]?.value
  // Check Type Page-Database和Inline-Database
  if (
    rawMetadata?.type !== 'collection_view_page' &&
    rawMetadata?.type !== 'collection_view'
  ) {
    console.error(`pageId "${pageId}" is not a database`)
    return EmptyData(pageId)
  }
  const collection = Object.values(pageRecordMap.collection)[0]?.value || {}
  const collectionId = rawMetadata?.collection_id
  const collectionQuery = pageRecordMap.collection_query
  const collectionView = pageRecordMap.collection_view
  const schema = collection?.schema

  const viewIds = rawMetadata?.view_ids
  const collectionData = []

  const pageIds = getAllPageIds(
    collectionQuery,
    collectionId,
    collectionView,
    viewIds
  )

  if (pageIds?.length === 0) {
    console.error(
      '获取到的文章列表为空，请检查notion模板',
      collectionQuery,
      collection,
      collectionView,
      viewIds,
      pageRecordMap
    )
  } else {
    // console.log('有效Page数量', pageIds?.length)
  }

  // 抓取主数据库最多抓取1000个blocks，溢出的数block这里统一抓取一遍
  const blockIdsNeedFetch = []
  for (let i = 0; i < pageIds.length; i++) {
    const id = pageIds[i]
    const value = block[id]?.value
    if (!value) {
      blockIdsNeedFetch.push(id)
    }
  }
  const fetchedBlocks = await fetchInBatches(blockIdsNeedFetch)
  block = Object.assign({}, block, fetchedBlocks)

  // 获取每篇文章基础数据
  for (let i = 0; i < pageIds.length; i++) {
    const id = pageIds[i]
    const value = block[id]?.value || fetchedBlocks[id]?.value
    const properties =
      (await getPageProperties(
        id,
        value,
        schema,
        null,
        getTagOptions(schema)
      )) || null

    if (properties) {
      collectionData.push(properties)
    }
  }

  // 站点配置优先读取配置表格，否则读取blog.config.js 文件
  const NOTION_CONFIG = (await getConfigMapFromConfigPage(collectionData)) || {}

  // 处理每一条数据的字段
  collectionData.forEach(function (element) {
    adjustPageProperties(element, NOTION_CONFIG)
  })

  // 站点基础信息
  const siteInfo = getSiteInfo({ collection, block, pageId })

  // 文章计数
  let postCount = 0

  // 查找所有的Post和Page
  const allPages = collectionData.filter(post => {
    if (post?.type === 'Post' && post.status === 'Published') {
      postCount++
    }
    return (
      post &&
      post?.slug &&
      //   !post?.slug?.startsWith('http') &&
      (post?.status === 'Invisible' || post?.status === 'Published')
    )
  })

  // Sort by date
  if (siteConfig('POSTS_SORT_BY', '', NOTION_CONFIG) === 'date') {
    allPages.sort((a, b) => {
      return b?.publishDate - a?.publishDate
    })
  }

  const notice = await getNotice(
    collectionData.filter(post => {
      return (
        post &&
        post?.type &&
        post?.type === 'Notice' &&
        post.status === 'Published'
      )
    })?.[0]
  )
  // 所有分类
  const categoryOptions = getAllCategories({
    allPages,
    categoryOptions: getCategoryOptions(schema)
  })
  // 所有标签
  const tagOptions = getAllTags({
    allPages,
    tagOptions: getTagOptions(schema),
    NOTION_CONFIG
  })
  // 旧的菜单
  const customNav = getCustomNav({
    allPages: collectionData.filter(
      post => post?.type === 'Page' && post.status === 'Published'
    )
  })
  // 新的菜单
  const customMenu = await getCustomMenu({ collectionData, NOTION_CONFIG })
  const latestPosts = getLatestPosts({ allPages, from, latestPostCount: 6 })
  const allNavPages = getNavPages({ allPages })

  return {
    NOTION_CONFIG,
    notice,
    siteInfo,
    allPages,
    allNavPages,
    collection,
    collectionQuery,
    collectionId,
    collectionView,
    viewIds,
    block,
    schema,
    tagOptions,
    categoryOptions,
    rawMetadata,
    customNav,
    customMenu,
    postCount,
    pageIds,
    latestPosts
  }
}

/**
 * 返回给浏览器前端的数据处理
 * 适当脱敏
 * 减少体积
 * 其它处理
 * @param {*} db
 */
function handleDataBeforeReturn(db) {
  // 清理多余数据
  delete db.block
  delete db.schema
  delete db.rawMetadata
  delete db.pageIds
  delete db.viewIds
  delete db.collection
  delete db.collectionQuery
  delete db.collectionId
  delete db.collectionView

  // 清理多余的块
  if (db?.notice) {
    db.notice = cleanBlock(db?.notice)
    delete db.notice?.id
  }

  db.tagOptions = cleanIds(db?.tagOptions)
  db.categoryOptions = cleanIds(db?.categoryOptions)
  db.customMenu = cleanIds(db?.customMenu)

  //   db.latestPosts = shortenIds(db?.latestPosts)
  db.allNavPages = shortenIds(db?.allNavPages)
  //   db.allPages = cleanBlocks(db?.allPages)

  db.allNavPages = cleanPages(db?.allNavPages, db.tagOptions)
  db.allPages = cleanPages(db.allPages, db.tagOptions)
  db.latestPosts = cleanPages(db.latestPosts, db.tagOptions)
  return db
}

/**
 * 处理文章列表中的异常数据
 * @param {Array} allPages - 所有页面数据
 * @param {Array} tagOptions - 标签选项
 * @returns {Array} 处理后的 allPages
 */
function cleanPages(allPages, tagOptions) {
  // 校验参数是否为数组
  if (!Array.isArray(allPages) || !Array.isArray(tagOptions)) {
    console.warn('Invalid input: allPages and tagOptions should be arrays.')
    return allPages || [] // 返回空数组或原始值
  }

  // 提取 tagOptions 中所有合法的标签名
  const validTags = new Set(
    tagOptions
      .map(tag => (typeof tag.name === 'string' ? tag.name : null))
      .filter(Boolean) // 只保留合法的字符串
  )

  // 遍历所有的 pages
  allPages.forEach(page => {
    // 确保 tagItems 是数组
    if (Array.isArray(page.tagItems)) {
      // 对每个 page 的 tagItems 进行过滤
      page.tagItems = page.tagItems.filter(
        tagItem =>
          validTags.has(tagItem?.name) && typeof tagItem.name === 'string' // 校验 tagItem.name 是否是字符串
      )
    }
  })

  return allPages
}

/**
 * 清理一组数据的id
 * @param {*} items
 * @returns
 */
function shortenIds(items) {
  if (items && Array.isArray(items)) {
    return deepClone(
      items.map(item => {
        item.short_id = getShortId(item.id)
        delete item.id
        return item
      })
    )
  }
  return items
}

/**
 * 清理一组数据的id
 * @param {*} items
 * @returns
 */
function cleanIds(items) {
  if (items && Array.isArray(items)) {
    return deepClone(
      items.map(item => {
        delete item.id
        return item
      })
    )
  }
  return items
}

/**
 * 清理block数据
 */
function cleanBlock(item) {
  const post = deepClone(item)
  const pageBlock = post?.blockMap?.block
  //   delete post?.id
  //   delete post?.blockMap?.collection

  if (pageBlock) {
    for (const i in pageBlock) {
      pageBlock[i] = cleanBlock(pageBlock[i])
      delete pageBlock[i]?.role
      delete pageBlock[i]?.value?.version
      delete pageBlock[i]?.value?.created_by_table
      delete pageBlock[i]?.value?.created_by_id
      delete pageBlock[i]?.value?.last_edited_by_table
      delete pageBlock[i]?.value?.last_edited_by_id
      delete pageBlock[i]?.value?.space_id
      delete pageBlock[i]?.value?.version
      delete pageBlock[i]?.value?.format?.copied_from_pointer
      delete pageBlock[i]?.value?.format?.block_locked_by
      delete pageBlock[i]?.value?.parent_table
      delete pageBlock[i]?.value?.copied_from_pointer
      delete pageBlock[i]?.value?.copied_from
      delete pageBlock[i]?.value?.created_by_table
      delete pageBlock[i]?.value?.created_by_id
      delete pageBlock[i]?.value?.last_edited_by_table
      delete pageBlock[i]?.value?.last_edited_by_id
      delete pageBlock[i]?.value?.permissions
      delete pageBlock[i]?.value?.alive
    }
  }
  return post
}

/**
 * 获取最新文章 根据最后修改时间倒序排列
 * @param {*}} param0
 * @returns
 */
function getLatestPosts({ allPages, from, latestPostCount }) {
  const allPosts = allPages?.filter(
    page => page.type === 'Post' && page.status === 'Published'
  )

  const latestPosts = Object.create(allPosts).sort((a, b) => {
    const dateA = new Date(a?.lastEditedDate || a?.publishDate)
    const dateB = new Date(b?.lastEditedDate || b?.publishDate)
    return dateB - dateA
  })
  return latestPosts.slice(0, latestPostCount)
}

/**
 * 获取用户自定义单页菜单
 * 旧版本，不读取Menu菜单，而是读取type=Page生成菜单
 * @param notionPageData
 * @returns {Promise<[]|*[]>}
 */
function getCustomNav({ allPages }) {
  const customNav = []
  if (allPages && allPages.length > 0) {
    allPages.forEach(p => {
      p.to = p.slug
      customNav.push({
        icon: p.icon || null,
        name: p.title,
        href: p.href,
        target: p.target,
        show: true
      })
    })
  }
  return customNav
}

/**
 * 获取自定义菜单
 * @param {*} allPages
 * @returns
 */
function getCustomMenu({ collectionData, NOTION_CONFIG }) {
  const menuPages = collectionData.filter(
    post =>
      post.status === 'Published' &&
      (post?.type === 'Menu' || post?.type === 'SubMenu')
  )
  const menus = []
  if (menuPages && menuPages.length > 0) {
    menuPages.forEach(e => {
      e.show = true
      if (e.type === 'Menu') {
        menus.push(e)
      } else if (e.type === 'SubMenu') {
        const parentMenu = menus[menus.length - 1]
        if (parentMenu) {
          if (parentMenu.subMenus) {
            parentMenu.subMenus.push(e)
          } else {
            parentMenu.subMenus = [e]
          }
        }
      }
    })
  }
  return menus
}

/**
 * 获取标签选项
 * @param schema
 * @returns {undefined}
 */
function getTagOptions(schema) {
  if (!schema) return {}
  const tagSchema = Object.values(schema).find(
    e => e.name === BLOG.NOTION_PROPERTY_NAME.tags
  )
  return tagSchema?.options || []
}

/**
 * 获取分类选项
 * @param schema
 * @returns {{}|*|*[]}
 */
function getCategoryOptions(schema) {
  if (!schema) return {}
  const categorySchema = Object.values(schema).find(
    e => e.name === BLOG.NOTION_PROPERTY_NAME.category
  )
  return categorySchema?.options || []
}

/**
 * 站点信息
 * @param notionPageData
 * @param from
 * @returns {Promise<{title,description,pageCover,icon}>}
 */
function getSiteInfo({ collection, block, NOTION_CONFIG }) {
  const defaultTitle = NOTION_CONFIG?.TITLE || BLOG.TITLE
  const defaultDescription = NOTION_CONFIG?.DESCRIPTION || BLOG.DESCRIPTION
  const defaultPageCover =
    NOTION_CONFIG?.HOME_BANNER_IMAGE || BLOG.HOME_BANNER_IMAGE
  const defaultIcon = NOTION_CONFIG?.AVATAR || BLOG.AVATAR
  const defaultLink = NOTION_CONFIG?.LINK || BLOG.LINK
  if (!collection && !block) {
    return {
      title: defaultTitle,
      description: defaultDescription,
      pageCover: defaultPageCover,
      icon: defaultIcon,
      link: defaultLink
    }
  }

  const title = collection?.name?.[0][0] || defaultTitle
  const description = collection?.description
    ? Object.assign(collection).description[0][0]
    : defaultDescription

  const pageCover = collection?.cover
    ? mapImgUrl(collection?.cover, collection, 'collection')
    : defaultPageCover

  // 用户头像压缩一下
  let icon = compressImage(
    collection?.icon
      ? mapImgUrl(collection?.icon, collection, 'collection')
      : defaultIcon
  )
  // 站点网址
  const link = NOTION_CONFIG?.LINK || defaultLink

  // 站点图标不能是emoji
  const emojiPattern = /\uD83C[\uDF00-\uDFFF]|\uD83D[\uDC00-\uDE4F]/g
  if (!icon || emojiPattern.test(icon)) {
    icon = defaultIcon
  }
  return { title, description, pageCover, icon, link }
}

/**
 * 获取导航用的精减文章列表
 * gitbook主题用到，只保留文章的标题分类标签分类信息，精减掉摘要密码日期等数据
 * 导航页面的条件，必须是Posts
 * @param {*} param0
 */
export function getNavPages({ allPages }) {
  const allNavPages = allPages?.filter(post => {
    return (
      post &&
      post?.slug &&
      post?.type === 'Post' &&
      post?.status === 'Published'
    )
  })

  return allNavPages.map(item => ({
    id: item.id,
    title: item.title || '',
    pageCoverThumbnail: item.pageCoverThumbnail || '',
    category: item.category || null,
    tags: item.tags || null,
    summary: item.summary || null,
    slug: item.slug,
    href: item.href,
    pageIcon: item.pageIcon || '',
    lastEditedDate: item.lastEditedDate,
    publishDate: item.publishDate,
    ext: item.ext || {}
  }))
}