/**
 * 卡若创业派对 - 内容解析工具
 * 解析 TipTap HTML 为阅读页可展示的 segments
 *
 * segment 类型：
 *   { type: 'text',    text }
 *   { type: 'mention', userId, nickname }        — @某人，点击加好友（提交存客宝见 utils/soulBridge.submitCkbLead）
 *   { type: 'linkTag', label, url, ... }         — #链接标签，点击跳转（阅读页 onLinkTagTap：外链→link-preview、小程序→navigateToMiniProgram）
 *   { type: 'image',   src, alt }               — 图片
 *   { type: 'video',   src }                    — 正文内嵌视频（管理端 rich-video-wrap / <video>）
 */

/** 判断内容是否为 HTML */
function isHtmlContent(content) {
  if (!content || typeof content !== 'string') return false
  const trimmed = content.trim()
  return trimmed.includes('<') && trimmed.includes('>') && /<[a-z][^>]*>/i.test(trimmed)
}

/** 解码常见 HTML 实体 */
function decodeEntities(str) {
  return str
    .replace(/&nbsp;/g, ' ')
    .replace(/&amp;/g, '&')
    .replace(/&lt;/g, '<')
    .replace(/&gt;/g, '>')
    .replace(/&quot;/g, '"')
    .replace(/&#39;/g, "'")
}

/**
 * 正文里 img 的 src：管理端常见为 /uploads/...（相对 API 根），浏览器会自动补全域名；
 * 小程序 <image> 不会拼接 baseUrl，需在此用 assetBase 转成可请求的绝对地址。
 * @param {string} src
 * @param {string} [assetBase] - 如 https://soulapi.example.com（无末尾 /）
 */
function resolveArticleImageSrc(src, assetBase) {
  if (!src || typeof src !== 'string') return src
  const s = src.trim()
  if (!s) return s
  if (/^(https?:|wxfile:|data:|blob:)/i.test(s)) return s
  if (s.startsWith('//')) return 'https:' + s
  if (s.startsWith('/')) {
    const base = String(assetBase || '').replace(/\/$/, '')
    if (!base) return s
    return base + s
  }
  return s
}

/** 从 <img ...> 标签取出 src（双引号/单引号） */
function extractImgSrcFromTag(tag) {
  const d = tag.match(/src\s*=\s*"([^"]*)"/i)
  if (d && d[1] != null) return d[1]
  const q = tag.match(/src\s*=\s*'([^']*)'/i)
  if (q && q[1] != null) return q[1]
  return ''
}

/**
 * 单行展示用：昵称、#标签文案、章节外标题类字段 — 合并换行、<br>、连续空白（避免 TipTap/粘贴带入异常断行）
 */
function cleanSingleLineField(s) {
  if (!s && s !== 0) return ''
  let t = decodeEntities(String(s))
    .replace(/<br\s*\/?>/gi, ' ')
    .replace(/\r\n|\r|\n/g, ' ')
    .replace(/[\s\u00a0\u200b\u200c\u200d\ufeff\u3000]+/g, ' ')
    .trim()
  return t
}

/** @提及昵称：去首尾空白、零宽、全角空格；合并内部换行/<br> */
function cleanMentionNickname(n) {
  return cleanSingleLineField(n)
}

/** 纯文本在 mention 节点前若已有「@」，去掉末尾 @，避免渲染成「找@@阿浪」 */
function stripTrailingAtForMention(before) {
  return before.replace(/[@＠][\s\u00a0\u200b]*$/u, '')
}

/**
 * 将一个 HTML block 字符串解析为 segments 数组
 * 处理三种内联元素：mention / linkTag(span) / linkTag(a) / img
 */
function parseBlockToSegments(block, config) {
  const segs = []
  const normalize = s => (s || '').trim().toLowerCase()
  const personTokenSet = new Set()
  for (const p of ((config && config.persons) || [])) {
    const token = normalize((p && p.token) || '')
    if (token) personTokenSet.add(token)
  }
  // 合并匹配所有内联元素
  const tokenRe = /<span[^>]*data-type="mention"[^>]*>[\s\S]*?<\/span>|<span[^>]*data-type="linkTag"[^>]*>[\s\S]*?<\/span>|<a[^>]*href="([^"]*)"[^>]*>(#[^<]*)<\/a>|<img[^>]*\/?>/gi
  let lastEnd = 0
  let m

  while ((m = tokenRe.exec(block)) !== null) {
    // 前置纯文本（mention 紧挨手写「找@」时去掉重复 @）
    let before = decodeEntities(block.slice(lastEnd, m.index).replace(/<[^>]+>/g, ''))
    const tag = m[0]
    if (/data-type="mention"/i.test(tag)) {
      before = stripTrailingAtForMention(before)
    }
    if (before.trim()) segs.push({ type: 'text', text: before })

    if (/data-type="mention"/i.test(tag)) {
      // @mention — TipTap mention span（span 内常见「@ 昵称」多空格，统一紧挨显示）
      const idMatch = tag.match(/data-id="([^"]*)"/)
      const labelMatch = tag.match(/data-label="([^"]*)"/)
      const innerText = tag.replace(/<[^>]+>/g, '')
      const userId = idMatch ? idMatch[1].trim() : ''
      let nickname = labelMatch ? labelMatch[1] : innerText.replace(/^[@＠]\s*/, '')
      nickname = cleanMentionNickname((nickname || '').trim())
      const userExists = !!normalize(userId) && personTokenSet.has(normalize(userId))
      if (userExists && nickname) {
        segs.push({ type: 'mention', userId, nickname, mentionDisplay: '@' + nickname })
      } else if (nickname) {
        // 被 @ 人物不存在时降级为普通文本，保持“静态 @某人”展示
        segs.push({ type: 'text', text: '@' + nickname })
      }

    } else if (/data-type="linkTag"/i.test(tag)) {
      // #linkTag — 自定义 span 格式（data-type="linkTag" data-url="..." data-tag-type="..." data-page-path="..." data-app-id="..."）
      const urlMatch      = tag.match(/data-url="([^"]*)"/)
      const tagTypeMatch  = tag.match(/data-tag-type="([^"]*)"/)
      const pagePathMatch = tag.match(/data-page-path="([^"]*)"/)
      const tagIdMatch    = tag.match(/data-tag-id="([^"]*)"/)
      const appIdMatch    = tag.match(/data-app-id="([^"]*)"/)
      const mpKeyMatch    = tag.match(/data-mp-key="([^"]*)"/)
      const passPhoneMatch = tag.match(/data-pass-phone="([^"]*)"/)
      const phoneParamMatch = tag.match(/data-phone-param-name="([^"]*)"/)
      const innerText     = cleanSingleLineField(tag.replace(/<[^>]+>/g, '').replace(/^#/, ''))
      const url           = urlMatch      ? urlMatch[1]      : ''
      const tagType       = tagTypeMatch  ? tagTypeMatch[1]  : 'url'
      const pagePath      = pagePathMatch ? pagePathMatch[1] : ''
      const tagId         = tagIdMatch    ? tagIdMatch[1]    : ''
      const appId         = appIdMatch    ? appIdMatch[1]   : ''
      const mpKey         = mpKeyMatch    ? mpKeyMatch[1]   : ''
      const passPhone = !!(passPhoneMatch && (passPhoneMatch[1] === '1' || /^true$/i.test(passPhoneMatch[1])))
      const phoneParamName = (phoneParamMatch && phoneParamMatch[1]) ? phoneParamMatch[1] : 'phone'
      segs.push({ type: 'linkTag', label: innerText || '#', url, tagType, pagePath, tagId, appId, mpKey, passPhone, phoneParamName })

    } else if (/^<a /i.test(tag)) {
      // #linkTag — 旧格式 <a href>（insertLinkTag 旧版产生，url 可能为空）
      // m[1] = href, m[2] = innerText（以 # 开头）
      const url   = m[1] || ''
      const label = cleanSingleLineField((m[2] || '').replace(/^#/, ''))
      // 旧格式没有 tagType，在 onLinkTagTap 中会按 label 匹配缓存的 linkTags 配置降级处理
      segs.push({ type: 'linkTag', label: label || '#', url, tagType: '', pagePath: '', tagId: '' })

    } else if (/^<img /i.test(tag)) {
      // 图片（src 可能为相对路径，需结合 config.assetBase）
      const rawSrc = extractImgSrcFromTag(tag)
      const altMatch = tag.match(/alt\s*=\s*"([^"]*)"/i) || tag.match(/alt\s*=\s*'([^']*)'/i)
      if (rawSrc) {
        const decoded = decodeEntities(rawSrc)
        const src =
          config && config.assetBase
            ? resolveArticleImageSrc(decoded, config.assetBase)
            : resolveArticleImageSrc(decoded, '')
        segs.push({ type: 'image', src, alt: altMatch ? decodeEntities(altMatch[1]) : '' })
      }
    }

    lastEnd = m.index + tag.length
  }

  // 尾部纯文本
  const after = decodeEntities(block.slice(lastEnd).replace(/<[^>]+>/g, ''))
  if (after.trim()) segs.push({ type: 'text', text: after })

  return segs
}

/**
 * 解析 <table> HTML 为结构化数据 { headers: string[], rows: string[][] }
 */
function parseTableHtml(tableHtml) {
  const stripTags = s => decodeEntities((s || '').replace(/<[^>]+>/g, '').trim())
  const headers = []
  const rows = []

  const headMatch = tableHtml.match(/<thead[^>]*>([\s\S]*?)<\/thead>/i)
  if (headMatch) {
    const thRe = /<th[^>]*>([\s\S]*?)<\/th>/gi
    let m
    while ((m = thRe.exec(headMatch[1])) !== null) {
      headers.push(stripTags(m[1]))
    }
  }

  const bodyMatch = tableHtml.match(/<tbody[^>]*>([\s\S]*?)<\/tbody>/i)
  const bodyContent = bodyMatch ? bodyMatch[1] : tableHtml
  const trRe = /<tr[^>]*>([\s\S]*?)<\/tr>/gi
  let trM
  let isFirst = !headMatch
  while ((trM = trRe.exec(bodyContent)) !== null) {
    if (headMatch && trM[0] === (headMatch[0].match(/<tr[^>]*>[\s\S]*?<\/tr>/i) || [])[0]) continue
    const cells = []
    const cellRe = /<t[dh][^>]*>([\s\S]*?)<\/t[dh]>/gi
    let cM
    while ((cM = cellRe.exec(trM[1])) !== null) {
      cells.push(stripTags(cM[1]))
    }
    if (!cells.length) continue
    if (isFirst && !headers.length) {
      headers.push(...cells)
      isFirst = false
    } else {
      rows.push(cells)
    }
  }
  return { headers, rows }
}

/**
 * 从 HTML 中解析出 lines（纯文本行）和 segments（含富文本片段）
 * segment 新增类型：heading / quote / listItem
 * @param {string} html
 * @param {object} [config] - { persons: [], linkTags: [], assetBase?: string }
 */
function parseHtmlToSegments(html, config) {
  const lines = []
  const segments = []

  // 0. 提取 <table> 块
  const tables = []
  let text = html.replace(/<table[^>]*>[\s\S]*?<\/table>/gi, (match) => {
    const idx = tables.length
    tables.push(parseTableHtml(match))
    return '\n__TABLE_' + idx + '__\n'
  })

  // 1. 提取 <h2>/<h3> → heading 占位
  const headings = []
  text = text.replace(/<h([2-6])[^>]*>([\s\S]*?)<\/h\1>/gi, function (_, lvl, inner) {
    const idx = headings.length
    headings.push({ level: parseInt(lvl, 10), raw: inner.trim() })
    return '\n__H_' + idx + '__\n'
  })

  // 2. 提取 <blockquote> → quote 占位
  const quotes = []
  text = text.replace(/<blockquote[^>]*>([\s\S]*?)<\/blockquote>/gi, function (_, inner) {
    const idx = quotes.length
    quotes.push(inner.trim())
    return '\n__Q_' + idx + '__\n'
  })

  // 2.5 内嵌视频：去掉管理端「rich-video-caption」说明（仅编辑器用）；整段 <video> 抽成占位
  // 须在列表拆解之前执行，否则 li/段落内 video 会在 parseBlockToSegments 里被 strip 成纯文本
  const videos = []
  text = text.replace(/<div[^>]*\brich-video-caption\b[^>]*>[\s\S]*?<\/div>/gi, '')
  text = text.replace(/<video\b[^>]*>[\s\S]*?<\/video>/gi, function (match) {
    const sm = match.match(/\bsrc\s*=\s*"([^"]*)"/i) || match.match(/\bsrc\s*=\s*'([^']*)'/i)
    const rawSrc = sm ? String(sm[1] || '').trim() : ''
    if (!rawSrc) return match
    const decoded = decodeEntities(rawSrc)
    const src =
      config && config.assetBase
        ? resolveArticleImageSrc(decoded, config.assetBase)
        : resolveArticleImageSrc(decoded, '')
    const idx = videos.length
    videos.push({ src })
    return '\n__VIDEO_' + idx + '__\n'
  })

  // 3. 列表 → listItem 占位（保留 ordered 标记）
  var olDepth = 0
  var olCounter = 0
  text = text.replace(/<ol[^>]*>/gi, function () { olDepth++; olCounter = 0; return '\n' })
  text = text.replace(/<\/ol>/gi, function () { olDepth = Math.max(0, olDepth - 1); return '\n' })
  text = text.replace(/<ul[^>]*>/gi, '\n')
  text = text.replace(/<\/ul>/gi, '\n')
  text = text.replace(/<li[^>]*>([\s\S]*?)<\/li>/gi, function (_, inner) {
    var cleaned = inner.replace(/<p[^>]*>/gi, '').replace(/<\/p>/gi, '').trim()
    if (!cleaned) return '\n'
    // 列表项内仅有视频占位时，提升为独立视频块（避免 __VIDEO_n__ 当正文展示）
    var onlyVid = cleaned.match(/^__VIDEO_(\d+)__$/)
    if (onlyVid) {
      return '\n__VIDEO_' + onlyVid[1] + '__\n'
    }
    if (olDepth > 0) {
      olCounter++
      return '\n__LI_O_' + olCounter + '__ ' + cleaned + '\n'
    }
    return '\n__LI_U__ ' + cleaned + '\n'
  })

  // 4. 块级标签换行
  text = text.replace(/<\/p>\s*<p[^>]*>/gi, '\n\n')
  text = text.replace(/<p[^>]*>/gi, '')
  text = text.replace(/<\/p>/gi, '\n')
  text = text.replace(/<div[^>]*>/gi, '')
  text = text.replace(/<\/div>/gi, '\n')
  text = text.replace(/<br\s*\/?>/gi, '\n')
  text = text.replace(/<hr\s*\/?>/gi, '\n')

  // 5. 逐段解析
  var blocks = text.split(/\n+/)
  for (var bi = 0; bi < blocks.length; bi++) {
    var block = blocks[bi]
    if (!block.trim()) continue

    // table
    var tblM = block.trim().match(/^__TABLE_(\d+)__$/)
    if (tblM) {
      var tbl = tables[parseInt(tblM[1], 10)]
      if (tbl) {
        lines.push([tbl.headers.join(' | '), ...tbl.rows.map(function (r) { return r.join(' | ') })].join('\n'))
        segments.push([{ type: 'table', headers: tbl.headers, rows: tbl.rows }])
      }
      continue
    }

    // video
    var vidM = block.trim().match(/^__VIDEO_(\d+)__$/)
    if (vidM) {
      var vid = videos[parseInt(vidM[1], 10)]
      if (vid && vid.src) {
        lines.push('')
        segments.push([{ type: 'video', src: vid.src }])
      }
      continue
    }

    // heading
    var hM = block.trim().match(/^__H_(\d+)__$/)
    if (hM) {
      var h = headings[parseInt(hM[1], 10)]
      if (h) {
        var hText = decodeEntities(h.raw.replace(/<[^>]+>/g, '').trim())
        lines.push(hText)
        segments.push([{ type: 'heading', level: h.level, text: hText }])
      }
      continue
    }

    // quote
    var qM = block.trim().match(/^__Q_(\d+)__$/)
    if (qM) {
      var qHtml = quotes[parseInt(qM[1], 10)]
      if (qHtml) {
        var qText = decodeEntities(
          qHtml.replace(/<\/?p[^>]*>/gi, '\n').replace(/<[^>]+>/g, '').trim()
        ).replace(/\n{2,}/g, '\n')
        lines.push(qText)
        segments.push([{ type: 'quote', text: qText }])
      }
      continue
    }

    // ordered list item
    var liOM = block.match(/^__LI_O_(\d+)__\s*(.*)/)
    if (liOM) {
      var liNum = parseInt(liOM[1], 10)
      var liContent = liOM[2].trim()
      var liSegs = parseBlockToSegments(liContent, config)
      var liText = decodeEntities(liContent.replace(/<[^>]+>/g, '').trim())
      lines.push(liNum + '. ' + liText)
      segments.push([{ type: 'listItem', ordered: true, number: liNum, text: liText, segs: liSegs }])
      continue
    }

    // unordered list item
    var liUM = block.match(/^__LI_U__\s*(.*)/)
    if (liUM) {
      var liContent2 = liUM[1].trim()
      var liSegs2 = parseBlockToSegments(liContent2, config)
      var liText2 = decodeEntities(liContent2.replace(/<[^>]+>/g, '').trim())
      lines.push(liText2)
      segments.push([{ type: 'listItem', ordered: false, text: liText2, segs: liSegs2 }])
      continue
    }

    // 普通段落
    var blockSegs = parseBlockToSegments(block, config)
    if (!blockSegs.length) continue

    if (blockSegs.length === 1 && blockSegs[0].type === 'image') {
      lines.push('')
      segments.push(blockSegs)
      continue
    }

    if (config && (config.persons?.length || config.linkTags?.length)) {
      var expanded = []
      for (var si = 0; si < blockSegs.length; si++) {
        var seg = blockSegs[si]
        if (seg.type === 'text' && seg.text) {
          expanded.push.apply(expanded, matchLineToSegments(seg.text, config))
        } else {
          expanded.push(seg)
        }
      }
      blockSegs = expanded
    }

    var lineText = decodeEntities(block.replace(/<[^>]+>/g, '')).trim()
    lines.push(lineText)
    segments.push(blockSegs)
  }

  return { lines, segments }
}

/** 清理 Markdown 格式标记（**加粗** *斜体* __加粗__ _斜体_ ~~删除线~~ `代码` 等）*/
function stripMarkdownFormatting(text) {
  if (!text) return text
  let s = text
  s = s.replace(/^#{1,6}\s+/gm, '')
  s = s.replace(/\*\*(.+?)\*\*/g, '$1')
  s = s.replace(/__(.+?)__/g, '$1')
  s = s.replace(/(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)/g, '$1')
  s = s.replace(/(?<!_)_(?!_)(.+?)(?<!_)_(?!_)/g, '$1')
  s = s.replace(/~~(.+?)~~/g, '$1')
  s = s.replace(/`([^`]+)`/g, '$1')
  s = s.replace(/^>\s+/gm, '')
  s = s.replace(/^---$/gm, '')
  s = s.replace(/^\* /gm, '• ')
  s = s.replace(/^- /gm, '• ')
  s = s.replace(/^\d+\.\s/gm, '')
  return s
}

/**
 * 对一行纯文本进行 @人名 / #标签 自动匹配，返回 segments 数组
 * config: { persons: [{ personId, token, name, label, aliases }], linkTags: [...] }
 * 点击加好友时须传 persons.token（与 CKB lead 的 targetUserId 一致），不能用 personId。
 */
function matchLineToSegments(line, config) {
  if (!config || (!config.persons?.length && !config.linkTags?.length)) {
    return [{ type: 'text', text: line }]
  }
  // 编辑器/系统在 @ 与人名之间插入的普通空格，合并为紧挨 @（避免「找@ 阿浪」无法匹配人名）
  line = line.replace(/([@＠])\s+(?=[\u4e00-\u9fffA-Za-z0-9_\u00b7])/g, '$1')
  const normalize = s => (s || '').trim().toLowerCase()
  const personMap = {}
  const tagMap = {}
  for (const p of (config.persons || [])) {
    const token = (p.token || '').trim()
    if (!token) continue
    const display = (p.name || p.label || '').trim()
    const aliasStr = p.aliases != null ? String(p.aliases) : ''
    const keys = [display, p.label, ...(aliasStr ? aliasStr.split(',') : [])]
      .map((x) => (x != null ? String(x) : '').trim())
      .filter(Boolean)
      .map(normalize)
      .filter(Boolean)
    for (const k of keys) {
      if (!personMap[k]) personMap[k] = p
    }
  }
  for (const t of (config.linkTags || [])) {
    const keys = [t.label, ...(t.aliases ? t.aliases.split(',') : [])].map(normalize).filter(Boolean)
    for (const k of keys) { if (!tagMap[k]) tagMap[k] = t }
  }
  const esc = n => n.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
  const personNames = Object.keys(personMap).sort((a, b) => b.length - a.length).map(esc)
  const tagLabels = Object.keys(tagMap).sort((a, b) => b.length - a.length).map(esc)
  if (!personNames.length && !tagLabels.length) return [{ type: 'text', text: line }]

  const parts = []
  if (personNames.length) parts.push('[@＠]\\s*(' + personNames.join('|') + ')')
  if (tagLabels.length) parts.push('[#＃]\\s*(' + tagLabels.join('|') + ')')
  const pattern = new RegExp(parts.join('|'), 'gi')

  const segs = []
  let lastEnd = 0
  let m
  while ((m = pattern.exec(line)) !== null) {
    if (m.index > lastEnd) {
      segs.push({ type: 'text', text: line.slice(lastEnd, m.index) })
    }
    const full = m[0]
    if (/^[@＠]/u.test(full)) {
      const body = full.replace(/^[@＠]\s*/u, '')
      const person = personMap[normalize(body)]
      if (person) {
        const nick = cleanSingleLineField(person.name || person.label || body)
        const uid = (person.token || '').trim()
        if (uid) {
          segs.push({ type: 'mention', userId: uid, nickname: nick, mentionDisplay: '@' + nick })
        } else {
          segs.push({ type: 'text', text: full })
        }
      } else {
        segs.push({ type: 'text', text: full })
      }
    } else {
      const body = full.replace(/^[#＃]\s*/u, '')
      const tag = tagMap[normalize(body)]
      if (tag) {
        segs.push({
          type: 'linkTag',
          label: tag.label || body,
          url: tag.url || '',
          tagType: tag.type || 'url',
          pagePath: tag.pagePath || '',
          tagId: tag.tagId || '',
          appId: tag.appId || '',
          mpKey: tag.mpKey || '',
          passPhone: !!tag.passPhone,
          phoneParamName: tag.phoneParamName || 'phone'
        })
      } else {
        segs.push({ type: 'text', text: full })
      }
    }
    lastEnd = m.index + full.length
  }
  if (lastEnd < line.length) {
    segs.push({ type: 'text', text: line.slice(lastEnd) })
  }
  return segs.length ? segs : [{ type: 'text', text: line }]
}

/** 纯文本/Markdown 按行解析 */
function parsePlainTextToSegments(text, config) {
  const cleaned = stripMarkdownFormatting(text)
  const lines = cleaned.split('\n').map(l => l.trim()).filter(l => l.length > 0)
  const segments = lines.map(line => matchLineToSegments(line, config))
  return { lines, segments }
}

/** 清理残留的 Markdown 图片引用文本（如 "image.png![](xxx)" ） */
function stripOrphanImageRefs(text) {
  if (!text) return text
  text = text.replace(/[^\s]*\.(?:png|jpg|jpeg|gif|webp|svg|bmp)!\[[^\]]*\]\([^)]*\)/gi, '')
  text = text.replace(/!\[[^\]]*\]\([^)]*\)/g, '')
  return text
}

/**
 * 将原始内容解析为 contentSegments（用于阅读页展示）
 * @param {string} rawContent
 * @param {object} [config] - { persons: [], linkTags: [] }
 * @returns {{ lines: string[], segments: Array<Array<segment>> }}
 */
function parseContent(rawContent, config) {
  if (!rawContent || typeof rawContent !== 'string') {
    return { lines: [], segments: [] }
  }
  let content = stripOrphanImageRefs(rawContent)
  if (isHtmlContent(content)) {
    return parseHtmlToSegments(content, config)
  }
  return parsePlainTextToSegments(content, config)
}

module.exports = {
  parseContent,
  isHtmlContent,
  cleanSingleLineField,
}