Discourse 导入 IPS4 脚本

协助完善私信的导入功能

已有针对旧版 IPB3 的代码,但在新版 IPS4 中,部分表名已被更改。

摘要

def import_private_messages
puts “”, “正在导入私信…”

topic_count = mysql_query("SELECT COUNT(msg_id) count FROM #{TABLE_PREFIX}message_posts").first["count"]

last_private_message_topic_id = -1

batches(BATCH_SIZE) do |offset|
  private_messages = mysql_query(<<-SQL
      SELECT msg_id pmtextid,
             msg_topic_id topic_id,
             msg_author_id fromuserid,
             mt_title title,
             msg_post message,
             
             mt_invited_members touserarray,
             
             mt_to_member_id to_user_id,
             msg_is_first_post first_post,
             msg_date dateline
        FROM #{TABLE_PREFIX}message_topics, #{TABLE_PREFIX}message_posts
       WHERE msg_topic_id = mt_id
         AND msg_date > UNIX_TIMESTAMP(STR_TO_DATE('#{IMPORT_AFTER}', '%Y-%m-%d'))
    ORDER BY msg_topic_id, msg_id
       LIMIT #{BATCH_SIZE}
      OFFSET #{offset}
  SQL
                                )

  puts "正在处理 #{private_messages.count} 条消息"
  break if private_messages.count < 1
  puts "正在处理 . . . "
  private_messages = private_messages.reject { |pm| @lookup.post_already_imported?("pm-#{pm['pmtextid']}") }

  title_username_of_pm_first_post = {}

  create_posts(private_messages, total: topic_count, offset: offset) do |m|
    skip = false
    mapped = {}

    mapped[:id] = "pm-#{m['pmtextid']}"
    mapped[:user_id] = user_id_from_imported_user_id(m['fromuserid']) || Discourse::SYSTEM_USER_ID
    mapped[:raw] = clean_up(m['message']) rescue nil
    mapped[:created_at] = Time.zone.at(m['dateline'])
    title = @htmlentities.decode(m['title']).strip[0...255]
    topic_id = nil

    next if mapped[:raw].blank?

    # 参与此私信的用户。
    target_usernames = []
    target_userids = []
    begin
      to_user_array = [ m['to_user_id'] ] + array_from_members_string(m['touserarray'])
    rescue
      puts "#{m['pmtextid']} -- #{m['touserarray']}"
      skip = true
    end

    begin
      to_user_array.each do |to_user|
        user_id = user_id_from_imported_user_id(to_user)
        username = User.find_by(id: user_id).try(:username)
        target_userids << user_id || Discourse::SYSTEM_USER_ID
        target_usernames << username if username
        if user_id
          puts "找到用户:#{to_user} -- #{user_id} -- #{username}"
        else
          puts "无法找到用户:#{to_user}"
        end
      end
    rescue
      puts "跳过 pm-#{m['pmtextid']},`to_user_array` 已损坏 -- #{to_user_array.inspect}"
      skip = true
    end

    participants = target_userids
    participants << mapped[:user_id]
    begin
      participants.sort!
    rescue
      puts "其中一位参与者的 ID 为 nil -- #{participants.inspect}"
    end

    if last_private_message_topic_id != m['topic_id']
      last_private_message_topic_id = m['topic_id']
      puts "新消息:#{m['topic_id']}: #{title} 来自 #{m['fromuserid']} (#{mapped[:user_id]})" unless QUIET
      # 主题帖消息
      topic_id = m['topic_id']
      mapped[:title] = title
      mapped[:archetype] = Archetype.private_message
      mapped[:target_usernames] = target_usernames.join(',')
      if mapped[:target_usernames].size < 1 # 给自己发私信?
        # skip = true
        mapped[:target_usernames] = "system"
        puts "pm-#{m['pmtextid']} 没有目标用户 (#{m['touserarray']})"
      end
    else # 回复
      topic_id = topic_lookup_from_imported_post_id("pm-#{topic_id}")
      if !topic_id
        skip = true
      end
      mapped[:topic_id] = topic_id
      puts "回复消息 #{topic_id}: #{m['topic_id']}: 来自 #{m['fromuserid']} (#{mapped[:user_id]})"  unless QUIET
    end
    #        puts "#{target_usernames} -- #{mapped[:target_usernames]}"
    #        puts "添加 #{mapped}"
    skip ? nil : mapped
    #        puts "#{'-'*50}> 已添加"
  end
end

这是当前表的样式:

问题出在 mt_invited_members 字段上。

如果在查询请求中移除这一行,消息可以导入,但回复链会丢失。我几乎不清楚需要进行哪些修改才能正确导入私信。
曾尝试将 mt_invited_members 替换为 mt_starter_id,但结果消息完全无法导入。

目前,我成功编写了两个从 IPS 4.5 迁移到 Discourse 的导入脚本中的一个。

用户
头像
私信(部分)
分类
论坛
话题

在导入附件时遇到了一些问题。脚本完成后,我肯定会将其公开分享给大众。

目录