-- 从反向索引解析多维度计数(用户/客户端/每用户客户端) -- KEYS[1] = affinity_rev:{groupID}:{accountID} -- ARGV[1] = 过期阈值时间戳 (now - ttl) -- ARGV[2] = 目标 userID(传 "" 则不计算 perUserClients) -- 返回: {totalMembers, uniqueUsers, uniqueClients, perUserClients} redis.call('ZREMRANGEBYSCORE', KEYS[1], '-inf', ARGV[1]) local members = redis.call('ZRANGE', KEYS[1], 0, -1) local total = #members if total == 0 then return {0, 0, 0, 0} end local target_user = ARGV[2] local users = {} local clients = {} local user_count = 0 local client_count = 0 local per_user_count = 0 for _, member in ipairs(members) do local sep = string.find(member, '/', 1, true) if sep then local uid = string.sub(member, 1, sep - 1) local cid = string.sub(member, sep + 1) if not users[uid] then users[uid] = true user_count = user_count + 1 end if not clients[cid] then clients[cid] = true client_count = client_count + 1 end if target_user ~= '' and uid == target_user then per_user_count = per_user_count + 1 end end end return {total, user_count, client_count, per_user_count}