C# 使用Queue<T>代替递归算法遍历树
递归时候每次调用自身在堆栈上要记录返回地址,而堆栈的空间很少,调用次数多了后会产生堆栈溢出,以下代码是实际项目中,通过Queue<T>来避免递归算法的代码:
/// <summary>
/// 获取某个节点下特定属性的所有子孙节点
/// </summary>
/// <param name="groupId"></param>
/// <returns></returns>
public IList<OfficeGroupNodeDto> GetSelfAndChildOfficesByGroupId(int groupId)
{
Func<int, string, BusinessType, int, string, OfficeGroupNodeDto> createGroupNodeFunc =
(id, name, bizType, parentId, parentName) =>
new OfficeGroupNodeDto()
{
OfficeId = id,
Name = name,
BizType = bizType,
ParentId = parentId,
ParentName = parentName
};
var offices = GetTenantOffices().ToArray();
Func<int, OfficeDto[]> getChildOfficesFunc = id => offices.Where(x => x.ParentId == id).ToArray();
//创建队列
var groupNodeCacheQueue = new Queue<OfficeGroupNodeDto>();
var currentGroup = groupId == 0 || groupId == Office.AdminOffice.Id
? ToDto(Office.AdminOffice)
: offices.FirstOrDefault(x => x.Id == groupId && x.OfficeType == OfficeType.Group);
if (currentGroup == null) return null;
var officeGroupNodeDto =
createGroupNodeFunc(currentGroup.Id, currentGroup.Abbreviation, currentGroup.BizType, 0, string.Empty);
//初始进入队列一个元素
groupNodeCacheQueue.Enqueue(officeGroupNodeDto);
//循环读取队列内元素,直到读取完
while (groupNodeCacheQueue.Count > 0)
{
//出队列一个元素节点
var currentGroupNode = groupNodeCacheQueue.Dequeue();
//获取该节点的所有孩子节点
var childOffices = getChildOfficesFunc(currentGroupNode.OfficeId);
currentGroupNode.IsGroup = true;
var hasChildren = childOffices.SafeAny();
currentGroupNode.HasChildren = hasChildren;
if (!hasChildren) continue;
//遍历当前节点的孩子节点
foreach (var office in childOffices)
{
if (office.BizType == BusinessType.FranchiseChain) continue;
//创建符合条件的节点
var newNode = createGroupNodeFunc(office.Id, office.Abbreviation, BusinessType.RegularChain,
currentGroupNode.OfficeId, currentGroupNode.Name);
currentGroupNode.Items.Add(newNode);
if (office.OfficeType == OfficeType.Office) continue;
//把还有子节点的孩子元素节点到队列,待下次循环继续
groupNodeCacheQueue.Enqueue(newNode);
}
}
return new List<OfficeGroupNodeDto>() { officeGroupNodeDto };
}
注:主要看注释部分的总体思路,其他次要细节不用太关注
当然这里也可以用其他数据结构如Stack<T>,根据实际需要选择,如有没有顺序要求。