原文:XML数据读取方式性能比较(一)
几个月来,疑被SOA,一直在和XML操作打交道,SQL差不多又忘光了。现在已经知道,至少有四种常用人XML数据操作方式(好像Java差不多),不过还没有实际比较过这些方式各有哪些特点或优劣。正好看到网上也没有这方面的实验,偶来总结一下。
测试开始先读取XML源,用一个比较大的RSS文件链接,复制到项目bin/debug目录下。
Stream xmlStream =new MemoryStream(File.ReadAllBytes(path));
一、XmlDocument 方式
代码
1 static IList testXmlDocument()
2 {
3 var doc =new XmlDocument();
4 doc.Load(xmlStream);
5 var nodeList = doc.DocumentElement.ChildNodes;
6 var lstChannel =new List<Object>(nodeList.Count );
7 foreach (XmlNode node in nodeList)
8 {
9 var channel =new
10 {
11 Title = node.SelectSingleNode("title").InnerText,
12 Link = node.SelectSingleNode("link").InnerText,
13 Description = node.SelectSingleNode("description").InnerText,
14 Content = node.SelectSingleNode("content").InnerText,
15 PubDate = node.SelectSingleNode("pubDate").InnerText,
16 Author = node.SelectSingleNode("author").InnerText,
17 Category = node.SelectSingleNode("category").InnerText
18 };
19 lstChannel.Add(channel);
20 }
21 return lstChannel;
22 }
二、XPathNavigator 方式
代码
1 static IList testXmlNavigator()
2 {
3 var doc =new XmlDocument();
4 doc.Load(xmlStream);
5 var nav = doc.CreateNavigator();
6 nav.MoveToRoot();
7 var nodeList = nav.Select("/channel/item");
8 var lstChannel =new List<Object>(nodeList.Count);
9 foreach (XPathNavigator node in nodeList)
10 {
11 var channel =new
12 {
13 Title = node.SelectSingleNode("title").Value,
14 Link = node.SelectSingleNode("link").Value,
15 Description = node.SelectSingleNode("description").Value,
16 Content = node.SelectSingleNode("content").Value,
17 PubDate = node.SelectSingleNode("pubDate").Value,
18 Author = node.SelectSingleNode("author").Value,
19 Category = node.SelectSingleNode("category").Value
20 };
21 lstChannel.Add(channel);
22 }
23 return lstChannel;
24 }
三、XmlTextReader 方式
代码
1 static List<Channel> testXmlReader()
2 {
3 var lstChannel =new List<Channel>();
4 var reader = XmlReader.Create(xmlStream);
5 while (reader.Read())
6 {
7 if (reader.Name =="item"&& reader.NodeType == XmlNodeType.Element)
8 {
9 var channel =new Channel();
10 lstChannel.Add(channel);
11 while (reader.Read())
12 {
13 if (reader.Name =="item") break;
14 if (reader.NodeType != XmlNodeType.Element) continue;
15 switch (reader.Name)
16 {
17 case"title":
18 channel.Title = reader.ReadString();
19 break;
20 case"link":
21 channel.Link = reader.ReadString();
22 break;
23 case"description":
24 channel.Description = reader.ReadString();
25 break;
26 case"content":
27 channel.Content = reader.ReadString();
28 break;
29 case"pubDate":
30 channel.PubDate = reader.ReadString();
31 break;
32 case"author":
33 channel.Author = reader.ReadString();
34 break;
35 case"category":
36 channel.Category = reader.ReadString();
37 break;
38 default:
39 break;
40 }
41 }
42 }
43 }
44 return lstChannel;
45 }
四、Linq to XML 方式
代码
1 static IList testXmlLinq()
2 {
3 var xd = XDocument.Load(xmlStream);
4 var list = from node in xd.Elements("channel").Descendants("item")
5 select new
6 {
7 Title = node.Element("title").Value,
8 Link = node.Element("link").Value,
9 Description = node.Element("description").Value,
10 Content = node.Element("content").Value,
11 PubDate = node.Element("pubDate").Value,
12 Author = node.Element("author").Value,
13 Category = node.Element("category").Value
14 };
15 return list.ToList();
16 }
测试结果:
XmlDocment | 47ms |
XPathNavigator | 42ms |
XmlTextReader | 23ms |
Xml Linq | 28ms |
小结一下自己的认识,XmlDocument的操作基本按W3C的DOM操作方式,不过要将全部节点解析成对象加载到内存中,往往造成很大浪费。所以微软自己的编程规范也不推荐用它。这里由于读取了所有节点,可能因此性能和Navigator方式相差不大。在三种随机读取方式中,Xml Linq性能最高,只是方法名有点别扭。XmlTextReader方式是所谓的SAX,只读向前,无疑性能最高,不过实现上麻烦了不少,要比较精确的控制访问逻辑,也无法用匿名类存储数据。
.Net 3.5发布Xml Linq可以很好地取代前两种方式,通常情况下,最好用它。只有个别场合,如果对性能要求极高,或者读取Xml数据量太大不能一下子下载或读取到内存中,那就只好痛苦委身于XmlTextReader了。