我想找到用转义版本替换字符串中所有保留字符的最快方法。
我脑海中自然浮现出两种幼稚的方式(注意,保留字符集只是一个例子):
A:使用查找字典和 String.Replace
private Dictionary<string, string> _someEscapeTokens = new Dictionary<string, string>()
{
{"\t", @"\t"},
{"\n", @"\n"},
{"\r", @"\r"}
};
public string GetEscapedStringByNaiveLookUp(string s)
{
foreach (KeyValuePair<string, string> escapeToken in _someEscapeTokens.Where(kvp => s.Contains(kvp.Key)))
{
s = s.Replace(escapeToken.Key, escapeToken.Value);
}
return s;
}
B:遍历字符串中的每个字符
public string GetEscapedStringByTraversingCharArray(string s)
{
char[] chars = s.ToCharArray();
StringBuilder sb = new StringBuilder();
for (int i = 0; i < chars.Length; i++)
{
switch (chars[i])
{
case '\t':
sb.Append(@"\t"); break;
case '\n':
sb.Append(@"\n"); break;
case '\r':
sb.Append(@"\r"); break;
default:
sb.Append(chars[i]); break;
}
}
return sb.ToString();
}
正如我已经测试过的,版本 B 轻松优于第一个版本。
注意:我已经考虑过Regex.Escape,但由于字符集与我的不匹配,所以不适合。
但是,还有其他方法可以解决这个问题(考虑到性能)吗?
请参阅下面的代码。
测试是在两个不同的系统上完成的,目标是.NET Framework 4.0。无论如何,结果几乎是一样的:
Char Array (short string) average time: 38600 ns
Foreach (short string) average time: 26680 ns
Char Array (long string) average time: 48,1 ms
Foreach (long string) average time: 64,2 ms
Char Array (escaping only) average time: 13,6 ms
Foreach (escaping only) average time: 17,3 ms
这让我得出这样的结论:对于短字符串来说,
foreach
版本似乎稍快一些,但对于较长的字符串来说,它不知何故“下降”了。然而,我们在这里讨论的是非常小的差异。
测试代码:
private static void Main(string[] args)
{
//around 700 characters
string shortString = new StackTrace().ToString();
string longString;
string pureEscape;
//loading from a file with 1000000 words http://loremipsum.de/
using (StreamReader sr = new StreamReader(@"C:\users\ekrueger\desktop\LoremIpsum.txt"))
{
longString = sr.ReadToEnd();
}
//text file containing only escapable characters (length ~1000000)
using (StreamReader sr = new StreamReader(@"C:\users\ekrueger\desktop\PureEscape.txt"))
{
pureEscape = sr.ReadToEnd();
}
List<double> timesCharArrayShortString = new List<double>();
List<double> timesForeachShortString = new List<double>();
List<long> timesCharArrayLongString = new List<long>();
List<long> timesForeachLongString = new List<long>();
List<long> timesCharArrayPureEscape = new List<long>();
List<long> timesForeachPureEscape = new List<long>();
Stopwatch sw = new Stopwatch();
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringByTraversingCharArray(shortString);
sw.Stop();
timesCharArrayShortString.Add(sw.Elapsed.TotalMilliseconds * 1000000);
}
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringForeach(shortString);
sw.Stop();
timesForeachShortString.Add(sw.Elapsed.TotalMilliseconds * 1000000);
}
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringByTraversingCharArray(longString);
sw.Stop();
timesCharArrayLongString.Add(sw.ElapsedMilliseconds);
}
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringForeach(longString);
sw.Stop();
timesForeachLongString.Add(sw.ElapsedMilliseconds);
}
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringByTraversingCharArray(pureEscape);
sw.Stop();
timesCharArrayPureEscape.Add(sw.ElapsedMilliseconds);
}
for (int i = 0; i < 10; i++)
{
sw.Restart();
GetEscapedStringForeach(pureEscape);
sw.Stop();
timesForeachPureEscape.Add(sw.ElapsedMilliseconds);
}
Console.WriteLine("Char Array (short string) average time: {0} ns", timesCharArrayShortString.Average());
Console.WriteLine("Foreach (short string) average time: {0} ns", timesForeachShortString.Average());
Console.WriteLine("Char Array (long string) average time: {0} ms", timesCharArrayLongString.Average());
Console.WriteLine("Foreach (long string) average time: {0} ms", timesForeachLongString.Average());
Console.WriteLine("Char Array (escaping only) average time: {0} ms", timesCharArrayPureEscape.Average());
Console.WriteLine("Foreach (escaping only) average time: {0} ms", timesForeachPureEscape.Average());
Console.Read();
}
private static string GetEscapedStringByTraversingCharArray(string s)
{
if (String.IsNullOrEmpty(s))
return s;
char[] chars = s.ToCharArray();
StringBuilder sb = new StringBuilder(s.Length);
for (int i = 0; i < chars.Length; i++)
{
switch (chars[i])
{
case '\t':
sb.Append(@"\t"); break;
case '\n':
sb.Append(@"\n"); break;
case '\r':
sb.Append(@"\r"); break;
case '\f':
sb.Append(@"\f"); break;
default:
sb.Append(chars[i]); break;
}
}
return sb.ToString();
}
public static string GetEscapedStringForeach(string s)
{
if (String.IsNullOrEmpty(s))
return s;
StringBuilder sb = new StringBuilder(s.Length);
foreach (Char ch in s)
{
switch (ch)
{
case '\t':
sb.Append(@"\t"); break;
case '\n':
sb.Append(@"\n"); break;
case '\r':
sb.Append(@"\r"); break;
default:
sb.Append(ch); break;
}
}
return sb.ToString();
}
您无需将
string
转换为Char[]
,因此解决方案可以稍微改进:
public string GetEscapedStringByTraversingCharArray(string s) {
// do not forget about null...
if (String.IsNullOrEmpty(s))
return s;
// we can be sure, that it requires at least s.Length symbols, let's allocate then
// in case we want to trade memory for speed we can even put
// StringBuilder sb = new StringBuilder(2 * s.Length);
// for the worst case when all symbols should be escaped
StringBuilder sb = new StringBuilder(s.Length);
foreach(Char ch in s) {
switch (ch) {
case '\t':
sb.Append(@"\t");
break;
case '\n':
sb.Append(@"\n");
break;
case '\r':
sb.Append(@"\r");
break;
default:
sb.Append(ch);
break;
}
}
return sb.ToString();
}
第一个选项较慢是有道理的,因为您使用以下命令创建大量字符串对象:
s = s.Replace(escapeToken.Key, escapeToken.Value);
在第二种方法中,不需要创建 char[],因为 string 也有索引器。 也许提高性能唯一可以做的就是使用容量初始化 StringBuilder,这样就不需要调整大小。 您仍然可以在第二种方法中使用字典。