最初的问题是“去除非法字符”:

public string RemoveInvalidChars(string filename)
{
    return string.Concat(filename.Split(Path.GetInvalidFileNameChars()));
}

相反，你可能想要替换它们:

public string ReplaceInvalidChars(string filename)
{
    return string.Join("_", filename.Split(Path.GetInvalidFileNameChars()));    
}

这个答案在Ceres的另一个帖子里，我真的很喜欢它的简洁。

这些都是很好的解决方案，但它们都依赖于Path。getinvalidfilenamecars，它可能不像您想象的那么可靠。请注意MSDN文档中关于Path的以下注释。GetInvalidFileNameChars:

此方法返回的数组不保证包含文件和目录名中无效的完整字符集。完整的无效字符集可能因文件系统而异。例如，在基于windows的桌面平台上，无效路径字符可能包括ASCII/Unicode字符1到31，以及引号(")、小于(<)、大于(>)、管道(|)、退格(\b)、null(\0)和制表符(\t)。

Path的情况也好不到哪里去。GetInvalidPathChars方法。它包含了完全相同的评论。

这是我的一点小小的贡献。在同一字符串中替换而不创建新字符串或stringbuilder的方法。它快速，易于理解，是本文中所有提到的一个很好的选择。

private static HashSet<char> _invalidCharsHash;
private static HashSet<char> InvalidCharsHash
{
  get { return _invalidCharsHash ?? (_invalidCharsHash = new HashSet<char>(Path.GetInvalidFileNameChars())); }
}

private static string ReplaceInvalidChars(string fileName, string newValue)
{
  char newChar = newValue[0];

  char[] chars = fileName.ToCharArray();
  for (int i = 0; i < chars.Length; i++)
  {
    char c = chars[i];
    if (InvalidCharsHash.Contains(c))
      chars[i] = newChar;
  }

  return new string(chars);
}

你可以这样调用它:

string illegal = "\"M<>\"\\a/ry/ h**ad:>> a\\/:*?\"<>| li*tt|le|| la\"mb.?";
string legal = ReplaceInvalidChars(illegal);

并返回:

_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

值得注意的是，该方法总是将无效字符替换为给定值，但不会删除它们。如果你想删除无效字符，这个替代方法可以做到:

private static string RemoveInvalidChars(string fileName, string newValue)
{
  char newChar = string.IsNullOrEmpty(newValue) ? char.MinValue : newValue[0];
  bool remove = newChar == char.MinValue;

  char[] chars = fileName.ToCharArray();
  char[] newChars = new char[chars.Length];
  int i2 = 0;
  for (int i = 0; i < chars.Length; i++)
  {
    char c = chars[i];
    if (InvalidCharsHash.Contains(c))
    {
      if (!remove)
        newChars[i2++] = newChar;
    }
    else
      newChars[i2++] = c;

  }

  return new string(newChars, 0, i2);
}

基准

如果性能是你所追求的，我用这篇文章中发现的大多数方法执行了定时测试运行。其中一些方法不会替换为给定的char，因为OP要求清除字符串。我添加了用给定字符替换的测试，如果您的预期场景只需要删除不需要的字符，还添加了一些用空字符替换的测试。用于此基准测试的代码位于末尾，因此您可以运行自己的测试。

注意:方法Test1和Test2都在本文中提出。

第一次运行

替换为'_'，1000000次迭代

结果:

============Test1===============
Elapsed=00:00:01.6665595
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test2===============
Elapsed=00:00:01.7526835
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test3===============
Elapsed=00:00:05.2306227
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test4===============
Elapsed=00:00:14.8203696
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test5===============
Elapsed=00:00:01.8273760
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test6===============
Elapsed=00:00:05.4249985
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test7===============
Elapsed=00:00:07.5653833
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test8===============
Elapsed=00:12:23.1410106
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test9===============
Elapsed=00:00:02.1016708
Result=_M ____a_ry_ h__ad___ a_________ li_tt_le__ la_mb._

============Test10===============
Elapsed=00:00:05.0987225
Result=M ary had a little lamb.

============Test11===============
Elapsed=00:00:06.8004289
Result=M ary had a little lamb.

第二次运行

去除无效字符，100万次迭代

注意:Test1不会移除，只会替换。

结果:

============Test1===============
Elapsed=00:00:01.6945352
Result= M     a ry  h  ad    a          li tt le   la mb.

============Test2===============
Elapsed=00:00:01.4798049
Result=M ary had a little lamb.

============Test3===============
Elapsed=00:00:04.0415688
Result=M ary had a little lamb.

============Test4===============
Elapsed=00:00:14.3397960
Result=M ary had a little lamb.

============Test5===============
Elapsed=00:00:01.6782505
Result=M ary had a little lamb.

============Test6===============
Elapsed=00:00:04.9251707
Result=M ary had a little lamb.

============Test7===============
Elapsed=00:00:07.9562379
Result=M ary had a little lamb.

============Test8===============
Elapsed=00:12:16.2918943
Result=M ary had a little lamb.

============Test9===============
Elapsed=00:00:02.0770277
Result=M ary had a little lamb.

============Test10===============
Elapsed=00:00:05.2721232
Result=M ary had a little lamb.

============Test11===============
Elapsed=00:00:05.2802903
Result=M ary had a little lamb.

基准测试结果

方法Test1、Test2和Test5是最快的。方法Test8是最慢的。

CODE

以下是基准测试的完整代码:

private static HashSet<char> _invalidCharsHash;
private static HashSet<char> InvalidCharsHash
{
  get { return _invalidCharsHash ?? (_invalidCharsHash = new HashSet<char>(Path.GetInvalidFileNameChars())); }
}

private static string _invalidCharsValue;
private static string InvalidCharsValue
{
  get { return _invalidCharsValue ?? (_invalidCharsValue = new string(Path.GetInvalidFileNameChars())); }
}

private static char[] _invalidChars;
private static char[] InvalidChars
{
  get { return _invalidChars ?? (_invalidChars = Path.GetInvalidFileNameChars()); }
}

static void Main(string[] args)
{
  string testPath = "\"M <>\"\\a/ry/ h**ad:>> a\\/:*?\"<>| li*tt|le|| la\"mb.?";

  int max = 1000000;
  string newValue = "";

  TimeBenchmark(max, Test1, testPath, newValue);
  TimeBenchmark(max, Test2, testPath, newValue);
  TimeBenchmark(max, Test3, testPath, newValue);
  TimeBenchmark(max, Test4, testPath, newValue);
  TimeBenchmark(max, Test5, testPath, newValue);
  TimeBenchmark(max, Test6, testPath, newValue);
  TimeBenchmark(max, Test7, testPath, newValue);
  TimeBenchmark(max, Test8, testPath, newValue);
  TimeBenchmark(max, Test9, testPath, newValue);
  TimeBenchmark(max, Test10, testPath, newValue);
  TimeBenchmark(max, Test11, testPath, newValue);

  Console.Read();
}

private static void TimeBenchmark(int maxLoop, Func<string, string, string> func, string testString, string newValue)
{
  var sw = new Stopwatch();
  sw.Start();
  string result = string.Empty;

  for (int i = 0; i < maxLoop; i++)
    result = func?.Invoke(testString, newValue);

  sw.Stop();

  Console.WriteLine($"============{func.Method.Name}===============");
  Console.WriteLine("Elapsed={0}", sw.Elapsed);
  Console.WriteLine("Result={0}", result);
  Console.WriteLine("");
}

private static string Test1(string fileName, string newValue)
{
  char newChar = string.IsNullOrEmpty(newValue) ? char.MinValue : newValue[0];

  char[] chars = fileName.ToCharArray();
  for (int i = 0; i < chars.Length; i++)
  {
    if (InvalidCharsHash.Contains(chars[i]))
      chars[i] = newChar;
  }

  return new string(chars);
}

private static string Test2(string fileName, string newValue)
{
  char newChar = string.IsNullOrEmpty(newValue) ? char.MinValue : newValue[0];
  bool remove = newChar == char.MinValue;

  char[] chars = fileName.ToCharArray();
  char[] newChars = new char[chars.Length];
  int i2 = 0;
  for (int i = 0; i < chars.Length; i++)
  {
    char c = chars[i];
    if (InvalidCharsHash.Contains(c))
    {
      if (!remove)
        newChars[i2++] = newChar;
    }
    else
      newChars[i2++] = c;

  }

  return new string(newChars, 0, i2);
}

private static string Test3(string filename, string newValue)
{
  foreach (char c in InvalidCharsValue)
  {
    filename = filename.Replace(c.ToString(), newValue);
  }

  return filename;
}

private static string Test4(string filename, string newValue)
{
  Regex r = new Regex(string.Format("[{0}]", Regex.Escape(InvalidCharsValue)));
  filename = r.Replace(filename, newValue);
  return filename;
}

private static string Test5(string filename, string newValue)
{
  return string.Join(newValue, filename.Split(InvalidChars));
}

private static string Test6(string fileName, string newValue)
{
  return InvalidChars.Aggregate(fileName, (current, c) => current.Replace(c.ToString(), newValue));
}

private static string Test7(string fileName, string newValue)
{
  string regex = string.Format("[{0}]", Regex.Escape(InvalidCharsValue));
  return Regex.Replace(fileName, regex, newValue, RegexOptions.Compiled);
}

private static string Test8(string fileName, string newValue)
{
  string regex = string.Format("[{0}]", Regex.Escape(InvalidCharsValue));
  Regex removeInvalidChars = new Regex(regex, RegexOptions.Singleline | RegexOptions.Compiled | RegexOptions.CultureInvariant);
  return removeInvalidChars.Replace(fileName, newValue);
}

private static string Test9(string fileName, string newValue)
{
  StringBuilder sb = new StringBuilder(fileName.Length);
  bool changed = false;

  for (int i = 0; i < fileName.Length; i++)
  {
    char c = fileName[i];
    if (InvalidCharsHash.Contains(c))
    {
      changed = true;
      sb.Append(newValue);
    }
    else
      sb.Append(c);
  }

  if (sb.Length == 0)
    return newValue;

  return changed ? sb.ToString() : fileName;
}

private static string Test10(string fileName, string newValue)
{
  if (!fileName.Any(c => InvalidChars.Contains(c)))
  {
    return fileName;
  }

  return new string(fileName.Where(c => !InvalidChars.Contains(c)).ToArray());
}

private static string Test11(string fileName, string newValue)
{
  string invalidCharsRemoved = new string(fileName
    .Where(x => !InvalidChars.Contains(x))
    .ToArray());

  return invalidCharsRemoved;
}

浏览这里的答案，它们似乎都涉及使用无效文件名字符的char数组。

当然，这可能是一种微观优化——但对于那些希望检查大量值是否为有效文件名的人来说，值得注意的是，构建一个由无效字符组成的哈希集将带来明显更好的性能。

过去，我对哈希集(或字典)比遍历列表的速度快得多感到非常惊讶。对于字符串，它是一个低得可笑的数字(从记忆中大约5-7项)。对于大多数其他简单数据(对象引用，数字等)，神奇的交叉似乎是大约20个项目。

路径中有40个无效字符。InvalidFileNameChars“列表”。今天搜索了一下，在StackOverflow上有一个很好的基准测试，显示hashset将花费40个数组/列表一半多一点的时间:https://stackoverflow.com/a/10762995/949129

下面是我用于清理路径的helper类。我现在忘记了为什么我在里面有漂亮的替换选项，但这是一个可爱的奖励。

额外的奖励方法“IsValidLocalPath”:)

(**那些不使用正则表达式的)

public static class PathExtensions
{
    private static HashSet<char> _invalidFilenameChars;
    private static HashSet<char> InvalidFilenameChars
    {
        get { return _invalidFilenameChars ?? (_invalidFilenameChars = new HashSet<char>(Path.GetInvalidFileNameChars())); }
    }


    /// <summary>Replaces characters in <c>text</c> that are not allowed in file names with the 
    /// specified replacement character.</summary>
    /// <param name="text">Text to make into a valid filename. The same string is returned if 
    /// it is valid already.</param>
    /// <param name="replacement">Replacement character, or NULL to remove bad characters.</param>
    /// <param name="fancyReplacements">TRUE to replace quotes and slashes with the non-ASCII characters ” and ⁄.</param>
    /// <returns>A string that can be used as a filename. If the output string would otherwise be empty, "_" is returned.</returns>
    public static string ToValidFilename(this string text, char? replacement = '_', bool fancyReplacements = false)
    {
        StringBuilder sb = new StringBuilder(text.Length);
        HashSet<char> invalids = InvalidFilenameChars;
        bool changed = false;

        for (int i = 0; i < text.Length; i++)
        {
            char c = text[i];
            if (invalids.Contains(c))
            {
                changed = true;
                char repl = replacement ?? '\0';
                if (fancyReplacements)
                {
                    if (c == '"') repl = '”'; // U+201D right double quotation mark
                    else if (c == '\'') repl = '’'; // U+2019 right single quotation mark
                    else if (c == '/') repl = '⁄'; // U+2044 fraction slash
                }
                if (repl != '\0')
                    sb.Append(repl);
            }
            else
                sb.Append(c);
        }

        if (sb.Length == 0)
            return "_";

        return changed ? sb.ToString() : text;
    }


    /// <summary>
    /// Returns TRUE if the specified path is a valid, local filesystem path.
    /// </summary>
    /// <param name="pathString"></param>
    /// <returns></returns>
    public static bool IsValidLocalPath(this string pathString)
    {
        // From solution at https://stackoverflow.com/a/11636052/949129
        Uri pathUri;
        Boolean isValidUri = Uri.TryCreate(pathString, UriKind.Absolute, out pathUri);
        return isValidUri && pathUri != null && pathUri.IsLoopback;
    }
}

我绝对更喜欢杰夫·耶茨的想法。它将完美地工作，如果你稍微修改它:

string regex = String.Format("[{0}]", Regex.Escape(new string(Path.GetInvalidFileNameChars())));
Regex removeInvalidChars = new Regex(regex, RegexOptions.Singleline | RegexOptions.Compiled | RegexOptions.CultureInvariant);

改进仅仅是转义自动生成的正则表达式。

我使用Linq来清理文件名。您也可以很容易地扩展它来检查有效路径。

private static string CleanFileName(string fileName)
{
    return Path.GetInvalidFileNameChars().Aggregate(fileName, (current, c) => current.Replace(c.ToString(), string.Empty));
}

更新

一些评论指出这个方法不适合他们，所以我包含了一个到DotNetFiddle片段的链接，这样你就可以验证这个方法了。

https://dotnetfiddle.net/nw1SWY