c# - 在 C# 中将包含命令行参数的字符串拆分为 string[]

Question

我有一个字符串，其中包含要传递给另一个可执行文件的命令行参数，我需要提取包含各个参数的字符串 []，就像在命令行上指定命令时 C# 一样。当通过反射执行另一个程序集入口点时，将使用 string[]。

这有标准功能吗？或者是否有正确拆分参数的首选方法（正则表达式？）？它必须正确处理可能包含空格的 '"' 分隔字符串，所以我不能只在 ' ' 上拆分。

示例字符串：

string parameterString = @"/src:""C:\tmp\Some Folder\Sub Folder"" /users:""abcdefg@hijkl.com"" tasks:""SomeTask,Some Other Task"" -someParam foo";

示例结果：

string[] parameterArray = new string[] { 
  @"/src:C:\tmp\Some Folder\Sub Folder",
  @"/users:abcdefg@hijkl.com",
  @"tasks:SomeTask,Some Other Task",
  @"-someParam",
  @"foo"
};

我不需要命令行解析库，只是一种获取应该生成的 String[] 的方法。

更新：我必须更改预期结果以匹配 C# 实际生成的结果（删除了拆分字符串中的额外“”）

Answer 1

令我烦恼的是，没有基于检查每个字符的函数来拆分字符串的函数。如果有，你可以这样写：

    public static IEnumerable<string> SplitCommandLine(string commandLine)
    {
        bool inQuotes = false;

        return commandLine.Split(c =>
                                 {
                                     if (c == '\"')
                                         inQuotes = !inQuotes;

                                     return !inQuotes && c == ' ';
                                 })
                          .Select(arg => arg.Trim().TrimMatchingQuotes('\"'))
                          .Where(arg => !string.IsNullOrEmpty(arg));
    }

虽然已经写了，为什么不写必要的扩展方法。好吧，你说服了我...

首先，我自己的 Split 版本采用一个必须决定指定字符是否应该拆分字符串的函数：

    public static IEnumerable<string> Split(this string str, 
                                            Func<char, bool> controller)
    {
        int nextPiece = 0;

        for (int c = 0; c < str.Length; c++)
        {
            if (controller(str[c]))
            {
                yield return str.Substring(nextPiece, c - nextPiece);
                nextPiece = c + 1;
            }
        }

        yield return str.Substring(nextPiece);
    }

根据情况，它可能会产生一些空字符串，但也许这些信息在其他情况下会很有用，所以我不会删除这个函数中的空条目。

其次（更普通的）一个小助手，它将从字符串的开头和结尾修剪一对匹配的引号。它比标准的 Trim 方法更繁琐——它只会从每一端修剪一个字符，并且不会只从一端修剪：

    public static string TrimMatchingQuotes(this string input, char quote)
    {
        if ((input.Length >= 2) && 
            (input[0] == quote) && (input[input.Length - 1] == quote))
            return input.Substring(1, input.Length - 2);

        return input;
    }

我想你也会想要一些测试。好吧，那好吧。但这绝对是最后一件事！首先是一个辅助函数，它将拆分结果与预期的数组内容进行比较：

    public static void Test(string cmdLine, params string[] args)
    {
        string[] split = SplitCommandLine(cmdLine).ToArray();

        Debug.Assert(split.Length == args.Length);

        for (int n = 0; n < split.Length; n++)
            Debug.Assert(split[n] == args[n]);
    }

然后我可以编写这样的测试：

        Test("");
        Test("a", "a");
        Test(" abc ", "abc");
        Test("a b ", "a", "b");
        Test("a b \"c d\"", "a", "b", "c d");

这是针对您的要求的测试：

        Test(@"/src:""C:\tmp\Some Folder\Sub Folder"" /users:""abcdefg@hijkl.com"" tasks:""SomeTask,Some Other Task"" -someParam",
             @"/src:""C:\tmp\Some Folder\Sub Folder""", @"/users:""abcdefg@hijkl.com""", @"tasks:""SomeTask,Some Other Task""", @"-someParam");

请注意，该实现具有额外的功能，即如果有意义，它将删除参数周围的引号（感谢 TrimMatchingQuotes 函数）。我相信这是正常命令行解释的一部分。

Answer 2

除了Earwicker提供的良好且纯托管的解决方案之外，为了完整起见，可能值得一提的是，Windows 还提供了将字符串分解为字符串数组的功能：CommandLineToArgvW

LPWSTR *CommandLineToArgvW(
    LPCWSTR lpCmdLine, int *pNumArgs);

以类似于标准 C 运行时 argv 和 argc 值的方式解析 Unicode 命令行字符串并返回指向命令行参数的指针数组以及此类参数的计数。

从 C# 调用此 API 并在托管代码中解压缩生成的字符串数组的示例可在以下位置找到：“<a href="http://intellitect.com/converting-command-line-string-to-args-using- commandlinetoargvw-api/" rel="noreferrer">使用 CommandLineToArgvW() API 将命令行字符串转换为 Args[]。” 下面是相同代码的稍微简单的版本：

[DllImport("shell32.dll", SetLastError = true)]
static extern IntPtr CommandLineToArgvW(
    [MarshalAs(UnmanagedType.LPWStr)] string lpCmdLine, out int pNumArgs);

public static string[] CommandLineToArgs(string commandLine)
{
    int argc;
    var argv = CommandLineToArgvW(commandLine, out argc);        
    if (argv == IntPtr.Zero)
        throw new System.ComponentModel.Win32Exception();
    try
    {
        var args = new string[argc];
        for (var i = 0; i < args.Length; i++)
        {
            var p = Marshal.ReadIntPtr(argv, i * IntPtr.Size);
            args[i] = Marshal.PtrToStringUni(p);
        }

        return args;
    }
    finally
    {
        Marshal.FreeHGlobal(argv);
    }
}

Answer 3

Windows 命令行解析器的行为就像你说的那样，在空格上拆分，除非它前面有一个未闭合的引号。我建议自己编写解析器。可能是这样的：

    static string[] ParseArguments(string commandLine)
    {
        char[] parmChars = commandLine.ToCharArray();
        bool inQuote = false;
        for (int index = 0; index < parmChars.Length; index++)
        {
            if (parmChars[index] == '"')
                inQuote = !inQuote;
            if (!inQuote && parmChars[index] == ' ')
                parmChars[index] = '\n';
        }
        return (new string(parmChars)).Split('\n');
    }

Answer 4

我从 Jeffrey L Whitledge 那里得到了答案，并对其进行了一些改进。

它现在支持单引号和双引号。您可以通过使用其他类型的引号在参数本身中使用引号。

它还从参数中去除引号，因为这些对参数信息没有贡献。

    public static string[] SplitArguments(string commandLine)
    {
        var parmChars = commandLine.ToCharArray();
        var inSingleQuote = false;
        var inDoubleQuote = false;
        for (var index = 0; index < parmChars.Length; index++)
        {
            if (parmChars[index] == '"' && !inSingleQuote)
            {
                inDoubleQuote = !inDoubleQuote;
                parmChars[index] = '\n';
            }
            if (parmChars[index] == '\'' && !inDoubleQuote)
            {
                inSingleQuote = !inSingleQuote;
                parmChars[index] = '\n';
            }
            if (!inSingleQuote && !inDoubleQuote && parmChars[index] == ' ')
                parmChars[index] = '\n';
        }
        return (new string(parmChars)).Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries);
    }

Answer 5

因为我想要与 OP 相同的行为（拆分字符串与 windows cmd 完全相同）我写了一堆测试用例并测试了这里发布的答案：

    Test( 0, m, "One",                    new[] { "One" });
    Test( 1, m, "One ",                   new[] { "One" });
    Test( 2, m, " One",                   new[] { "One" });
    Test( 3, m, " One ",                  new[] { "One" });
    Test( 4, m, "One Two",                new[] { "One", "Two" });
    Test( 5, m, "One  Two",               new[] { "One", "Two" });
    Test( 6, m, "One   Two",              new[] { "One", "Two" });
    Test( 7, m, "\"One Two\"",            new[] { "One Two" });
    Test( 8, m, "One \"Two Three\"",      new[] { "One", "Two Three" });
    Test( 9, m, "One \"Two Three\" Four", new[] { "One", "Two Three", "Four" });
    Test(10, m, "One=\"Two Three\" Four", new[] { "One=Two Three", "Four" });
    Test(11, m, "One\"Two Three\" Four",  new[] { "OneTwo Three", "Four" });
    Test(12, m, "One\"Two Three   Four",  new[] { "OneTwo Three   Four" });
    Test(13, m, "\"One Two\"",            new[] { "One Two" });
    Test(14, m, "One\" \"Two",            new[] { "One Two" });
    Test(15, m, "\"One\"  \"Two\"",       new[] { "One", "Two" });
    Test(16, m, "One\\\"  Two",           new[] { "One\"", "Two" });
    Test(17, m, "\\\"One\\\"  Two",       new[] { "\"One\"", "Two" });
    Test(18, m, "One\"",                  new[] { "One" });
    Test(19, m, "\"One",                  new[] { "One" });
    Test(20, m, "One \"\"",               new[] { "One", "" });
    Test(21, m, "One \"",                 new[] { "One", "" });
    Test(22, m, "1 A=\"B C\"=D 2",        new[] { "1", "A=B C=D", "2" });
    Test(23, m, "1 A=\"B \\\" C\"=D 2",   new[] { "1", "A=B \" C=D", "2" });
    Test(24, m, "1 \\A 2",                new[] { "1", "\\A", "2" });
    Test(25, m, "1 \\\" 2",               new[] { "1", "\"", "2" });
    Test(26, m, "1 \\\\\" 2",             new[] { "1", "\\\"", "2" });
    Test(27, m, "\"",                     new[] { "" });
    Test(28, m, "\\\"",                   new[] { "\"" });
    Test(29, m, "'A B'",                  new[] { "'A", "B'" });
    Test(30, m, "^",                      new[] { "^" });
    Test(31, m, "^A",                     new[] { "A" });
    Test(32, m, "^^",                     new[] { "^" });
    Test(33, m, "\\^^",                   new[] { "\\^" });
    Test(34, m, "^\\\\",                  new[] { "\\\\" });
    Test(35, m, "^\"A B\"",               new[] { "A B" });

    // Test cases Anton

    Test(36, m, @"/src:""C:\tmp\Some Folder\Sub Folder"" /users:""abcdefg@hijkl.com"" tasks:""SomeTask,Some Other Task"" -someParam foo", new[] { @"/src:C:\tmp\Some Folder\Sub Folder", @"/users:abcdefg@hijkl.com", @"tasks:SomeTask,Some Other Task", @"-someParam", @"foo" });

    // Test cases Daniel Earwicker 

    Test(37, m, "",            new string[] { });
    Test(38, m, "a",           new[] { "a" });
    Test(39, m, " abc ",       new[] { "abc" });
    Test(40, m, "a b ",        new[] { "a", "b" });
    Test(41, m, "a b \"c d\"", new[] { "a", "b", "c d" });

    // Test cases Fabio Iotti 

    Test(42, m, "this is a test ",    new[] { "this", "is", "a", "test" });
    Test(43, m, "this \"is a\" test", new[] { "this", "is a", "test" });

    // Test cases Kevin Thach

    Test(44, m, "\"C:\\Program Files\"",                       new[] { "C:\\Program Files" });
    Test(45, m, "\"He whispered to her \\\"I love you\\\".\"", new[] { "He whispered to her \"I love you\"." });

“预期”值来自在我的机器（Win10 x64）上使用 cmd.exe 和一个简单的打印程序直接测试它：

static void Main(string[] args) => Console.Out.WriteLine($"Count := {args.Length}\n{string.Join("\n", args.Select((v,i) => $"[{i}] => '{v}'"))}");

这些是结果：

Solution                      | Failed Tests
------------------------------|------------------------------------- 
Atif Aziz (749653)            | 2, 3, 10, 11, 12, 14, 16, 17, 18, 26, 28, 31, 32, 33, 34, 35, 36, 37, 39, 45
Jeffrey L Whitledge (298968)  | 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45
Daniel Earwicker (298990)     | 10, 11, 12, 14, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 45
Anton (299795)                | 12, 16, 17, 18, 19, 21, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 45
CS. (467313)                  | 12, 18, 19, 21, 27, 31, 32, 33, 34, 35
Vapour in the Alley (2132004) | 10, 11, 12, 14, 16, 17, 20, 21, 22, 23, 25, 26, 27, 28, 29, 31, 32, 33, 34, 35, 36, 45
Monoman (7774211)             | 14, 16, 17, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 45
Thomas Petersson (19091999)   | 2, 3, 10, 11, 12, 14, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 39, 45
Fabio Iotti (19725880)        | 1, 2, 3, 7, 10, 11, 12, 13, 14, 15, 16, 17, 19, 21, 22, 23, 25, 26, 28, 29, 30, 35, 36, 37, 39, 40, 42, 44, 45
ygoe (23961658)               | 26, 31, 32, 33, 34, 35
Kevin Thach (24829691)        | 10, 11, 12, 14, 18, 19, 20, 21, 22, 23, 26, 27, 31, 32, 33, 34, 35, 36
Lucas De Jesus (31621370)     | 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45
HarryP (48008872)             | 24, 26, 31, 32, 33, 34, 35
TylerY86 (53290784)           | 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 41, 43, 44, 45
Louis Somers (55903304)       | 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 29, 31, 32, 33, 34, 35, 36, 39, 41, 43, 44, 45
user2126375 (58233585)        | 5, 6, 15, 16, 17, 31, 32, 33, 34, 35
DilipNannaware (59131568)     | 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 31, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45
Mikescher (this)              | -

因为没有答案似乎是正确的（至少基于我的用例）这是我的解决方案，它目前通过了所有测试用例（但如果有人有其他（失败）极端情况，请发表评论）：

public static IEnumerable<string> SplitArgs(string commandLine)
{
    var result = new StringBuilder();

    var quoted = false;
    var escaped = false;
    var started = false;
    var allowcaret = false;
    for (int i = 0; i < commandLine.Length; i++)
    {
        var chr = commandLine[i];

        if (chr == '^' && !quoted)
        {
            if (allowcaret)
            {
                result.Append(chr);
                started = true;
                escaped = false;
                allowcaret = false;
            }
            else if (i + 1 < commandLine.Length && commandLine[i + 1] == '^')
            {
                allowcaret = true;
            }
            else if (i + 1 == commandLine.Length)
            {
                result.Append(chr);
                started = true;
                escaped = false;
            }
        }
        else if (escaped)
        {
            result.Append(chr);
            started = true;
            escaped = false;
        }
        else if (chr == '"')
        {
            quoted = !quoted;
            started = true;
        }
        else if (chr == '\\' && i + 1 < commandLine.Length && commandLine[i + 1] == '"')
        {
            escaped = true;
        }
        else if (chr == ' ' && !quoted)
        {
            if (started) yield return result.ToString();
            result.Clear();
            started = false;
        }
        else
        {
            result.Append(chr);
            started = true;
        }
    }

    if (started) yield return result.ToString();
}

我用来生成测试结果的代码可以在这里找到

Answer 6

Earwicker的良好且纯粹的托管解决方案未能处理这样的论点：

Test("\"He whispered to her \\\"I love you\\\".\"", "He whispered to her \"I love you\".");

它返回了 3 个元素：

"He whispered to her \"I
love
you\"."

所以这里有一个修复来支持“quoted \"escape\" quote”：

public static IEnumerable<string> SplitCommandLine(string commandLine)
{
    bool inQuotes = false;
    bool isEscaping = false;

    return commandLine.Split(c => {
        if (c == '\\' && !isEscaping) { isEscaping = true; return false; }

        if (c == '\"' && !isEscaping)
            inQuotes = !inQuotes;

        isEscaping = false;

        return !inQuotes && Char.IsWhiteSpace(c)/*c == ' '*/;
        })
        .Select(arg => arg.Trim().TrimMatchingQuotes('\"').Replace("\\\"", "\""))
        .Where(arg => !string.IsNullOrEmpty(arg));
}

用另外 2 个案例进行了测试：

Test("\"C:\\Program Files\"", "C:\\Program Files");
Test("\"He whispered to her \\\"I love you\\\".\"", "He whispered to her \"I love you\".");

还注意到使用CommandLineToArgvW的Atif Aziz接受的答案也失败了。它返回了 4 个元素：

He whispered to her \ 
I 
love 
you". 

希望这对将来寻找此类解决方案的人有所帮助。

Answer 7

我喜欢迭代器，现在LINQ就像字符串数组一样容易IEnumerable<String>使用，所以我遵循Jeffrey L Whitledge 的回答的精神是（作为 的扩展方法string）：

public static IEnumerable<string> ParseArguments(this string commandLine)
{
    if (string.IsNullOrWhiteSpace(commandLine))
        yield break;

    var sb = new StringBuilder();
    bool inQuote = false;
    foreach (char c in commandLine) {
        if (c == '"' && !inQuote) {
            inQuote = true;
            continue;
        }

        if (c != '"' && !(char.IsWhiteSpace(c) && !inQuote)) {
            sb.Append(c);
            continue;
        }

        if (sb.Length > 0) {
            var result = sb.ToString();
            sb.Clear();
            inQuote = false;
            yield return result;
        }
    }

    if (sb.Length > 0)
        yield return sb.ToString();
}

Answer 8

Environment.GetCommandLineArgs()

Answer 9

在您的问题中，您要求使用正则表达式，而我是它们的忠实粉丝和用户，因此当我需要与您进行相同的论点拆分时，我在谷歌搜索并没有找到简单的解决方案后编写了自己的正则表达式。我喜欢简短的解决方案，所以我做了一个，它是：

            var re = @"\G(""((""""|[^""])+)""|(\S+)) *";
            var ms = Regex.Matches(CmdLine, re);
            var list = ms.Cast<Match>()
                         .Select(m => Regex.Replace(
                             m.Groups[2].Success
                                 ? m.Groups[2].Value
                                 : m.Groups[4].Value, @"""""", @"""")).ToArray();

它处理引号内的空格和引号，并将封闭的“”转换为“。随意使用代码！

Answer 10

哦，见鬼。这一切……哎呀。但这是合法的官方。来自 Microsoft 的 C# for .NET Core，可能仅适用于 Windows，也可能是跨平台的，但已获得 MIT 许可。

精选花絮、方法声明和值得注意的评论；

internal static unsafe string[] InternalCreateCommandLine(bool includeArg0)
private static unsafe int SegmentCommandLine(char * pCmdLine, string[] argArray, bool includeArg0)
private static unsafe int ScanArgument0(ref char* psrc, char[] arg)
private static unsafe int ScanArgument(ref char* psrc, ref bool inquote, char[] arg)

-

// First, parse the program name (argv[0]). Argv[0] is parsed under special rules. Anything up to 
// the first whitespace outside a quoted subtring is accepted. Backslashes are treated as normal 
// characters.

-

// Rules: 2N backslashes + " ==> N backslashes and begin/end quote
//      2N+1 backslashes + " ==> N backslashes + literal "
//         N backslashes     ==> N backslashes

这是从 .NET Framework 移植到 .NET Core 的代码，我假设是 MSVC C 库或CommandLineToArgvW.

这是我用正则表达式处理一些恶作剧并忽略零位参数的半心半意尝试。这有点神奇。

private static readonly Regex RxWinArgs
  = new Regex("([^\\s\"]+\"|((?<=\\s|^)(?!\"\"(?!\"))\")+)(\"\"|.*?)*\"[^\\s\"]*|[^\\s]+",
    RegexOptions.Compiled
    | RegexOptions.Singleline
    | RegexOptions.ExplicitCapture
    | RegexOptions.CultureInvariant);

internal static IEnumerable<string> ParseArgumentsWindows(string args) {
  var match = RxWinArgs.Match(args);

  while (match.Success) {
    yield return match.Value;
    match = match.NextMatch();
  }
}

对古怪的生成输出进行了相当多的测试。它的输出与猴子输入和运行的内容的相当比例相匹配CommandLineToArgvW。

Answer 11

有一个 NuGet 包，其中包含您需要的功能：

Microsoft.CodeAnalysis.Common包含具有方法SplitCommandLineIntoArguments的类CommandLineParser。

你像这样使用它：

using Microsoft.CodeAnalysis;
// [...]
var cli = @"/src:""C:\tmp\Some Folder\Sub Folder"" /users:""abcdefg@hijkl.com"" tasks:""SomeTask,Some Other Task"" -someParam foo";
var cliArgs = CommandLineParser.SplitCommandLineIntoArguments(cli, true);

Console.WriteLine(string.Join('\n', cliArgs));
// prints out:
// /src:"C:\tmp\Some Folder\Sub Folder"
// /users:"abcdefg@hijkl.com"
// tasks:"SomeTask,Some Other Task"
// -someParam
// foo

Answer 12

这篇The Code Project 文章是我过去使用的。这是一段很好的代码，但它可能会起作用。

这篇MSDN 文章是我能找到的唯一一篇解释 C# 如何解析命令行参数的文章。

Answer 13

采用：

public static string[] SplitArguments(string args) {
    char[] parmChars = args.ToCharArray();
    bool inSingleQuote = false;
    bool inDoubleQuote = false;
    bool escaped = false;
    bool lastSplitted = false;
    bool justSplitted = false;
    bool lastQuoted = false;
    bool justQuoted = false;

    int i, j;

    for(i=0, j=0; i<parmChars.Length; i++, j++) {
        parmChars[j] = parmChars[i];

        if(!escaped) {
            if(parmChars[i] == '^') {
                escaped = true;
                j--;
            } else if(parmChars[i] == '"' && !inSingleQuote) {
                inDoubleQuote = !inDoubleQuote;
                parmChars[j] = '\n';
                justSplitted = true;
                justQuoted = true;
            } else if(parmChars[i] == '\'' && !inDoubleQuote) {
                inSingleQuote = !inSingleQuote;
                parmChars[j] = '\n';
                justSplitted = true;
                justQuoted = true;
            } else if(!inSingleQuote && !inDoubleQuote && parmChars[i] == ' ') {
                parmChars[j] = '\n';
                justSplitted = true;
            }

            if(justSplitted && lastSplitted && (!lastQuoted || !justQuoted))
                j--;

            lastSplitted = justSplitted;
            justSplitted = false;

            lastQuoted = justQuoted;
            justQuoted = false;
        } else {
            escaped = false;
        }
    }

    if(lastQuoted)
        j--;

    return (new string(parmChars, 0, j)).Split(new[] { '\n' });
}

根据Alley的回答中的 Vapor ，这个也支持 ^ 转义。

例子：

• 这是一个测试
  • 这
  • 是
  • 一种
  • 测试
• 这是一个测试
  • 这
  • 是一个
  • 测试
• 这个 ^"is a^" 测试
  • 这
  • “是
  • 一种”
  • 测试
• 这个“”“是一个^^测试”
  • 这
  • ​​​
  • 是一个 ^ 测试

它还支持多个空格（每个空格块只中断一次参数）。

Answer 14

纯托管的解决方案可能会有所帮助。WINAPI 函数的“问题”注释太多，在其他平台上不可用。这是我的代码，它具有明确定义的行为（您可以根据需要进行更改）。

它应该与 .NET/Windows 在提供该参数时所做的相同string[] args，并且我已经将它与许多“有趣”的值进行了比较。

这是一个经典的状态机实现，它从输入字符串中获取每个单个字符并将其解释为当前状态，产生输出和新状态。状态在变量escape、和中定义inQuote，输出收集在和中。hadQuoteprevChcurrentArgargs

我在真实命令提示符（Windows 7）上通过实验发现的一些特性：\\produces \、\"produces "、""在引用范围内produces "。

这个^角色似乎也很神奇：它总是在不加倍时消失。否则它对真正的命令行没有影响。我的实现不支持这一点，因为我还没有找到这种行为的模式。也许有人对此了解更多。

不适合这种模式的是以下命令：

cmd /c "argdump.exe "a b c""

该cmd命令似乎捕获了外部引号并逐字记录其余部分。这里面一定有什么特殊的魔法酱。

我没有对我的方法进行基准测试，但认为它相当快。它不使用Regex也不做任何字符串连接，而是使用 aStringBuilder来收集参数的字符并将它们放在一个列表中。

/// <summary>
/// Reads command line arguments from a single string.
/// </summary>
/// <param name="argsString">The string that contains the entire command line.</param>
/// <returns>An array of the parsed arguments.</returns>
public string[] ReadArgs(string argsString)
{
    // Collects the split argument strings
    List<string> args = new List<string>();
    // Builds the current argument
    var currentArg = new StringBuilder();
    // Indicates whether the last character was a backslash escape character
    bool escape = false;
    // Indicates whether we're in a quoted range
    bool inQuote = false;
    // Indicates whether there were quotes in the current arguments
    bool hadQuote = false;
    // Remembers the previous character
    char prevCh = '\0';
    // Iterate all characters from the input string
    for (int i = 0; i < argsString.Length; i++)
    {
        char ch = argsString[i];
        if (ch == '\\' && !escape)
        {
            // Beginning of a backslash-escape sequence
            escape = true;
        }
        else if (ch == '\\' && escape)
        {
            // Double backslash, keep one
            currentArg.Append(ch);
            escape = false;
        }
        else if (ch == '"' && !escape)
        {
            // Toggle quoted range
            inQuote = !inQuote;
            hadQuote = true;
            if (inQuote && prevCh == '"')
            {
                // Doubled quote within a quoted range is like escaping
                currentArg.Append(ch);
            }
        }
        else if (ch == '"' && escape)
        {
            // Backslash-escaped quote, keep it
            currentArg.Append(ch);
            escape = false;
        }
        else if (char.IsWhiteSpace(ch) && !inQuote)
        {
            if (escape)
            {
                // Add pending escape char
                currentArg.Append('\\');
                escape = false;
            }
            // Accept empty arguments only if they are quoted
            if (currentArg.Length > 0 || hadQuote)
            {
                args.Add(currentArg.ToString());
            }
            // Reset for next argument
            currentArg.Clear();
            hadQuote = false;
        }
        else
        {
            if (escape)
            {
                // Add pending escape char
                currentArg.Append('\\');
                escape = false;
            }
            // Copy character from input, no special meaning
            currentArg.Append(ch);
        }
        prevCh = ch;
    }
    // Save last argument
    if (currentArg.Length > 0 || hadQuote)
    {
        args.Add(currentArg.ToString());
    }
    return args.ToArray();
}

Answer 15

目前，这是我拥有的代码：

    private String[] SplitCommandLineArgument(String argumentString)
    {
        StringBuilder translatedArguments = new StringBuilder(argumentString);
        bool escaped = false;
        for (int i = 0; i < translatedArguments.Length; i++)
        {
            if (translatedArguments[i] == '"')
            {
                escaped = !escaped;
            }
            if (translatedArguments[i] == ' ' && !escaped)
            {
                translatedArguments[i] = '\n';
            }
        }

        string[] toReturn = translatedArguments.ToString().Split(new char[] { '\n' }, StringSplitOptions.RemoveEmptyEntries);
        for(int i = 0; i < toReturn.Length; i++)
        {
            toReturn[i] = RemoveMatchingQuotes(toReturn[i]);
        }
        return toReturn;
    }

    public static string RemoveMatchingQuotes(string stringToTrim)
    {
        int firstQuoteIndex = stringToTrim.IndexOf('"');
        int lastQuoteIndex = stringToTrim.LastIndexOf('"');
        while (firstQuoteIndex != lastQuoteIndex)
        {
            stringToTrim = stringToTrim.Remove(firstQuoteIndex, 1);
            stringToTrim = stringToTrim.Remove(lastQuoteIndex - 1, 1); //-1 because we've shifted the indicies left by one
            firstQuoteIndex = stringToTrim.IndexOf('"');
            lastQuoteIndex = stringToTrim.LastIndexOf('"');
        }
        return stringToTrim;
    }

它不适用于转义引号，但它适用于我迄今为止遇到的情况。

Answer 16

这是对 Anton 代码的回复，该代码不适用于转义引号。我修改了3个地方。

1. SplitCommandLineArguments中StringBuilder的构造函数，将任何\"替换为\r
2. 在SplitCommandLineArguments的for 循环中，我现在将\r字符替换回\"。
3. 将SplitCommandLineArgument方法从private更改为public static。

---

public static string[] SplitCommandLineArgument( String argumentString )
{
    StringBuilder translatedArguments = new StringBuilder( argumentString ).Replace( "\\\"", "\r" );
    bool InsideQuote = false;
    for ( int i = 0; i < translatedArguments.Length; i++ )
    {
        if ( translatedArguments[i] == '"' )
        {
            InsideQuote = !InsideQuote;
        }
        if ( translatedArguments[i] == ' ' && !InsideQuote )
        {
            translatedArguments[i] = '\n';
        }
    }

    string[] toReturn = translatedArguments.ToString().Split( new char[] { '\n' }, StringSplitOptions.RemoveEmptyEntries );
    for ( int i = 0; i < toReturn.Length; i++ )
    {
        toReturn[i] = RemoveMatchingQuotes( toReturn[i] );
        toReturn[i] = toReturn[i].Replace( "\r", "\"" );
    }
    return toReturn;
}

public static string RemoveMatchingQuotes( string stringToTrim )
{
    int firstQuoteIndex = stringToTrim.IndexOf( '"' );
    int lastQuoteIndex = stringToTrim.LastIndexOf( '"' );
    while ( firstQuoteIndex != lastQuoteIndex )
    {
        stringToTrim = stringToTrim.Remove( firstQuoteIndex, 1 );
        stringToTrim = stringToTrim.Remove( lastQuoteIndex - 1, 1 ); //-1 because we've shifted the indicies left by one
        firstQuoteIndex = stringToTrim.IndexOf( '"' );
        lastQuoteIndex = stringToTrim.LastIndexOf( '"' );
    }
    return stringToTrim;
}

Answer 17

你可以看看我昨天发布的代码：

[C#] 路径和参数字符串

它将文件名 + 参数拆分为字符串 []。处理短路径、环境变量和丢失的文件扩展名。

（最初是用于注册表中的 UninstallString。）

Answer 18

试试这个代码：

    string[] str_para_linha_comando(string str, out int argumentos)
    {
        string[] linhaComando = new string[32];
        bool entre_aspas = false;
        int posicao_ponteiro = 0;
        int argc = 0;
        int inicio = 0;
        int fim = 0;
        string sub;

        for(int i = 0; i < str.Length;)
        {
            if (entre_aspas)
            {
                // Está entre aspas
                sub = str.Substring(inicio+1, fim - (inicio+1));
                linhaComando[argc - 1] = sub;

                posicao_ponteiro += ((fim - posicao_ponteiro)+1);
                entre_aspas = false;
                i = posicao_ponteiro;
            }
            else
            {
            tratar_aspas:
                if (str.ElementAt(i) == '\"')
                {
                    inicio = i;
                    fim = str.IndexOf('\"', inicio + 1);
                    entre_aspas = true;
                    argc++;
                }
                else
                {
                    // Se não for aspas, então ler até achar o primeiro espaço em branco
                    if (str.ElementAt(i) == ' ')
                    {
                        if (str.ElementAt(i + 1) == '\"')
                        {
                            i++;
                            goto tratar_aspas;
                        }

                        // Pular os espaços em branco adiconais
                        while(str.ElementAt(i) == ' ') i++;

                        argc++;
                        inicio = i;
                        fim = str.IndexOf(' ', inicio);
                        if (fim == -1) fim = str.Length;
                        sub = str.Substring(inicio, fim - inicio);
                        linhaComando[argc - 1] = sub;
                        posicao_ponteiro += (fim - posicao_ponteiro);

                        i = posicao_ponteiro;
                        if (posicao_ponteiro == str.Length) break;
                    }
                    else
                    {
                        argc++;
                        inicio = i;
                        fim = str.IndexOf(' ', inicio);
                        if (fim == -1) fim = str.Length;

                        sub = str.Substring(inicio, fim - inicio);
                        linhaComando[argc - 1] = sub;
                        posicao_ponteiro += fim - posicao_ponteiro;
                        i = posicao_ponteiro;
                        if (posicao_ponteiro == str.Length) break;
                    }
                }
            }
        }

        argumentos = argc;

        return linhaComando;
    }

它是用葡萄牙语写的。

Answer 19

这是完成工作的一条线（请参阅在 BurstCmdLineArgs(...) 方法中完成所有工作的一条线）。

不是我所说的最易读的代码行，但为了可读性，你可以把它分开。这很简单，并且不适用于所有参数情况（例如其中包含拆分字符串字符分隔符的文件名参数）。

该解决方案在我使用它的解决方案中运行良好。就像我说的那样，它可以在没有大量代码的情况下完成工作，以处理所有可能的参数格式 n 阶乘。

using System;
using System.Collections.Generic;
using System.Linq;

namespace CmdArgProcessor
{
    class Program
    {
        static void Main(string[] args)
        {
            // test switches and switches with values
            // -test1 1 -test2 2 -test3 -test4 -test5 5

            string dummyString = string.Empty;

            var argDict = BurstCmdLineArgs(args);

            Console.WriteLine("Value for switch = -test1: {0}", argDict["test1"]);
            Console.WriteLine("Value for switch = -test2: {0}", argDict["test2"]);
            Console.WriteLine("Switch -test3 is present? {0}", argDict.TryGetValue("test3", out dummyString));
            Console.WriteLine("Switch -test4 is present? {0}", argDict.TryGetValue("test4", out dummyString));
            Console.WriteLine("Value for switch = -test5: {0}", argDict["test5"]);

            // Console output:
            //
            // Value for switch = -test1: 1
            // Value for switch = -test2: 2
            // Switch -test3 is present? True
            // Switch -test4 is present? True
            // Value for switch = -test5: 5
        }

        public static Dictionary<string, string> BurstCmdLineArgs(string[] args)
        {
            var argDict = new Dictionary<string, string>();

            // Flatten the args in to a single string separated by a space.
            // Then split the args on the dash delimiter of a cmd line "switch".
            // E.g. -mySwitch myValue
            //  or -JustMySwitch (no value)
            //  where: all values must follow a switch.
            // Then loop through each string returned by the split operation.
            // If the string can be split again by a space character,
            // then the second string is a value to be paired with a switch,
            // otherwise, only the switch is added as a key with an empty string as the value.
            // Use dictionary indexer to retrieve values for cmd line switches.
            // Use Dictionary::ContainsKey(...) where only a switch is recorded as the key.
            string.Join(" ", args).Split('-').ToList().ForEach(s => argDict.Add(s.Split()[0], (s.Split().Count() > 1 ? s.Split()[1] : "")));

            return argDict;
        }
    }
}

Answer 20

我认为 C# 应用程序没有单引号或 ^ 引号。以下功能对我来说很好用：

public static IEnumerable<String> SplitArguments(string commandLine)
{
    Char quoteChar = '"';
    Char escapeChar = '\\';
    Boolean insideQuote = false;
    Boolean insideEscape = false;

    StringBuilder currentArg = new StringBuilder();

    // needed to keep "" as argument but drop whitespaces between arguments
    Int32 currentArgCharCount = 0;                  

    for (Int32 i = 0; i < commandLine.Length; i++)
    {
        Char c = commandLine[i];
        if (c == quoteChar)
        {
            currentArgCharCount++;

            if (insideEscape)
            {
                currentArg.Append(c);       // found \" -> add " to arg
                insideEscape = false;
            }
            else if (insideQuote)
            {
                insideQuote = false;        // quote ended
            }
            else
            {
                insideQuote = true;         // quote started
            }
        }
        else if (c == escapeChar)
        {
            currentArgCharCount++;

            if (insideEscape)   // found \\ -> add \\ (only \" will be ")
                currentArg.Append(escapeChar + escapeChar);       

            insideEscape = !insideEscape;
        }
        else if (Char.IsWhiteSpace(c))
        {
            if (insideQuote)
            {
                currentArgCharCount++;
                currentArg.Append(c);       // append whitespace inside quote
            }
            else
            {
                if (currentArgCharCount > 0)
                    yield return currentArg.ToString();

                currentArgCharCount = 0;
                currentArg.Clear();
            }
        }
        else
        {
            currentArgCharCount++;
            if (insideEscape)
            {
                // found non-escaping backslash -> add \ (only \" will be ")
                currentArg.Append(escapeChar);                       
                currentArgCharCount = 0;
                insideEscape = false;
            }
            currentArg.Append(c);
        }
    }

    if (currentArgCharCount > 0)
        yield return currentArg.ToString();
}

Answer 21

在这里找不到我喜欢的东西。我讨厌用一个小命令行的 yield 魔法来搞乱堆栈（如果它是一个 TB 的流，那就是另一回事了）。

这是我的看法，它支持带双引号的引号转义，如下所示：

param="a 15"" 屏幕还不错" param2='a 15" 屏幕还不错' param3="" param4= /param5

结果：

param="a 15" 屏幕还不错"

param2='15" 屏幕还不错'

参数 3=""

参数4=

/参数5

public static string[] SplitArguments(string commandLine)
{
    List<string> args         = new List<string>();
    List<char>   currentArg   = new List<char>();
    char?        quoteSection = null; // Keeps track of a quoted section (and the type of quote that was used to open it)
    char[]       quoteChars   = new[] {'\'', '\"'};
    char         previous     = ' '; // Used for escaping double quotes

    for (var index = 0; index < commandLine.Length; index++)
    {
        char c = commandLine[index];
        if (quoteChars.Contains(c))
        {
            if (previous == c) // Escape sequence detected
            {
                previous = ' '; // Prevent re-escaping
                if (!quoteSection.HasValue)
                {
                    quoteSection = c; // oops, we ended the quoted section prematurely
                    continue;         // don't add the 2nd quote (un-escape)
                }

                if (quoteSection.Value == c)
                    quoteSection = null; // appears to be an empty string (not an escape sequence)
            }
            else if (quoteSection.HasValue)
            {
                if (quoteSection == c)
                    quoteSection = null; // End quoted section
            }
            else
                quoteSection = c; // Start quoted section
        }
        else if (char.IsWhiteSpace(c))
        {
            if (!quoteSection.HasValue)
            {
                args.Add(new string(currentArg.ToArray()));
                currentArg.Clear();
                previous = c;
                continue;
            }
        }

        currentArg.Add(c);
        previous = c;
    }

    if (currentArg.Count > 0)
        args.Add(new string(currentArg.ToArray()));

    return args.ToArray();
}

Answer 22

我已经实现了状态机以具有相同的解析器结果，就好像 args 将被传递到 .NET 应用程序并在static void Main(string[] args)方法中处理一样。

    public static IList<string> ParseCommandLineArgsString(string commandLineArgsString)
    {
        List<string> args = new List<string>();

        commandLineArgsString = commandLineArgsString.Trim();
        if (commandLineArgsString.Length == 0)
            return args;

        int index = 0;
        while (index != commandLineArgsString.Length)
        {
            args.Add(ReadOneArgFromCommandLineArgsString(commandLineArgsString, ref index));
        }

        return args;
    }

    private static string ReadOneArgFromCommandLineArgsString(string line, ref int index)
    {
        if (index >= line.Length)
            return string.Empty;

        var sb = new StringBuilder(512);
        int state = 0;
        while (true)
        {
            char c = line[index];
            index++;
            switch (state)
            {
                case 0: //string outside quotation marks
                    if (c == '\\') //possible escaping character for quotation mark otherwise normal character
                    {
                        state = 1;
                    }
                    else if (c == '"') //opening quotation mark for string between quotation marks
                    {
                        state = 2;
                    }
                    else if (c == ' ') //closing arg
                    {
                        return sb.ToString();
                    }
                    else
                    {
                        sb.Append(c);
                    }

                    break;
                case 1: //possible escaping \ for quotation mark or normal character
                    if (c == '"') //If escaping quotation mark only quotation mark is added into result
                    {
                        state = 0;
                        sb.Append(c);
                    }
                    else // \ works as not-special character
                    {
                        state = 0;
                        sb.Append('\\');
                        index--;
                    }

                    break;
                case 2: //string between quotation marks
                    if (c == '"') //quotation mark in string between quotation marks can be escape mark for following quotation mark or can be ending quotation mark for string between quotation marks
                    {
                        state = 3;
                    }
                    else if (c == '\\') //escaping \ for possible following quotation mark otherwise normal character
                    {
                        state = 4;
                    }
                    else //text in quotation marks
                    {
                        sb.Append(c);
                    }

                    break;
                case 3: //quotation mark in string between quotation marks
                    if (c == '"') //Quotation mark after quotation mark - that means that this one is escaped and can added into result and we will stay in string between quotation marks state
                    {
                        state = 2;
                        sb.Append(c);
                    }
                    else //we had two consecutive quotation marks - this means empty string but the following chars (until space) will be part of same arg result as well
                    {
                        state = 0;
                        index--;
                    }

                    break;
                case 4: //possible escaping \ for quotation mark or normal character in string between quotation marks
                    if (c == '"') //If escaping quotation mark only quotation mark added into result
                    {
                        state = 2;
                        sb.Append(c);
                    }
                    else
                    {
                        state = 2;
                        sb.Append('\\');
                        index--;
                    }

                    break;
            }

            if (index == line.Length)
                return sb.ToString();
        }
    }

Answer 23

这是将空格（单个或多个空格）视为命令行参数分隔符并返回真正的命令行参数的解决方案：

static string[] ParseMultiSpacedArguments(string commandLine)
{
    var isLastCharSpace = false;
    char[] parmChars = commandLine.ToCharArray();
    bool inQuote = false;
    for (int index = 0; index < parmChars.Length; index++)
    {
        if (parmChars[index] == '"')
            inQuote = !inQuote;
        if (!inQuote && parmChars[index] == ' ' && !isLastCharSpace)
            parmChars[index] = '\n';

        isLastCharSpace = parmChars[index] == '\n' || parmChars[index] == ' ';
    }

    return (new string(parmChars)).Split('\n');
}

Answer 24

我写了一个方法来分离文件名和它的参数，使用ProcessStartInfo它需要分离文件名和参数字符串。

例如"C:\Users\Me\Something.exe" -a 会给出{ "C:\Users\Me\Something.exe", "-a" }结果

下面的代码：

    public static string[] SplitCommandFromArgs(string commandLine)
    {
        commandLine = commandLine.Trim();
        if (commandLine[0] == '"')
        {
            bool isEscaped = false;
            for (int c = 1; c < commandLine.Length; c++)
            {
                if (commandLine[c] == '"' && !isEscaped)
                {
                    return new string[] { commandLine.Substring(1, c - 1), commandLine.Substring(c + 1).Trim() };
                }
                isEscaped = commandLine[c] == '\\';
            }
        }
        else
        {
            for (int c = 1; c < commandLine.Length; c++) {
                if (commandLine[c] == ' ')
                {
                    return new string[] { commandLine.Substring(0, c), commandLine.Substring(c).Trim() };
                }
            }
        }
        return new string[] { commandLine, "" };
    }

Answer 25

我不确定我是否理解您，但是用作分隔符的字符是否也可以在文本中找到？（除了它用双“？”转义）

如果是这样，我会创建一个for循环，并用 <|> （或另一个“安全”字符，但确保它只替换 <"> 而不是 <"">

在迭代字符串之后，我会像之前发布的那样，拆分字符串，但现在在字符 <|> 上。

Answer 26

是的，字符串对象有一个名为的内置函数Split()，该函数接受一个参数，指定要查找的字符作为分隔符，并返回一个字符串数组 (string[])，其中包含各个值。