使用正则表达式在C＃扩展方法中解析字符串

Question

I need to create extension method which pars(split) my string.

For example: If I have string

COMMAND 1 PROCESSED "JOB command" 20160801 09:05:24

It should be split like this

COMMAND

1

PROCESSED

"JOB command"

20160801

09:05:24

Other example. If I have string:

COMMAND 2 ERROR 06 00000032 "Message window is still active." 20160801 09:05:24

It should be split like this:

COMMAND

2

ERROR

06

00000032

"Message window is still active."

20160801 09:05:24

I have solution for this. But I am sure that there is much cleaner solution.

My solution:

 public static List<string> GetTokens(this string line)
        {
            // TODO: Code refactoring:
            var res = new List<string>();
            var parts = Regex.Split(line, "/[^\\s\"']+|\"([^\"]*)\"|'([^']*)'/g");

            var subParts = parts[0].Split(' ');
            foreach (var val in subParts)
            {
                res.Add(val);
            }
            res.Add(parts[1]);
            subParts = parts[2].Split(' ');
            foreach (var val in subParts)
            {
                res.Add(val);
            }

            res.RemoveAll(f => f.Trim() == "");
            return res;
        }

I would like to implement cleaner solution. Any ideas?

Answer 1

I suggest implementing a simple loop instead of complex regular expression :

public static IEnumerable<String> GetTokens(this string value) {
  if (string.IsNullOrEmpty(value))
    yield break; // or throw exception in case of value == null

  bool inQuotation = false;
  int index = 0;

  for (int i = 0; i < value.Length; ++i) {
    char ch = value[i];

    if (ch == '"')
      inQuotation = !inQuotation;
    else if ((ch == ' ') && (!inQuotation)) {
      yield return value.Substring(index, i - index);

      index = i + 1;
    }
  }

  if (index < value.Length)
    yield return value.Substring(index, value.Length - index);
}

Test

var source = 
  "COMMAND 2 ERROR 06 00000032 \"Message window is still active.\" 20160801 09:05:24";

Console.Write(string.Join(Environment.NewLine, GetTokens(source)));

Output

 COMMAND
 2
 ERROR
 06
 00000032
 "Message window is still active."
 20160801
 09:05:24

Edit : in case you want two quotation types with " (double) as well as ' (single):

public static IEnumerable<String> GetTokens(string value) {
  if (string.IsNullOrEmpty(value))
    yield break;

  bool inQuotation = false;
  bool inApostroph = false;

  int index = 0;

  for (int i = 0; i < value.Length; ++i) {
    char ch = value[i];

    if (inQuotation) 
      inQuotation = ch != '"';
    else if (inApostroph) 
      inApostroph = ch != '\'';
    else if (ch == '"')
      inQuotation = true;
    else if (ch == '\'')
      inApostroph = true;
    else if ((ch == ' ') && (!inQuotation)) {
      yield return value.Substring(index, i - index);

      index = i + 1;
    }
  }

  if (index < value.Length)
    yield return value.Substring(index, value.Length - index);
}

Answer 2

After a while a figured out some simple code:

public static List<string> GetTokens(this string line)
{
    return Regex.Matches(line, @"([^\s""]+|""([^""]*)"")").OfType<Match>().Select(l => l.Groups[1].Value).ToList();
}

I tested the code with a MessageBox which showed the List with | in-between each item:

Answer 3

You can use regex like : ([^\\s"]+|"[^"]*") with globlal identifier

Demo and Explaination

Answer 4

A pure regex solution:

public static List<string> GetTokens(this string line)
{
    return Regex.Matches(line,
        @""".*?""|\S+").Cast<Match>().Select(m => m.Value).ToList();
}

The ".*?"|\\S+ regex matches either a quoted string or a non-space char sequence. These matches then can be returned as collection in one go.

Here is a demo: https://ideone.com/hmLQIt .