Java string.split vs. C# Regex.split - limit to certain number of fields

Question

I am a Java developer, but am working on a C# project. What I need to do is split a String by a delimiter, but limit it to a certain number of fields. In Java, I can do this:

String message = "xx/xx - xxxxxxxxxxxxxxxxxxx - xxxxxxx";
String[] splitMessage = message.split("\\s-", 3);

In this case, it will split it by the - , but I want to also have it check for any space before the dash, and limit it to 3 fields of the String. The String coming through is broken down into ___ - ____________ - _________ with the first space being a date (like 12/31 ) the second space being a message about the string, and the third space being a location tied to the message. The reason I limit it to 3 fields so the array only has 3 elements. The reason I do this is because sometimes the message can have dashes in it to look like this: 12/31 - Test message - test - Test City, 11111 . So my Java code above would split it into this:

0: 12/31
1: Test message - test
2: Test City, 11111

I am trying to achieve something similar in C#, but am not sure how to limit it to a certain number of fields. This is my C# code:

var splitMessage = Regex.Split(Message, " -");

The problem is that without a limit, it splits it into 4 or 5 fields, instead of just the 3. For example, if this were the message: 12/31 - My test - don't use - just a test - Test City, 11111 , it would return a string[] with 5 indexes:

0: 12/31
1: My test
2: don't use
3: just a test
4: Test City, 11111

When I want it to return this:

0: 12/31
1: My test - don't use - just a test
2: Test City, 11111

Before you ask, I can't change the incoming String. I have to parse it the same why I did in Java. So is there an equivalent to limiting it to 3 fields? Is there a better way to do it besides using Regex.Split() ?

Answer 1

If you want to split based on the first and last instance of - , such that you get exactly three fields (so long as there are at least two dashes in the string), C# does actually have a neat trick for this. C# Regex allows for non-fixed-width lookbehinds. So the following regex:

(?<=^[^-]*)-|-(?=[^-]*$)

(<=      //start lookbehind
   ^     //look for start of string
   [^-]* //followed by any amount of non-dash characters
)        //end lookbehind
-        //match the dash
|        //OR
-        //match a dash
(?=      //lookahead for
   [^-]* //any amount of non-dash characters
   $     //then the end of the string
)        //end lookahead

Will match the first and last dash, and allow you to split the string the way you want to.

var splitMessage = Regex.Split(Message, "(?<=^[^-]*)-|-(?=[^-]*$)");

Note that this also has no problem splitting into fewer than three groups, if there are less dashes, but will not split into more than three.

Answer 2

You can't split like with the delimiter inside the one of the desired grouped, except when that is the last group.

You can however use a custom regex that consume as much as possible in the 2nd group to parse the said input:

var splitMessage = Regex.Match("12/31 - Test message - test - Test City, 11111", "^(.+?) - (.+) - (.+)$")
    .Groups
    .Cast<Group>()
    // skip first group which is the entire match
    .Skip(1)
    .Select(x => x.Value)
    .ToArray();

Given that the first group is "xx/xx", you can also opt to use this regex instead:

"^(../..) - (.+) - (.+)$"
// or, assuming they are date
"^(\d{2}/\d{2}) - (.+) - (.+)$"

---

EDIT: Or, you can just split by " - ", and then concatenate everything in the middle together when there is more than 3 matches:

var groups = "12/31 - Test message - test - Test City, 11111".Split(new[] { " - " }, StringSplitOptions.None);
if (groups.Length > 3)
{
    groups = new[]
    {
        groups[0],
        string.Join(" - ", groups.Skip(1).Take(groups.Length - 2)),
        groups[groups.Length - 1]
    };
}

Answer 3

Whe I have to split a string at certain delimiters including optional spaces, I do it usually this way:

String message = "xx/xx - xxxxxxxxxxxxxxxxxxx - xxxxxxx";
String[] splitMessage = message.split(" *- *", 3);    
System.out.println(Arrays.asList(splitMessage));

Outputs: [xx/xx, xxxxxxxxxxxxxxxxxxx, xxxxxxx]

String message = "12/31 - My test - don't use - just a test - Test City; 11111";
String[] splitMessage = message.split(" *- *", 3);    
System.out.println(Arrays.asList(splitMessage));

Outputs: [12/31, My test, don't use - just a test - Test City; 11111]

But you seem to want that something different:

splitMessage[0] shall contain the first part
splitMessage[1] shall contain the second and third part
splitMessage[2] shall contain the rest

How do you want to tell your computer that the second output element shall contain two parts? I think this is impossible except by splitting the string into all 5 parts and then re-concatenating the parts together as you want.

Maybe it's not clear what result you want. Can you specify the requirement more clearly: What shall happen if the input string contains more than 3 elements?