Thread: Need help with a specific regular expression

Hi there,

I need a little help with a regular expression which I want to use to split a string.

The string could look similar to this:

"a = \"Hello World\" b = \"Some other String\" c = \"test\""

The String is read from a file where the file contents would be:

a = "Hello World" b = "Some other String" c = "test"

After splitting I would like to get the following array:

String[] splitString = new String[] {"a", "=", "\"Hello World\"", "b", "=", "\"Some other String\"", "c", "=", "\"test\""}

I know I could just write a simple text parser to go over the string character by character, but I would rather not do it to keep my code clean and simple. No need to reinvent the wheel.
However, I just cant seem to be able to find the right regular expression for this task. I know that a RE must exist because this can be solved by a finite automaton.

If somebody was so kind to explain to me how the RE is supposed to look like this would be great.
Thank you all very much.

Hi,

I can think of 2 ways to do this:

1. Use String.split() directly

String[] splitStr = str.split("(?:(?<=\\w) (?==))|(?:(?<==) (?=\"))|(?:(?<=\") (?=\\w))");

I'm not a regex expert, so it may be possible to refine the above. Note the use of:

• (?<=X) - zero-width positive lookbehind
• (?=X) - zero-width positive lookahead
• (?:X) - non-capturing group
• | - OR Logical operator

The basic idea is to split against the space character conditional upon the character before and after the space character to avoid splitting up "Hello World" and other similar quoted strings.

See Pattern (Java Platform SE 6) and Regular Expression Tutorial - Learn How to Use Regular Expressions.

2. Use Pattern and Matcher

	String[] splitMultiSteps(String str) {
		System.out.println(str);
		Pattern pattern = Pattern.compile("(\\w+) (=) (\"[^\"]+\")");
		Matcher matcher = pattern.matcher(str);
 
		List<String> matches = new ArrayList<>();
		while (matcher.find()) {
			for (int i = 1; i <= matcher.groupCount(); i++) {
				String matchedStr = matcher.group(i);
				matches.add(matchedStr);
			}
		}
 
		return matches.toArray(new String[matches.size()]);
	}

This uses a far simpler regex with the aim of matching var = "non-double quote chars", with a capturing group for each of the required elements, and iterate through each 'find' that the regex matches.

Personally I'd prefer the latter because it's more maintainable.

I have tested the second approach and it works.
Thank you very much.

However, I should have said earlier, that the file also contains lines like:

i = 5
j = 3.677
k = true
l = false

Which now can not be parsed anymore.

I changed your code to account for this, if anybody is interested in my solution, here it is:

	public static String[] splitConfig(String str) {
		System.out.println("splitConfig"+str);
		Pattern pattern = Pattern.compile("(\\w+) (=) ([^\"]+|\"[^\"]+\")");
		Matcher matcher = pattern.matcher(str);
 
		List<String> matches = new ArrayList<>();
		while (matcher.find()) {
			for (int i = 1; i <= matcher.groupCount(); i++) {
				String matchedStr = matcher.group(i);
				matches.add(matchedStr);
			}
		}
		String[] result = matches.toArray(new String[matches.size()]);
		System.out.println(Arrays.toString(result));
		return result;
	}

This will parse the following text, line by line:

a = 5
b = 3.3331
c = false
d = "Hello World"

to this String array:

String[] {
"a", "=", "5",
"b", "=", "3.3331",
"c", "=", "false",
"d", "=", "\"Hello World\""
}

Thank you again, very useful answer.