使用正则表达式捕获与不包括空格的字符混合的数字作为数字

Question

I have a text that can contain numbers, letters and special characters and I want to extract all the numbers in it using a regular expression.

The tricky part is that any two numbers with characters in between should be extracted as a whole number. Any two numbers with spaces in between should yield as two separate numbers.

Example: ds[44]%6c should yield 446 but 2021 ds[44]%6c should yield 2021 , 446

I have tried the following Regex

(-?\d+)

Which works fine to some extent, but I don't know how to match until I see a whitespace and ignore the chars between the numbers.

Answer 1

Matcher.results()

We can create a regular expression that captures a sequence containing at least one digit enclosed with zero or more non-white-space characters both on the left and on the right.

And with this regular expression using Java 9 Matcher.results() we can generate a Stream of MatchResult s, which is an object containing information about the matching group.

The only thing left is to extract the matching groups, eliminate non-digit characters, and collect the result.

public static final Pattern TEXT_WITH_DIGITS = Pattern.compile("[^\\s]*\\d+[^\\s]*");

public static List<Integer> getInts(String str) {
    
    return TEXT_WITH_DIGITS.matcher(str).results() // Stream<MatchResult>
        .map(MatchResult::group)                   // Stream<String> - extract the matching string
        .map(s -> s.replaceAll("\\D+", ""))        // remove non-digit characters
        .map(Integer::valueOf)                     // Stream<Integer> - parse the string
        .toList();
}

Pattern.splitAsStream()

Another option is to split the given string on white spaces. For that, we can make use of the Java 8 Pattern.splitAsStream() which generates a stream of elements identical to the ones produced by String.split() . The difference is that Pattern.splitAsStream() creates a stream directly from the regex engine without allocating an intermediate array in memory.

Then to apply the same transformation as in the previous example with one small addition: we need to address the edge-case when the givens string starts with a white-space. In such case, the very first element would be an empty string, and we can use dropWhile() to discard such string.

public static final Pattern WHITE_SPACES = Pattern.compile("\\s+");

public static List<Integer> getInts(String str) {
    
    return WHITE_SPACES.splitAsStream(str)
        .dropWhile(String::isEmpty)         // very first element might be empty, and if that's the case it needs to be skipped
        .map(s -> s.replaceAll("\\D+", "")) // remove non-digit characters
        .map(Integer::valueOf)              // Stream<Integer> - parse the string
        .toList();
}

main()

public static void main(String[] args) {
    System.out.println(getInts("ds[44]%6c"));
    System.out.println(getInts("2021 ds[44]%6c"));
}

Output:

[446]       // "ds[44]%6c"
[2021, 446] // "2021 ds[44]%6c"