CODEWITHSUNDEEP
javaregex
Eduard Paraschivescu
Eduard Paraschivescu

Reputation: 13

How to use capture group as start of expression in regex?

The task at hand: I am trying to prepare for printing a specific list of automatically generated ids. They are of the format aa-bb-cc-dd-ee-ff-gg... every tuple is selectable by [a-zA-Z0-9]+ (of indeterminate length), the delimiters are [-] (max one).

There are anywhere between one and nine tuples in every id. If the id is 3 tuples or less, I would return one group. If the id is more than 3 tuples (4+) then I would return two groups, the first one being composed of 3 tuples, and the second one of the rest.

Only one string would be treated at a time. Here is the test set:

 one1
 one1-two2
 one1-two2-three3
 one1-two2-three3-4a
 one1-two2-three3-4a-5a
 one1-two2-three3-4a-5a-6a
 one1-two2-three3-4a-5a-6a-7a

Concretely that would mean: 

 one1 -> {"one1"}
 one1-two2 -> {"one1-two2"}
 one1-two2-three3 -> {"one1-two2-three3"}
 one1-two2-three3-4a -> {"one1-two2-three3", "4a"}
 one1-two2-three3-4a-5a -> {"one1-two2-three3", "4a-5a"}
 one1-two2-three3-4a-5a-6a -> {"one1-two2-three3", "4a-5a-6a"}
 one1-two2-three3-4a-5a-6a-7a -> {"one1-two2-three3", "4a-5a-6a-7a"}

Work done up until now (this always properly selects the first group)

(^[a-zA-Z0-9]+$)|(^[a-zA-Z0-9]+[-][a-zA-Z0-9]+$)|(^[a-zA-Z0-9]+[-][a-zA-Z0-9]+[-][a-zA-Z0-9]+$)|(^[a-zA-Z0-9]+[-][a-zA-Z0-9]+[-][a-zA-Z0-9]+)

What I am trying to achieve: start at the end of the capture group, check if it is not the end of the line, start reading after the first '-' char following that point, match until the end of the line.

Additional information : I am using Java's native regex engine.

Upvotes: 1

Views: 54

Answers (2)

Andreas
Andreas

Reputation: 159135

The following regex will match only valid strings, and returns 2 capture groups:

([a-zA-Z0-9]+(?:-[a-zA-Z0-9]+){0,2})(?:-([a-zA-Z0-9]+(?:-[a-zA-Z0-9]+)*))?

Explanation

(                            Start capture of group 1:
  [a-zA-Z0-9]+                 Match first tuple of group 1
  (?:-[a-zA-Z0-9]+){0,2}       Match 0-2 delimiter+tuple pairs for a total of 1-3 tuples
)                            End capture of group 1
(?:                          Optional:
  -                            Match delimiter initiating group 2
  (                            Start capture of group 2:
    [a-zA-Z0-9]+                 Match first tuple of group 2
    (?:-[a-zA-Z0-9]+)*           Match 0+ delimiter+tuple pairs for a total of 1+ tuples
  )                            End capture of group 2
)?                           End optional

Demo

public static void main(String... args) {
    test("one1",
         "one1-two2",
         "one1-two2-three3",
         "one1-two2-three3-4a",
         "one1-two2-three3-4a-5a",
         "one1-two2-three3-4a-5a-6a",
         "one1-two2-three3-4a-5a-6a-7a",
         "one1-two2-three3-4a-5a-6a-7a-8a",
         "one1_two2"); // fail: invalid character
}

private static void test(String... values) {
    Pattern p = Pattern.compile("([a-zA-Z0-9]+(?:-[a-zA-Z0-9]+){0,2})(?:-([a-zA-Z0-9]+(?:-[a-zA-Z0-9]+){0,3}))?");
    for (String value : values) {
        Matcher m = p.matcher(value);
        if (! m.matches())
            System.out.printf("%s -> NO MATCH%n", value);
        else if (m.start(2) == -1) // capture group 2 not found
            System.out.printf("%s -> {\"%s\"}%n", value, m.group(1));
        else
            System.out.printf("%s -> {\"%s\", \"%s\"}%n", value, m.group(1), m.group(2));
    }
}

Output

one1 -> {"one1"}
one1-two2 -> {"one1-two2"}
one1-two2-three3 -> {"one1-two2-three3"}
one1-two2-three3-4a -> {"one1-two2-three3", "4a"}
one1-two2-three3-4a-5a -> {"one1-two2-three3", "4a-5a"}
one1-two2-three3-4a-5a-6a -> {"one1-two2-three3", "4a-5a-6a"}
one1-two2-three3-4a-5a-6a-7a -> {"one1-two2-three3", "4a-5a-6a-7a"}
one1-two2-three3-4a-5a-6a-7a-8a -> {"one1-two2-three3", "4a-5a-6a-7a-8a"}
one1_two2 -> NO MATCH

Upvotes: 0

revo
revo

Reputation: 48741

You don't need to over-complicate things to get around the problem: 

(?m)^(\w+(?:-\w+){0,2})(?:-(\w+(?:-\w+)*))?$

(?m) enables multiline flag which makes ^ and $ anchors to match beginning and end of each line respectively. A match starts with matching word characters \w+ then up to two more of -\w+ patterns which builds first capturing group information.

Second capturing group contains whatever comes after. If you are sure about formatting you could do this too:

(?m)^(\w+(?:-\w+){0,2})(.+)?$

Test it on live demo

Upvotes: 1

Related Questions