internally recursive regular expression java

Question

I have the following sql queries I'd like to be able to group in one regular expression:

CREATE INDEX blah_idx ON blah (id ASC)

CREATE INDEX blah2_idx ON blah2 (foo ASC, id DEC)

I'd like to be able to use a java regex to group these so that I get:

blah_idx, blah, id, ASC

blah2_idx, blah2, foo, ASC, id, DEC

I can get the first with CREATE INDEX (\\\w+) ON (\\\w+) \\((\w+) (\w+) \\) but I'd like to be able to also group the second but I can't see to define \\((\w+) (\w+) \\) to match repeatedly.

Is this even possible?

Joop Eggen · Accepted Answer · 2013-03-08 17:11:35Z

1

I left out some parentheses for readability. And spaces can be \\s+ or *.

"CREATE INDEX \\w+ ON \\w+ \\((\\w+ (ASC|DESC)(, \\w+ (ASC|DEC))*))\\)"
                              1     2        23       4       43 21

Nested groups ( ( ) ) are allowed and are numbered from left to right. For retrieval see the javadoc.

    final String[] sqls = {
        "CREATE INDEX blah_idx ON blah (id ASC)",
        "CREATE INDEX blah2_idx ON blah2 (foo ASC, id DEC)",
        "CREATE INDEX blah2_idx ON blah2 (foo ASC, id DEC, name ASC)",
    };

    final Pattern createIndexPattern = Pattern.compile(
      "CREATE INDEX (\\w+) ON (\\w+) \\(((\\w+) (ASC|DESC)(, (\\w+) (ASC|DEC))*)\\)");
    for (String sql : sqls) {
        System.out.println("SQL: " + sql);
        Matcher m = createIndexPattern.matcher(sql);
        if (!m.matches()) {
            System.out.println("No match!");
        } else {
            System.out.println("Match!");
            int groupCount = m.groupCount();
            for (int groupI = 1; groupI <= groupCount; ++groupI) {
                System.out.printf("[%d] %s%n", groupI, m.group(groupI));
            }
            String[] fieldsWithOrdering = m.group(3).split(",\\s*");
            System.out.println(Arrays.toString(fieldsWithOrdering));
        }
    }

edited Mar 8, 2013 at 17:11

answered Mar 8, 2013 at 15:46

Joop Eggen

110k8 gold badges89 silver badges142 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

tophersmith116 Over a year ago

oh that's very cool. I think I can work with this, since I know I have at most 3 sets of that foo ASC, section. Thank you!

Community · Accepted Answer · 2017-05-23 12:28:29Z

1

Reminds me of a question I once asked:

How to match nested function invocations (bracket pairs) using a regular expression (recursive?)

Not possible in most Regexp languages including Java unfortunately.

edited May 23, 2017 at 12:28

CommunityBot

11 silver badge

answered Mar 8, 2013 at 15:42

Steve Chambers

39.8k29 gold badges179 silver badges222 bronze badges

4 Comments

Rahul Over a year ago

could have posted this as a comment I suppose.

tophersmith116 Over a year ago

ah bummer... I guess I'll just match the outer, and then match the stuff inside the parens individually?

Steve Chambers Over a year ago

Usually have to resort to doing that kind of thing with regexes - at least it keeps it simpler for those reading your code.

Qtax Over a year ago

@tophersmith116, or do it properly and parse it instead. @ Steve, most popular regex flavors support recursion (Perl/PCRE/etc), or similarly powerful constructs (balancing groups in .NET).

Collectives™ on Stack Overflow

internally recursive regular expression java

2 Answers 2

1 Comment

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related