CODEWITHSUNDEEP
javaregex
tsiki
tsiki

Reputation: 1799

Matching several URLs in a string using regex

I'm trying to match a URL in a string, using regex from here: Regular expression to match URLs in Java

It works fine with one URL, but when I have two URLs in the string, it only matched the latter.

Here's the code:

Pattern pat = Pattern.compile(".*((https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|])", Pattern.DOTALL);
Matcher matcher = pat.matcher("asdasd http://www.asd.as/asd/123 or http://qwe.qw/qwe");
// now matcher.groupCount() == 2, not 4

Edit: stuff I've tried:

// .* removed, now doesn't match anything // Another edit: actually works, see below
Pattern pat = Pattern.compile("((https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|])", Pattern.DOTALL);

// .* made lazy, still only matches one
Pattern pat = Pattern.compile(".*?((https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|])", Pattern.DOTALL);

Any ideas?

Upvotes: 0

Views: 3089

Answers (1)

Martin Ender
Martin Ender

Reputation: 44259

It's because .* is greedy. It will just consume as much as possible (the whole string) and then backtrack. I.e. it will throw away one character at a time until the remaining characters can make up a URL. Hence the first URL will already be matched, but not captured. And unfortunately, matches cannot overlap. The fix should be simple. Remove the .* at the beginning of your pattern. Then you can also remove the outer parentheses from your pattern - there is no need to capture anything any more, because the whole match will be the URL you are looking for.

Pattern pat = Pattern.compile("(https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|]", Pattern.DOTALL);
Matcher matcher = pat.matcher("asdasd http://www.asd.as/asd/123 or http://qwe.qw/qwe");
while (matcher.find()) {
  System.out.println(matcher.group());
}

By the way, matcher.groupCount() does not tell you anything, because it gives you the number of groups in your pattern and not the number of captures in your target string. That's why your second approach (using .*?) did not help. You still have two capturing groups in the patter. Before calling find or anything, matcher does not know how many captures it will find in total.

Upvotes: 5

Related Questions