Regex match between delimiters

Question

I am trying to figure out a way to match domain name in a config file. My file may look like this:

<abc="xyz">abc.com</abc>

I want to match abc.com and replace it with placeholder text. My current Java solution was replacing exact domain in text by calling StrName.replace("abc.com","random") However, it will also replace abc.commmmm with randommmmm. I don't want that. So, I tried regex. However, my regex [.><](abc.com)[.<>] will also select > and < in the string. I don't want that. I also realized that the text could be something like this:

<abc="xyz">text.abc.com.net</abc>

I still want to replace abc.com, so my regex won't work. How do I correct this?

Answer 1

You may want to consider using positive look-ahead and positive look-behind in your regex. See: http://www.regular-expressions.info/lookaround.html

Positive look-ahead is written like THING1(?=THING2) and it means find THING1 followed by THING2.

Positive look-behind is written like (?<=THING1)THING2 and it means find THING2 followed by THING1

In both of these cases the THING within the lookahead will not be consumed. For your first example you could do something like: (?<=>)abc\.com(?=<)

meaning abc.com preceded by > from (?<=>) and followed by < from (?=<).

If you are also looking to replace abc.com between periods as in text.abc.com.... you can try:

(?<=[.><])abc\.com(?=[.><])

meaning abc.com preceded by <,> or . from (?<=[.><]) and followed by <,> or . from (?=[.><]).

This will give you <abc="xyz">random</abc> and <abc="xyz">text.random.net</abc> for your two examples, as well as no match for <abc="xyz">abc.commmm</abc>

NOTE: escape the period by using \. instead of . otherwise you will match abcdcom as . matches any charachter. See: http://docs.oracle.com/javase/6/docs/api/java/util/regex/Pattern.html

Answer 2

Besides using look-arounds, you can do this with standard capturing groups.

First, define what you're looking for...

Find abc.com when it appears within tags — enclosed within > and <
and optionally has additional dotted prefixes and/or suffixes.
Replace abc.com with random when found in these circumstances.

This looks to me like a multi-level domain with two or more segments, which abc.com must be a part of.

What are the parts we're looking for?

1. The fixed string abc.com – the re for that is abc\.com escaping the dot so it's a literal instead of "any character".
2. Optional domain parts preceding "abc.com" – letters followed by a dot [a-z]+\.
   ... but there can be zero or more of them, so ([a-z]+\.)*
3. Optional domain parts following "abc.com" – a dot followed by letters, zero
   or more times (\.[a-z]+)*
4. All of that enclosed within the end of a start tag > and the start of an end tag < ... so >something<

Putting that all together we get >([a-z]+\.)*abc\.com(\.[a-z]+)*< which needs to be escaped to be a Java string ">([a-z]+\\.)*abc\\.com(\\.[a-z]+)*<"

Now, since we are matching and consuming the > and < we'll need them in the replacement string, and we're capturing the prefix and suffix so we need to put those in the replacement also using the capturing groups 1 and 2, giving >$1random$2<

I put this re in at regexplanet.com http://fiddle.re/fxhfxn using these test strings and producing the replacement strings:

<abc="xyz">abc.com</abc>           =>  <abc="xyz">random</abc>
<abc="xyz">abc.commmmm</abc>       =>  <abc="xyz">abc.commmmm</abc>
<abc="xyz">text.abc.com.net</abc>  =>  <abc="xyz">text.random.net</abc>
[abc="xyz"]text.abc.com.net[/abc]  =>  [abc="xyz"]text.abc.com.net[/abc]    
<abc="xyz">abc-com</abc>           =>  <abc="xyz">abc-com</abc>
<abc="xyz">abc.com.fr</abc>        =>  <abc="xyz">random.fr</abc>
<abc="xyz">www.abc.com</abc>       =>  <abc="xyz">www.random</abc>