Apex regex and non-capturing groups

Question

I have a regular expression that I want to use to pull ID's out of a JSON string. The expression itself appears to work fine based on the data I've given it, so that's good.

^(?:.*)(?:{"id":)(\d+)(?:.*)$

Unfortunately, it involves more than one non-capturing group ("(?:)"), and that seems to break Apex with the dreaded "Invalid regex at index 9." Is there actually a 1 non-capture group limit in Apex's regex flavor or is there something else wrong with this? I wasn't able to find any documentation on such a limit, so i'm inclined to think that there is some other error in my regex, but I can't find it!

Just for completeness, here's an example that is failing on my sandbox.

pattern myPattern = pattern.compile('^(?:.*)(?:{"id":)$'); 
matcher myMatcher = myPattern.matcher('{"user":{"id":11111,"userName":');
System.assert(myMatcher.matches() && myMatcher.hitEnd());

if(myMatcher.matches())
{
    system.debug(myMatcher.group(1));
}

* bolded for easier skimming

Answer 1

I was able to sort this out just now. Unfortunately, the error that the Force.com IDE gives is not terribly obvious, and the problem appears to be bug in the underlying Java interpreter (compiler? I forget).

Apex really doesn't like curly-brackets in strings. Double escaping them solves the problem!

pattern myPattern = pattern.compile('^(?:.*)(?:\\{"id":)(\\d+)(?:.*)$'); 
string examplestring = '\\{"user":\\{"id":11111,"userName":';
system.debug(examplestring);
matcher myMatcher = myPattern.matcher(examplestring);
System.assert(myMatcher.matches() && myMatcher.hitEnd());

if(myMatcher.matches())
{
    system.debug(myMatcher.group(1));
}

This is why it looked like the optional capture groups were the problem, my second one uses a curly-bracket!

Answer 2

Depending on the size of your payload, you might run out of heap space while parsing the string via regex.

If your payload is stable enough, this alternative method using the JSON serializer might allow you to fetch the User Id from the payload and ignore the rest as well as make it possible to quickly access attributes other than id without making your regex pattern more complicated.

String payload = '{"user":{"id":"11111","userName":"[email protected]", "photoUrl":"https://theurl.com/blah"}}';

Map<String, Object> jsonMap = (Map<String, Object>)JSON.deserializeUntyped(payload);

Map<String, Object> userAttributes = (Map<String, Object>)jsonMap.get('user');
system.debug(userAttributes);
// |DEBUG|{id=11111, photoUrl=https://theurl.com/blah, [email protected]}

String userId = (String)userAttributes.get('id');
system.debug(userId);
// |DEBUG|11111

---

Another approach if you would like to be a little more memory efficient, since the JSON attributes which are not defined in the concrete Types are ignored during deserialization. This method would be appropriate if your payload has a stable enough structure for you to build classes to represent the data that is important to your application.

public class UserWrapper {
    public UserElement user { get; set; }
}

public class UserElement {
    public string id       { get; set; }
    public string userName { get; set; }
    public string photoUrl { get; set; }
}

String payload = '{"user":{"id":"11111","userName":"[email protected]", "photoUrl":"https://theurl.com/blah"}}';

// deserialize into the outer object, discarding the elements that don't match
UserWrapper uWrapper = (UserWrapper)JSON.deserialize(payload, UserWrapper.class);

// easy reference to the user item
UserElement userElement = uWrapper.user;

String userId = userElement.id;

system.debug(userId);