Java regex pattern for a REST filter/search param?

Rory :

I have a filter param in my REST API that can look like this:

owner='bob'                            // owner is bob
owner!'bob'                            // owner is not bob
owner.town='bel*'                      // owner's town starts with bel
owner.town='*bel*'                     // owner's town contains bel
owner.car='{ some json blob here}'     // owner's car equals some json blob
owner.car~                             // owner has a property named car

So I want to capture:

A required String key
The can contain letters, numbers, underscores, dots, hyphens or colons
A required operator: +|!|~
An optional value, that can begin and/or end with an optional asterisk
If the optional value is present, it must be surrounded by single quotes

I have started with the following, buts it's not returning what I'd expect:

String filter = "owner='bob'";
Pattern.compile("(\\w+?)(=|!|~)(*?)(\\w+?)(*?)");
final Matcher matcher = pattern.matcher(filter);

// Results
matcher.group(1)   // "owner"
matcher.group(2)   // "="
matcher.group(3)   // "\"
matcher.group(4)   // ""
matcher.group(5)   // "bob"
matcher.group(6)   // ""
matcher.group(7)   // "\"

Problems I'm having:

I don't think (*?) is correctly capturing zero or one asterisks
Passing "owner~" results in a failure when invoking matcher.group(anything)

I'm pretty sure there are other problems with the regex I havn't found yet...

Aaron :

I would use the following regex :

^([\w.:-]+)([=!~])('.*')?$

It is composed of three consecutive groups defined as follows :

a first group matches the key, which is a non-empty sequence of letters, numbers, underscores, dots, hyphens or colons.
a second group matches the operator, which is one of "=", "!" or "~"
a third optional group matches the rest, which would be the property value when the operator isn't "~"

You can try it here.

Note that if the mention that the value can start or end with a * is supposed to imply that it can't contain * in other places you will want to change the third group into ('\*?[^*]*\*?').

Java regex pattern for a REST filter/search param?

Guess you like