Groups

Multiple expressions can be grouped together by wrapping them in (). This is useful when we have multiple alternatives that all start or end with the same thing:

'tang' ('ible' | 'ent' | 'o')

This matches the words tangible, tangent and tango.

Capturing groups #

Groups can also be used to capture their content, for example to replace it with something else. In a regex, every group is a capturing group by default. This is not the case in Pomsky: Capturing groups must be prefixed with :.

:('foo')

Capturing groups are consecutively numbered, to be able to refer to them later:

:('Max' | 'Laura')
(' is ' | ' was ')
:('asleep' | 'awake')

The first group, containing the name, has index 1, the third group with the adverb has the index 2. The second group is skipped because it isn’t capturing (it isn’t prefixed with :).

Named capturing groups #

Because groups are non-capturing by default, you can add parentheses freely without accidentally changing the capturing group indices. However, it’s usually better to use named capturing groups, so you don’t need to count groups and instead refer to each group by a name:

:name('Max' | 'Laura')
(' is ' | ' was ')
:adverb('asleep' | 'awake')

Now, the capturing groups are named name and adverb.

Atomic groups #

If we put the atomic keyword in front of a group, the RegExp engine can’t backtrack into the group. This can improve matching performance and prevent “catastrophic backtracking”:

% atomic('if' | 'else' | 'while' | 'for') %

You can find out more about backtracking here.

Note that atomic groups are only supported in the Java, PCRE, Ruby and .NET flavors.