Once or not at all regex group returns null

Question
1 
val p = """^(\d+)(?:-(\d+))?$""".r
val m = p.findAllIn("2")
for i <- 1 to m.groupCount do println(m.group(i))

Output:

2
null

This is unexpected, perhaps a bug? There should only be one group, not two. regex101 behaves as expected, and only returns one group matching 2.

2

#groupCount returns all groups present in the matcher’s backing pattern, not just the matched groups, and this feels quite reasonable. Calling #group(Int) on a non-matching group yields null - which is unfortunate, but probably an artifact of directly mapping to the underlying Java regex API.

3

But Scala is not Java.

4

Yes, so…? :slightly_smiling_face: The links go to Scala documentation. And reporting two groups makes sense, no matter the language. Cf.

$ python -c 'import re; print(re.search(re.compile("^(\\d+)(?:-(\\d+))?$"), "2").groups())'
('2', None)

The only complaint I have with the Scala behavior is getting a null rather than a proper Option for an unmatched group - and this is likely due to Java heritage.

1 Like
5

Example lib for safety

3 Likes