I have a simple example in GNU Smalltalk 3.2.5 of attempting to group match on a key value setting:
st> m := 'a=b' =~ '(.*?)=(.*)'
MatchingRegexResults:'a=b'('a','b')
The above example works just as expected. However, if there is no match to the second group (.*), an exception is generated:
st> m := 'a=' =~ '(.*?)=(.*)'
Object: Interval new "<-0x4ce2bdf0>" error: Invalid index 1: index out of range
SystemExceptions.IndexOutOfRange(Exception)>>signal (ExcHandling.st:254)
SystemExceptions.IndexOutOfRange class>>signalOn:withIndex: (SysExcept.st:660)
Interval>>first (Interval.st:245)
Kernel.MatchingRegexResults>>at: (Regex.st:382)
Kernel.MatchingRegexResults>>printOn: (Regex.st:305)
Kernel.MatchingRegexResults(Object)>>printString (Object.st:534)
Kernel.MatchingRegexResults(Object)>>printNl (Object.st:571)
I don't understand this behavior. I would have expected the result to be ('a', nil) and that m at: 2 to be nil. I tried a different approach as follows:
st> 'a=' =~ '(.*?)=(.*)' ifMatched: [ :m | 'foo' printNl ]
'foo'
'foo'
Which determines properly that there's a match to the regex. But I still can't check if a specific group is nil:
st> 'a=' =~ '(.*?)=(.*)' ifMatched: [ :m | (m at: 2) ifNotNil: [ (m at: 2) printNl ] ]
Object: Interval new "<-0x4ce81b58>" error: Invalid index 1: index out of range
SystemExceptions.IndexOutOfRange(Exception)>>signal (ExcHandling.st:254)
SystemExceptions.IndexOutOfRange class>>signalOn:withIndex: (SysExcept.st:660)
Interval>>first (Interval.st:245)
Kernel.MatchingRegexResults>>at: (Regex.st:382)
optimized [] in UndefinedObject>>executeStatements (a String:1)
Kernel.MatchingRegexResults>>ifNotMatched:ifMatched: (Regex.st:322)
Kernel.MatchingRegexResults(RegexResults)>>ifMatched: (Regex.st:188)
UndefinedObject>>executeStatements (a String:1)
nil
st>
I don't understand this behavior. I would have expected the result to be ('a', nil) and that m at: 2 to be nil. At least that's the way it works in any other language I've used regex in. This makes me think maybe I'm not doing something correct with my syntax.
My question this is: do I have the correct syntax for attempting to match ASCII key value pairs like this (for example, in parsing environment settings)? And if I do, why is an exception being generated, or is there a way I can have it provide a result that I can check without generating an exception?
I found a related issue reported at gnu.org from Dec 2013 with no responses.
The issue had been fixed in master after the above report was received. The commit can be seen here. A stable release is currently blocked by the glib event loop integration.