xeger not handling certain non-capturing groups

[Pierlou]

Hi, thanks for this lib!
I have come across a weird behaviour using rstr.xeger: this pattern "(?:(?:^|;)(AAA|BBB|CCC))+$" aims to capture any string that contains groups of strings within ["AAA", "BBB", "CCC"], separated by semicolons. These are valid strings:

• BBB;AAA
• BBB;AAA;BBB
• CCC

However xeger returns things like:

• BBBCCC;CCCBBB : some groups are not separated
• ;AAA;BBBCCC : the separator is also at the start of the string

[bjmc]

Interesting. Someone else opened an issue asking about non-capturing groups, so I wonder if there's something we're handling incorrectly in that situation. I'll try to investigate when I have some time, but I'm also happy to review a PR if you're feeling ambitious. The group-handling is here.

[bjmc]

I think the issue here is how we're handling the ^ character in combination with the + repeat.

A slightly modified version of your regex, '(?:(?:;)(AAA|BBB|CCC))+$' produces correctly matching strings. For example:

• ;AAA;BBB;BBB;CCC;BBB
• ;BBB;CCC;BBB;BBB;CCC;BBB;AAA;CCC;AAA;BBB;BBB

the separator is also at the start of the string

I think that's correct? In the group (?:^|;) this regex is saying match either the start-of-input anchor ^ or a semicolon ;. Unless there's something I'm misunderstanding, it's legitimate to start with ;

some groups are not separated

I think this is the bug, more or less: the "not separated" cases are the result of the situation where random.choice() has picked the ^ instead of the ; but we don't output any printable character for "start of input" so we wind up with repeats that are missing separators.