approved

Absolutely brilliant!

Approved

• Damn, you're right – thanks for noticing.
• Eh, I don't think it's critical.
• Oh, you're correct, it is exploitable indeed. When writing large test cases, it looks like I just copy-pasted the small test parts and forgot to modify them. I just assumed something about myself that I didn't do. It's not my code to begin with; I just did some cleaning 😅

I think it should be good now. Please review it again.

yeah sure

Fixed in this fork

lookup table gang

• rand_ascii_remove_count's return type is accidentally set to bool rather than std::size_t
• Consider raising the upper bound for dist_ascii_remove_count from 25 to 26 (minor boost in coverage)
• Having alternating sets of 25 valid and invalid random tests is a bit of a predictable pattern. It could be exploited, although I doubt someone is willing to go through all that trouble. It's up to you whether you think this a valid concern; if so, consider shuffling to intersperse valid and invalid tests.

Approved, but I don't have the privileges to close this.

Approved