Normalization can make a mismatch around regex

[m1kit]

I found a very minor corner case in dynamically generated test cases.

Given a regex .{,20} the random test generator generates a string like this:

B_V$n<?2.M0$@1CjSjyt

I expected this string to match the regular expression.

However, the normalizer normalizes the string like this:

b-`v$n<?2.`m0$@1cjsjyt

Note that there are two spaces inserted after _ and ..

This normalization process itself is expected behavior, but this normalized string does not match to the regex. This is because the length of the string is increased to 22 by two spaces.

Hi @goneall @anshuldutt21, please let me know your feelings on this issue.
I think this is a very minor issue and we may ignore. Instead, we can change a way to generate random tests (like treat a pattern . as [a-zA-Z0-9]).

[goneall]

I think this is a very minor issue and we may ignore. Instead, we can change a way to generate random tests (like treat a pattern . as [a-zA-Z0-9]).

Agree it is a very minor issue - it would be better to resolve so that we get a cleaner set of test results. Changing the pattern for the generated tests seems quite reasonable as it would test the most likely scenarios.