Exclude all unicode line terminator characters from matching dot

[anthony-chang]

Signed-off-by: Anthony Chang antchang@nvidia.com

Fixes #5415

The wildcard . matching should exclude all line terminators defined in Java (Line terminator section). This PR adds exclusions for next-line (\u0085), line-separator (\u2028), and paragraph-separator (\u2029) characters.

[NVnavkumar]

Can we add more unit tests to test these cases? In particular add cases where these characters are in the input string, and the regular expression should match in CPU and GPU?

[NVnavkumar]

Changes look good. One more thing, can you try updating the variable here in RegularExpressionTranspilerSuite.scala, add the 3 unicode line terminators to it, and re-run the suite? This should enable the fuzz testing to generate more examples with line terminators to confirm that there aren't any edge cases missed.

  private val REGEXP_LIMITED_CHARS_COMMON = "|()[]{},.^$*+?abc123x\\ \t\r\n\f\u000bBsdwSDWzZ"

[anthony-chang]

build