[Parse] Refine UTF8 validation-related aspects

[pinkjuice66]

Refined some UTF8 validation-related aspects.

• '0x80' is not a valid byte for the start of a UTF8 character; it's a continuation byte.(RFC 3629)
  Ensure that the function call isStartOfUTF8Character(0x80) returns false.
• Removed a duplicate condition check : (EncodedBytes == 1 || !isStartOfUTF8Character(CurByte))
  The expression !isStartOfUTF8Character(CurByte) should handle the pattern of 0b10xxxxxx when the EncodedBytes evaluates to 1.
• Removed CLO8(:) function.
  Guess there have been API changes in the LLVM function, and we no longer require the type conversion wrapper function for it. This is because the llvm::countl_one function now takes a parameter as a template type.

[ahoppen]

Thanks. Looks good to me 👍🏽

Could you also open a corresponding PR to fix this in the new lexer that’s implemented in Swift? https://github.com/apple/swift-syntax/blob/44f69680412c7802a7512f4e34f143c210b4dee2/Sources/SwiftParser/Lexer/UnicodeScalarExtensions.swift

[ahoppen]

Out of curiosity: Do you know of any character that uses a 0x80 continuation byte? If so, it would be nice to add a test case containing it.

[pinkjuice66]

@ahoppen
Added a test case that uses edge continuation bytes (0x80 and 0xBF), and another case for the lexer to accurately diagnose when there's a 0x80 in the source.

[pinkjuice66]

@ahoppen
Sure, will be following up on it in the Swift version.

BTW, recently started studying Swift compiler infrastructure and am a big fan of your playground series introducing the compiler. I appreciate your efforts for the community; it helps a lot.

[ahoppen]

Thanks for the extensive test cases!

[ahoppen]

@swift-ci Please smoke test

[ahoppen]

@swift-ci Please smoke test Linux