[Clang] Do not check for underscores in isAllowedInitiallyIDChar

[cor3ntin]

Summary

isAllowedInitiallyIDChar is only used with non-ASCII codepoints,
which are handled by isAsciiIdentifierStart.
To make that clearer, remove the check for _ from
isAllowedInitiallyIDChar, and assert on ASCII - to ensure neither
_ or $ are passed to this function.

[aaron.ballman]

The change is fine by me, but you should put NFC in the patch title so it's more clear that this is simplifying code in a way that won't change behavior (and doesn't need tests).

clang/lib/Lex/Lexer.cpp

1486

nit: 80-col limit

Also, shouldn't this say it was called with an ASCII codepoint?

[cor3ntin]

Fix assert message

[cor3ntin]

clang/lib/Lex/Lexer.cpp

1486

It should. I started with "isAllowedInitiallyIDChar should not be called with a non-ASCII codepoint") and... yeah, thanks for catching that!
It fits nicely in 80 cols too now :)

[aaron.ballman]

LGTM assuming @tahonermann doesn't spot concerns.

[tahonermann]

LGTM, thanks Corentin.

[cor3ntin]

Thank you both!

clang/lib/Lex/Lexer.cpp

static bool isAllowedIDChar(uint32_t C, const LangOptions &LangOpts) {
   } else {
     static const llvm::sys::UnicodeCharSet C99AllowedIDChars(
         C99AllowedIDCharRanges);
     return C99AllowedIDChars.contains(C);
   }
 }
 
 static bool isAllowedInitiallyIDChar(uint32_t C, const LangOptions &LangOpts) {
+  assert(C > 0x7F && "isAllowedInitiallyIDChar called with an ASCII codepoint");
   if (LangOpts.AsmPreprocessor) {
     return false;
   }
   if (LangOpts.CPlusPlus || LangOpts.C2x) {
     static const llvm::sys::UnicodeCharSet XIDStartChars(XIDStartRanges);
-    // '_' doesn't have the XID_Start property but is allowed in C++.
-    return C == '_' || XIDStartChars.contains(C);
+    return XIDStartChars.contains(C);
   }
   if (!isAllowedIDChar(C, LangOpts))
     return false;
   if (LangOpts.C11) {
     static const llvm::sys::UnicodeCharSet C11DisallowedInitialIDChars(
         C11DisallowedInitialIDCharRanges);
     return !C11DisallowedInitialIDChars.contains(C);
   }