[Demangle][Rust] Parse array type

[dblaikie]

llvm/test/Demangle/rust.test

160–165

Is it worth having two tests here? The parser seems fairly orthogonally designed and generic args ('p') don't seem more likely to be buggy/weird than any other argument type, such as the u8/0 in the second case? Might be inclined towards only having the second test, unless there's some interesting interaction with the generic args I'm missing.

[tmiasko]

• Remove redundant test case.

[tmiasko]

llvm/test/Demangle/rust.test

160–165

I left only the second test case (I wanted to test that code is parsing a <type> and a <const>, which is unclear with placeholders accepted by both).

llvm/lib/Demangle/RustDemangle.cpp

 //          | <backref>                   // backref
 void Demangler::demangleType() {
   char C = consume();
   BasicType Type;
   if (parseBasicType(C, Type))
     return printBasicType(Type);
 
   switch (C) {
+  case 'A':
+    print("[");
+    demangleType();
+    print("; ");
+    demangleConst();
+    print("]");
+    break;
   case 'C':
   case 'M':
   case 'X':
   case 'Y':
   case 'N':
   case 'I':
     Position -= 1;
     demanglePath(true /* InType */);

llvm/test/Demangle/rust.test

CHECK: <<_ as Name>>
        _RMC5crateYpC4Name
 
 CHECK: <name::Name>
        _RMC5crateNvC4name4Name
 
 CHECK: <Name<>>
        _RMC5crateIC4NameE
 
+; Types
+
+CHECK: types::<[_; _]>
+       _RIC5typesAppE
+
+CHECK: types::<[u8; 0]>
+       _RIC5typesAhj0_E
+
 ; Integer constants. Test value demangling.
 
 CHECK: integer::<0>
        _RIC7integerKi0_E
 
 CHECK: integer::<1>
        _RIC7integerKi1_E