fix: Parse MD5 mangled names

They are parsed correctly, but just returned as their mangled form.

Author: Swatinem

src/lib.rs

@@ -255,6 +255,7 @@ pub enum VarStorageKind {
 // Represents an identifier which may be a template.
 #[derive(Clone, PartialEq)]
 pub enum Name<'a> {
+    Md5(&'a [u8]),
     Operator(Operator<'a>),
     NonTemplate(&'a [u8]),
     AsInterface(&'a [u8]),
@@ -267,6 +268,10 @@ pub enum Name<'a> {
 impl<'a> fmt::Debug for Name<'a> {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         match *self {
+            Name::Md5(s) => f
+                .debug_tuple("Md5")
+                .field(&String::from_utf8_lossy(s))
+                .finish(),
             Name::Operator(ref op) => f.debug_tuple("Operator").field(&op).finish(),
             Name::NonTemplate(s) => f
                 .debug_tuple("NonTemplate")
@@ -500,6 +505,17 @@ impl<'a> ParserState<'a> {
             return Err(self.fail("does not start with b'?'"));
         }
 
+        if self.consume(b"?@") {
+            let name = self.read_md5_name()?;
+            return Ok(ParseResult {
+                symbol: Symbol {
+                    name,
+                    scope: NameSequence { names: Vec::new() },
+                },
+                symbol_type: Type::None,
+            });
+        }
+
         if self.consume(b"$") {
             if self.consume(b"TSS") {
                 let mut guard_num: i32 = i32::from(
@@ -687,6 +703,21 @@ impl<'a> ParserState<'a> {
         }
     }
 
+    /// An MD5 mangled name is `??@` followed by 32 characters and a terminating `@`.
+    ///
+    /// See https://github.com/llvm/llvm-project/blob/818cf30b83305fa4a2f75821349210b0f7aff4a4/llvm/lib/Demangle/MicrosoftDemangle.cpp#L754
+    fn read_md5_name(&mut self) -> Result<Name<'a>> {
+        let start_offset = self.offset;
+
+        while self.read_hex_digit().is_some() {}
+        let end_offset = self.offset;
+
+        if self.offset - start_offset != 32 || !self.consume(b"@") {
+            return Err(self.fail("expected MD5 mangled name of length 32"));
+        }
+        Ok(Name::Md5(&self.input.as_bytes()[start_offset..end_offset]))
+    }
+
     fn read_digit(&mut self) -> Option<u8> {
         match self.peek() {
             Some(first) => {
@@ -2180,6 +2211,11 @@ impl<'a> Serializer<'a> {
 
     fn write_one_name(&mut self, name: &Name) -> Result<()> {
         match *name {
+            Name::Md5(ref name) => {
+                write!(self.w, "??@")?;
+                self.w.write_all(name)?;
+                write!(self.w, "@")?;
+            }
             Name::Operator(ref op) => {
                 self.write_space()?;
                 self.write_operator_name(op)?;
@@ -2244,6 +2280,11 @@ impl<'a> Serializer<'a> {
         }
 
         match names.name {
+            Name::Md5(ref name) => {
+                write!(self.w, "??@")?;
+                self.w.write_all(name)?;
+                write!(self.w, "@")?;
+            }
             Name::Operator(ref op) => {
                 match *op {
                     Operator::Ctor => {

tests/test_basics.rs

@@ -184,7 +184,12 @@ fn other_tests() {
 
     expect(
         "?Present1@?QIDXGISwapChain4@@CDXGISwapChain@@UAGJIIPBUDXGI_PRESENT_PARAMETERS@@@Z", 
-        "public: virtual long __stdcall CDXGISwapChain::[IDXGISwapChain4]::Present1(unsigned int,unsigned int,struct DXGI_PRESENT_PARAMETERS const *)"
+        "public: virtual long __stdcall CDXGISwapChain::[IDXGISwapChain4]::Present1(unsigned int,unsigned int,struct DXGI_PRESENT_PARAMETERS const *)");
+
+    // An MD5 mangled name is "valid" but output as-is
+    expect(
+        "??@8ba8d245c9eca390356129098dbe9f73@",
+        "??@8ba8d245c9eca390356129098dbe9f73@",
     );
 }