GT-3409 MDMang MD5-hashed symbols

Ghidra/Features/MicrosoftDmang/src/main/java/mdemangler/MDMang.java

@@ -480,6 +480,17 @@ public class MDMang {
 	public MDObjectCPP getEmbeddedObject(MDObjectCPP obj) {
 		return obj;
 	}
+
+	/**
+	 * This method is meant to be overridden as needed to process a hashed
+	 *  object.  In this default case (MDMang), we properly process the
+	 *  hashed object.  Overridden methods might just throw an exception,
+	 *  allowing a failed demangling.
+	 */
+	public void processHashedObject(MDObjectCPP obj) throws MDException {
+		obj.processHashedObject();
+	}
+
 }
 
 /******************************************************************************/

Ghidra/Features/MicrosoftDmang/src/main/java/mdemangler/MDMangGhidra.java

@@ -46,10 +46,6 @@ public class MDMangGhidra extends MDMang {
 		return dataTypeResult;
 	}
 
-	private DemangledType processNamespace(MDQualifiedBasicName qualifiedBasicName) {
-		return processNamespace(qualifiedBasicName.getQualification());
-	}
-
 	public DemangledType processNamespace(MDQualifiedName qualifiedName) {
 		return processNamespace(qualifiedName.getQualification());
 	}
@@ -183,8 +179,8 @@ public class MDMangGhidra extends MDMang {
 				}
 				else {
 					variable = new DemangledVariable(
-						objectCPP.getQualifiedName().getBasicName().toString());
-					variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+						objectCPP.getName());
+					variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				}
 				variable.setDatatype(dt);
 				resultObject = variable;
@@ -204,8 +200,8 @@ public class MDMangGhidra extends MDMang {
 			}
 			else if (typeinfo instanceof MDFunctionInfo) {
 				DemangledFunction function =
-					new DemangledFunction(objectCPP.getQualifiedName().getBasicName().toString());
-				function.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+					new DemangledFunction(objectCPP.getName());
+				function.setNamespace(processNamespace(objectCPP.getQualfication()));
 				resultObject = function;
 				objectResult = processFunction((MDFunctionInfo) typeinfo, function);
 				// Any other special values to be set?
@@ -233,8 +229,8 @@ public class MDMangGhidra extends MDMang {
 			else if (typeinfo instanceof MDVxTable) { //Includes VFTable, VBTable, and RTTI4
 				MDVxTable vxtable = (MDVxTable) typeinfo;
 				DemangledVariable variable =
-					new DemangledVariable(objectCPP.getQualifiedName().getBasicName().toString());
-				variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+					new DemangledVariable(objectCPP.getName());
+				variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				variable.setConst(vxtable.isConst());
 				variable.setVolatile(vxtable.isVolatile());
 				variable.setPointer64(vxtable.isPointer64());
@@ -245,8 +241,8 @@ public class MDMangGhidra extends MDMang {
 			}
 			else if (typeinfo instanceof AbstractMDMetaClass) { //Includes all RTTI, except RTTI4
 				DemangledVariable variable =
-					new DemangledVariable(objectCPP.getQualifiedName().getBasicName().toString());
-				variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+					new DemangledVariable(objectCPP.getName());
+				variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				resultObject = variable;
 				// The following code would be an alternative, depending on whether we get
 				//  customer complaints or other fall-out from having created a variable here.
@@ -254,8 +250,8 @@ public class MDMangGhidra extends MDMang {
 			}
 			else if (typeinfo instanceof MDGuard) {
 				DemangledVariable variable =
-					new DemangledVariable(objectCPP.getQualifiedName().getBasicName().toString());
-				variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+					new DemangledVariable(objectCPP.getName());
+				variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				resultObject = variable;
 				// The following code would be an alternative, depending on whether we get
 				//  customer complaints or other fall-out from having created a variable here.
@@ -264,8 +260,8 @@ public class MDMangGhidra extends MDMang {
 			else {
 				// Any others (e.g., case '9')
 				DemangledVariable variable =
-					new DemangledVariable(objectCPP.getQualifiedName().getBasicName().toString());
-				variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+					new DemangledVariable(objectCPP.getName());
+				variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				resultObject = variable;
 				// The following code would be an alternative, depending on whether we get
 				//  customer complaints or other fall-out from having created a variable here.
@@ -288,9 +284,9 @@ public class MDMangGhidra extends MDMang {
 			}
 		}
 		else {
-			String baseName = objectCPP.getQualifiedName().getBasicName().toString();
-			if (objectCPP.getQualifiedName().isString()) {
-				MDString mstring = objectCPP.getQualifiedName().getBasicName().getMDString();
+			String baseName = objectCPP.getName();
+			if (objectCPP.isString()) {
+				MDString mstring = objectCPP.getMDString();
 				DemangledString demangledString = new DemangledString(mstring.getName(),
 					mstring.toString(), mstring.getLength(), mstring.isUnicode());
 				resultObject = demangledString;
@@ -298,7 +294,7 @@ public class MDMangGhidra extends MDMang {
 			else if (baseName.length() != 0) {
 				DemangledVariable variable;
 				variable = new DemangledVariable(baseName);
-				variable.setNamespace(processNamespace(objectCPP.getQualifiedName()));
+				variable.setNamespace(processNamespace(objectCPP.getQualfication()));
 				resultObject = variable;
 			}
 		}
@@ -803,13 +799,10 @@ public class MDMangGhidra extends MDMang {
 
 	@Override
 	public MDObjectCPP getEmbeddedObject(MDObjectCPP obj) {
-		MDObjectCPP embedded = obj.getQualifiedName().getBasicName().getEmbeddedObject();
-		if (embedded != null) {
-			return embedded;
-		}
-		return obj;
+		return obj.getEmbeddedObject();
 	}
 }
 
 /******************************************************************************/
 /******************************************************************************/
+

Ghidra/Features/MicrosoftDmang/src/main/java/mdemangler/MDMangVS2015.java

@@ -111,6 +111,12 @@ public class MDMangVS2015 extends MDMang {
 	public MDObjectCPP getEmbeddedObject(MDObjectCPP obj) {
 		return obj.getEmbeddedObject();
 	}
+
+	@Override
+	public void processHashedObject(MDObjectCPP obj) throws MDException {
+		obj.processHashedObjectMSVC();
+	}
+
 }
 
 /******************************************************************************/

Ghidra/Features/MicrosoftDmang/src/main/java/mdemangler/naming/MDQualifiedBasicName.java

@@ -61,6 +61,10 @@ public class MDQualifiedBasicName extends MDParsableItem {
 		return -1;
 	}
 
+	/**
+	 * Returns {@code true} if the symbol's Basic Name is of a {@link MDString} type.
+	 *  @return {@code true} if Basic Name is of {@link MDString} type.
+	 */
 	public boolean isString() {
 		if (basicName == null) {
 			return false;
@@ -68,6 +72,18 @@ public class MDQualifiedBasicName extends MDParsableItem {
 		return basicName.isString();
 	}
 
+	/**
+	 * Returns the {@link MDString} from the Basic Name if it is a symbol of that type; else
+	 *  returns null.
+	 *  @return the {@link MDString} or null if does not exist.
+	 */
+	public MDString getMDString() {
+		if (isString()) {
+			return basicName.getMDString();
+		}
+		return null;
+	}
+
 	@Override
 	public void insert(StringBuilder builder) {
 		basicName.insert(builder);

Ghidra/Features/MicrosoftDmang/src/main/java/mdemangler/object/MDObjectCPP.java

@@ -15,10 +15,8 @@
  */
 package mdemangler.object;
 
-import mdemangler.MDException;
-import mdemangler.MDMang;
-import mdemangler.naming.MDBasicName;
-import mdemangler.naming.MDQualifiedBasicName;
+import mdemangler.*;
+import mdemangler.naming.*;
 import mdemangler.typeinfo.MDTypeInfo;
 import mdemangler.typeinfo.MDTypeInfoParser;
 
@@ -26,9 +24,11 @@ import mdemangler.typeinfo.MDTypeInfoParser;
  * This class represents a derivative of an <b><code>MDObject</code></b> which is a C++ object.
  */
 public class MDObjectCPP extends MDObject {
+	protected MDHashedObject hashedObject;
 	protected MDQualifiedBasicName qualifiedName;
 	protected MDTypeInfo typeInfo;
 	protected boolean embeddedObjectFlag;
+	protected boolean hashedObjectFlag;
 
 	public MDObjectCPP(MDMang dmang) {
 		super(dmang);
@@ -49,16 +49,66 @@ public class MDObjectCPP extends MDObject {
 	 */
 	public MDObjectCPP getEmbeddedObject() {
 		if (embeddedObjectFlag) {
-			return getQualifiedName().getBasicName().getEmbeddedObject();
+			return qualifiedName.getBasicName().getEmbeddedObject();
 		}
 		return this;
 	}
 
+	/**
+	 * Returns the name of the symbol, minus any namespace component.
+	 * @return the name.
+	 */
+	public String getName() {
+		if (hashedObjectFlag) {
+			return hashedObject.toString();
+		}
+		return getQualifiedName().getBasicName().toString();
+	}
+
+	/**
+	 * Returns the {@link MDQualification} component that represents the namespace.
+	 * @return the namespace information.
+	 */
+	public MDQualification getQualfication() {
+		if (hashedObjectFlag) {
+			return hashedObject.getQualification();
+		}
+		return getQualifiedName().getQualification();
+	}
+
+	/**
+	 * Returns {@code true} if the symbol's Basic Name is of a {@link MDString} type.
+	 *  @return {@code true} if Basic Name is of {@link MDString} type.
+	 */
+	public boolean isString() {
+		if (qualifiedName == null) {
+			return false;
+		}
+		return qualifiedName.isString();
+	}
+
+	/**
+	 * Returns the {@link MDString} from the Basic Name if it is a symbol of that type; else
+	 *  returns null.
+	 *  @return the {@link MDString} or null if does not exist.
+	 */
+	public MDString getMDString() {
+		if (isString()) {
+			return qualifiedName.getMDString();
+		}
+		return null;
+	}
+
 	@Override
 	public void insert(StringBuilder builder) {
-		qualifiedName.insert(builder);
-		if (typeInfo != null) {
-			typeInfo.insert(builder);
+		if (hashedObjectFlag) {
+			hashedObject.insert(builder);
+		}
+		else {
+			qualifiedName.insert(builder);
+			if (typeInfo != null) {
+				typeInfo.insert(builder);
+			}
 		}
 	}
 
@@ -71,23 +121,123 @@ public class MDObjectCPP extends MDObject {
 		if ((dmang.peek(0) == '?') && (dmang.peek(1) == '?')) { //??? prefix
 			embeddedObjectFlag = true;
 		}
-		qualifiedName = new MDQualifiedBasicName(dmang);
-		qualifiedName.parse();
-		if (qualifiedName.isString()) {
-			return;
+
+		if ((dmang.peek(0) == '?') && (dmang.peek(1) == '@')) { //??@ prefix
+			// MDMANG SPECIALIZATION USED.
+			dmang.processHashedObject(this);
 		}
-		if (dmang.peek() != MDMang.DONE) {
-			int RTTINum = qualifiedName.getRTTINumber();
-			typeInfo = MDTypeInfoParser.parse(dmang, RTTINum);
-			if (qualifiedName.isTypeCast()) {
-				typeInfo.setTypeCast();
+		else {
+			qualifiedName = new MDQualifiedBasicName(dmang);
+			qualifiedName.parse();
+			if (qualifiedName.isString()) {
+				return;
 			}
-			typeInfo.parse();
-			if (!typeInfo.getNameModifier().isEmpty()) {
-				qualifiedName.setNameModifier(typeInfo.getNameModifier());
+			if (dmang.peek() != MDMang.DONE) {
+				int RTTINum = qualifiedName.getRTTINumber();
+				typeInfo = MDTypeInfoParser.parse(dmang, RTTINum);
+				if (qualifiedName.isTypeCast()) {
+					typeInfo.setTypeCast();
+				}
+				typeInfo.parse();
+				if (!typeInfo.getNameModifier().isEmpty()) {
+					qualifiedName.setNameModifier(typeInfo.getNameModifier());
+				}
 			}
 		}
 	}
+
+	public void processHashedObject() throws MDException {
+		hashedObject = new MDHashedObject(dmang);
+		hashedObject.parse();
+		hashedObjectFlag = true;
+	}
+
+	public void processHashedObjectMSVC() throws MDException {
+		throw new MDException("cannot parse hashed symbol");
+	}
+
+	/**
+	 * Represents the MD5 hashed representation of the internals of the {@link MDObjectCPP}.
+	 *  It takes the place of the {@link MDQualifiedBasicName}.  We have included an unused
+	 *  (except to be able to return one that is empty) {@link MDQualification} so that the
+	 *  {@link MDObjectCPP} has one to return.
+	 * <p>
+	 * Not sure that we will keep this class in the long run or find a way to include it
+	 *  inside of the {@link MDObjectCPP} or if this {@link MDHashedObject} should be pulled
+	 *  out separately and detected/parsed by the {@link MDMangObjectParser}.  If this last
+	 *  thing is done, then we would have to find places in the package where we explicitly
+	 *  create {@link MDObjectCPP MDObjectCPPs}, such as in {@link MDNestedName}, and check to
+	 *  see if we can modify those locations (in all cases) to perform the detecting/parsing
+	 *  using the {@link MDMangObjectParser}.  This will need more study when more time is
+	 *  available.
+	 */
+	public class MDHashedObject extends MDParsableItem {
+		private String hashString = "";
+		private MDQualification qualification; // We are making this dummy object
+
+		/**
+		 * Constructor for {@link MDHashedObject}
+		 * @param dmang The {@link MDMang} for which the work is performed and from from which
+		 *  the information is parsed.
+		 */
+		public MDHashedObject(MDMang dmang) {
+			super(dmang);
+			qualification = new MDQualification(dmang);
+		}
+
+		/**
+		 * Returns the hashed string.
+		 * @return the hashed string.
+		 */
+		public String getHashString() {
+			return hashString;
+		}
+
+		/**
+		 * Returns an empty {@link MDQualification} that represents the namespace of the symbol.
+		 *  Note: we have yet to decide whether we should do anything but this.
+		 * @return the namespace information (empty for now).
+		 */
+		public MDQualification getQualification() {
+			return qualification;
+		}
+
+		@Override
+		protected void parseInternal() throws MDException {
+
+			if ((dmang.peek() != '?') && (dmang.peek(1) != '@')) {
+				throw new MDException("Invalid HashedObject");
+			}
+			dmang.increment(2);
+
+			StringBuilder builder = new StringBuilder();
+			char ch;
+			int start = dmang.getIndex();
+			while ((ch = dmang.peek()) != MDMang.DONE) {
+				if (ch == '@') {
+					break;
+				}
+				if (!(Character.isLetter(ch) || Character.isDigit(ch))) {
+					break;
+				}
+				builder.append(ch);
+				dmang.next();
+			}
+			int end = dmang.getIndex();
+			if ((end - start) != 32 || ch != '@') {
+				throw new MDException("Invalid HashedObject");
+			}
+			dmang.increment();
+			hashString = builder.toString();
+		}
+
+		@Override
+		public void insert(StringBuilder builder) {
+			// We have made up the output format.  Nothing is sacrosanct about this output.
+			builder.append("`" + hashString + "'");
+		}
+	}
+
 }
 
 /******************************************************************************/

Ghidra/Features/MicrosoftDmang/src/test/java/mdemangler/MDMangBaseTest.java

@@ -14472,6 +14472,26 @@ public class MDMangBaseTest extends AbstractGenericTest {
 		demangleAndTest();
 	}
 
+	//Issue 1344: Long symbols get MD5-hashed.
+	// We have made up the output format.  Nothing is sacrosanct about this output.
+	@Test
+	public void testHashedSymbolComponentsLongerThan5096_1() throws Exception {
+		mangled = "??@f4873c94f485cd6716c2319fc51ac714@";
+		msTruth = "";
+		mdTruth = "`f4873c94f485cd6716c2319fc51ac714'";
+		demangleAndTest();
+	}
+
+	//Issue 1344: Long symbols get MD5-hashed.
+	// We have made up the output format.  Nothing is sacrosanct about this output.
+	@Test
+	public void testHashedSymbolComponentsLongerThan5096_2() throws Exception {
+		mangled = "?catch$0@?0???@f4873c94f485cd6716c2319fc51ac714@@4HA";
+		msTruth = "";
+		mdTruth = "int ``f4873c94f485cd6716c2319fc51ac714''::`1'::catch$0";
+		demangleAndTest();
+	}
+
 	//TODO: ignore for now.
 	@Ignore
 	public void testFuzzyFit() throws Exception {