Split method body to basic blocks

Decompiler.csproj

@@ -50,6 +50,7 @@
     <Compile Include="src\ByteCode.StackAnalysis.cs" />
     <Compile Include="src\StackExpression.cs" />
     <Compile Include="src\StackExpressionCollection.cs" />
+    <Compile Include="src\ControlFlow.cs" />
     <Compile Include="src\Util.cs" />
     <EmbeddedResource Include="src\MainForm.resx">
       <DependentUpon>MainForm.cs</DependentUpon>

src/AstMetodBodyBuilder.cs

@@ -22,8 +22,10 @@ namespace Decompiler
 			methodDef.Body.Simplify();
 			
 			ByteCodeCollection body = new ByteCodeCollection(methodDef);
-			StackExpressionCollection exprCol = new StackExpressionCollection(body);
-			exprCol.Optimize();
+			StackExpressionCollection exprCollection = new StackExpressionCollection(body);
+			exprCollection.Optimize();
+			
+			BasicBlockSet basicBlockSet = new BasicBlockSet(exprCollection);
 			
 			foreach(VariableDefinition varDef in methodDef.Body.Variables) {
 				localVarTypes[varDef.Name] = varDef.VariableType;
@@ -35,8 +37,11 @@ namespace Decompiler
 //				astBlock.Children.Add(astLocalVar);
 			}
 			
-			for(int i = 0; i < exprCol.Count; i++) {
-				StackExpression expr = exprCol[i];
+			for(int b = 0; b < basicBlockSet.Elements.Count; b++) {
+				BasicBlock basicBlock = (BasicBlock)basicBlockSet.Elements[b];
+				astBlock.Children.Add(MakeComment(basicBlock.ToString()));
+				for(int i = 0; i < basicBlock.Body.Count; i++) {
+					StackExpression expr = basicBlock.Body[i];
 					Ast.Statement astStatement = null;
 					try {
 						List<Ast.Expression> args = new List<Ast.Expression>();
@@ -47,8 +52,8 @@ namespace Decompiler
 						object codeExpr = MakeCodeDomExpression(methodDef, expr, args.ToArray());
 						if (codeExpr is Ast.Expression) {
 							if (expr.PushCount == 1) {
-							string type = expr.ExpressionByteCode.Type.FullName;
-							string name = string.Format("expr{0:X2}", expr.ExpressionByteCode.Offset);
+								string type = expr.LastByteCode.Type.FullName;
+								string name = string.Format("expr{0:X2}", expr.LastByteCode.Offset);
 								Ast.LocalVariableDeclaration astLocal = new Ast.LocalVariableDeclaration(new Ast.TypeReference(type.ToString()));
 								astLocal.Variables.Add(new Ast.VariableDeclaration(name, (Ast.Expression)codeExpr));
 								astStatement = astLocal;
@@ -59,27 +64,28 @@ namespace Decompiler
 							astStatement = (Ast.Statement)codeExpr;
 						}
 					} catch (NotImplementedException) {
-					astStatement = MakeComment(expr.ExpressionByteCode.Description);
+						astStatement = MakeComment(expr.LastByteCode.Description);
 					}
 					if (expr.IsBranchTarget) {
 						astBlock.Children.Add(new Ast.LabelStatement(string.Format("IL_{0:X2}", expr.FirstByteCode.Offset)));
 					}
 					// Skip last return statement
-				if (i == exprCol.Count - 1 && 
-				    expr.ExpressionByteCode.OpCode.Code == Code.Ret &&
-				    expr.ExpressionByteCode.PopCount == 0 &&
-				    !expr.IsBranchTarget) {
-					continue;
-				}
+//					if (i == exprCol.Count - 1 && 
+//					    expr.LastByteCode.OpCode.Code == Code.Ret &&
+//					    expr.LastByteCode.PopCount == 0 &&
+//					    !expr.IsBranchTarget) {
+//						continue;
+//					}
 					astBlock.Children.Add(astStatement);
 				}
+			}
 			
 			return astBlock;
 		}
 		
 		static Ast.ExpressionStatement MakeComment(string text)
 		{
-			text = "/*" + text + "*/";
+			text = "/* " + text + "*/";
 			return new Ast.ExpressionStatement(new PrimitiveExpression(text, text));
 		}
 		
@@ -97,7 +103,7 @@ namespace Decompiler
 					allArgs.Add(astExpr);
 				}
 			}
-			return MakeCodeDomExpression(methodDef, expr.ExpressionByteCode, allArgs.ToArray());
+			return MakeCodeDomExpression(methodDef, expr.LastByteCode, allArgs.ToArray());
 		}
 		
 		static object MakeCodeDomExpression(MethodDefinition methodDef, ByteCode byteCode, params Ast.Expression[] args)

src/ByteCode.cs

@@ -9,6 +9,7 @@ namespace Decompiler
 {
 	public partial class ByteCode
 	{
+		StackExpression owner;
 		ByteCode previous;
 		ByteCode next;
 		
@@ -17,6 +18,11 @@ namespace Decompiler
 		OpCode opCode;
 		object operand;
 		
+		public StackExpression Owner {
+			get { return owner; }
+			set { owner = value; }
+		}
+		
 		public ByteCode Previous {
 			get { return previous; }
 			set { previous = value; }
@@ -44,7 +50,7 @@ namespace Decompiler
 		
 		public ByteCode BranchTarget {
 			get {
-				return (ByteCode)operand;
+				return operand as ByteCode;
 			}
 		}
 		
@@ -66,7 +72,7 @@ namespace Decompiler
 		public string Description {
 			get {
 				return string.Format(
-					" {1, -22} # {2}->{3} {4} {5}",
+					"{1, -22} # {2}->{3} {4} {5}",
 					this.Offset,
 					this.OpCode + " " + FormatByteCodeOperand(this.Operand),
 					this.OpCode.StackBehaviourPop,

src/ControlFlow.cs

@@ -0,0 +1,187 @@
+using System;
+using System.Collections;
+using System.Collections.Generic;
+
+namespace Decompiler
+{
+	public class Set<T>: List<T>
+	{
+		
+	}
+	
+	
+	public class BasicBlock
+	{
+		int id;
+		BasicBlockSet owner;
+		Set<BasicBlock> predecessors;
+		public Set<BasicBlock> successors;
+		BasicBlock fallThroughSuccessor;
+		BasicBlock branchSuccessor;
+		List<StackExpression> body = new List<StackExpression>();
+		
+		#region Peoperties
+		
+		public int Id {
+			get { return id; }
+			set { id = value; }
+		}
+		
+		public BasicBlockSet Owner {
+			get { return owner; }
+			set { owner = value; }
+		}
+		
+		public Set<BasicBlock> Predecessors {
+			get { return predecessors; }
+			set { predecessors = value; }
+		}
+		
+		public Set<BasicBlock> Successors {
+			get { return successors; }
+			set { successors = value; }
+		}
+		
+		public BasicBlock FallThroughSuccessor {
+			get { return fallThroughSuccessor; }
+			set { fallThroughSuccessor = value; }
+		}
+		
+		public BasicBlock BranchSuccessor {
+			get { return branchSuccessor; }
+			set { branchSuccessor = value; }
+		}
+		
+		public List<StackExpression> Body {
+			get { return body; }
+			set { body = value; }
+		}
+		
+		#endregion
+		
+		public override string ToString()
+		{
+			//return string.Format("BackBlock {0} ({1} expressions)", id, body.Count);
+			return string.Format("BackBlock {0}", id, body.Count);
+		}
+	}
+	
+	public enum BasicBlockSetType {
+		MethodBody,
+		Acyclic,
+		Loop,
+	}
+	
+	public class BasicBlockSet
+	{
+		BasicBlockSet owner;
+		BasicBlockSetType type;
+		
+		object head;
+		Set<object> elements = new Set<object>();
+		
+		Set<BasicBlock> BasicBlockSuccessors;
+		
+		BasicBlock headBasicBlock {
+			get {
+				return null;
+			}
+		}
+		
+		public BasicBlockSet Owner {
+			get { return owner; }
+		}
+		
+		public BasicBlockSetType Type {
+			get { return type; }
+		}
+		
+		public object Head {
+			get { return head; }
+		}
+		
+		public Set<object> Elements {
+			get { return elements; }
+		}
+		
+		
+		BasicBlockSet()
+		{
+			
+		}
+		
+		public BasicBlockSet(object head, object tail)
+		{
+			if (head == null) throw new ArgumentNullException("head");
+			if (tail == null) throw new ArgumentNullException("tail");
+			
+			BasicBlockSet headAsSet = head as BasicBlockSet;
+			BasicBlockSet tailAsSet = tail as BasicBlockSet;
+			
+			// Add head
+			if (head is BasicBlock) {
+				this.head = head;
+				this.elements.Add(head);
+			} else if (headAsSet != null && headAsSet.type == BasicBlockSetType.Acyclic) {
+				this.head = headAsSet.head;
+				this.elements.AddRange(headAsSet.elements);
+			} else if (headAsSet != null && headAsSet.type == BasicBlockSetType.Loop) {
+				this.head = headAsSet;
+				this.elements.Add(headAsSet);
+			} else {
+				throw new Exception("Invalid head");
+			}
+			
+			// Add tail
+			if (tail is BasicBlock) {
+				this.elements.Add(tail);
+			} else if (tailAsSet != null && tailAsSet.type == BasicBlockSetType.Acyclic) {
+				this.elements.AddRange(tailAsSet.elements);
+			} else if (tailAsSet != null && tailAsSet.type == BasicBlockSetType.Loop) {
+				this.elements.Add(tailAsSet);
+			} else {
+				throw new Exception("Invalid tail");
+			}
+			
+			// Get type
+			if (tail is BasicBlock) {
+				if (((BasicBlock)tail).successors.Contains(this.headBasicBlock)) {
+					this.type = BasicBlockSetType.Loop;
+				} else {
+					this.type = BasicBlockSetType.Acyclic;
+				}
+			} else if (tailAsSet != null) {
+				if (tailAsSet.BasicBlockSuccessors.Contains(this.headBasicBlock)) {
+					
+				}
+			} else {
+				throw new Exception("Invalid tail");
+			}
+		}
+		
+		public BasicBlockSet(StackExpressionCollection exprs)
+		{
+			if (exprs.Count == 0) throw new ArgumentException("Count == 0", "exprs");
+			
+			this.owner = null;
+			this.type = BasicBlockSetType.MethodBody;
+			
+			BasicBlock basicBlock = null;
+			int basicBlockId = 1;
+			for(int i = 0; i < exprs.Count; i++) {
+				// Start new basic block if
+				//  - this is first expression
+				//  - last expression was branch
+				//  - this expression is branch target
+				if (i == 0 || exprs[i - 1].BranchTarget != null || exprs[i].BranchesHere.Count > 0){
+					basicBlock = new BasicBlock();
+					this.elements.Add(basicBlock);
+					basicBlock.Id = basicBlockId++;
+				}
+				basicBlock.Body.Add(exprs[i]);
+			}
+			
+			this.head = this.elements[0];
+		}
+	}
+}

src/StackExpression.cs

@@ -9,11 +9,16 @@ namespace Decompiler
 {
 	public class StackExpression
 	{
-		ByteCode expressionByteCode;
+		StackExpressionCollection owner;
+		ByteCode lastByteCode;
 		List<StackExpression> lastArguments = new List<StackExpression>();
 		
-		public ByteCode ExpressionByteCode {
-			get { return expressionByteCode; }
+		public StackExpressionCollection Owner {
+			get { return owner; }
+		}
+		
+		public ByteCode LastByteCode {
+			get { return lastByteCode; }
 		}
 		
 		// A list of closed expression for last arguments
@@ -29,7 +34,7 @@ namespace Decompiler
 		
 		public CilStack StackAfter {
 			get {
-				return this.ExpressionByteCode.StackAfter;
+				return this.LastByteCode.StackAfter;
 			}
 		}
 		
@@ -43,6 +48,26 @@ namespace Decompiler
 			}
 		}
 		
+		public List<StackExpression> BranchesHere {
+			get {
+				List<StackExpression> branchesHere = new List<StackExpression>();
+				foreach(ByteCode byteCode in this.FirstByteCode.BranchesHere) {
+					branchesHere.Add(byteCode.Owner);
+				}
+				return branchesHere;
+			}
+		}
+		
+		public StackExpression BranchTarget {
+			get {
+				if (this.lastByteCode.BranchTarget == null) {
+					return null;
+				} else {
+					return this.lastByteCode.BranchTarget.Owner;
+				}
+			}
+		}
+		
 		public bool IsBranchTarget {
 			get {
 				return this.FirstByteCode.BranchesHere.Count > 0;
@@ -77,9 +102,9 @@ namespace Decompiler
 				stackSize += expr.PushCount;
 			}
 			{
-				stackSize -= expressionByteCode.PopCount;
+				stackSize -= lastByteCode.PopCount;
 				minStackSize = Math.Min(minStackSize, stackSize);
-				stackSize += expressionByteCode.PushCount;
+				stackSize += lastByteCode.PushCount;
 			}
 			popCount = -minStackSize;
 			pushCount = stackSize - minStackSize;
@@ -90,19 +115,21 @@ namespace Decompiler
 				if (lastArguments.Count > 0) {
 					return lastArguments[0].FirstByteCode;
 				} else {
-					return this.ExpressionByteCode;
+					return this.LastByteCode;
 				}
 			}
 		}
 		
-		public StackExpression(ByteCode expressionByteCode)
+		public StackExpression(StackExpressionCollection owner, ByteCode lastByteCode)
 		{
-			this.expressionByteCode = expressionByteCode;
+			this.owner = owner;
+			this.lastByteCode = lastByteCode;
+			this.lastByteCode.Owner = this;
 		}
 		
 		public bool MustBeParenthesized {
 			get {
-				switch(this.ExpressionByteCode.OpCode.Code) {
+				switch(this.LastByteCode.OpCode.Code) {
 					#region Arithmetic
 						case Code.Neg:
 						case Code.Not:

src/StackExpressionCollection.cs

@@ -12,7 +12,7 @@ namespace Decompiler
 		public StackExpressionCollection(ByteCodeCollection byteCodeCol)
 		{
 			foreach(ByteCode bc in byteCodeCol) {
-				this.Add(new StackExpression(bc));
+				this.Add(new StackExpression(this, bc));
 			}
 		}