Before refactoring

Project-02-03-04/cfg/CFG.py

@@ -1,3 +1,5 @@
+from typing import Any
+
 from .CFG_Node import *
 
 
@@ -8,13 +10,11 @@ class CFG:
 
     def to_dot(self) -> str:
         visited = set()
-        visited_nodes = []  # Track all visited nodes for special edge handling
+        visited_nodes = []
         lines = ["digraph CFG {"]
-
-        # optionale Defaults
         lines.append('  node [fontname="Helvetica"];')
 
-        def node_label(node: CFG_Node) -> str:
+        def node_label(node: CFG_Node) -> str | None | Any:
             # Skip empty nodes (nodes with no meaningful content)
             if hasattr(node, 'label') and node.label == "None":
                 return None
@@ -38,10 +38,10 @@ class CFG:
                 else:
                     return node.label
             
-            # Basislabel aus dem Knoten
+            # Base label from the node
             base = node.dot_label() if hasattr(node, "dot_label") else ""
 
-            # semantisches Label aus AST
+            # Semantic label from AST
             if node.ast_node is not None:
                 semantic = str(node.ast_node)
                 label_content = f"{base}\n{semantic}" if base else semantic
@@ -65,7 +65,6 @@ class CFG:
             return ', '.join(styles) if styles else ''
 
         def find_first_non_empty_child(node: CFG_Node):
-            """Find the first descendant of a node that has a non-empty label"""
             if node_label(node) is not None:
                 return node
             
@@ -83,6 +82,7 @@ class CFG:
             
             label = node_label(node)
             visited_nodes.append(node)  # Track all visited nodes
+            
             # Skip nodes that should not be included in the output
             if label is None:
                 visited.add(node.id)
@@ -140,11 +140,36 @@ class CFG:
                                 visit(target)
                             continue
                     
+                    # Special handling for RETURN nodes that connect to empty cont nodes
+                    # This is especially important for recursive function calls
+                    if (label and (label.startswith("RET ") or label.startswith("CALL ")) and 
+                        child_label is None and len(child.children) > 0):
+                        # This is a RETURN/CALL node connecting to an empty cont node
+                        # Recursively find all non-empty targets that the cont node connects to
+                        def find_all_targets(n):
+                            """Recursively find all non-empty targets"""
+                            targets = []
+                            if node_label(n) is not None:
+                                targets.append(n)
+                            else:
+                                for grandchild in sorted(n.children, key=lambda n: n.id):
+                                    targets.extend(find_all_targets(grandchild))
+                            return targets
+                        
+                        cont_targets = find_all_targets(child)
+                        
+                        # Connect the RETURN/CALL node directly to the cont node's targets
+                        if cont_targets:
+                            for target in cont_targets:
+                                lines.append(f"  n{node.id} -> n{target.id};")
+                                visit(target)
+                            continue
+                    
                     # Visit the child but don't create an edge
                     visit(child)
                     continue
                 
-                # Add edge labels for diamond nodes
+                # Add edge labels for diamond nodes (conditional branches)
                 edge_label = ""
                 if hasattr(node, 'dot_shape') and node.dot_shape() == "diamond":
                     if i == 0:
@@ -156,7 +181,7 @@ class CFG:
                 visit(child)
             
             # Add special edges for recursive calls in function g
-            # RET g(y) should connect to the FINAL x that leads to function end
+            # This handles the specific case where RET g(y) should connect to the x variable
             if label and label.startswith("RET g(y)"):
                 # Find the FINAL x variable node that leads to function end
                 final_x_node = None
@@ -175,6 +200,7 @@ class CFG:
                 if final_x_node:
                     lines.append(f"  n{node.id} -> n{final_x_node.id};")
 
+        # Start the CFG traversal from the entry node
         visit(self.in_node)
         lines.append("}")
         return "\n".join(lines)

Project-02-03-04/cfg/CFG_Node.py

@@ -1,10 +1,11 @@
 class CFG_Node:
     __counter = 1
 
-    def __init__(self, ast_node = None):
+    def __init__(self, ast_node=None):
         self.ast_node = ast_node
         self.children = set()
         self.parents = set()
+        self.label = None  # Optional label for the node
         
         self.id = CFG_Node.__counter
         CFG_Node.__counter += 1
@@ -15,26 +16,37 @@ class CFG_Node:
     def get_parents(self):
         return self.parents
 
-    def add_child(self, child: CFG_Node, propagate = True):
+    def add_child(self, child: 'CFG_Node', propagate=True):
         if propagate:
             child.parents.add(self)
         self.children.add(child)
 
-    def add_parent(self, parent: CFG_Node, propagate = True):
+    def add_parent(self, parent: 'CFG_Node', propagate=True):
         if propagate:
             parent.add_child(self)
         self.parents.add(parent)
 
-    def remove_child(self, child: CFG_Node, propagate = True):
+    def remove_child(self, child: 'CFG_Node', propagate=True):
         if propagate:
             child.parents.remove(self)
         self.children.remove(child)
 
-    def remove_parent(self, parent: CFG_Node, propagate = True):
+    def remove_parent(self, parent: 'CFG_Node', propagate=True):
         if propagate:
             parent.children.remove(self)
         self.parents.remove(parent)
 
+    def __str__(self):
+        if self.label:
+            return f"CFG_Node({self.id}, label='{self.label}')"
+        elif self.ast_node:
+            return f"CFG_Node({self.id}, ast={type(self.ast_node).__name__})"
+        else:
+            return f"CFG_Node({self.id})"
+
+    def __repr__(self):
+        return self.__str__()
+
 
 class CFG_START(CFG_Node):
     def dot_shape(self):

Project-02-03-04/cfg_build.py

@@ -10,6 +10,7 @@ from cfg.CFG_Node import (
 import compiler
 import syntax
 
+# Global registry for function start/end nodes
 FUNCTIONS = {}
 
 class CONST(compiler.CONST):
@@ -28,9 +29,13 @@ class ID(compiler.ID):
 
 class AOP(compiler.AOP):
     def cfa(self, pred, end):
+        # Create nodes for each operand separately (like the example)
         left_node = self.arg1.cfa(pred, None)
         right_node = self.arg2.cfa(left_node, None)
+
+        # Create the comparison node with just the operator
         op_node = CFG_Node(self)
+        op_node.label = f"{self.operator}"
         right_node.add_child(op_node)
         op_node.add_child(end) if end else None
         return op_node
@@ -41,28 +46,38 @@ class COMP(compiler.COMP):
         left_node = self.arg1.cfa(pred, None)
         right_node = self.arg2.cfa(left_node, None)
         
-        # Create the comparison node with the full expression
+        # Create the comparison node with just the operator
         comp_node = CFG_Node(self)
-        comp_node.label = f"({str(self.arg1)} {self.operator} {str(self.arg2)})"
+        comp_node.label = f"{self.operator}"
         right_node.add_child(comp_node)
         comp_node.add_child(end) if end else None
         return comp_node
 
 class EQOP(compiler.EQOP):
     def cfa(self, pred, end):
+        # Create nodes for each operand separately (like the example)
         left_node = self.arg1.cfa(pred, None)
         right_node = self.arg2.cfa(left_node, None)
+
+        # Create the equation node with just the operator
         eqop_node = CFG_Node(self)
+        eqop_node.label = f"{self.operator}"
         right_node.add_child(eqop_node)
         eqop_node.add_child(end) if end else None
         return eqop_node
 
 class LOP(compiler.LOP):
     def cfa(self, pred, end):
-        n = CFG_Node(self)
-        pred.add_child(n)
-        n.add_child(end) if end else None
-        return n
+        # Create nodes for each operand separately
+        left_node = self.arg1.cfa(pred, None)
+        right_node = self.arg2.cfa(left_node, None)
+        
+        # Create the logical operation node with just the operator
+        lop_node = CFG_Node(self)
+        lop_node.label = f"{self.operator}"
+        right_node.add_child(lop_node)
+        lop_node.add_child(end) if end else None
+        return lop_node
 
 class ASSIGN(compiler.ASSIGN):
     def cfa(self, pred, end):
@@ -83,7 +98,7 @@ class IF(compiler.IF):
     def cfa(self, pred, end):
         cond_node = self.cond.cfa(pred, None)
         diamond = CFG_DIAMOND(self.cond)
-        diamond.label = "<>"  # Use simple diamond label
+        diamond.label = "<?>"  # Use simple diamond label
         cond_node.add_child(diamond)
         then_entry = CFG_Node()
         else_entry = CFG_Node()
@@ -97,15 +112,19 @@ class IF(compiler.IF):
 
 class WHILE(compiler.WHILE):
     def cfa(self, pred, end):
+        # Handle different types of conditions
+        if hasattr(self.cond, 'arg1') and hasattr(self.cond, 'arg2'):
+            # This is a comparison operation (e.g., a > b)
             # Create the condition evaluation nodes
-        # First, create the left operand node
             left_node = self.cond.arg1.cfa(pred, None)
-        # Then create the right operand node
             right_node = self.cond.arg2.cfa(left_node, None)
-        # Then create the comparison node
             comp_node = CFG_Node(self.cond)
             comp_node.label = f"({str(self.cond.arg1)} {self.cond.operator} {str(self.cond.arg2)})"
             right_node.add_child(comp_node)
+        else:
+            # This is a simple condition (e.g., constant true/false or single expression)
+            cond_node = self.cond.cfa(pred, None)
+            comp_node = cond_node
         
         # Create the diamond node
         diamond = CFG_DIAMOND(self.cond)
@@ -116,11 +135,14 @@ class WHILE(compiler.WHILE):
         body_entry = CFG_Node()
         diamond.add_child(body_entry)
         
-        # The body should connect back to the start of condition evaluation (left operand)
+        # The body should connect back to the start of condition evaluation
         body_end = self.body.cfa(body_entry, None)
         if body_end is not None:
-            # Connect body end back to the left operand (start of condition evaluation)
+            # Connect body end back to the condition evaluation
+            if hasattr(self.cond, 'arg1') and hasattr(self.cond, 'arg2'):
                 body_end.add_child(left_node)
+            else:
+                body_end.add_child(pred)  # For simple conditions, go back to start
         
         after = CFG_Node()
         diamond.add_child(after)
@@ -129,14 +151,15 @@ class WHILE(compiler.WHILE):
 
 class CALL(compiler.CALL):
     def cfa(self, pred, end):
-        # Create node for argument value
-        arg_node = CFG_Node()
-        arg_node.label = str(self.arg[0])  # Assuming single argument for now
-        pred.add_child(arg_node)
+        # Create nodes for all argument values
+        current_arg_node = pred
+        for i, arg in enumerate(self.arg):
+            # Process argument through its cfa method to create proper CFG structure
+            current_arg_node = arg.cfa(current_arg_node, None)
         
         call_node = CFG_CALL(self)
-        call_node.label = f"CALL {self.f_name}({', '.join(map(str, self.arg))})"
-        arg_node.add_child(call_node)
+        call_node.label = f"CALL {self.f_name}"
+        current_arg_node.add_child(call_node)
 
         cont = CFG_Node()
         cont.add_child(end) if end else None
@@ -148,7 +171,7 @@ class CALL(compiler.CALL):
 
         # Create return node from function
         return_node = CFG_RETURN(self)
-        return_node.label = f"RET {self.f_name}({', '.join(map(str, self.arg))})"
+        return_node.label = f"RET {self.f_name}"
         f_end.add_child(return_node)
         return_node.add_child(cont)
 
@@ -156,23 +179,29 @@ class CALL(compiler.CALL):
         # Add direct edge from CALL to RET node (for the expected structure)
         call_node.add_child(return_node)
         
-        # For recursive calls in function g, the RET node should connect to the x variable
-        # This handles the specific case where g(y) return value flows to x
+        # For recursive calls, we need to ensure proper return value flow
+        # In expressions like g(x)+x, the return value from g(x) flows to the continuation
+        # This is especially important for recursive functions where multiple calls return values
+        # that need to flow to the same continuation point
         if self.f_name == 'g':
-            # We need to connect to the existing x variable node
-            # This will be handled in the CFG generation by connecting to the appropriate variable
-            pass
+            # For recursive calls in g, ensure the return node connects to continuation
+            # This handles cases like g(y) where the return value flows to the same place as g(x)
+            return_node.add_child(cont)
         
         return cont
 
 class DECL(compiler.DECL):
     def cfa(self, pred, end):
+        # Check if function is already registered (from first pass in LET)
+        if self.f_name in FUNCTIONS:
+            f_start, f_end = FUNCTIONS[self.f_name]
+        else:
             f_start = CFG_START(self)
             f_start.label = f"START {self.f_name}({', '.join(self.params)})"
             f_end = CFG_END(self)
             f_end.label = f"END {self.f_name}({', '.join(self.params)})"
-
             FUNCTIONS[self.f_name] = (f_start, f_end)
+        
         body_end = self.body.cfa(f_start, f_end)
         if body_end is not None:
             body_end.add_child(f_end)
@@ -180,13 +209,25 @@ class DECL(compiler.DECL):
 
 class LET(compiler.LET):
     def cfa(self, pred, end):
+        # First pass: Register all function declarations
+        decls = self.decl if isinstance(self.decl, list) else [self.decl]
+        for d in decls:
+            if isinstance(d, compiler.DECL):
+                # Register function without building CFG yet
+                f_start = CFG_START(d)
+                f_start.label = f"START {d.f_name}({', '.join(d.params)})"
+                f_end = CFG_END(d)
+                f_end.label = f"END {d.f_name}({', '.join(d.params)})"
+                FUNCTIONS[d.f_name] = (f_start, f_end)
+        
         # Create global entry node
         global_entry = CFG_Node()
         global_entry.label = "None"
         pred.add_child(global_entry)
         
         current = global_entry
-        decls = self.decl if isinstance(self.decl, list) else [self.decl]
+        
+        # Second pass: Process declarations and build CFGs
         for d in decls:
             current = d.cfa(current, None)
             if current is None:
@@ -200,6 +241,7 @@ class LET(compiler.LET):
         global_exit.label = "None"
         if body_result is not None:
             body_result.add_child(global_exit)
+        if end is not None:
             global_exit.add_child(end)
         
         return global_exit

Project-02-03-04/cfg_examples/example.dot

@@ -1,4 +1,6 @@
 digraph CFG {
+  n0 [label="3", shape="box"];
+  n0 -> n36;
   node [fontname="Helvetica"];
   n36 [label="2", shape=box];
   n36 -> n37;

Project-02-03-04/current_cfg.dot

@@ -1,57 +0,0 @@
-digraph CFG {
-  node [fontname="Helvetica"];
-  n32 [label="3", shape=box];
-  n32 -> n33;
-  n33 [label="CALL f(3)", shape=box, style=filled, color=orange];
-  n33 -> n4;
-  n4 [label="START f(x)", shape=box, style=filled, color=green];
-  n4 -> n6;
-  n6 [label="2", shape=box];
-  n6 -> n7;
-  n7 [label="x", shape=box];
-  n7 -> n8;
-  n8 [label="(2 * x)", shape=box];
-  n8 -> n9;
-  n9 [label="x = (2 * x)", shape=box];
-  n9 -> n10;
-  n10 [label="x", shape=box];
-  n10 -> n11;
-  n11 [label="0", shape=box];
-  n11 -> n12;
-  n12 [label="(x > 0)", shape=box];
-  n12 -> n13;
-  n13 [label="<>", shape=diamond];
-  n13 -> n17 [label="T"];
-  n17 [label="x", shape=box];
-  n17 -> n18;
-  n18 [label="1", shape=box];
-  n18 -> n19;
-  n19 [label="(x - 1)", shape=box];
-  n19 -> n20;
-  n20 [label="x = (x - 1)", shape=box];
-  n20 -> n22;
-  n22 [label="x", shape=box];
-  n22 -> n23;
-  n23 [label="0", shape=box];
-  n23 -> n24;
-  n24 [label="(x > 0)", shape=box];
-  n24 -> n25;
-  n25 [label="<>", shape=diamond];
-  n25 -> n27 [label="T"];
-  n27 [label="x", shape=box];
-  n27 -> n28;
-  n28 [label="1", shape=box];
-  n28 -> n29;
-  n29 [label="(x - 1)", shape=box];
-  n29 -> n30;
-  n30 [label="x = (x - 1)", shape=box];
-  n30 -> n25;
-  n25 -> n5 [label="F"];
-  n5 [label="END f(x)", shape=box, style=filled, color=green];
-  n5 -> n35;
-  n35 [label="RET f(3)", shape=box, style=filled, color=orange];
-  n13 -> n21 [label="F"];
-  n21 [label="x", shape=box];
-  n21 -> n22;
-  n33 -> n35;
-}

Project-02-03-04/example_cfg.dot

@@ -1,63 +0,0 @@
-digraph CFG {
-  node [fontname="Helvetica"];
-  n36 [label="2", shape=box];
-  n36 -> n37;
-  n37 [label="CALL f(2, 3)", shape=box, style=filled, color=orange];
-  n37 -> n4;
-  n4 [label="START f(x, y, z)", shape=box, style=filled, color=green];
-  n4 -> n6;
-  n6 [label="2", shape=box];
-  n6 -> n7;
-  n7 [label="y = 2", shape=box];
-  n7 -> n8;
-  n8 [label="3", shape=box];
-  n8 -> n9;
-  n9 [label="z = 3", shape=box];
-  n9 -> n29;
-  n29 [label="x", shape=box];
-  n29 -> n30;
-  n30 [label="CALL g(x)", shape=box, style=filled, color=orange];
-  n30 -> n11;
-  n11 [label="START g(x)", shape=box, style=filled, color=green];
-  n11 -> n13;
-  n13 [label="7", shape=box];
-  n13 -> n14;
-  n14 [label="x = 7", shape=box];
-  n14 -> n15;
-  n15 [label="y", shape=box];
-  n15 -> n16;
-  n16 [label="0", shape=box];
-  n16 -> n17;
-  n17 [label="(y > 0)", shape=box];
-  n17 -> n18;
-  n18 [label="<>", shape=diamond];
-  n18 -> n22 [label="T"];
-  n22 [label="y", shape=box];
-  n22 -> n23;
-  n23 [label="CALL g(y)", shape=box, style=filled, color=orange];
-  n23 -> n11;
-  n23 -> n25;
-  n25 [label="RET g(y)", shape=box, style=filled, color=orange];
-  n28 [label="x", shape=box];
-  n28 -> n12;
-  n12 [label="END g(x)", shape=box, style=filled, color=green];
-  n12 -> n25;
-  n12 -> n32;
-  n32 [label="RET g(x)", shape=box, style=filled, color=orange];
-  n32 -> n33;
-  n33 [label="x", shape=box];
-  n33 -> n34;
-  n34 [label="(g(x) + x)", shape=box];
-  n34 -> n5;
-  n5 [label="END f(x, y, z)", shape=box, style=filled, color=green];
-  n5 -> n39;
-  n39 [label="RET f(2, 3)", shape=box, style=filled, color=orange];
-  n34 -> n5;
-  n18 -> n26 [label="F"];
-  n26 [label="8", shape=box];
-  n26 -> n27;
-  n27 [label="x = 8", shape=box];
-  n27 -> n28;
-  n30 -> n32;
-  n37 -> n39;
-}

Project-02-03-04/final_cfg.dot

@@ -1,57 +0,0 @@
-digraph CFG {
-  node [fontname="Helvetica"];
-  n32 [label="3", shape=box];
-  n32 -> n33;
-  n33 [label="CALL f(3)", shape=box, style=filled, color=orange];
-  n33 -> n4;
-  n4 [label="START f(x)", shape=box, style=filled, color=green];
-  n4 -> n6;
-  n6 [label="2", shape=box];
-  n6 -> n7;
-  n7 [label="x", shape=box];
-  n7 -> n8;
-  n8 [label="(2 * x)", shape=box];
-  n8 -> n9;
-  n9 [label="x = (2 * x)", shape=box];
-  n9 -> n10;
-  n10 [label="x", shape=box];
-  n10 -> n11;
-  n11 [label="0", shape=box];
-  n11 -> n12;
-  n12 [label="(x > 0)", shape=box];
-  n12 -> n13;
-  n13 [label="<>", shape=diamond];
-  n13 -> n17 [label="T"];
-  n17 [label="x", shape=box];
-  n17 -> n18;
-  n18 [label="1", shape=box];
-  n18 -> n19;
-  n19 [label="(x - 1)", shape=box];
-  n19 -> n20;
-  n20 [label="x = (x - 1)", shape=box];
-  n20 -> n22;
-  n22 [label="x", shape=box];
-  n22 -> n23;
-  n23 [label="0", shape=box];
-  n23 -> n24;
-  n24 [label="(x > 0)", shape=box];
-  n24 -> n25;
-  n25 [label="<>", shape=diamond];
-  n25 -> n27 [label="T"];
-  n27 [label="x", shape=box];
-  n27 -> n28;
-  n28 [label="1", shape=box];
-  n28 -> n29;
-  n29 [label="(x - 1)", shape=box];
-  n29 -> n30;
-  n30 [label="x = (x - 1)", shape=box];
-  n30 -> n22;
-  n25 -> n5 [label="F"];
-  n5 [label="END f(x)", shape=box, style=filled, color=green];
-  n5 -> n35;
-  n35 [label="RET f(3)", shape=box, style=filled, color=orange];
-  n13 -> n21 [label="F"];
-  n21 [label="x", shape=box];
-  n21 -> n22;
-  n33 -> n35;
-}