Update app.py

app.py

@@ -309,13 +309,23 @@ def format_python_code(code):
             ('->', '.'),  # Convert arrow operator to dot
             ('::', '.'),  # Convert scope resolution to dot
 
-            # Braces and brackets
-            ('{', ':'),  # Convert opening braces to colons
+            # Braces and brackets - convert to Python indentation
+            ('{', ':\n'),  # Convert opening braces to colons with newline
             ('}', ''),  # Remove closing braces
 
-            # Semicolons (remove them)
+            # Semicolons (convert to newlines)
             ('; ', '\n'),  # Convert semicolons to newlines
             (';', '\n'),  # Convert semicolons to newlines
+
+            # Ternary operators (C++ style to Python)
+            (' ? ', ' if '),  # Convert ternary ? to if
+            (' : ', ' else '),  # Convert ternary : to else
+
+            # Comparison operators
+            ('< SEP', '<'),  # Fix common typo
+            ('< SEp', '<'),  # Fix common typo
+            ('<SEP', '<'),  # Fix common typo
+            ('<SEp', '<'),  # Fix common typo
         ]
 
         for old, new in corrections:
@@ -362,6 +372,14 @@ def format_python_code(code):
         # Add colons where missing for control structures
         formatted_code = re.sub(r'\b(if|for|while|def|class|try|except|with)\s+([^:]+)(?!\s*:)', r'\1 \2:', formatted_code)
 
+        # Fix ternary operators that became malformed
+        # Convert "return a if condition else b" to proper Python ternary
+        formatted_code = re.sub(r'return\s+(.+?)\s+if\s+(.+?)\s+else\s+(.+)', r'return \1 if \2 else \3', formatted_code)
+
+        # Ensure functions have proper structure
+        # If a line starts with "def" and doesn't end with ":", add it
+        formatted_code = re.sub(r'(def\s+\w+\([^)]*)\s*$', r'\1:', formatted_code, flags=re.MULTILINE)
+
         return formatted_code
 
     except Exception as e:
@@ -383,7 +401,7 @@ def generate_code_from_pseudo(pseudo_code, max_length, temperature, top_k, top_p
         start_time = time.time()
         
         # Format input with Python-specific instructions
-        prompt = f"<PSEUDO> {pseudo_code.strip()} <SEP> Generate Python code with proper indentation and syntax: <CODE>"
+        prompt = f"<PSEUDO> {pseudo_code.strip()} <SEP> Generate proper Python code with correct syntax, indentation, and separate lines. Use def for functions, if/else for conditionals, for/while for loops, and proper Python syntax only. No C++ or other languages. <CODE>"
         
         # Tokenize with error handling
         device = next(loaded_model.parameters()).device
@@ -608,6 +626,11 @@ def generate_code_from_pseudo(pseudo_code, max_length, temperature, top_k, top_p
                     # Format as proper Python code
                     code = format_python_code(code)
                     
+                    # For debugging: include raw generated code
+                    raw_code = generated.strip()
+                    if not code.startswith('#'):
+                        code = f"# Raw generated code:\n# {raw_code}\n\n# Formatted Python code:\n{code}"
+                    
                     # Ensure we have some content
                     if not code or code.isspace():
                         code = f"# Generated sequence {i+1} was empty after cleaning"