Update app.py

app.py

@@ -33,35 +33,23 @@ def format_prompt(topic, description, difficulty):
 # Function to clean and format the AI output
 def clean_and_format_output(output):
     """
-    Cleans and validates the output to ensure it is valid JSON.
+    Cleans, validates, and attempts to parse JSON output. Handles truncation and formatting errors.
     """
-    # Remove non-ASCII characters
-    cleaned_output = re.sub(r'[^\x00-\x7F]+', '', output)
-    
-    # Remove extraneous symbols like ◀ and backticks
-    cleaned_output = re.sub(r'`|<s>|</s>|◀', '', cleaned_output)
-    
-    # Remove text before the first '{' and after the last '}'
-    cleaned_output = re.sub(r'^[^{]*', '', cleaned_output)
-    cleaned_output = re.sub(r'[^}]*$', '', cleaned_output)
-    
-    # Replace improperly escaped characters (e.g., \_)
-    cleaned_output = re.sub(r'\\_', '_', cleaned_output)
-    
-    # Normalize whitespace
-    cleaned_output = re.sub(r'\s+', ' ', cleaned_output).strip()
-    
-    # Remove improperly escaped quotes
-    cleaned_output = cleaned_output.replace('\\"', '"')
-    
-    # Remove trailing commas
-    cleaned_output = re.sub(r',\s*(\}|\])', r'\1', cleaned_output)
-    
     try:
-        # Attempt to parse the cleaned string as JSON
+        # Step 1: Clean the raw output
+        cleaned_output = re.sub(r'[^\x00-\x7F]+', '', output)  # Remove non-ASCII characters
+        cleaned_output = re.sub(r'`|<s>|</s>|◀', '', cleaned_output)  # Remove extraneous symbols
+        cleaned_output = re.sub(r'^[^{]*', '', cleaned_output)  # Remove text before first '{'
+        cleaned_output = re.sub(r'[^}]*$', '', cleaned_output)  # Remove text after last '}'
+        cleaned_output = re.sub(r'\\_', '_', cleaned_output)  # Replace improperly escaped underscores
+        cleaned_output = re.sub(r'\s+', ' ', cleaned_output).strip()  # Normalize whitespace
+        cleaned_output = cleaned_output.replace('\\"', '"')  # Fix improperly escaped quotes
+        cleaned_output = re.sub(r',\s*(\}|\])', r'\1', cleaned_output)  # Remove trailing commas
+
+        # Step 2: Attempt to parse as JSON
         json_output = json.loads(cleaned_output)
-        
-        # Check for required structure
+
+        # Step 3: Validate required structure
         if "title" not in json_output or "sections" not in json_output:
             raise ValueError("Missing required keys: 'title' or 'sections'.")
         if not isinstance(json_output["sections"], list):
@@ -69,9 +57,18 @@ def clean_and_format_output(output):
         for section in json_output["sections"]:
             if "subheading" not in section or "content" not in section:
                 raise ValueError("Each section must contain 'subheading' and 'content'.")
-        
+
         return json_output
-    except (json.JSONDecodeError, ValueError) as e:
+
+    except json.JSONDecodeError as e:
+        # Handle JSON decoding errors
+        return {
+            "error": "Failed to parse or validate output as JSON",
+            "details": f"JSON decoding error: {str(e)}",
+            "output": cleaned_output
+        }
+    except ValueError as e:
+        # Handle validation errors
         return {
             "error": "Failed to parse or validate output as JSON",
             "details": str(e),