malformed deepseek response handling for formatted json

generate_prompts.py

@@ -310,10 +310,15 @@ class JournalPromptGenerator:
         """
         Parse the AI response to extract new prompts.
         Expected format: JSON array with keys "newprompt0" to "newpromptN" where N = num_prompts-1
+
+        Handles DeepSeek API responses that may include backticks and leading "json" string.
         """
+        # First, try to clean up the response content
+        cleaned_content = self._clean_ai_response(response_content)
+
         try:
             # Try to parse as JSON
-            data = json.loads(response_content)
+            data = json.loads(cleaned_content)
 
             # Convert to list of prompt dictionaries
             new_prompts = []
@@ -333,6 +338,7 @@ class JournalPromptGenerator:
             self.console.print("[yellow]Warning: AI response is not valid JSON, attempting to extract prompts...[/yellow]")
             self.console.print(f"[yellow]Full response content for debugging:[/yellow]")
             self.console.print(f"[yellow]{response_content}[/yellow]")
+            self.console.print(f"[yellow]Cleaned content: {cleaned_content}[/yellow]")
 
             # Look for patterns in the text
             lines = response_content.strip().split('\n')
@@ -346,8 +352,55 @@ class JournalPromptGenerator:
                     }
                     new_prompts.append(prompt_obj)
 
+            # If still no prompts could be parsed, dump the full payload for debugging
+            if not new_prompts:
+                self.console.print("[red]Error: Could not extract any prompts from AI response[/red]")
+                self.console.print("[red]Full payload dump for debugging:[/red]")
+                self.console.print(f"[red]{response_content}[/red]")
+
             return new_prompts
 
+    def _clean_ai_response(self, response_content: str) -> str:
+        """
+        Clean up AI response content to handle common formatting issues from DeepSeek API.
+
+        Handles:
+        1. Leading/trailing backticks (```json ... ```)
+        2. Leading "json" string on its own line
+        3. Extra whitespace and newlines
+        """
+        content = response_content.strip()
+
+        # Remove leading/trailing backticks (```json ... ```)
+        if content.startswith('```'):
+            # Find the first newline after the opening backticks
+            lines = content.split('\n')
+            if len(lines) > 1:
+                # Check if first line contains "json" or other language specifier
+                first_line = lines[0].strip()
+                if 'json' in first_line.lower() or first_line == '```':
+                    # Remove the first line (```json or ```)
+                    content = '\n'.join(lines[1:])
+
+            # Remove trailing backticks if present
+            if content.endswith('```'):
+                content = content[:-3].rstrip()
+
+        # Remove leading "json" string on its own line (case-insensitive)
+        lines = content.split('\n')
+        if len(lines) > 0:
+            first_line = lines[0].strip().lower()
+            if first_line == 'json':
+                content = '\n'.join(lines[1:])
+
+        # Also handle the case where "json" might be at the beginning of the first line
+        # but not the entire line (e.g., "json\n{...}")
+        content = content.strip()
+        if content.lower().startswith('json\n'):
+            content = content[4:].strip()
+
+        return content.strip()
+
     def generate_prompts(self) -> List[Dict[str, str]]:
         """Generate new journal prompts using AI."""
         self.console.print("\n[cyan]Generating new journal prompts...[/cyan]")