feat(intelligent_moderator): Reasoning levels

1 files changed, 82 insertions, 6 deletions

diff --git a/test_moderator.py b/test_moderator.py
index 0725642..70b57f2 100755
--- a/test_moderator.py
+++ b/test_moderator.py
@@ -175,9 +175,10 @@ class RedditDownloader:
 class TestIntelligentModerator(IntelligentModeratorBase):
     """Test version of the intelligent roleplay moderator."""
     
-    def __init__(self, openai_api_key: str):
+    def __init__(self, openai_api_key: str, reasoning_level: int = 2):
         """Initialize the test moderator."""
         super().__init__(openai_api_key)
+        self.reasoning_level = reasoning_level
     
     def test_file(self, file_path: Path, author_name: str = "testuser") -> Dict[str, Any]:
         """Test a single file against the moderator."""
@@ -249,6 +250,53 @@ class TestIntelligentModerator(IntelligentModeratorBase):
         
         return results
     
+    def _format_reasoning(self, evaluation: dict) -> str:
+        """Format reasoning based on the configured reasoning level."""
+        original_reasoning = evaluation.get('reasoning', '')
+        
+        if self.reasoning_level == 0:
+            # No reasoning included
+            return ""
+        elif self.reasoning_level == 1:
+            # Brief reasoning - extract key points
+            return self._extract_brief_reasoning(original_reasoning)
+        else:
+            # Full reasoning (default)
+            return original_reasoning
+    
+    def _extract_brief_reasoning(self, reasoning: str) -> str:
+        """Extract brief key points from the full reasoning."""
+        # Convert to lowercase for easier processing
+        reasoning_lower = reasoning.lower()
+        
+        # Extract key issues
+        issues = []
+        
+        # Check for common low-effort indicators
+        if any(word in reasoning_lower for word in ['short', 'brief', 'minimal', 'little']):
+            issues.append('short')
+        if any(word in reasoning_lower for word in ['word count', 'words', 'length']):
+            issues.append('low word count')
+        if any(word in reasoning_lower for word in ['plot', 'story', 'narrative']):
+            if 'no plot' in reasoning_lower or 'lack' in reasoning_lower:
+                issues.append('no plot development')
+        if any(word in reasoning_lower for word in ['effort', 'substance', 'content']):
+            if 'low' in reasoning_lower or 'lack' in reasoning_lower:
+                issues.append('low effort')
+        if any(word in reasoning_lower for word in ['creativity', 'originality']):
+            if 'lack' in reasoning_lower or 'no' in reasoning_lower:
+                issues.append('lacks creativity')
+        if any(word in reasoning_lower for word in ['artwork', 'art', 'visual', 'image']):
+            if 'primarily' in reasoning_lower or 'showcasing' in reasoning_lower:
+                issues.append('artwork showcase')
+        
+        # If no specific issues found, return a generic brief message
+        if not issues:
+            return "Content quality below standards"
+        
+        # Return comma-separated list of issues
+        return ", ".join(issues)
+    
     # Abstract method implementations for IntelligentModeratorBase
     def _get_submission_id(self, submission: MockSubmission) -> str:
         """Get the submission ID."""
@@ -291,7 +339,7 @@ class TestIntelligentModerator(IntelligentModeratorBase):
         print(f"ERROR: {message}")
 
 
-def print_results(results: List[Dict[str, Any]], pause: bool = False) -> None:
+def print_results(results: List[Dict[str, Any]], pause: bool = False, moderator: TestIntelligentModerator = None) -> None:
     """Print test results in a formatted way."""
     print("\n" + "="*80)
     print("TEST RESULTS")
@@ -335,13 +383,27 @@ def print_results(results: List[Dict[str, Any]], pause: bool = False) -> None:
             print(f"\n📧 Mod Mail (Art Flair Change):")
             print(f"  Subject: Your post flair has been changed to Art")
             print(f"  Message: Your roleplay post has been automatically re-flaired as 'Art' because it appears to be primarily showcasing artwork or visual content rather than roleplay.")
-            print(f"  Reasoning: {evaluation['reasoning']}")
+            
+            # Format reasoning based on moderator's reasoning level
+            if moderator:
+                formatted_reasoning = moderator._format_reasoning(evaluation)
+                if formatted_reasoning:
+                    print(f"  Reasoning: {formatted_reasoning}")
+            else:
+                print(f"  Reasoning: {evaluation['reasoning']}")
         
         if "REMOVE_POST" in actions:
             print(f"\n📧 Mod Mail (Low Effort Removal):")
             print(f"  Subject: Your roleplay post has been removed for low effort")
             print(f"  Message: Your roleplay post has been removed because it was determined to be low effort content.")
-            print(f"  Reasoning: {evaluation['reasoning']}")
+            
+            # Format reasoning based on moderator's reasoning level
+            if moderator:
+                formatted_reasoning = moderator._format_reasoning(evaluation)
+                if formatted_reasoning:
+                    print(f"  Reasoning: {formatted_reasoning}")
+            else:
+                print(f"  Reasoning: {evaluation['reasoning']}")
         
         # Add pause between results if requested
         if pause and i < len(results):
@@ -370,6 +432,12 @@ Examples:
   # Test with pause between results
   python test_moderator.py --directory real_test_posts/ --random 3 --pause
   
+  # Test with brief reasoning in mod mail
+  python test_moderator.py --directory real_test_posts/ --random 2 --reasoning-level 1
+  
+  # Test with no reasoning in mod mail
+  python test_moderator.py --file sample_post.txt --reasoning-level 0
+  
   # Download 20 roleplay posts from Reddit
   python test_moderator.py --download-roleplay 20
   
@@ -424,6 +492,14 @@ Examples:
     )
     
     parser.add_argument(
+        "--reasoning-level", "-rl",
+        type=int,
+        choices=[0, 1, 2],
+        default=2,
+        help="Reasoning level for mod mail (0=none, 1=brief, 2=full, default: 2)"
+    )
+    
+    parser.add_argument(
         "--download-roleplay",
         type=int,
         metavar="COUNT",
@@ -485,7 +561,7 @@ Examples:
     
     # Initialize moderator
     try:
-        moderator = TestIntelligentModerator(api_key)
+        moderator = TestIntelligentModerator(api_key, reasoning_level=args.reasoning_level)
     except Exception as e:
         print(f"❌ Error initializing moderator: {e}")
         sys.exit(1)
@@ -504,7 +580,7 @@ Examples:
         results.extend(dir_results)
     
     # Print results
-    print_results(results, pause=args.pause)
+    print_results(results, pause=args.pause, moderator=moderator)
     
     # Summary
     total = len(results)