fix: prevent Ollama tool summary from being overwritten by empty response

- Add condition to check if final_response_text is already set before overwriting
- Fix applies to both sync (get_response) and async (get_response_async) methods
- Prevents tool summary from being lost when Ollama returns empty response after tool execution
- Maintains backward compatibility for non-Ollama providers
- Resolves issue where agent.start() returns None despite successful tool execution

Fixes the remaining issue from PR #943 where tool summary was generated but overwritten

Co-authored-by: Mervin Praison <[email protected]>

Author: github-actions[bot]

src/praisonai-agents/praisonaiagents/llm/llm.py

@@ -1120,7 +1120,7 @@ def get_response(
                     else:
                         # No tool calls, we're done with this iteration
                         # If we've executed tools in previous iterations, this response contains the final answer
-                        if iteration_count > 0:
+                        if iteration_count > 0 and not final_response_text:
                             final_response_text = response_text.strip() if response_text else ""
                         break
 
@@ -1869,7 +1869,7 @@ async def get_response_async(
                 else:
                     # No tool calls, we're done with this iteration
                     # If we've executed tools in previous iterations, this response contains the final answer
-                    if iteration_count > 0:
+                    if iteration_count > 0 and not final_response_text:
                         final_response_text = response_text.strip()
                     break
 

test_ollama_response_fix.py

@@ -0,0 +1,132 @@
+#!/usr/bin/env python3
+"""
+Test script to verify Ollama response overwrite fix
+"""
+
+import sys
+import os
+
+# Add the source directory to the path
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), 'src', 'praisonai-agents'))
+
+def test_ollama_tool_summary_logic():
+    """Test the logic that was causing the bug"""
+    print("Testing Ollama tool summary logic...")
+    
+    try:
+        from praisonaiagents.llm.llm import LLM
+        
+        # Create an Ollama LLM instance
+        llm = LLM(model="ollama/test")
+        
+        # Test the tool summary generation method directly
+        tool_results = ["The stock price of Google is 100"]
+        empty_response = ""  # This is what Ollama often returns
+        
+        tool_summary = llm._generate_ollama_tool_summary(tool_results, empty_response)
+        
+        if tool_summary:
+            print(f"✅ Tool summary generated correctly: {repr(tool_summary)}")
+            
+            # Simulate the fixed logic:
+            # 1. Set final_response_text to tool_summary (this happens in the fix)
+            final_response_text = tool_summary
+            print(f"✅ final_response_text set to: {repr(final_response_text)}")
+            
+            # 2. Test the condition that was causing the bug
+            iteration_count = 1  # Simulate that tools were executed
+            response_text = ""   # Simulate empty Ollama response
+            
+            # OLD BUGGY LOGIC:
+            # if iteration_count > 0:
+            #     final_response_text = response_text.strip() if response_text else ""
+            # This would overwrite final_response_text with ""
+            
+            # NEW FIXED LOGIC:
+            if iteration_count > 0 and not final_response_text:
+                final_response_text = response_text.strip() if response_text else ""
+                print("❌ This condition should NOT execute because final_response_text is already set")
+            else:
+                print("✅ Condition correctly skipped - final_response_text preserved")
+            
+            # 3. Verify final_response_text is still the tool summary
+            if final_response_text == tool_summary:
+                print("✅ SUCCESS: Tool summary preserved through the fix!")
+                return True
+            else:
+                print(f"❌ FAILED: Tool summary was overwritten. Expected: {repr(tool_summary)}, Got: {repr(final_response_text)}")
+                return False
+        else:
+            print("❌ Tool summary generation failed")
+            return False
+            
+    except Exception as e:
+        print(f"❌ Test failed with error: {e}")
+        return False
+
+def test_non_ollama_compatibility():
+    """Test that the fix doesn't break non-Ollama models"""
+    print("\nTesting non-Ollama compatibility...")
+    
+    try:
+        from praisonaiagents.llm.llm import LLM
+        
+        # Create a non-Ollama LLM instance
+        llm = LLM(model="gpt-3.5-turbo")
+        
+        # Test the tool summary generation method with non-Ollama model
+        tool_results = ["The stock price of Google is 100"]
+        response_text = "Based on the tools, Google's stock price is 100"
+        
+        tool_summary = llm._generate_ollama_tool_summary(tool_results, response_text)
+        
+        if tool_summary is None:
+            print("✅ Non-Ollama model correctly returns None for tool summary")
+            
+            # Simulate the logic for non-Ollama models
+            final_response_text = ""
+            iteration_count = 1
+            
+            # This should execute for non-Ollama models since final_response_text is empty
+            if iteration_count > 0 and not final_response_text:
+                final_response_text = response_text.strip() if response_text else ""
+                print("✅ Non-Ollama logic executed correctly")
+            
+            if final_response_text == response_text:
+                print("✅ SUCCESS: Non-Ollama compatibility maintained!")
+                return True
+            else:
+                print(f"❌ FAILED: Non-Ollama logic broken. Expected: {repr(response_text)}, Got: {repr(final_response_text)}")
+                return False
+        else:
+            print(f"❌ Non-Ollama model incorrectly generated summary: {tool_summary}")
+            return False
+            
+    except Exception as e:
+        print(f"❌ Test failed with error: {e}")
+        return False
+
+def main():
+    """Run all tests"""
+    print("=" * 60)
+    print("Testing Ollama Response Overwrite Fix")
+    print("=" * 60)
+    
+    test1_passed = test_ollama_tool_summary_logic()
+    test2_passed = test_non_ollama_compatibility()
+    
+    print("\n" + "=" * 60)
+    print("TEST RESULTS:")
+    print(f"Ollama tool summary logic: {'✅ PASSED' if test1_passed else '❌ FAILED'}")
+    print(f"Non-Ollama compatibility: {'✅ PASSED' if test2_passed else '❌ FAILED'}")
+    
+    if test1_passed and test2_passed:
+        print("\n🎉 ALL TESTS PASSED! The fix should resolve the issue.")
+        return True
+    else:
+        print("\n💥 SOME TESTS FAILED! The fix needs more work.")
+        return False
+
+if __name__ == "__main__":
+    success = main()
+    sys.exit(0 if success else 1)