fix: resolve all SonarCloud issues across Python + React frontend

Author: nikolay-e

benchmarks/ablation_benchmark.py

@@ -132,8 +132,129 @@ async def run_certamen_tournament(
     }
 
 
+_SECTION_DIVIDER = "---\n\n"
+
+
+async def _run_single_models(
+    certamen: Certamen, question: str
+) -> list[dict[str, Any]]:
+    results = []
+    for model_name in certamen.healthy_models:
+        print("\n" + "=" * 80)
+        print(f"APPROACH 1: Single Model ({model_name}) + Chain-of-Thought")
+        print("=" * 80 + "\n")
+        try:
+            result = await run_single_model_with_cot(
+                question, model_name, certamen
+            )
+            results.append(result)
+            print("\n✅ Complete!")
+            print(f"   Duration: {result['duration_seconds']:.1f}s")
+            print(f"   Cost: ${result['cost_estimate']:.4f}")
+            if not result["cost_estimate"]:
+                print("   (Local/free model - no API costs)")
+        except Exception as e:
+            logger.error(f"Failed to run {model_name}: {e}")
+            print(f"\n❌ Failed: {e}")
+            print("   Continuing with other models...")
+    return results
+
+
+def _print_cost_summary(
+    single_model_results: list[dict[str, Any]],
+    tournament_result: dict[str, Any],
+) -> None:
+    if not single_model_results:
+        return
+    print("\n📊 Cost Summary:")
+    print("-" * 80)
+    for i, result in enumerate(single_model_results):
+        print(
+            f"  Single Model #{i + 1} ({result['model']}): ${result['cost_estimate']:.4f}"
+        )
+    print(f"  Certamen Tournament: ${tournament_result['cost_actual']:.4f}")
+    avg_single_cost = sum(
+        r["cost_estimate"] for r in single_model_results
+    ) / len(single_model_results)
+    if avg_single_cost > 0:
+        print(
+            f"\n  Tournament vs Avg Single Model: {tournament_result['cost_actual'] / avg_single_cost:.2f}x cost"
+        )
+    print("-" * 80)
+
+
+def _write_micro_report(
+    output_path: Path,
+    question: str,
+    config_path: str,
+    single_model_results: list[dict[str, Any]],
+    tournament_result: dict[str, Any],
+) -> None:
+    with open(output_path, "w") as f:
+        f.write(
+            "# Micro-Benchmark Results: Single Model vs Certamen Framework\n\n"
+        )
+        f.write(f"Date: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n")
+        f.write(f"Config: `{config_path}`\n\n")
+        f.write("## Test Question\n\n")
+        f.write(f"{question}\n\n")
+        f.write(_SECTION_DIVIDER)
+
+        for i, single_result in enumerate(single_model_results):
+            f.write(f"## Approach 1.{i + 1}: Single Model\n\n")
+            f.write(f"Model: {single_result['model']}\n\n")
+            f.write(
+                f"Duration: {single_result['duration_seconds']:.1f} seconds\n\n"
+            )
+            f.write(f"Cost: ${single_result['cost_estimate']:.4f}\n\n")
+            f.write("### Response\n\n")
+            f.write(f"{single_result['response']}\n\n")
+
+        f.write(_SECTION_DIVIDER)
+        f.write("## Approach 2: Certamen Framework Tournament\n\n")
+        f.write(f"Champion: {tournament_result['champion_model']}\n\n")
+        f.write(
+            f"Duration: {tournament_result['duration_seconds']:.1f} seconds "
+            f"({tournament_result['duration_seconds'] / 60:.1f} minutes)\n\n"
+        )
+        f.write(f"Total Cost: ${tournament_result['cost_actual']:.4f}\n\n")
+
+        if tournament_result.get("cost_by_model"):
+            f.write("### Cost Breakdown by Model\n\n")
+            for model_key, cost in tournament_result["cost_by_model"].items():
+                f.write(f"- {model_key}: ${cost:.4f}\n")
+            f.write("\n")
+
+        if single_model_results:
+            cost_estimate = single_model_results[0]["cost_estimate"]
+            cost_actual = tournament_result["cost_actual"]
+            cost_multiple = (
+                f"{cost_actual / cost_estimate:.1f}x"
+                if cost_estimate > 0
+                else "N/A (free/local baseline)"
+            )
+            f.write(f"Cost Multiple vs Single Model: {cost_multiple}\n\n")
+            f.write(
+                f"Time Multiple vs Single Model: "
+                f"{tournament_result['duration_seconds'] / single_model_results[0]['duration_seconds']:.1f}x\n\n"
+            )
+
+        if tournament_result["eliminated_models"]:
+            f.write(
+                f"Eliminated Models: {', '.join(str(m) for m in tournament_result['eliminated_models'])}\n\n"
+            )
+
+        f.write("### Champion Response\n\n")
+        f.write(f"{tournament_result['response']}\n\n")
+        f.write(_SECTION_DIVIDER)
+
+        model_names = [res["model"] for res in single_model_results] + [
+            "Certamen Framework"
+        ]
+        f.write(generate_manual_evaluation_template(model_names))
+
+
 async def main(args: dict[str, Any] | None = None) -> None:
-    """Run micro-benchmark."""
     if args is None:
         parser = argparse.ArgumentParser(
             description="Micro-Benchmark: Single Model vs Certamen Framework"
@@ -145,7 +266,7 @@ async def main(args: dict[str, Any] | None = None) -> None:
             help="Path to configuration file (REQUIRED - no defaults)",
         )
         parsed_args = parser.parse_args()
-        args = vars(parsed_args)  # Convert to dict
+        args = vars(parsed_args)
 
     config_path = args.get("config")
     if not config_path:
@@ -155,21 +276,16 @@ async def main(args: dict[str, Any] | None = None) -> None:
             "The framework does not use fallback configurations."
         )
 
-    # Load config and inject outputs_dir
     from certamen.config.loader import Config
 
     config_obj = Config(config_path)
     if not config_obj.load():
         raise RuntimeError(f"Failed to load configuration from {config_path}")
 
-    # Benchmark explicitly sets outputs_dir to current directory
     config_obj.config_data["outputs_dir"] = "."
-
-    # Initialize Certamen (loads config, secrets, models, runs health check)
     logger.info("Initializing Certamen...")
     certamen = await Certamen.from_settings(settings=config_obj.config_data)
 
-    # Get question from config
     question = certamen.config.config_data.get("question")
     if not question:
         raise ValueError(
@@ -190,7 +306,6 @@ async def main(args: dict[str, Any] | None = None) -> None:
     print(f"\nTest Question:\n{question}")
     print()
 
-    # Check if we have any healthy models
     if not certamen.is_ready:
         print("\n❌ ERROR: No healthy models available")
         if certamen.failed_models:
@@ -205,31 +320,8 @@ async def main(args: dict[str, Any] | None = None) -> None:
         for model_key, error in certamen.failed_models.items():
             print(f"  - {model_key}: {error}")
 
-    single_model_results = []
+    single_model_results = await _run_single_models(certamen, question)
 
-    # Run single models
-    for model_name in certamen.healthy_models:
-        print("\n" + "=" * 80)
-        print(f"APPROACH 1: Single Model ({model_name}) + Chain-of-Thought")
-        print("=" * 80 + "\n")
-
-        try:
-            single_result = await run_single_model_with_cot(
-                question, model_name, certamen
-            )
-            single_model_results.append(single_result)
-
-            print("\n✅ Complete!")
-            print(f"   Duration: {single_result['duration_seconds']:.1f}s")
-            print(f"   Cost: ${single_result['cost_estimate']:.4f}")
-            if single_result["cost_estimate"] == 0.0:
-                print("   (Local/free model - no API costs)")
-        except Exception as e:
-            logger.error(f"Failed to run {model_name}: {e}")
-            print(f"\n❌ Failed: {e}")
-            print("   Continuing with other models...")
-
-    # Run tournament
     print("\n" + "=" * 80)
     print("APPROACH 2: Certamen Framework Tournament")
     print("=" * 80 + "\n")
@@ -238,114 +330,34 @@ async def main(args: dict[str, Any] | None = None) -> None:
 
     print("\n✅ Complete!")
     print(
-        f"   Duration: {tournament_result['duration_seconds']:.1f}s ({tournament_result['duration_seconds'] / 60:.1f} min)"
+        f"   Duration: {tournament_result['duration_seconds']:.1f}s "
+        f"({tournament_result['duration_seconds'] / 60:.1f} min)"
     )
     print(f"   Total Cost: ${tournament_result['cost_actual']:.4f}")
 
-    # Show cost breakdown
     if tournament_result.get("cost_by_model"):
         print("\n   Cost Breakdown:")
         for model_key, cost in tournament_result["cost_by_model"].items():
             print(f"     - {model_key}: ${cost:.4f}")
 
-    # Save results
     print("\n" + "=" * 80)
     print("BENCHMARK RESULTS")
     print("=" * 80)
 
-    # Print cost summary
-    if single_model_results:
-        print("\n📊 Cost Summary:")
-        print("-" * 80)
-        for i, result in enumerate(single_model_results):
-            print(
-                f"  Single Model #{i + 1} ({result['model']}): ${result['cost_estimate']:.4f}"
-            )
-        print(
-            f"  Certamen Tournament: ${tournament_result['cost_actual']:.4f}"
-        )
+    _print_cost_summary(single_model_results, tournament_result)
 
-        # Show comparison
-        avg_single_cost = sum(
-            r["cost_estimate"] for r in single_model_results
-        ) / len(single_model_results)
-        if avg_single_cost > 0:
-            print(
-                f"\n  Tournament vs Avg Single Model: {tournament_result['cost_actual'] / avg_single_cost:.2f}x cost"
-            )
-        print("-" * 80)
-
-    # Save reports relative to project root
     reports_dir = Path(__file__).parent / "reports"
     reports_dir.mkdir(parents=True, exist_ok=True)
     output_path = reports_dir / "micro_benchmark_results.md"
 
-    with open(output_path, "w") as f:
-        f.write(
-            "# Micro-Benchmark Results: Single Model vs Certamen Framework\n\n"
-        )
-        f.write(f"Date: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n")
-        f.write(f"Config: `{config_path}`\n\n")
-
-        f.write("## Test Question\n\n")
-        f.write(f"{question}\n\n")
-
-        f.write("---\n\n")
-
-        for i, single_result in enumerate(single_model_results):
-            f.write(f"## Approach 1.{i + 1}: Single Model\n\n")
-            f.write(f"Model: {single_result['model']}\n\n")
-            f.write(
-                f"Duration: {single_result['duration_seconds']:.1f} seconds\n\n"
-            )
-            f.write(f"Cost: ${single_result['cost_estimate']:.4f}\n\n")
-            f.write("### Response\n\n")
-            f.write(f"{single_result['response']}\n\n")
-
-        f.write("---\n\n")
-
-        f.write("## Approach 2: Certamen Framework Tournament\n\n")
-        f.write(f"Champion: {tournament_result['champion_model']}\n\n")
-        f.write(
-            f"Duration: {tournament_result['duration_seconds']:.1f} seconds ({tournament_result['duration_seconds'] / 60:.1f} minutes)\n\n"
-        )
-        f.write(f"Total Cost: ${tournament_result['cost_actual']:.4f}\n\n")
-
-        # Cost breakdown by model
-        if tournament_result.get("cost_by_model"):
-            f.write("### Cost Breakdown by Model\n\n")
-            for model_key, cost in tournament_result["cost_by_model"].items():
-                f.write(f"- {model_key}: ${cost:.4f}\n")
-            f.write("\n")
-
-        # Cost and time comparisons
-        if single_model_results:
-            cost_estimate = single_model_results[0]["cost_estimate"]
-            cost_actual = tournament_result["cost_actual"]
-            if cost_estimate > 0:
-                cost_multiple = f"{cost_actual / cost_estimate:.1f}x"
-            else:
-                cost_multiple = "N/A (free/local baseline)"
-            f.write(f"Cost Multiple vs Single Model: {cost_multiple}\n\n")
-
-            f.write(
-                f"Time Multiple vs Single Model: {tournament_result['duration_seconds'] / single_model_results[0]['duration_seconds']:.1f}x\n\n"
-            )
-
-        if tournament_result["eliminated_models"]:
-            f.write(
-                f"Eliminated Models: {', '.join(str(m) for m in tournament_result['eliminated_models'])}\n\n"
-            )
-
-        f.write("### Champion Response\n\n")
-        f.write(f"{tournament_result['response']}\n\n")
-
-        f.write("---\n\n")
-
-        model_names = [res["model"] for res in single_model_results] + [
-            "Certamen Framework"
-        ]
-        f.write(generate_manual_evaluation_template(model_names))
+    await asyncio.to_thread(
+        _write_micro_report,
+        output_path,
+        question,
+        config_path,
+        single_model_results,
+        tournament_result,
+    )
 
     print(f"\n📄 Results saved to: {output_path}")
     print()