refactor: fix critical code inconsistencies (types, protocols, logging)

.github/workflows/diff-context.yml

@@ -0,0 +1,10 @@
+name: Diff Context
+on: [pull_request]
+
+permissions:
+  contents: read
+  pull-requests: write
+
+jobs:
+  diff:
+    uses: nikolay-e/treemapper-action/.github/workflows/diff-context.yml@v1

src/arbitrium_core/application/execution/executor.py

@@ -6,6 +6,7 @@
 
 from arbitrium_core.application.workflow.nodes.base import BaseNode
 from arbitrium_core.application.workflow.registry import registry
+from arbitrium_core.domain.errors import GraphValidationError
 from arbitrium_core.shared.logging import get_contextual_logger
 
 logger = get_contextual_logger(__name__)
@@ -158,7 +159,9 @@ def _build_graph(
                 errors.append(f"Unknown node type: {node_type}")
 
         if errors:
-            raise ValueError(f"Graph build errors: {'; '.join(errors)}")
+            raise GraphValidationError(
+                f"Graph build errors: {'; '.join(errors)}"
+            )
 
         dependencies: dict[str, list[str]] = defaultdict(list)
         connections: dict[str, list[tuple[str, str, str]]] = defaultdict(list)
@@ -178,11 +181,11 @@ def _build_graph(
             )
 
             if source not in node_instances:
-                raise ValueError(
+                raise GraphValidationError(
                     f"Edge references unknown source node: {source}"
                 )
             if target not in node_instances:
-                raise ValueError(
+                raise GraphValidationError(
                     f"Edge references unknown target node: {target}"
                 )
 
@@ -226,7 +229,9 @@ def _topological_sort(
                         queue.append(other_id)
 
         if len(result) != len(nodes):
-            raise ValueError("Graph contains a cycle - check node connections")
+            raise GraphValidationError(
+                "Graph contains a cycle - check node connections"
+            )
 
         return result
 
@@ -260,7 +265,7 @@ def _build_execution_layers(
             ]
 
             if not current_layer:
-                raise ValueError(
+                raise GraphValidationError(
                     "Graph contains a cycle - check node connections"
                 )
 
@@ -358,7 +363,7 @@ def _validate_workflow(
         node_instances: dict[str, BaseNode],
     ) -> None:
         if not node_instances:
-            raise ValueError("No valid nodes in graph")
+            raise GraphValidationError("No valid nodes in graph")
 
     def _build_execution_graph(
         self,

src/arbitrium_core/domain/errors.py

@@ -16,6 +16,7 @@
     "ExceptionClassifier",
     "FatalError",
     "FileSystemError",
+    "GraphValidationError",
     "InputError",
     "ModelError",
     "ModelResponseError",
@@ -112,6 +113,23 @@ def __init__(
         super().__init__(enhanced_message, *args)
 
 
+class GraphValidationError(ArbitriumError):
+    """Exception for workflow graph validation errors."""
+
+    def __init__(
+        self,
+        message: str,
+        node_id: str | None = None,
+        *args: object,
+        **kwargs: object,
+    ) -> None:
+        self.node_id = node_id
+        enhanced_message = message
+        if node_id:
+            enhanced_message = f"[Node: {node_id}] {enhanced_message}"
+        super().__init__(enhanced_message, *args)
+
+
 class ErrorClassification:
     __slots__ = ("error_type", "is_retryable", "message")
 

src/arbitrium_core/domain/knowledge/bank.py

@@ -63,14 +63,16 @@ def _determine_extractor_model_key(
                     else None
                 )
                 self.logger.warning(
-                    f"Leader not determined yet, using fallback model: {extractor_model_key}"
+                    "Leader not determined yet, using fallback model: %s",
+                    extractor_model_key,
                 )
             else:
                 leader_display = self.comparison.anon_mapping.get(
                     extractor_model_key, extractor_model_key
                 )
                 self.logger.info(
-                    f"Using tournament leader {leader_display} for insight extraction"
+                    "Using tournament leader %s for insight extraction",
+                    leader_display,
                 )
             return extractor_model_key
         else:
@@ -129,13 +131,16 @@ def _parse_claims_from_response(
         self, response_content: str, extractor_model_key: str
     ) -> list[str]:
         self.logger.debug(
-            f"[{extractor_model_key}] Raw insight extraction response: {response_content}"
+            "[%s] Raw insight extraction response: %s",
+            extractor_model_key,
+            response_content,
         )
 
         if detect_apology_or_refusal(response_content):
             self.logger.error(
-                f"[{extractor_model_key}] Model returned apology/refusal instead of insight extraction. "
-                f"Response: {response_content}"
+                "[%s] Model returned apology/refusal instead of insight extraction. Response: %s",
+                extractor_model_key,
+                response_content,
             )
             return []
 
@@ -145,7 +150,9 @@ def _parse_claims_from_response(
 
         if not claims:
             self.logger.warning(
-                f"[{extractor_model_key}] No valid insights found in response. Response: {response_content}"
+                "[%s] No valid insights found in response. Response: %s",
+                extractor_model_key,
+                response_content,
             )
 
         return claims

src/arbitrium_core/domain/tournament/tournament.py

@@ -1,12 +1,17 @@
 "Core comparison functionality for Arbitrium Framework."
 
 import asyncio
+import logging
 import re
 import statistics
 from abc import ABC, abstractmethod
 from collections.abc import Callable
 from datetime import datetime
-from typing import Any
+from pathlib import Path
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    from arbitrium_core.shared.logging.structured import ContextualLogger
 
 from arbitrium_core.domain.errors import (
     BudgetExceededError,
@@ -19,6 +24,9 @@
     PromptBuilder,
     PromptFormatter,
 )
+from arbitrium_core.domain.tournament.anonymizer import ModelAnonymizer
+from arbitrium_core.domain.tournament.report import ReportGenerator
+from arbitrium_core.domain.tournament.scoring import ScoreExtractor
 from arbitrium_core.ports.llm import BaseModel, ModelResponse
 from arbitrium_core.ports.similarity import SimilarityEngine
 from arbitrium_core.shared.constants import (
@@ -29,10 +37,6 @@
 from arbitrium_core.shared.logging import get_contextual_logger
 from arbitrium_core.shared.text import indent_text, strip_meta_commentary
 
-from .anonymizer import ModelAnonymizer
-from .report import ReportGenerator
-from .scoring import ScoreExtractor
-
 
 # Internal interfaces for ModelComparison
 class EventHandler(ABC):
@@ -42,7 +46,7 @@ def publish(self, _event_name: str, _data: dict[str, Any]) -> None:
 
 
 class HostEnvironment(ABC):
-    base_dir: Any  # Output directory path (required by implementations)
+    base_dir: Path | str  # Output directory path (required by implementations)
 
     @abstractmethod
     async def read_file(self, path: str) -> str:
@@ -116,7 +120,9 @@ def reset(self) -> None:
 
 
 class CostTracker:
-    def __init__(self, logger: Any = None):
+    def __init__(
+        self, logger: "logging.Logger | ContextualLogger | None" = None
+    ):
         self.total_cost = 0.0
         self.cost_by_model: dict[str, float] = {}
         self.logger = logger or get_contextual_logger("arbitrium.cost_tracker")
@@ -129,7 +135,10 @@ def add_cost(self, model_name: str, cost: float) -> None:
         self.cost_by_model[model_name] += cost
 
         self.logger.debug(
-            f"💰 Added ${cost:.4f} for {model_name}, total now: ${self.total_cost:.4f}"
+            "Added $%.4f for %s, total now: $%.4f",
+            cost,
+            model_name,
+            self.total_cost,
         )
 
     def get_summary(self) -> dict[str, Any]:
@@ -171,7 +180,7 @@ def display_summary(self) -> None:
                 extra={"display_type": "colored_text"},
             )
 
-        self.logger.info(f"💰 Tournament total cost: ${self.total_cost:.4f}")
+        self.logger.info("Tournament total cost: $%.4f", self.total_cost)
 
 
 class TournamentRunner:
@@ -196,7 +205,7 @@ async def run(self, initial_question: str) -> str:
         )
 
         self.logger.info(
-            f"Starting model comparison tournament: {initial_question}"
+            "Starting model comparison tournament: %s", initial_question
         )
         self.comp.previous_answers = []
         self.comp.eliminated_models = []
@@ -210,9 +219,7 @@ async def run(self, initial_question: str) -> str:
             self.logger.warning("Process interrupted by user.")
             return "Process interrupted by user."
         except Exception as e:
-            self.logger.error(
-                f"Unexpected error in tournament: {e!s}", exc_info=True
-            )
+            self.logger.exception("Unexpected error in tournament: %s", e)
             return f"Tournament error: {e!s}"
 
     async def _run_initial_phase(self, initial_question: str) -> bool:

src/arbitrium_core/engine.py

@@ -9,7 +9,7 @@
 from arbitrium_core.ports.similarity import SimilarityEngine
 from arbitrium_core.shared.logging import get_contextual_logger
 
-logger = get_contextual_logger("arbitrium")
+logger = get_contextual_logger(__name__)
 
 
 class _InternalEventHandler:
@@ -157,7 +157,7 @@ async def run_tournament(
                 f"Failed models: {list(self._failed_models.keys())}"
             )
 
-        logger.info(f"Starting tournament with {len(models)} models")
+        logger.info("Starting tournament with %d models", len(models))
 
         comparison = self._create_comparison(models)
         self._last_comparison = comparison
@@ -198,16 +198,16 @@ async def run_all_models(self, prompt: str) -> dict[str, ModelResponse]:
             raise RuntimeError("No healthy models available")
 
         logger.info(
-            f"Running prompt through {len(self._healthy_models)} models"
+            "Running prompt through %d models", len(self._healthy_models)
         )
 
         results = {}
         for model_key in self._healthy_models:
             try:
                 response = await self.run_single_model(model_key, prompt)
                 results[model_key] = response
-            except Exception as e:
-                logger.error(f"Failed to run {model_key}: {e}")
+            except Exception:
+                logger.exception("Failed to run %s", model_key)
                 # Continue with other models
                 continue
 

src/arbitrium_core/infrastructure/llm/model_factory.py

@@ -3,15 +3,15 @@
 from arbitrium_core.ports.llm import BaseModel
 from arbitrium_core.shared.logging import get_contextual_logger
 
-logger = get_contextual_logger("arbitrium.infrastructure.llm.model_factory")
+logger = get_contextual_logger(__name__)
 
 
 async def ensure_model_instances(
     models: dict[str, Any],
-) -> dict[str, Any]:
+) -> dict[str, BaseModel]:
     from arbitrium_core.infrastructure.llm.litellm_adapter import LiteLLMModel
 
-    result = {}
+    result: dict[str, BaseModel] = {}
     for key, model in models.items():
         if isinstance(model, BaseModel):
             result[key] = model
@@ -22,14 +22,16 @@ async def ensure_model_instances(
                     model_config=model,
                 )
                 result[key] = instance
-            except Exception as e:
-                logger.error(f"Failed to create model from config {key}: {e}")
+            except Exception:
+                logger.exception("Failed to create model from config %s", key)
         else:
-            logger.warning(f"Unknown model type for {key}: {type(model)}")
+            logger.warning("Unknown model type for %s: %s", key, type(model))
     return result
 
 
-async def ensure_single_model_instance(model: Any, key: str = "model") -> Any:
+async def ensure_single_model_instance(
+    model: Any, key: str = "model"
+) -> BaseModel | None:
     from arbitrium_core.infrastructure.llm.litellm_adapter import LiteLLMModel
 
     if isinstance(model, BaseModel):
@@ -40,9 +42,9 @@ async def ensure_single_model_instance(model: Any, key: str = "model") -> Any:
                 model_key=model.get("name", key),
                 model_config=model,
             )
-        except Exception as e:
-            logger.error(f"Failed to create model from config {key}: {e}")
+        except Exception:
+            logger.exception("Failed to create model from config %s", key)
             return None
     else:
-        logger.warning(f"Unknown model type for {key}: {type(model)}")
+        logger.warning("Unknown model type for %s: %s", key, type(model))
         return None

src/arbitrium_core/ports/cache.py

@@ -1,14 +1,11 @@
-from abc import ABC, abstractmethod
+from typing import Protocol
 
 
-class CacheProtocol(ABC):
-    @abstractmethod
+class CacheProtocol(Protocol):
     def get(
         self, model_name: str, prompt: str, temperature: float, max_tokens: int
-    ) -> tuple[str, float] | None:
-        pass
+    ) -> tuple[str, float] | None: ...
 
-    @abstractmethod
     def set(
         self,
         model_name: str,
@@ -17,13 +14,8 @@ def set(
         max_tokens: int,
         response: str,
         cost: float,
-    ) -> None:
-        pass
+    ) -> None: ...
 
-    @abstractmethod
-    def clear(self) -> None:
-        pass
+    def clear(self) -> None: ...
 
-    @abstractmethod
-    def close(self) -> None:
-        pass
+    def close(self) -> None: ...

src/arbitrium_core/ports/llm.py

@@ -1,6 +1,9 @@
 from abc import ABC, abstractmethod
 from collections.abc import Awaitable
-from typing import Any, Protocol
+from typing import TYPE_CHECKING, Any, Protocol
+
+if TYPE_CHECKING:
+    from arbitrium_core.ports.cache import CacheProtocol
 
 
 class ModelResponse:
@@ -94,5 +97,5 @@ def from_config(
         cls,
         model_key: str,
         config: dict[str, Any],
-        response_cache: Any | None = None,
+        response_cache: "CacheProtocol | None" = None,
     ) -> Awaitable[BaseModel]: ...