fix(logging): use structured log

Author: dineshpinto

pyproject.toml

@@ -13,6 +13,7 @@ dependencies = [
     "pandas>=2.2.3",
     "python-dotenv>=1.0.1",
     "requests>=2.32.3",
+    "structlog>=25.1.0",
 ]
 
 [dependency-groups]
@@ -33,7 +34,7 @@ target-version = "py312"
 
 [tool.ruff.lint]
 select = ["ALL"]
-ignore = ["D203", "D212", "COM812", "D", "S105", "ANN401", "T201", "ISC003"]
+ignore = ["D203", "D212", "COM812", "D", "S105", "ANN401", "ISC003"]
 
 [tool.ruff.lint.extend-per-file-ignores]
 "tests/**/*.py" = ["S101", "ARG"]

src/data/final_consensus.json

@@ -1,8 +1,8 @@
 {
-    "aggregated_response": "Here's a combined response based on the majority features of the previous answers:\n\n**Why a freshly painted room may feel warmer than it actually is**\n\nWhen you enter a freshly painted room, you may notice that it feels warmer than it actually is. This warming effect is due to several factors:\n\n1. **Volatile Organic Compounds (VOCs)**: Paints and coatings often contain VOCs, which are chemicals that evaporate quickly at room temperature. These VOCs can contribute to the warming effect by releasing heat as they transition from a liquid to a gas state.\n2. **Moisture**: Paints can contain moisture, which is released as the paint dries. As the moisture warms up, it condenses into water vapor, releasing latent heat into the air.\n3. **Thermal Mass**: Painted surfaces can have a higher thermal mass than unpainted surfaces, absorbing and releasing heat more slowly. As the paint dries, it warms up the room by absorbing",
+    "aggregated_response": "Here's a combined response based on the majority of the models:\n\nHeat is emitted in a freshly painted room due to two primary reasons: evaporation of solvents and release of volatile organic compounds (VOCs) from the paint, and the exothermic chemical reactions that occur during the drying process.\n\nWhen paint is applied to a surface, it contains solvents and other chemicals that are volatile, meaning they can evaporate at room temperature. As the paint dries, these solvents and VOCs evaporate into the air, carrying heat with them. This process of evaporation is endothermic, meaning it absorbs heat from the surrounding environment, resulting in a cooler temperature in the room.\n\nAdditionally, the paint itself also emits heat as it dries due to the exothermic chemical reactions that occur during the drying process. These reactions involve the breakdown of the paint's components, releasing energy in the form of heat. This heat is then distributed throughout the room, contributing to the overall warmth",
     "responses": {
-        "qwen/qwen-vl-plus:free": "When you enter a freshly painted room, you may notice that it feels warmer than it actually is. This warming effect is due to several factors:\n\n1. **Volatile Organic Compounds (VOCs)**: Paints and coatings often contain VOCs, which are chemicals that evaporate quickly at room temperature. These VOCs can contribute to the warming effect.\n2. **Moisture**: Paints can contain moisture, either from the paint itself or from the air in the room. As the paint dries, it releases moisture into the air, which can be cooled as it condenses, releasing heat.\n3. **Thermal Mass**: Painted surfaces can have a higher thermal mass than unpainted surfaces, meaning they can absorb and release heat more slowly. As the paint dries, it warms up the room by absorbing and storing heat.\n4. **Evaporation**: The process of paint drying releases heat as the solvent in the paint evaporates. This heat contributes to the",
-        "sophosympatheia/rogue-rose-103b-v0.2:free": " When you enter a freshly painted room, it may feel warmer than it actually is due to several factors.\n\n1. **Volatile Organic Compounds (VOCs)**: Paints and coatings often contain VOCs, which are chemicals that evaporate quickly at room temperature. As these VOCs evaporate, they release heat, contributing to the warming effect in the room.\n2. **Moisture**: Paints can contain moisture, either from the paint itself or from the air in the room. As the paint dries, it releases moisture into the air, which can be cooled as it condenses, releasing heat.\n3. **Thermal Mass**: Painted surfaces can have a higher thermal mass than unpainted surfaces, meaning they can absorb and release heat more slowly. As the paint dries, it warms up",
-        "meta-llama/llama-3.2-3b-instruct:free": "When you enter a freshly painted room, heat is emitted due to several factors:\n\n1. **Volatile Organic Compounds (VOCs)**: Paints and coatings often contain VOCs, which evaporate quickly at room temperature and release heat as they transition from a liquid to a gas state.\n2. **Moisture**: Paints can contain moisture, which is released as the paint dries. As the moisture warms up, it condenses into water vapor, releasing latent heat into the air.\n3. **Thermal Mass**: Painted surfaces can have a higher thermal mass than unpainted surfaces, absorbing and releasing heat more slowly. As the paint dries, it warms up the room by absorbing and storing heat.\n4. **Evaporation**: As the paint dries, the solvents and other volatile components evaporate, releasing heat into the air.\n\nThese factors combined contribute to the warming effect you may experience when entering a freshly painted room."
+        "qwen/qwen-vl-plus:free": "Heat is emitted in a freshly painted room due to the evaporation of solvents and volatile organic compounds (VOCs) from the paint, which absorbs heat from the surrounding environment, and the chemical reactions that occur during the drying process, which release energy in the form of heat.",
+        "sophosympatheia/rogue-rose-103b-v0.2:free": " Heat is emitted in a freshly painted room due to two primary reasons: evaporation of solvents and release of volatile organic compounds (VOCs) from the paint, and the exothermic chemical reactions that occur during the drying process.\n\n1. Evaporation of solvents and VOCs: When paint is applied to a surface, it contains solvents and other chemicals that are volatile, meaning they can evaporate at room temperature. As the paint dries, these solvents and VOCs evaporate into the air, carrying heat with them. This process of evaporation is endothermic, meaning it absorbs heat from the surrounding environment, resulting in a cooler temperature in the room.\n2. Exothermic chemical reactions: The paint itself also emits heat as it dries due to the exothermic chemical reactions that occur during",
+        "meta-llama/llama-3.2-3b-instruct:free": "Heat is emitted in a freshly painted room because the paint contains volatile organic compounds (VOCs) that evaporate as the paint dries, absorbing heat from the surrounding environment, and the chemical reactions that occur during drying release energy in the form of heat."
     }
 }

src/flare_ai_consensus/consensus/consensus.py

@@ -1,9 +1,13 @@
 import asyncio
 
+import structlog
+
 from flare_ai_consensus.consensus.config import ConsensusConfig, ModelConfig
 from flare_ai_consensus.router.client import AsyncOpenRouterClient
 from flare_ai_consensus.utils.parser import parse_chat_response
 
+logger = structlog.get_logger(__name__)
+
 
 def build_improvement_conversation(
     consensus_config: ConsensusConfig, aggregated_response: str
@@ -50,13 +54,13 @@ async def get_response_for_model(
     if aggregated_response is None:
         # Use initial prompt for the first round.
         conversation = consensus_config.initial_prompt
-        print(f"Sending initial prompt to {model.model_id}.")
+        logger.info("sending initial prompt", model_id=model.model_id)
     else:
         # Build the improvement conversation.
         conversation = build_improvement_conversation(
             consensus_config, aggregated_response
         )
-        print(f"Sending improvement prompt to {model.model_id}.")
+        logger.info("sending improvement prompt", model_id=model.model_id)
 
     payload = {
         "model": model.model_id,
@@ -66,8 +70,7 @@ async def get_response_for_model(
     }
     response = await client.send_chat_completion(payload)
     text = parse_chat_response(response)
-    print(f"{model.model_id} has provided a new response.")
-
+    logger.info("new response", model_id=model.model_id, response=text)
     return model.model_id, text
 
 

src/flare_ai_consensus/main.py

@@ -1,5 +1,7 @@
 import asyncio
 
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.consensus import aggregator, consensus
 from flare_ai_consensus.consensus.config import ConsensusConfig
@@ -9,6 +11,8 @@
     saver,
 )
 
+logger = structlog.get_logger(__name__)
+
 
 async def run_consensus(
     client: AsyncOpenRouterClient,
@@ -26,7 +30,7 @@ async def run_consensus(
     aggregated_response = await aggregator.async_centralized_llm_aggregator(
         client, consensus_config.aggregator_config, responses
     )
-    print("\nInitial responses have been aggregated.")
+    logger.info("initial response aggregation complete")
 
     # Step 2: Improvement rounds.
     for i in range(consensus_config.iterations):
@@ -36,15 +40,15 @@ async def run_consensus(
         aggregated_response = await aggregator.async_centralized_llm_aggregator(
             client, consensus_config.aggregator_config, responses
         )
-        print(f"\nThe responses have been aggregated after iteration {i + 1}:")
+        logger.info("responses aggregated", iteration=i + 1)
 
     # Step 3: Save final consensus.
     output_file = config.data_path / "final_consensus.json"
     saver.save_json(
         {"aggregated_response": aggregated_response, "responses": responses},
         output_file,
     )
-    print(f"\nFinal consensus saved to {output_file}")
+    logger.info("saved consensus", output_file=output_file)
 
     # Close the async client to release resources.
     await client.close()

src/flare_ai_consensus/utils/saver.py

@@ -1,10 +1,14 @@
 import json
 from pathlib import Path
 
+import structlog
+
+logger = structlog.get_logger(__name__)
+
 
 def save_json(contents: dict, file_path: Path) -> None:
     """Save json files to specified path."""
 
     with file_path.open("w") as f:
         json.dump(contents, f, indent=4)
-    print(f"Data saved to {file_path}.")
+    logger.info("saved data", file_path=file_path)

tests/chat_completion.py

@@ -1,12 +1,16 @@
 import argparse
 
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.router import (
     requests,  # This module should expose send_chat_completion
 )
 from flare_ai_consensus.router.client import OpenRouterClient
 from flare_ai_consensus.utils.saver import save_json
 
+logger = structlog.get_logger(__name__)
+
 
 def parse_arguments() -> argparse.Namespace:
     parser = argparse.ArgumentParser(
@@ -43,40 +47,43 @@ def default_mode(
             "temperature": 0.7,
         }
         try:
-            print(f"\nIteration {i + 1}: Sending conversation to model {model_id} ...")
+            logger.info("sending conversation", model_id=model_id, iteration=i + 1)
             response = requests.send_chat_completion(client, payload)
             # Extract the assistant's response.
             assistant_response = (
                 response.get("choices", [])[0].get("message", {}).get("content", "")
             )
-            print(f"Assistant Response (Iteration {i + 1}):")
-            print(assistant_response)
+            logger.info(
+                "assistant response",
+                iteration=i + 1,
+                assistant_response=assistant_response,
+            )
 
             # Append the assistant's response to the conversation history.
             conversation.append({"role": "assistant", "content": assistant_response})
 
             # Ask for improved response
             if i < num_iterations - 1:
                 conversation.append({"role": "user", "content": improvement_prompt})
-        except Exception as e:  # noqa: BLE001
-            print(f"Error for model {model_id} in iteration {i + 1}: {e}")
+        except Exception:
+            logger.exception("error", model_id=model_id, iteration=i + 1)
             break
 
     # Save the final conversation to a file.
     output_file = config.data_path / "chat_response.json"
     save_json({"conversation": conversation}, output_file)
-    print(f"\nFinal conversation saved to {output_file}")
+    logger.info("final conversation saved", output_file=output_file)
 
 
 def interactive_mode(client: OpenRouterClient, model_id: str) -> None:
     """Run the chat in interactive mode."""
     conversation = []
-    print("Interactive mode. Type 'exit' to quit.")
+    print("Interactive mode. Type 'exit' to quit.")  # noqa: T201
 
     while True:
         user_input = input("\nEnter your 'user' prompt: ")
         if user_input.strip().lower() == "exit":
-            print("Exiting interactive mode.")
+            print("Exiting interactive mode.")  # noqa: T201
             break
 
         conversation.append({"role": "user", "content": user_input})
@@ -93,11 +100,11 @@ def interactive_mode(client: OpenRouterClient, model_id: str) -> None:
             assistant_msg = (
                 response.get("choices", [])[0].get("message", {}).get("content", "")
             )
-            print("\nAssistant:")
-            print(assistant_msg)
+            print("\nAssistant:")  # noqa: T201
+            print(assistant_msg)  # noqa: T201
             conversation.append({"role": "assistant", "content": assistant_msg})
         except Exception as e:  # noqa: BLE001
-            print(f"Error for model {model_id}: {e}")
+            print(f"Error for model {model_id}: {e}")  # noqa: T201
 
 
 def main() -> None:
@@ -138,7 +145,7 @@ def main() -> None:
     elif args.mode == "interactive":
         interactive_mode(client, model_id)
     else:
-        print("Invalid mode. Please choose 'default' or 'interactive'.")
+        logger.info("invalid mode: choose 'default' or 'interactive'")
 
 
 if __name__ == "__main__":

tests/completion.py

@@ -1,10 +1,14 @@
 import argparse
 
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.router import requests
 from flare_ai_consensus.router.client import OpenRouterClient
 from flare_ai_consensus.utils.saver import save_json
 
+logger = structlog.get_logger(__name__)
+
 
 def parse_arguments() -> argparse.Namespace:
     """Parse command-line arguments and return the parsed namespace."""
@@ -58,17 +62,19 @@ def start_chat(args: argparse.Namespace) -> None:
     )
 
     try:
-        print(f"Sending prompt to model {model_id} ...")
+        logger.info("sending prompt", model_id=model_id)
         response = send_prompt(client, model_id, prompt)
 
         # Save the full JSON response to a file.
         output_file = config.data_path / "response.json"
         save_json(response, output_file)
 
         # Print response
-        print(response.get("choices", [])[0].get("text", ""))
-    except Exception as e:  # noqa: BLE001
-        print(f"Error for model {model_id}: {e}")
+        response_text = response.get("choices", [])[0].get("text", "")
+        logger.info("model response", response_text=response_text)
+
+    except Exception as e:
+        logger.exception("error", model_id=model_id, error=e)
 
 
 def main() -> None:

tests/credits.py

@@ -1,11 +1,15 @@
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.router.client import OpenRouterClient
 
+logger = structlog.get_logger(__name__)
+
 
 def get_credits(client: OpenRouterClient) -> None:
     # Retrieve available credits
     current_credits = client.get_credits()
-    print(current_credits)
+    logger.info("current credits", current_credits=current_credits)
 
 
 if __name__ == "__main__":

tests/model_endpoints.py

@@ -1,11 +1,15 @@
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.router.client import OpenRouterClient
 from flare_ai_consensus.utils.openrouter import extract_author
 
+logger = structlog.get_logger(__name__)
+
 
 def get_model_endpoints(client: OpenRouterClient, author: str, slug: str) -> None:
     endpoints = client.get_model_endpoints(author, slug)
-    print(endpoints)
+    logger.info("model endpoints", endpoints=endpoints)
 
 
 if __name__ == "__main__":

tests/working_models.py

@@ -1,11 +1,15 @@
 import asyncio
 
+import structlog
+
 from flare_ai_consensus.config import config
 from flare_ai_consensus.consensus.config import ModelConfig
 from flare_ai_consensus.router.client import AsyncOpenRouterClient
 from flare_ai_consensus.utils.loader import load_json
 from flare_ai_consensus.utils.saver import save_json
 
+logger = structlog.get_logger(__name__)
+
 
 async def _test_model_completion(
     client: AsyncOpenRouterClient,
@@ -56,15 +60,17 @@ async def _test_model_completion(
     try:
         response = await send_func(payload)
         if "error" not in response:
-            print(f"Model {model_id} works with {api_endpoint}!")
+            logger.info("model works", model_id=model_id, api_endpoint=api_endpoint)
             return (model, True)
         error_info = response.get("error", {})
-        print(
-            f"Model {model_id} returned error in {api_endpoint}: "
-            f"{error_info.get('message', 'Unknown error')}"
+        logger.error(
+            "testing model",
+            model_id=model_id,
+            api_endpoint=api_endpoint,
+            error=error_info.get("message", "Unknown error"),
         )
-    except Exception as e:  # noqa: BLE001
-        print(f"Error testing model {model_id} with {api_endpoint}: {e}")
+    except Exception:
+        logger.exception("testing model", model_id=model_id, api_endpoint=api_endpoint)
         return (model, False)
     else:
         return (model, False)
@@ -121,7 +127,11 @@ async def main() -> None:
             config.data_path / f"free_working_{endpoint}_models.json"
         )
         save_json({"data": working_models}, completion_output_file)
-        print(f"\nWorking {endpoint} models saved to {completion_output_file}.\n")
+        logger.info(
+            "working models saved",
+            endpoint=endpoint,
+            completion_output_file=completion_output_file,
+        )
 
     await client.close()