adjust logging system

.env.example

@@ -9,6 +9,7 @@ USE_EMBEDDING_CACHE=True
 LOG_LLM_CHAT_CONTENT=False
 CHAT_FREQUENCY_PENALTY=0.0
 CHAT_PRESENCE_PENALTY=0.0
+LOG_TRACE_PATH=log_traces
 
 # embedding model configs:
 EMBEDDING_OPENAI_API_KEY=your_api_key

rdagent/app/factor_extraction_and_implementation/factor_extract_and_implement.py

@@ -7,6 +7,7 @@
 from rdagent.scenarios.qlib.factor_task_implementation import (
     COSTEERFG_QUANT_FACTOR_IMPLEMENTATION,
 )
+from rdagent.log import rdagent_logger as logger
 
 assert load_dotenv()
 
@@ -20,3 +21,5 @@ def extract_factors_and_implement(report_file_path: str) -> None:
 
 if __name__ == "__main__":
     extract_factors_and_implement("/home/xuyang1/workspace/report.pdf")
+
+# %%

rdagent/components/coder/factor_coder/CoSTEER/evolvable_subjects.py

@@ -4,7 +4,7 @@
     FileBasedFactorImplementation,
 )
 from rdagent.core.evolving_framework import EvolvableSubjects
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 
 
 class FactorEvolvingItem(FactorExperiment, EvolvableSubjects):
@@ -23,7 +23,7 @@ def __init__(
             sub_gt_implementations,
         ) != len(self.sub_tasks):
             self.sub_gt_implementations = None
-            RDAgentLog().warning(
+            logger.warning(
                 "The length of sub_gt_implementations is not equal to the length of sub_tasks, set sub_gt_implementations to None",
             )
         else:

rdagent/components/coder/factor_coder/CoSTEER/knowledge_management.py

@@ -28,7 +28,7 @@
     RAGStrategy,
 )
 from rdagent.core.experiment import Implementation
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 from rdagent.core.prompts import Prompts
 from rdagent.oai.llm_utils import (
     APIBackend,
@@ -336,7 +336,7 @@ def analyze_component(
             )["component_no_list"]
             return [all_component_nodes[index - 1] for index in sorted(list(set(component_no_list)))]
         except:
-            RDAgentLog().warning("Error when analyzing components.")
+            logger.warning("Error when analyzing components.")
             analyze_component_user_prompt = "Your response is not a valid component index list."
 
         return []
@@ -714,7 +714,7 @@ def __init__(self, init_component_list=None) -> None:
         Load knowledge, offer brief information of knowledge and common handle interfaces
         """
         self.graph: UndirectedGraph = UndirectedGraph.load(Path.cwd() / "graph.pkl")
-        RDAgentLog().info(f"Knowledge Graph loaded, size={self.graph.size()}")
+        logger.info(f"Knowledge Graph loaded, size={self.graph.size()}")
 
         if init_component_list:
             for component in init_component_list:

rdagent/components/coder/factor_coder/CoSTEER/scheduler.py

@@ -9,7 +9,7 @@
 )
 from rdagent.components.coder.factor_coder.utils import get_data_folder_intro
 from rdagent.core.conf import RD_AGENT_SETTINGS
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 from rdagent.core.prompts import Prompts
 from rdagent.core.scenario import Scenario
 from rdagent.oai.llm_utils import APIBackend
@@ -25,7 +25,7 @@ def RandomSelect(to_be_finished_task_index, implementation_factors_per_round):
         implementation_factors_per_round,
     )
 
-    RDAgentLog().info(f"The random selection is: {to_be_finished_task_index}")
+    logger.info(f"The random selection is: {to_be_finished_task_index}")
     return to_be_finished_task_index
 
 

rdagent/components/coder/factor_coder/factor.py

@@ -16,7 +16,7 @@
     RuntimeErrorException,
 )
 from rdagent.core.experiment import Experiment, FBImplementation, Task
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 from rdagent.oai.llm_utils import md5_hash
 
 
@@ -74,7 +74,7 @@ def __init__(
     ) -> None:
         super().__init__(*args, **kwargs)
         self.executed_factor_value_dataframe = executed_factor_value_dataframe
-        self.logger = RDAgentLog()
+        self.logger = logger
         self.raise_exception = raise_exception
 
     @staticmethod

rdagent/components/coder/model_coder/task_loader.py

@@ -9,13 +9,12 @@
     load_and_process_pdfs_by_langchain,
 )
 from rdagent.components.loader.task_loader import ModelTaskLoader
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 from rdagent.core.prompts import Prompts
 from rdagent.oai.llm_utils import APIBackend
 
 document_process_prompts = Prompts(file_path=Path(__file__).parent / "prompts.yaml")
 
-
 def extract_model_from_doc(doc_content: str) -> dict:
     """
     Extract model information from document content.
@@ -62,7 +61,7 @@ def extract_model_from_doc(doc_content: str) -> dict:
             else:
                 break
 
-    RDAgentLog().info(f"已经完成{len(model_dict)}个模型的提取")
+    logger.info(f"已经完成{len(model_dict)}个模型的提取")
 
     return model_dict
 

rdagent/components/knowledge_management/vector_base.py

@@ -5,10 +5,9 @@
 import pandas as pd
 from scipy.spatial.distance import cosine
 
-from rdagent.core.log import RDAgentLog
+from rdagent.log import rdagent_logger as logger
 from rdagent.oai.llm_utils import APIBackend
 
-
 class KnowledgeMetaData:
     def __init__(self, content: str = "", label: str = None, embedding=None, identity=None):
         self.label = label
@@ -127,7 +126,7 @@ def __init__(self, vector_df_path: Union[str, Path] = None):
         else:
             self.vector_df = pd.DataFrame(columns=["id", "label", "content", "embedding"])
 
-        RDAgentLog().info(f"VectorBase loaded, shape={self.vector_df.shape}")
+        logger.info(f"VectorBase loaded, shape={self.vector_df.shape}")
 
     def shape(self):
         return self.vector_df.shape
@@ -205,4 +204,4 @@ def load(self, vector_df_path, **kwargs):
 
     def save(self, vector_df_path, **kwargs):
         self.vector_df.to_pickle(vector_df_path)
-        RDAgentLog().info(f"Save vectorBase vector_df to: {vector_df_path}")
+        logger.info(f"Save vectorBase vector_df to: {vector_df_path}")

rdagent/core/conf.py

@@ -16,6 +16,11 @@
 
 class RDAgentSettings(BaseSettings):
     # TODO: (xiao) I think most of the config should be in oai.config
+    # Log configs
+    # TODO: (xiao) think it can be a seperate config.
+    log_trace_path: str | None = None
+    log_llm_chat_content: bool = True
+
     use_azure: bool = True
     use_azure_token_provider: bool = False
     managed_identity_client_id: str | None = None
@@ -28,7 +33,6 @@ class RDAgentSettings(BaseSettings):
     prompt_cache_path: str = str(Path.cwd() / "prompt_cache.db")
     session_cache_folder_location: str = str(Path.cwd() / "session_cache_folder/")
     max_past_message_include: int = 10
-    log_llm_chat_content: bool = True
 
     # Chat configs
     chat_openai_api_key: str = ""

rdagent/core/utils.py

@@ -19,7 +19,10 @@ class RDAgentException(Exception):  # noqa: N818
 
 
 class SingletonMeta(type):
-    _instance_dict: ClassVar[dict] = {}
+    # _instance_dict: ClassVar[dict] = {}
+    def __init__(cls, *args, **kwargs):
+        cls._instance_dict = {}  # 为每个类创建一个唯一的_instance_dict字典
+        super().__init__(*args, **kwargs)
 
     def __call__(cls, *args: Any, **kwargs: Any) -> Any:
         # Since it's hard to align the difference call using args and kwargs, we strictly ask to use kwargs in Singleton

rdagent/log/__init__.py

@@ -0,0 +1,4 @@
+from .logger import RDAgentLog
+from .utils import LogColors
+
+rdagent_logger: RDAgentLog = RDAgentLog()

rdagent/log/base.py

@@ -0,0 +1,62 @@
+from __future__ import annotations
+
+from abc import abstractmethod
+from pathlib import Path
+
+
+class Storage:
+    """
+    Basic storage to support saving objects;
+
+    # Usage:
+
+    The storage has mainly two kind of users:
+    - The logging end: you can choose any of the following method to use the object
+        - We can use it directly with the native logging storage
+        - We can use it with other logging tools; For example, serve as a handler for loggers
+    - The view end:
+        - Mainly for the subclass of `logging.base.View`
+        - It should provide two kind of ways to provide content
+            - offline content provision.
+            - online content preovision.
+    """
+
+    @abstractmethod
+    def log(self, obj: object, name: str = "", **kwargs: dict) -> str | Path:
+        """
+
+        Parameters
+        ----------
+        obj : object
+            The object for logging.
+        name : str
+            The name of the object.  For example "a.b.c"
+            We may log a lot of objects to a same name
+
+        Returns
+        -------
+        str | Path
+            The storage identifier of the object.
+        """
+        ...
+
+
+class View:
+    """
+    Motivation:
+
+    Display the content in the storage
+    """
+    # TODO: pleas fix me
+    @abstractmethod
+    def display(s: Storage, watch: bool=False):
+        """
+
+        Parameters
+        ----------
+        s : Storage
+
+        watch : bool
+            should we watch the new content and display them
+        """
+        ...