Implementation:Sdv dev SDV Logging Utils

Knowledge Sources	Sdv_dev_SDV
Domains	Logging, Configuration
Last Updated	2026-02-14 19:00 GMT

Overview

Concrete tool for loading SDV logging configuration and managing logger lifecycle provided by the SDV library.

Description

The sdv.logging.utils module provides three public functions: get_sdv_logger_config reads the SDV logger configuration from a YAML file (sdv_logger_config.yml) stored in the user data directory or the package directory, handling permission and OS errors gracefully. disable_single_table_logger is a context manager that temporarily removes all handlers from the SingleTableSynthesizer logger (used by multi-table synthesizers to suppress duplicate logging during inner single-table fits). load_logfile_dataframe reads an SDV CSV log file into a pandas DataFrame with the correct column headers.

Usage

Use get_sdv_logger_config when initializing loggers (called internally by get_sdv_logger). Use disable_single_table_logger when running multi-table fits that internally fit single-table models. Use load_logfile_dataframe to analyze SDV log output programmatically.

Code Reference

Source Location

• Repository: Sdv_dev_SDV
• File: sdv/logging/utils.py
• Lines: 1-98

Signature

def get_sdv_logger_config() -> dict:
    """Return a dictionary with the logging configuration."""
    ...

@contextlib.contextmanager
def disable_single_table_logger():
    """Temporarily disables logging for the single table synthesizers."""
    ...

def load_logfile_dataframe(logfile: str) -> pd.DataFrame:
    """Load the SDV logfile as a pandas DataFrame with correct column headers."""
    ...

Import

from sdv.logging.utils import (
    get_sdv_logger_config,
    disable_single_table_logger,
    load_logfile_dataframe,
)

I/O Contract

Inputs (get_sdv_logger_config)

Name	Type	Required	Description
(none)	—	—	Reads from sdv_logger_config.yml in user data dir or package dir

Outputs (get_sdv_logger_config)

Name	Type	Description
logger_conf	dict	Logger configuration dictionary parsed from YAML

Inputs (load_logfile_dataframe)

Name	Type	Required	Description
logfile	str	Yes	Path to the SDV log CSV file

Outputs (load_logfile_dataframe)

Name	Type	Description
df	pd.DataFrame	DataFrame with columns: LEVEL, EVENT, TIMESTAMP, SYNTHESIZER CLASS NAME, SYNTHESIZER ID, TOTAL NUMBER OF TABLES, TOTAL NUMBER OF ROWS, TOTAL NUMBER OF COLUMNS

Usage Examples

Loading Logger Config

from sdv.logging.utils import get_sdv_logger_config

config = get_sdv_logger_config()
print(config.get('log_registry'))  # 'local' or None
print(config.get('loggers', {}).keys())

Disabling Single Table Logger

from sdv.logging.utils import disable_single_table_logger

# Used internally during multi-table fitting
with disable_single_table_logger():
    # Single-table logger is silenced here
    synthesizer._fit_table(table_name, table_data)

Loading Log Data

from sdv.logging.utils import load_logfile_dataframe

df = load_logfile_dataframe('sdv_logs.csv')
print(df.head())
# Shows: LEVEL, EVENT, TIMESTAMP, SYNTHESIZER CLASS NAME, ...

Related Pages

• Environment:Sdv_dev_SDV_Python_Runtime