Add usage logging (#1920)

Makefile

@@ -123,12 +123,8 @@ test-integration: ## run tests quickly with the default Python
 test-readme: ## run the readme snippets
 	invoke readme
 
-.PHONY: test-tutorials
-test-tutorials: ## run the tutorial notebooks
-	invoke tutorials
-
 .PHONY: test
-test: test-unit test-integration test-readme test-tutorials ## test everything that needs test dependencies
+test: test-unit test-integration test-readme ## test everything that needs test dependencies
 
 .PHONY: test-all
 test-all: ## run tests on every Python version with tox
@@ -265,5 +261,5 @@ release-major: check-release bumpversion-major release
 
 .PHONY: check-deps
 check-deps:
-	$(eval allow_list='cloudpickle=|graphviz=|numpy=|pandas=|tqdm=|copulas=|ctgan=|deepecho=|rdt=|sdmetrics=')
+	$(eval allow_list='cloudpickle=|graphviz=|numpy=|pandas=|tqdm=|copulas=|ctgan=|deepecho=|rdt=|sdmetrics=|platformdirs=')
 	pip freeze | grep -v "SDV.git" | grep -E $(allow_list) | sort > $(OUTPUT_FILEPATH)

latest_requirements.txt

@@ -8,3 +8,4 @@ pandas==2.2.2
 rdt==1.12.0
 sdmetrics==0.14.0
 tqdm==4.66.2
+platformdirs==4.2.0

pyproject.toml

@@ -38,6 +38,7 @@ dependencies = [
     'deepecho>=0.6.0',
     'rdt>=1.12.0',
     'sdmetrics>=0.14.0',
+    'platformdirs>=4.0'
 ]
 
 [project.urls]

sdv/__init__.py

@@ -16,8 +16,8 @@
 from types import ModuleType
 
 from sdv import (
-    constraints, data_processing, datasets, evaluation, io, lite, metadata, metrics, multi_table,
-    sampling, sequential, single_table, version)
+    constraints, data_processing, datasets, evaluation, io, lite, logging, metadata, metrics,
+    multi_table, sampling, sequential, single_table, version)
 
 __all__ = [
     'constraints',
@@ -26,6 +26,7 @@
     'evaluation',
     'io',
     'lite',
+    'logging',
     'metadata',
     'metrics',
     'multi_table',

sdv/logging/__init__.py

@@ -0,0 +1,9 @@
+"""Module for configuring loggers within the SDV library."""
+
+from sdv.logging.utils import disable_single_table_logger, get_sdv_logger, get_sdv_logger_config
+
+__all__ = (
+    'disable_single_table_logger',
+    'get_sdv_logger',
+    'get_sdv_logger_config',
+)

sdv/logging/sdv_logger_config.yml

@@ -0,0 +1,27 @@
+log_registry: 'local'
+version: 1
+loggers:
+  SingleTableSynthesizer:
+    level: INFO
+    propagate: false
+    handlers:
+      class: logging.FileHandler
+      filename: sdv_logs.log
+  MultiTableSynthesizer:
+    level: INFO
+    propagate: false
+    handlers:
+      class: logging.FileHandler
+      filename: sdv_logs.log
+    MultiTableMetadata:
+      level: INFO
+      propagate: false
+      handlers:
+        class: logging.FileHandler
+        filename: sdv_logs.log
+    SingleTableMetadata:
+      level: INFO
+      propagate: false
+      handlers:
+        class: logging.FileHandler
+        filename: sdv_logs.log

sdv/logging/utils.py

@@ -0,0 +1,98 @@
+"""Utilities for configuring logging within the SDV library."""
+
+import contextlib
+import logging
+from functools import lru_cache
+from pathlib import Path
+
+import platformdirs
+import yaml
+
+
+def get_sdv_logger_config():
+    """Return a dictionary with the logging configuration."""
+    logging_path = Path(__file__).parent
+    with open(logging_path / 'sdv_logger_config.yml', 'r') as f:
+        logger_conf = yaml.safe_load(f)
+
+    # Logfile to be in this same directory
+    store_path = Path(platformdirs.user_data_dir('sdv', 'sdv-dev'))
+    store_path.mkdir(parents=True, exist_ok=True)
+    for logger in logger_conf.get('loggers', {}).values():
+        handler = logger.get('handlers', {})
+        if handler.get('filename') == 'sdv_logs.log':
+            handler['filename'] = store_path / handler['filename']
+
+    return logger_conf
+
+
+@contextlib.contextmanager
+def disable_single_table_logger():
+    """Temporarily disables logging for the single table synthesizers.
+
+    This context manager temporarily removes all handlers associated with
+    the ``SingleTableSynthesizer`` logger, disabling logging for that module
+    within the current context. After the context exits, the
+    removed handlers are restored to the logger.
+    """
+    # Logging without ``SingleTableSynthesizer``
+    single_table_logger = logging.getLogger('SingleTableSynthesizer')
+    handlers = single_table_logger.handlers
+    single_table_logger.handlers = []
+    try:
+        yield
+    finally:
+        for handler in handlers:
+            single_table_logger.addHandler(handler)
+
+
+@lru_cache()
+def get_sdv_logger(logger_name):
+    """Get a logger instance with the specified name and configuration.
+
+    This function retrieves or creates a logger instance with the specified name
+    and applies configuration settings based on the logger's name and the logging
+    configuration.
+
+    Args:
+        logger_name (str):
+            The name of the logger to retrieve or create.
+
+    Returns:
+        logging.Logger:
+            A logger instance configured according to the logging configuration
+            and the specific settings for the given logger name.
+    """
+    logger_conf = get_sdv_logger_config()
+    if logger_conf.get('log_registry') is None:
+        # Return a logger without any extra settings and avoid writing into files or other streams
+        return logging.getLogger(logger_name)
+
+    if logger_conf.get('log_registry') == 'local':
+        logger = logging.getLogger(logger_name)
+        if logger_name in logger_conf.get('loggers'):
+            formatter = None
+            config = logger_conf.get('loggers').get(logger_name)
+            log_level = getattr(logging, config.get('level', 'INFO'))
+            if config.get('format'):
+                formatter = logging.Formatter(config.get('format'))
+
+            logger.setLevel(log_level)
+            logger.propagate = config.get('propagate', False)
+            handler = config.get('handlers')
+            handlers = handler.get('class')
+            handlers = [handlers] if isinstance(handlers, str) else handlers
+            for handler_class in handlers:
+                if handler_class == 'logging.FileHandler':
+                    logfile = handler.get('filename')
+                    file_handler = logging.FileHandler(logfile)
+                    file_handler.setLevel(log_level)
+                    file_handler.setFormatter(formatter)
+                    logger.addHandler(file_handler)
+                elif handler_class in ('logging.consoleHandler', 'logging.StreamHandler'):
+                    ch = logging.StreamHandler()
+                    ch.setLevel(log_level)
+                    ch.setFormatter(formatter)
+                    logger.addHandler(ch)
+
+        return logger

sdv/metadata/multi_table.py

@@ -1,5 +1,6 @@
 """Multi Table Metadata."""
 
+import datetime
 import json
 import logging
 import warnings
@@ -11,6 +12,7 @@
 
 from sdv._utils import _cast_to_iterable, _load_data_from_csv
 from sdv.errors import InvalidDataError
+from sdv.logging import get_sdv_logger
 from sdv.metadata.errors import InvalidMetadataError
 from sdv.metadata.metadata_upgrader import convert_metadata
 from sdv.metadata.single_table import SingleTableMetadata
@@ -19,6 +21,7 @@
     create_columns_node, create_summarized_columns_node, visualize_graph)
 
 LOGGER = logging.getLogger(__name__)
+MULTITABLEMETADATA_LOGGER = get_sdv_logger('MultiTableMetadata')
 WARNINGS_COLUMN_ORDER = ['Table Name', 'Column Name', 'sdtype', 'datetime_format']
 
 
@@ -1054,6 +1057,22 @@ def save_to_json(self, filepath):
         """
         validate_file_does_not_exist(filepath)
         metadata = self.to_dict()
+        total_columns = 0
+        for table in self.tables.values():
+            total_columns += len(table.columns)
+
+        MULTITABLEMETADATA_LOGGER.info(
+            '\nMetadata Save:\n'
+            '  Timestamp: %s\n'
+            '  Statistics about the metadata:\n'
+            '    Total number of tables: %s\n'
+            '    Total number of columns: %s\n'
+            '    Total number of relationships: %s',
+            datetime.datetime.now(),
+            len(self.tables),
+            total_columns,
+            len(self.relationships)
+        )
         with open(filepath, 'w', encoding='utf-8') as metadata_file:
             json.dump(metadata, metadata_file, indent=4)
 

sdv/metadata/single_table.py

@@ -16,13 +16,15 @@
     _cast_to_iterable, _format_invalid_values_string, _get_datetime_format, _is_boolean_type,
     _is_datetime_type, _is_numerical_type, _load_data_from_csv, _validate_datetime_format)
 from sdv.errors import InvalidDataError
+from sdv.logging import get_sdv_logger
 from sdv.metadata.errors import InvalidMetadataError
 from sdv.metadata.metadata_upgrader import convert_metadata
 from sdv.metadata.utils import read_json, validate_file_does_not_exist
 from sdv.metadata.visualization import (
     create_columns_node, create_summarized_columns_node, visualize_graph)
 
 LOGGER = logging.getLogger(__name__)
+SINGLETABLEMETADATA_LOGGER = get_sdv_logger('SingleTableMetadata')
 
 
 class SingleTableMetadata:
@@ -1206,6 +1208,16 @@ def save_to_json(self, filepath):
         validate_file_does_not_exist(filepath)
         metadata = self.to_dict()
         metadata['METADATA_SPEC_VERSION'] = self.METADATA_SPEC_VERSION
+        SINGLETABLEMETADATA_LOGGER.info(
+            '\nMetadata Save:\n'
+            '  Timestamp: %s\n'
+            '  Statistics about the metadata:\n'
+            '    Total number of tables: 1'
+            '    Total number of columns: %s'
+            '    Total number of relationships: 0',
+            datetime.now(),
+            len(self.columns)
+        )
         with open(filepath, 'w', encoding='utf-8') as metadata_file:
             json.dump(metadata, metadata_file, indent=4)