Fix multiple warnings

Author: fealho

sdv/metadata/single_table.py

@@ -1296,6 +1296,7 @@ def validate_data(self, data, sdtype_warnings=None):
             A warning is being raised if ``datetime_format`` is missing from a column represented
             as ``object`` in the dataframe and its sdtype is ``datetime``.
         """
+        _datetime_format_warning_flag = sdtype_warnings is not None
         sdtype_warnings = sdtype_warnings if sdtype_warnings is not None else defaultdict(list)
         if not isinstance(data, pd.DataFrame):
             raise ValueError(f'Data must be a DataFrame, not a {type(data)}.')
@@ -1315,7 +1316,7 @@ def validate_data(self, data, sdtype_warnings=None):
             errors += self._validate_column_data(data[column], sdtype_warnings)
 
         errors += self._validate_primary_key(data)
-        if sdtype_warnings is not None and len(sdtype_warnings):
+        if (not _datetime_format_warning_flag) and len(sdtype_warnings):
             df = pd.DataFrame(sdtype_warnings)
             message = (
                 "No 'datetime_format' is present in the metadata for the following columns:\n"

sdv/multi_table/base.py

@@ -90,6 +90,9 @@ def _initialize_models(self):
                 self._table_synthesizers[table_name] = self._synthesizer(
                     metadata=metadata, **synthesizer_parameters
                 )
+                # Mark synthesizer as embedded in a multi-table setting
+                # so it can suppres datetime_format warnings that are aggregated here
+                self._table_synthesizers[table_name]._suppress_datetime_format_warning = True
                 self._table_synthesizers[table_name]._data_processor.table_name = table_name
 
     def _get_pbar_args(self, **kwargs):

sdv/single_table/base.py

@@ -561,7 +561,18 @@ def validate(self, data):
             data (pandas.DataFrame):
                 The data to validate.
         """
-        self._original_metadata.validate_data({self._table_name: data})
+        # Suppress duplicate datetime_format warning only when this single-table synthesizer
+        # is embedded inside a multi-table synthesizer
+        if getattr(self, '_suppress_datetime_format_warning', False):
+            with warnings.catch_warnings():
+                warnings.filterwarnings(
+                    'ignore',
+                    message=r"No 'datetime_format' is present.*",
+                    category=UserWarning,
+                )
+                self._original_metadata.validate_data({self._table_name: data})
+        else:
+            self._original_metadata.validate_data({self._table_name: data})
         self._validate_transform_constraints(data, enforce_constraint_fitting=True)
 
         # Retaining the logic of returning errors and raising them here to maintain consistency

sdv/utils/utils.py

@@ -2,6 +2,7 @@
 
 import datetime
 import sys
+import warnings
 from copy import deepcopy
 
 import cloudpickle
@@ -55,7 +56,14 @@ def drop_unknown_references(data, metadata, drop_missing_values=False, verbose=T
     })
     metadata.validate()
     try:
-        metadata.validate_data(data)
+        # Suppress duplicate datetime_format warnings during referential integrity validation.
+        with warnings.catch_warnings():
+            warnings.filterwarnings(
+                'ignore',
+                message=r"No 'datetime_format' is present.*",
+                category=UserWarning,
+            )
+            metadata.validate_data(data)
         if drop_missing_values:
             _validate_foreign_keys_not_null(metadata, data)