sdv 1.33.1.dev0__tar.gz → 1.33.2.dev0__tar.gz

{sdv-1.33.1.dev0/sdv.egg-info → sdv-1.33.2.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sdv
-Version: 1.33.1.dev0
+Version: 1.33.2.dev0
 Summary: Generate synthetic data for single table, multi table and sequential data
 Author-email: "DataCebo, Inc." <info@sdv.dev>
 License-Expression: BUSL-1.1
@@ -9,7 +9,7 @@ Project-URL: Issue Tracker, https://github.com/sdv-dev/SDV/issues
 Project-URL: Changes, https://github.com/sdv-dev/SDV/blob/main/HISTORY.md
 Project-URL: Twitter, https://twitter.com/sdv_dev
 Project-URL: Chat, https://forum.datacebo.com
-Keywords: sdv,synthetic-data,synhtetic-data-generation,timeseries,single-table,multi-table
+Keywords: sdv,synthetic-data,synthetic-data-generation,timeseries,single-table,multi-table
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Natural Language :: English

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/pyproject.toml

@@ -15,7 +15,7 @@ classifiers = [
     'Programming Language :: Python :: 3.14',
     'Topic :: Scientific/Engineering :: Artificial Intelligence',
 ]
-keywords = ['sdv', 'synthetic-data', 'synhtetic-data-generation', 'timeseries', 'single-table', 'multi-table']
+keywords = ['sdv', 'synthetic-data', 'synthetic-data-generation', 'timeseries', 'single-table', 'multi-table']
 dynamic = ['version']
 license = 'BUSL-1.1'
 license-files = ['LICENSE']
@@ -150,7 +150,7 @@ namespaces = false
 version = {attr = 'sdv.__version__'}
 
 [tool.bumpversion]
-current_version = "1.33.1.dev0"
+current_version = "1.33.2.dev0"
 parse = '(?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(\.(?P<release>[a-z]+)(?P<candidate>\d+))?'
 serialize = [
     '{major}.{minor}.{patch}.{release}{candidate}',

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/sdv/__init__.py

@@ -6,7 +6,7 @@
 
 __author__ = 'DataCebo, Inc.'
 __email__ = 'info@sdv.dev'
-__version__ = '1.33.1.dev0'
+__version__ = '1.33.2.dev0'
 
 
 import sys

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/sdv/metadata/multi_table.py

@@ -160,16 +160,6 @@ class MultiTableMetadata:
                 f'tables {errors}.'
             )
 
-    def _validate_foreign_child_key(self, child_table_name, parent_table_name, child_foreign_key):
-        child_primary_key = _cast_to_iterable(self.tables[child_table_name].primary_key)
-        child_foreign_key = _cast_to_iterable(child_foreign_key)
-        if set(child_foreign_key).intersection(set(child_primary_key)):
-            raise InvalidMetadataError(
-                f"Invalid relationship between table '{parent_table_name}' and table "
-                f"'{child_table_name}'. A relationship must connect a primary key "
-                'with a non-primary key.'
-            )
-
     def _validate_new_foreign_key_is_not_reused(
         self, parent_table_name, parent_primary_key, child_table_name, child_foreign_key
     ):
@@ -238,8 +228,6 @@ class MultiTableMetadata:
             parent_table_name, parent_primary_key, child_table_name, child_foreign_key
         )
 
-        self._validate_foreign_child_key(child_table_name, parent_table_name, child_foreign_key)
-
         self._validate_relationship_sdtypes(
             parent_table_name, parent_primary_key, child_table_name, child_foreign_key
         )
@@ -312,7 +300,6 @@ class MultiTableMetadata:
               different
               ``sdtype``.
             - ``InvalidMetadataError`` if the relationship causes a circular dependency.
-            - ``InvalidMetadataError`` if ``child_foreign_key`` is a primary key.
         """
         self._validate_relationship(
             parent_table_name, child_table_name, parent_primary_key, child_foreign_key

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/sdv/multi_table/base.py

@@ -466,10 +466,18 @@ class BaseMultiTableSynthesizer:
             )
 
     def _assign_table_transformers(self, synthesizer, table_name, table_data):
-        """Update the ``synthesizer`` to ignore the foreign keys while preprocessing the data."""
+        """Update the ``synthesizer`` to ignore the foreign keys while preprocessing the data.
+
+        This function iterates through all foreign keys in the table. For each foreign key,
+        if it is not the primary key in the table, then the transformer is set to None (
+        meaning no transformer is assigned).
+        """
         synthesizer.auto_assign_transformers(table_data)
+        primary_key = self.metadata.tables[table_name].primary_key
         foreign_key_columns = self.metadata._get_all_foreign_keys(table_name)
-        column_name_to_transformers = {column_name: None for column_name in foreign_key_columns}
+        column_name_to_transformers = {
+            column_name: None for column_name in foreign_key_columns if column_name != primary_key
+        }
         synthesizer.update_transformers(column_name_to_transformers)
 
     def auto_assign_transformers(self, data):

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/sdv/multi_table/hma.py

@@ -348,9 +348,15 @@ class HMASynthesizer(BaseHierarchicalSampler, BaseMultiTableSynthesizer):
         table_meta = self._table_synthesizers[child_name].get_metadata()
 
         extension_rows = []
+        primary_key = self.metadata.tables[child_name].primary_key
         foreign_key_columns = self.metadata._get_all_foreign_keys(child_name)
-        foreign_key_values = child_table[foreign_key].unique()
-        child_table = child_table.set_index(foreign_key)
+        primary_key_is_a_foreign_key = primary_key and primary_key in foreign_key_columns
+        if primary_key_is_a_foreign_key and foreign_key == primary_key:
+            # data processor will set index of each table to the PK for table
+            foreign_key_values = child_table.index.unique()
+        else:
+            foreign_key_values = child_table[foreign_key].unique()
+            child_table = child_table.set_index(foreign_key)
 
         index = []
         scale_columns = None
@@ -447,6 +453,7 @@ class HMASynthesizer(BaseHierarchicalSampler, BaseMultiTableSynthesizer):
                 child_table = tables[child_name]
 
             foreign_keys = self.metadata._get_foreign_keys(table_name, child_name)
+
             for foreign_key in foreign_keys:
                 progress_bar_desc = (
                     f'({self._learned_relationships + 1}/{len(self.metadata.relationships)}) '
@@ -497,6 +504,7 @@ class HMASynthesizer(BaseHierarchicalSampler, BaseMultiTableSynthesizer):
         self._learned_relationships = 0
         parent_map = self.metadata._get_parent_map()
         self._print(text='Learning relationships:')
+
         for table_name in processed_data:
             if not parent_map.get(table_name):
                 self._augment_table(augmented_data[table_name], augmented_data, table_name)
@@ -517,7 +525,11 @@ class HMASynthesizer(BaseHierarchicalSampler, BaseMultiTableSynthesizer):
             keys (dict):
                 A dictionary mapping with the foreign key and it's values within the table.
         """
+        primary_key = self.metadata.tables[table_name].primary_key
         foreign_keys = self.metadata._get_all_foreign_keys(table_name)
+        if primary_key and primary_key in foreign_keys:
+            foreign_keys.remove(primary_key)
+
         keys = {}
         for fk in foreign_keys:
             keys[fk] = table_data.pop(fk).to_numpy()
@@ -767,8 +779,44 @@ class HMASynthesizer(BaseHierarchicalSampler, BaseMultiTableSynthesizer):
         return likelihoods.apply(self._find_parent_id, axis=1, num_rows=num_rows)
 
     def _add_foreign_key_columns(self, child_table, parent_table, child_name, parent_name):
+        """Add foreign key columns in the child table.
+
+        This function adds foreign key columns to a child table.
+        If the foreign key column does not exist in the child table, it adds the column.
+        If the foreign key column already exists in the child table (e.g., when it is also a PK)
+        and it contains invalid references (FKs not found in parent table), it overwrites the
+        foreign key values (from the parent table's PK).
+
+        Args:
+            child_table (pd.DataFrame):
+                The child table which may or may not contain the FK columns.
+            parent_table (pd.DataFrame):
+                The parent table which contains the primary key column.
+            child_name (str):
+                The name of the child table in the metadata.
+            parent_name (str):
+                The name of the parent table in the metadata.
+
+        Returns:
+            None: The child_table is modified in-place.
+        """
+        parent_primary_key = self.metadata.tables[parent_name].primary_key
+        parent_id_values = None
         for foreign_key in self.metadata._get_foreign_keys(parent_name, child_name):
-            if foreign_key not in child_table:
+            needs_assignment = foreign_key not in child_table
+
+            if not needs_assignment:
+                child_column = child_table[foreign_key].dropna()
+                if child_column.empty:
+                    needs_assignment = True
+                else:
+                    if parent_id_values is None:
+                        parent_id_values = parent_table[parent_primary_key].dropna().unique()
+
+                    if not child_column.isin(parent_id_values).all():
+                        needs_assignment = True
+
+            if needs_assignment:
                 parent_ids = self._find_parent_ids(
                     child_table=child_table,
                     parent_table=parent_table,

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0}/sdv/sampling/hierarchical_sampler.py

@@ -102,9 +102,8 @@ class BaseHierarchicalSampler:
         if len(sampled_rows):
             parent_key = self.metadata.tables[parent_name].primary_key
             if foreign_key in sampled_rows:
-                # If foreign key is in sampeld rows raises `SettingWithCopyWarning`
                 row_indices = sampled_rows.index
-                sampled_rows[foreign_key].iloc[row_indices] = parent_row[parent_key]
+                sampled_rows.loc[row_indices, foreign_key] = parent_row[parent_key]
             else:
                 sampled_rows[foreign_key] = (
                     parent_row[parent_key] if parent_row is not None else np.nan

{sdv-1.33.1.dev0 → sdv-1.33.2.dev0/sdv.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sdv
-Version: 1.33.1.dev0
+Version: 1.33.2.dev0
 Summary: Generate synthetic data for single table, multi table and sequential data
 Author-email: "DataCebo, Inc." <info@sdv.dev>
 License-Expression: BUSL-1.1
@@ -9,7 +9,7 @@ Project-URL: Issue Tracker, https://github.com/sdv-dev/SDV/issues
 Project-URL: Changes, https://github.com/sdv-dev/SDV/blob/main/HISTORY.md
 Project-URL: Twitter, https://twitter.com/sdv_dev
 Project-URL: Chat, https://forum.datacebo.com
-Keywords: sdv,synthetic-data,synhtetic-data-generation,timeseries,single-table,multi-table
+Keywords: sdv,synthetic-data,synthetic-data-generation,timeseries,single-table,multi-table
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Natural Language :: English