openmetadata-ingestion 1.9.3.0__py3-none-any.whl → 1.9.4.0__py3-none-any.whl

metadata/ingestion/source/database/oracle/connection.py

@@ -14,6 +14,7 @@ Source connection handler
 """
 import os
 import sys
+from copy import deepcopy
 from typing import Optional
 from urllib.parse import quote_plus
 
@@ -26,7 +27,9 @@ from metadata.generated.schema.entity.automations.workflow import (
     Workflow as AutomationWorkflow,
 )
 from metadata.generated.schema.entity.services.connections.database.oracleConnection import (
-    OracleConnection,
+    OracleConnection as OracleConnectionConfig,
+)
+from metadata.generated.schema.entity.services.connections.database.oracleConnection import (
     OracleDatabaseSchema,
     OracleServiceName,
     OracleTNSConnection,
@@ -39,6 +42,8 @@ from metadata.ingestion.connections.builders import (
     get_connection_args_common,
     get_connection_options_dict,
 )
+from metadata.ingestion.connections.connection import BaseConnection
+from metadata.ingestion.connections.secrets import connection_with_options_secrets
 from metadata.ingestion.connections.test_connections import test_connection_db_common
 from metadata.ingestion.ometa.ometa_api import OpenMetadata
 from metadata.ingestion.source.database.oracle.queries import (
@@ -55,123 +60,174 @@ LD_LIB_ENV = "LD_LIBRARY_PATH"
 logger = ingestion_logger()
 
 
-def get_connection_url(connection: OracleConnection) -> str:
-    """
-    Build the URL and handle driver version at system level
-    """
-
-    oracledb.version = CX_ORACLE_LIB_VERSION
-    sys.modules["cx_Oracle"] = oracledb
+class OracleConnection(BaseConnection[OracleConnectionConfig, Engine]):
+    def __init__(self, connection: OracleConnectionConfig):
+        super().__init__(connection)
 
-    url = f"{connection.scheme.value}://"
-    if connection.username:
-        url += f"{quote_plus(connection.username)}"
-        if not connection.password:
-            connection.password = SecretStr("")
-        url += f":{quote_plus(connection.password.get_secret_value())}"
-        url += "@"
-
-    url = _handle_connection_type(url=url, connection=connection)
-
-    options = get_connection_options_dict(connection)
-    if options:
-        params = "&".join(
-            f"{key}={quote_plus(value)}" for (key, value) in options.items() if value
+    def _get_client(self) -> Engine:
+        """
+        Create connection
+        """
+        try:
+            if self.service_connection.instantClientDirectory:
+                logger.info(
+                    f"Initializing Oracle thick client at {self.service_connection.instantClientDirectory}"
+                )
+                os.environ[LD_LIB_ENV] = self.service_connection.instantClientDirectory
+                oracledb.init_oracle_client(
+                    lib_dir=self.service_connection.instantClientDirectory
+                )
+        except DatabaseError as err:
+            logger.info(f"Could not initialize Oracle thick client: {err}")
+
+        return create_generic_db_connection(
+            connection=self.service_connection,
+            get_connection_url_fn=self.get_connection_url,
+            get_connection_args_fn=get_connection_args_common,
         )
-        if isinstance(connection.oracleConnectionType, OracleServiceName):
-            url = f"{url}&{params}"
-        else:
-            url = f"{url}?{params}"
-
-    return url
-
-
-def _handle_connection_type(url: str, connection: OracleConnection) -> str:
-    """
-    Depending on the oracle connection type, we need to handle the URL differently
-    """
-
-    if isinstance(connection.oracleConnectionType, OracleTNSConnection):
-        # ref https://stackoverflow.com/questions/14140902/using-oracle-service-names-with-sqlalchemy
-        url += connection.oracleConnectionType.oracleTNSConnection
-        return url
-
-    # If not TNS, we add the hostPort
-    url += connection.hostPort
 
-    if isinstance(connection.oracleConnectionType, OracleDatabaseSchema):
-        url += (
-            f"/{connection.oracleConnectionType.databaseSchema}"
-            if connection.oracleConnectionType.databaseSchema
-            else ""
+    def test_connection(
+        self,
+        metadata: OpenMetadata,
+        automation_workflow: Optional[AutomationWorkflow] = None,
+        timeout_seconds: Optional[int] = THREE_MIN,
+    ) -> TestConnectionResult:
+        """
+        Test connection. This can be executed either as part
+        of a metadata workflow or during an Automation Workflow
+        """
+
+        def test_oracle_package_access(engine):
+            try:
+                schema_name = engine.execute(ORACLE_GET_SCHEMA).scalar()
+                return ORACLE_GET_STORED_PACKAGES.format(schema=schema_name)
+            except Exception as e:
+                raise OraclePackageAccessError(
+                    f"Failed to access Oracle stored packages: {e}"
+                )
+
+        test_conn_queries = {
+            "CheckAccess": CHECK_ACCESS_TO_ALL,
+            "PackageAccess": test_oracle_package_access(self.client),
+        }
+
+        return test_connection_db_common(
+            metadata=metadata,
+            engine=self.client,
+            service_connection=self.service_connection,
+            automation_workflow=automation_workflow,
+            queries=test_conn_queries,
+            timeout_seconds=timeout_seconds,
         )
-        return url
-
-    if isinstance(connection.oracleConnectionType, OracleServiceName):
-        url = f"{url}/?service_name={connection.oracleConnectionType.oracleServiceName}"
-        return url
 
-    raise ValueError(f"Unknown connection type {connection.oracleConnectionType}")
+    def get_connection_dict(self) -> dict:
+        """
+        Return the connection dictionary for this service.
+        """
+        url = self.client.url
+        connection_copy = deepcopy(self.service_connection)
+
+        connection_dict = {
+            "driver": url.drivername,
+            "host": f"{url.host}:{url.port}",  # This is the format expected by data-diff. If we start using this for something else, we need to change it and modify the data-diff code.
+            "user": url.username,
+        }
+
+        # Add password if present in the connection
+        if connection_copy.password:
+            connection_dict["password"] = connection_copy.password.get_secret_value()
+
+        # Add connection type specific information
+        if isinstance(connection_copy.oracleConnectionType, OracleDatabaseSchema):
+            connection_dict[
+                "database"
+            ] = connection_copy.oracleConnectionType.databaseSchema
+        elif isinstance(connection_copy.oracleConnectionType, OracleServiceName):
+            connection_dict[
+                "database"
+            ] = connection_copy.oracleConnectionType.oracleServiceName
+        elif isinstance(connection_copy.oracleConnectionType, OracleTNSConnection):
+            connection_dict[
+                "host"
+            ] = connection_copy.oracleConnectionType.oracleTNSConnection
+
+        # Add connection options if present
+        if connection_copy.connectionOptions and connection_copy.connectionOptions.root:
+            connection_with_options_secrets(lambda: connection_copy)
+            connection_dict.update(connection_copy.connectionOptions.root)
+
+        # Add connection arguments if present
+        if (
+            connection_copy.connectionArguments
+            and connection_copy.connectionArguments.root
+        ):
+            connection_dict.update(get_connection_args_common(connection_copy))
+
+        return connection_dict
+
+    @staticmethod
+    def get_connection_url(connection: OracleConnectionConfig) -> str:
+        """
+        Build the URL and handle driver version at system level
+        """
+
+        oracledb.version = CX_ORACLE_LIB_VERSION
+        sys.modules["cx_Oracle"] = oracledb
+
+        url = f"{connection.scheme.value}://"
+        if connection.username:
+            url += f"{quote_plus(connection.username)}"
+            if not connection.password:
+                connection.password = SecretStr("")
+            url += f":{quote_plus(connection.password.get_secret_value())}"
+            url += "@"
+
+        url = OracleConnection._handle_connection_type(url=url, connection=connection)
+
+        options = get_connection_options_dict(connection)
+        if options:
+            params = "&".join(
+                f"{key}={quote_plus(value)}"
+                for (key, value) in options.items()
+                if value
+            )
+            if isinstance(connection.oracleConnectionType, OracleServiceName):
+                url = f"{url}&{params}"
+            else:
+                url = f"{url}?{params}"
 
+        return url
 
-def get_connection(connection: OracleConnection) -> Engine:
-    """
-    Create connection
-    """
-    try:
-        if connection.instantClientDirectory:
-            logger.info(
-                f"Initializing Oracle thick client at {connection.instantClientDirectory}"
+    @staticmethod
+    def _handle_connection_type(url: str, connection: OracleConnectionConfig) -> str:
+        """
+        Depending on the oracle connection type, we need to handle the URL differently
+        """
+
+        if isinstance(connection.oracleConnectionType, OracleTNSConnection):
+            # ref https://stackoverflow.com/questions/14140902/using-oracle-service-names-with-sqlalchemy
+            url += connection.oracleConnectionType.oracleTNSConnection
+            return url
+
+        # If not TNS, we add the hostPort
+        url += connection.hostPort
+
+        if isinstance(connection.oracleConnectionType, OracleDatabaseSchema):
+            url += (
+                f"/{connection.oracleConnectionType.databaseSchema}"
+                if connection.oracleConnectionType.databaseSchema
+                else ""
             )
-            os.environ[LD_LIB_ENV] = connection.instantClientDirectory
-            oracledb.init_oracle_client()
-    except DatabaseError as err:
-        logger.info(f"Could not initialize Oracle thick client: {err}")
+            return url
+
+        if isinstance(connection.oracleConnectionType, OracleServiceName):
+            url = f"{url}/?service_name={connection.oracleConnectionType.oracleServiceName}"
+            return url
 
-    return create_generic_db_connection(
-        connection=connection,
-        get_connection_url_fn=get_connection_url,
-        get_connection_args_fn=get_connection_args_common,
-    )
+        raise ValueError(f"Unknown connection type {connection.oracleConnectionType}")
 
 
 class OraclePackageAccessError(Exception):
     """
     Raised when unable to access Oracle stored packages
     """
-
-
-def test_connection(
-    metadata: OpenMetadata,
-    engine: Engine,
-    service_connection: OracleConnection,
-    automation_workflow: Optional[AutomationWorkflow] = None,
-    timeout_seconds: Optional[int] = THREE_MIN,
-) -> TestConnectionResult:
-    """
-    Test connection. This can be executed either as part
-    of a metadata workflow or during an Automation Workflow
-    """
-
-    def test_oracle_package_access(engine):
-        try:
-            schema_name = engine.execute(ORACLE_GET_SCHEMA).scalar()
-            return ORACLE_GET_STORED_PACKAGES.format(schema=schema_name)
-        except Exception as e:
-            raise OraclePackageAccessError(
-                f"Failed to access Oracle stored packages: {e}"
-            )
-
-    test_conn_queries = {
-        "CheckAccess": CHECK_ACCESS_TO_ALL,
-        "PackageAccess": test_oracle_package_access(engine),
-    }
-
-    return test_connection_db_common(
-        metadata=metadata,
-        engine=engine,
-        service_connection=service_connection,
-        automation_workflow=automation_workflow,
-        queries=test_conn_queries,
-        timeout_seconds=timeout_seconds,
-    )

metadata/ingestion/source/database/oracle/service_spec.py

@@ -1,3 +1,4 @@
+from metadata.ingestion.source.database.oracle.connection import OracleConnection
 from metadata.ingestion.source.database.oracle.lineage import OracleLineageSource
 from metadata.ingestion.source.database.oracle.metadata import OracleSource
 from metadata.ingestion.source.database.oracle.usage import OracleUsageSource
@@ -7,4 +8,5 @@ ServiceSpec = DefaultDatabaseSpec(
     metadata_source_class=OracleSource,
     lineage_source_class=OracleLineageSource,
     usage_source_class=OracleUsageSource,
+    connection_class=OracleConnection,
 )

metadata/ingestion/source/database/saphana/cdata_parser.py

@@ -105,6 +105,7 @@ class CDATAKeys(Enum):
     BASE_MEASURES = "baseMeasures"
     MEASURE = "measure"
     MEASURE_MAPPING = "measureMapping"
+    CALCULATED_MEASURES = "calculatedMeasures"
     PRIVATE_MEASURE_GROUP = "privateMeasureGroup"
     LOGICAL_MODEL = "logicalModel"
     DATA_SOURCES = "dataSources"
@@ -136,6 +137,9 @@ class DataSourceMapping(BaseModel):
     parents: Annotated[
         List[ParentSource], Field(..., description="Parent Sources for a target col")
     ]
+    formula: Annotated[
+        Optional[str], Field(None, description="Formula used to derive the column")
+    ]
 
 
 class DataSource(BaseModel):
@@ -372,10 +376,10 @@ def _get_column_datasources(
 
 def _get_column_datasources_with_names(
     entry: ET.Element, datasource_map: Optional[DataSourceMap] = None
-) -> List[Tuple[DataSource, str]]:
+) -> List[Tuple[DataSource, str, Optional[str]]]:
     """
     Get the DataSource and the actual source column name after traversal.
-    Returns a list of tuples (DataSource, column_name).
+    Returns a list of tuples (DataSource, column_name, formula).
     """
     if (
         datasource_map
@@ -387,6 +391,7 @@ def _get_column_datasources_with_names(
             ds_origin_list=[],
             current_ds=datasource_map[entry.get(CDATAKeys.COLUMN_OBJECT_NAME.value)],
             datasource_map=datasource_map,
+            formula=None,
         )
         return ds_col_pairs
 
@@ -399,6 +404,7 @@ def _get_column_datasources_with_names(
                 source_type=ViewType.DATA_BASE_TABLE,
             ),
             entry.get(CDATAKeys.COLUMN_NAME.value),
+            None,  # No formula for direct table sources
         )
     ]
 
@@ -441,7 +447,7 @@ def _traverse_ds(
         else:
             logger.info(
                 f"Can't find mapping for column [{current_column}] in [{current_ds}]. "
-                f"We still have to implement `calculatedViewAttributes`."
+                f"This might be a constant or derived column."
             )
 
     return ds_origin_list
@@ -449,19 +455,20 @@ def _traverse_ds(
 
 def _traverse_ds_with_columns(
     current_column: str,
-    ds_origin_list: List[Tuple[DataSource, str]],
+    ds_origin_list: List[Tuple[DataSource, str, Optional[str]]],
     current_ds: DataSource,
     datasource_map: Optional[DataSourceMap],
-) -> List[Tuple[DataSource, str]]:
+    formula: Optional[str] = None,
+) -> List[Tuple[DataSource, str, Optional[str]]]:
     """
     Traverse the ds dict jumping from target -> source columns and getting the right parent.
     We keep inspecting current datasources and will append to the origin list the ones
-    that are not LOGICAL, along with the final column name.
-    Returns a list of tuples (DataSource, column_name).
+    that are not LOGICAL, along with the final column name and formula.
+    Returns a list of tuples (DataSource, column_name, formula).
     """
     if current_ds.source_type != ViewType.LOGICAL:
-        # This is a final datasource, append it with the current column name
-        ds_origin_list.append((current_ds, current_column))
+        # This is a final datasource, append it with the current column name and formula
+        ds_origin_list.append((current_ds, current_column, formula))
 
     else:
         # Based on our current column, find the parents from the mappings in the current_ds
@@ -470,6 +477,10 @@ def _traverse_ds_with_columns(
         )
 
         if current_ds_mapping:
+            # Use this layer's formula if we don't have one yet
+            if current_ds_mapping.formula and not formula:
+                formula = current_ds_mapping.formula
+
             for parent in current_ds_mapping.parents:
                 parent_ds = datasource_map.get(parent.parent)
                 if not parent_ds:
@@ -484,17 +495,44 @@ def _traverse_ds_with_columns(
                     ds_origin_list=ds_origin_list,
                     current_ds=parent_ds,
                     datasource_map=datasource_map,
+                    formula=formula,
                 )
         else:
-            # Current column not in mapping. This can happen for calculated view attributes
             logger.info(
                 f"Can't find mapping for column [{current_column}] in [{current_ds}]. "
-                f"We still have to implement `calculatedViewAttributes`."
+                f"This might be a constant or derived column."
             )
 
     return ds_origin_list
 
 
+def _get_formula_from_logical_mapping(
+    entry: Optional[ET.Element], datasource_map: Optional[DataSourceMap]
+) -> Optional[str]:
+    """Extract formula from logical datasource mapping if it exists."""
+    if not entry or not datasource_map:
+        return None
+
+    column_object_name = entry.get(CDATAKeys.COLUMN_OBJECT_NAME.value)
+    column_name = entry.get(CDATAKeys.COLUMN_NAME.value)
+
+    if not column_object_name or not column_name:
+        return None
+
+    datasource = datasource_map.get(column_object_name)
+    if not datasource:
+        return None
+
+    if datasource.source_type != ViewType.LOGICAL or not datasource.mapping:
+        return None
+
+    mapping = datasource.mapping.get(column_name)
+    if not mapping:
+        return None
+
+    return mapping.formula
+
+
 def _read_attributes(
     tree: ET.Element, ns: dict, datasource_map: Optional[DataSourceMap] = None
 ) -> ParsedLineage:
@@ -506,8 +544,9 @@ def _read_attributes(
 
     for attribute in attribute_list.findall(CDATAKeys.ATTRIBUTE.value, ns):
         key_mapping = attribute.find(CDATAKeys.KEY_MAPPING.value, ns)
+        target_name = attribute.get(CDATAKeys.ID.value)
 
-        # Get the actual source datasources and their column names
+        # Get the actual source datasources, column names, and formulas
         data_sources_with_columns = _get_column_datasources_with_names(
             entry=key_mapping, datasource_map=datasource_map
         )
@@ -517,7 +556,8 @@ def _read_attributes(
                 ColumnMapping(
                     data_source=ds_info[0],  # The datasource
                     sources=[ds_info[1]],  # The actual source column name
-                    target=attribute.get(CDATAKeys.ID.value),
+                    target=target_name,
+                    formula=ds_info[2],  # Formula from traversal (if any)
                 )
                 for ds_info in data_sources_with_columns
             ]
@@ -541,16 +581,39 @@ def _read_calculated_attributes(
         return lineage
 
     for calculated_attr in calculated_attrs.findall(key.value, ns):
-        formula = (
-            calculated_attr.find(CDATAKeys.KEY_CALCULATION.value, ns)
-            .find(CDATAKeys.FORMULA.value, ns)
-            .text
-        )
-        lineage += _explode_formula(
-            target=calculated_attr.get(CDATAKeys.ID.value),
-            formula=formula,
-            base_lineage=base_lineage,
-        )
+        key_calc = calculated_attr.find(CDATAKeys.KEY_CALCULATION.value, ns)
+        if key_calc is not None:
+            formula_elem = key_calc.find(CDATAKeys.FORMULA.value, ns)
+            if formula_elem is not None and formula_elem.text:
+                lineage += _explode_formula(
+                    target=calculated_attr.get(CDATAKeys.ID.value),
+                    formula=formula_elem.text,
+                    base_lineage=base_lineage,
+                )
+
+    return lineage
+
+
+def _read_calculated_measures(
+    tree: ET.Element,
+    ns: dict,
+    base_lineage: ParsedLineage,
+) -> ParsedLineage:
+    """Compute the lineage based on the calculated measures"""
+    lineage = ParsedLineage()
+
+    calculated_measures = tree.find(CDATAKeys.CALCULATED_MEASURES.value, ns)
+    if not calculated_measures:
+        return lineage
+
+    for measure in calculated_measures.findall(CDATAKeys.MEASURE.value, ns):
+        formula_elem = measure.find(CDATAKeys.FORMULA.value, ns)
+        if formula_elem is not None and formula_elem.text:
+            lineage += _explode_formula(
+                target=measure.get(CDATAKeys.ID.value),
+                formula=formula_elem.text,
+                base_lineage=base_lineage,
+            )
 
     return lineage
 
@@ -573,8 +636,9 @@ def _read_base_measures(
 
     for measure in base_measures.findall(CDATAKeys.MEASURE.value, ns):
         measure_mapping = measure.find(CDATAKeys.MEASURE_MAPPING.value, ns)
+        target_name = measure.get(CDATAKeys.ID.value)
 
-        # Get the actual source datasources and their column names
+        # Get the actual source datasources, column names, and formulas
         data_sources_with_columns = _get_column_datasources_with_names(
             entry=measure_mapping, datasource_map=datasource_map
         )
@@ -584,7 +648,8 @@ def _read_base_measures(
                 ColumnMapping(
                     data_source=ds_info[0],  # The datasource
                     sources=[ds_info[1]],  # The actual source column name
-                    target=measure.get(CDATAKeys.ID.value),
+                    target=target_name,
+                    formula=ds_info[2],  # Formula from traversal (if any)
                 )
                 for ds_info in data_sources_with_columns
             ]
@@ -605,10 +670,16 @@ def _explode_formula(
     Returns:
         Parsed Lineage from the formula
     """
-    column_ds = {
-        match.group(1): base_lineage.find_target(match.group(1)).data_source
-        for match in FORMULA_PATTERN.finditer(formula)
-    }
+    column_ds = {}
+    for match in FORMULA_PATTERN.finditer(formula):
+        col_name = match.group(1)
+        mapping = base_lineage.find_target(col_name)
+        if mapping:
+            column_ds[col_name] = mapping.data_source
+
+    # If no columns found in base_lineage, it might be a constant formula
+    if not column_ds:
+        return ParsedLineage()
 
     # Group every datasource (key) with a list of the involved columns (values)
     ds_columns = defaultdict(list)
@@ -694,17 +765,42 @@ def _(cdata: str) -> ParsedLineage:
     attribute_lineage = _read_attributes(
         tree=logical_model, ns=ns, datasource_map=datasource_map
     )
-    calculated_attrs_lineage = _read_calculated_attributes(
+
+    base_measure_lineage = _read_base_measures(
+        tree=logical_model, ns=ns, datasource_map=datasource_map
+    )
+
+    # Combine base attributes and measures for calculated columns
+    combined_base_lineage = attribute_lineage + base_measure_lineage
+
+    # Read calculated attributes from calculationViews (if they exist)
+    cv_calculated_attrs_lineage = _read_calculated_attributes(
         tree=tree,
         ns=ns,
-        base_lineage=attribute_lineage,
+        base_lineage=combined_base_lineage,
         key=CalculatedAttrKey.CALCULATED_VIEW_ATTRIBUTE,
     )
-    base_measure_lineage = _read_base_measures(
-        tree=logical_model, ns=ns, datasource_map=datasource_map
+
+    # Read calculated attributes from logical model
+    logical_calculated_attrs_lineage = _read_calculated_attributes(
+        tree=logical_model,
+        ns=ns,
+        base_lineage=combined_base_lineage,
+        key=CalculatedAttrKey.CALCULATED_ATTRIBUTE,
     )
 
-    return attribute_lineage + calculated_attrs_lineage + base_measure_lineage
+    # Read calculated measures from logical model
+    calculated_measures_lineage = _read_calculated_measures(
+        tree=logical_model, ns=ns, base_lineage=combined_base_lineage
+    )
+
+    return (
+        attribute_lineage
+        + cv_calculated_attrs_lineage
+        + logical_calculated_attrs_lineage
+        + base_measure_lineage
+        + calculated_measures_lineage
+    )
 
 
 def _parse_cv_data_sources(tree: ET.Element, ns: dict) -> DataSourceMap:
@@ -812,11 +908,13 @@ def _build_mappings(calculation_view: ET.Element, ns: dict) -> List[DataSourceMa
     """
 
     input_mappings = _build_input_mappings(calculation_view=calculation_view, ns=ns)
-    # calculated_view_attrs = _build_cv_attributes(
-    #     calculation_view=calculation_view, ns=ns, input_mappings=input_mappings
-    # )
+    calculated_view_attrs = _build_cv_attributes(
+        calculation_view=calculation_view, ns=ns, input_mappings=input_mappings
+    )
 
-    return input_mappings
+    # Combine input mappings and calculated view attributes
+    all_mappings = input_mappings + calculated_view_attrs
+    return all_mappings
 
 
 def _build_input_mappings(
@@ -875,6 +973,8 @@ def _build_cv_attributes(
     if view_attrs is None:
         return mappings
 
+    cv_id = calculation_view.get(CDATAKeys.ID.value)
+
     for view_attr in view_attrs.findall(CDATAKeys.CALCULATION_VIEW_ATTRIBUTE.value, ns):
         formula = (
             view_attr.find(CDATAKeys.FORMULA.value, ns).text
@@ -886,25 +986,28 @@ def _build_cv_attributes(
             continue
 
         involved_columns = FORMULA_PATTERN.findall(formula)
+        # For calculated columns, all source columns should come from the same calculation view
+        # where the formula is defined
+        parents = []
         for col in involved_columns:
-            # Find the mapping for the involved column
-            mapping = next(
-                (mapping for mapping in input_mappings if mapping.target == col), None
-            )
-            if not mapping:
-                logger.debug(
-                    f"Can't find mapping for column [{col}] in [{input_mappings}]"
+            # The source columns for the formula are in the same calculation view
+            parents.append(
+                ParentSource(
+                    source=col,
+                    parent=cv_id,  # The parent is the current calculation view
                 )
-                continue
+            )
 
+        if parents:
             mappings.append(
                 DataSourceMapping(
                     target=view_attr.get(CDATAKeys.ID.value),
-                    parents=mapping.parents,
+                    parents=parents,
+                    formula=formula,
                 )
             )
 
-    return _group_mappings(mappings)
+    return mappings
 
 
 def _group_mappings(mappings: List[DataSourceMapping]) -> List[DataSourceMapping]: