ethyca-fides 2.66.2rc0__py2.py3-none-any.whl → 2.67.0__py2.py3-none-any.whl

fides/api/service/connectors/query_configs/bigquery_query_config.py

@@ -3,7 +3,7 @@ from typing import Any, Dict, List, Optional, Union, cast
 import pydash
 from fideslang.models import MaskingStrategies
 from loguru import logger
-from sqlalchemy import MetaData, Table, text
+from sqlalchemy import MetaData, Table, or_, text
 from sqlalchemy.engine import Engine
 from sqlalchemy.sql import Delete, Update
 from sqlalchemy.sql.elements import ColumnElement, TextClause
@@ -93,7 +93,7 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
 
         return where_clauses
 
-    def _generate_table_name(self) -> str:
+    def generate_table_name(self) -> str:
         """
         Prepends the dataset ID and project ID to the base table name
         if the BigQuery namespace meta is provided.
@@ -116,7 +116,7 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
         Returns a query string with backtick formatting for tables that have the same names as
         BigQuery reserved words.
         """
-        return f'SELECT {field_list} FROM `{self._generate_table_name()}` WHERE ({" OR ".join(clauses)})'
+        return f'SELECT {field_list} FROM `{self.generate_table_name()}` WHERE ({" OR ".join(clauses)})'
 
     def generate_masking_stmt(
         self,
@@ -125,6 +125,7 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
         policy: Policy,
         request: PrivacyRequest,
         client: Engine,
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> Union[List[Update], List[Delete]]:
         """
         Generate a masking statement for BigQuery.
@@ -137,7 +138,7 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
             logger.info(
                 f"Masking override detected for collection {node.address.value}: {masking_override.strategy.value}"
             )
-            return self.generate_delete(row, client)
+            return self.generate_delete(client, input_data or {})
         return self.generate_update(row, policy, request, client)
 
     def generate_update(
@@ -196,11 +197,18 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
             )
             return []
 
-        table = Table(self._generate_table_name(), MetaData(bind=client), autoload=True)
+        table = Table(self.generate_table_name(), MetaData(bind=client), autoload=True)
         where_clauses: List[ColumnElement] = [
-            getattr(table.c, k) == v for k, v in non_empty_reference_field_keys.items()
+            table.c[k] == v for k, v in non_empty_reference_field_keys.items()
         ]
 
+        # Create update values using Column objects as keys to handle column names with spaces
+        update_values = {}
+        for column_name, value in final_update_map.items():
+            # Use bracket notation to access columns with spaces in their names
+            column = table.c[column_name]
+            update_values[column] = value
+
         if self.partitioning:
             partition_clauses = self.get_partition_clauses()
             partitioned_queries = []
@@ -211,44 +219,55 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
                 partitioned_queries.append(
                     table.update()
                     .where(*(where_clauses + [text(partition_clause)]))
-                    .values(**final_update_map)
+                    .values(update_values)
                 )
 
             return partitioned_queries
 
-        return [table.update().where(*where_clauses).values(**final_update_map)]
+        return [table.update().where(*where_clauses).values(update_values)]
 
-    def generate_delete(self, row: Row, client: Engine) -> List[Delete]:
-        """Returns a List of SQLAlchemy DELETE statements for BigQuery. Does not actually execute the delete statement.
+    def generate_delete(
+        self,
+        client: Engine,
+        input_data: Optional[Dict[str, List[Any]]] = None,
+    ) -> List[Delete]:
+        """
+        Returns a List of SQLAlchemy DELETE statements for BigQuery. Does not actually execute the delete statement.
 
         Used when a collection-level masking override is present and the masking strategy is DELETE.
 
         A List of multiple DELETE statements are returned for partitioned tables; for a non-partitioned table,
         a single DELETE statement is returned in a List for consistent typing.
-
-        TODO: DRY up this method and `generate_update` a bit
         """
 
-        non_empty_reference_field_keys: Dict[str, Field] = filter_nonempty_values(
-            {
-                fpath.string_path: fld.cast(row[fpath.string_path])
-                for fpath, fld in self.reference_field_paths.items()
-                if fpath.string_path in row
-            }
-        )
+        if not input_data:
+            logger.warning(
+                "No input data provided for node '{}', skipping DELETE statement generation",
+                self.node.address,
+            )
+            return []
 
-        valid = len(non_empty_reference_field_keys) > 0
-        if not valid:
+        filtered_data = self.node.typed_filtered_values(input_data)
+
+        if not filtered_data:
             logger.warning(
                 "There is not enough data to generate a valid DELETE statement for {}",
                 self.node.address,
             )
             return []
 
-        table = Table(self._generate_table_name(), MetaData(bind=client), autoload=True)
-        where_clauses: List[ColumnElement] = [
-            getattr(table.c, k) == v for k, v in non_empty_reference_field_keys.items()
-        ]
+        table = Table(self.generate_table_name(), MetaData(bind=client), autoload=True)
+
+        # Build individual reference clauses
+        where_clauses: List[ColumnElement] = []
+        for column_name, values in filtered_data.items():
+            if len(values) == 1:
+                where_clauses.append(table.c[column_name] == values[0])
+            else:
+                where_clauses.append(table.c[column_name].in_(values))
+
+        # Combine reference clauses with OR instead of AND
+        combined_reference_clause = or_(*where_clauses)
 
         if self.partitioning:
             partition_clauses = self.get_partition_clauses()
@@ -259,12 +278,25 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
 
             for partition_clause in partition_clauses:
                 partitioned_queries.append(
-                    table.delete().where(*(where_clauses + [text(partition_clause)]))
+                    table.delete()
+                    .where(combined_reference_clause)
+                    .where(text(partition_clause))
                 )
 
             return partitioned_queries
 
-        return [table.delete().where(*where_clauses)]
+        return [table.delete().where(combined_reference_clause)]
+
+    def uses_delete_masking_strategy(self) -> bool:
+        """Check if this collection uses DELETE masking strategy.
+
+        Returns True if masking override is present and strategy is DELETE.
+        """
+        masking_override = self.node.collection.masking_strategy_override
+        return (
+            masking_override is not None
+            and masking_override.strategy == MaskingStrategies.DELETE
+        )
 
     def format_fields_for_query(
         self,
@@ -281,6 +313,25 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
                 formatted_fields.append(field_path.levels[0])
         return formatted_fields
 
+    def format_clause_for_query(
+        self, string_path: str, operator: str, operand: str
+    ) -> str:
+        """
+        Returns clauses with proper BigQuery backtick escaping for column names.
+        Handles column names with spaces and nested fields (dot-separated) by escaping each part individually.
+        """
+        # For nested fields (containing dots), escape each part individually
+        if "." in string_path:
+            parts = string_path.split(".")
+            escaped_field = ".".join(f"`{part}`" for part in parts)
+        else:
+            # For simple fields, wrap the entire name in backticks
+            escaped_field = f"`{string_path}`"
+
+        if operator == "IN":
+            return f"{escaped_field} IN ({operand})"
+        return f"{escaped_field} {operator} :{operand}"
+
     def generate_raw_query_without_tuples(
         self, field_list: List[str], filters: Dict[str, List[Any]]
     ) -> Optional[TextClause]:
@@ -290,27 +341,35 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
 
         This is an override of the base class method that supports nested fields for BigQuery.
 
-        Examples with dot-delimited field names, notice the periods are replaced with underscores in the parameter bindings:
+        Examples with field names containing dots and spaces, notice these are replaced with underscores in the parameter bindings:
 
         1. Single value filter:
            field_list = ["id", "name", "email"]
            filters = {"user.id": [123]}
 
-           Generates: SELECT id, name, email FROM `project_id.dataset_id.table_name` WHERE (user.id = :user_id)
+           Generates: SELECT id, name, email FROM `project_id.dataset_id.table_name` WHERE (`user`.`id` = :user_id)
            With parameter binding: user_id = 123
 
-        2. Multiple value filter:
+        2. Field with spaces:
+           field_list = ["id", "custom id", "email"]
+           filters = {"custom id": ["abc123"]}
+
+           Generates: SELECT id, `custom id`, email FROM `project_id.dataset_id.table_name` WHERE (`custom id` = :custom_id)
+           With parameter binding: custom_id = "abc123"
+
+        3. Multiple value filter with nested field:
            field_list = ["id", "name", "email"]
-           filters = {"user.status": ["active", "pending"]}
+           filters = {"contact_info.primary_email": ["active", "pending"]}
 
-           Generates: SELECT id, name, email FROM `project_id.dataset_id.table_name` WHERE (user.status IN (:user_status_in_stmt_generated_0, :user_status_in_stmt_generated_1))
-           With parameter bindings: user_status_in_stmt_generated_0 = "active", user_status_in_stmt_generated_1 = "pending"
+           Generates: SELECT id, name, email FROM `project_id.dataset_id.table_name` WHERE (`contact_info`.`primary_email` IN (:contact_info_primary_email_in_stmt_generated_0, :contact_info_primary_email_in_stmt_generated_1))
+           With parameter bindings: contact_info_primary_email_in_stmt_generated_0 = "active", contact_info_primary_email_in_stmt_generated_1 = "pending"
         """
         clauses = []
         query_data = {}
         for field_name, field_value in filters.items():
-            # Replace dots with underscores in field names for parameter binding
-            field_binding_name = field_name.replace(".", "_")
+            # Replace dots and spaces with underscores in field names for parameter binding
+            # SQLAlchemy parameter names cannot contain spaces or special characters
+            field_binding_name = field_name.replace(".", "_").replace(" ", "_")
             data = set(field_value)
             if len(data) == 1:
                 clauses.append(
@@ -333,7 +392,7 @@ class BigQueryQueryConfig(QueryStringWithoutTuplesOverrideQueryConfig):
                 clauses.append(self.format_clause_for_query(field_name, "IN", operand))
 
         if len(clauses) > 0:
-            formatted_fields = ", ".join(field_list)
+            formatted_fields = ", ".join([f"`{field}`" for field in field_list])
             query_str = self.get_formatted_query_string(formatted_fields, clauses)
             return text(query_str).params(query_data)
 

fides/api/service/connectors/query_configs/snowflake_query_config.py

@@ -30,7 +30,7 @@ class SnowflakeQueryConfig(SQLQueryConfig):
         """Returns field names in clauses surrounded by quotation marks as required by Snowflake syntax."""
         return f'"{string_path}" {operator} (:{operand})'
 
-    def _generate_table_name(self) -> str:
+    def generate_table_name(self) -> str:
         """
         Prepends the dataset name and schema to the base table name
         if the Snowflake namespace meta is provided.
@@ -57,7 +57,7 @@ class SnowflakeQueryConfig(SQLQueryConfig):
         clauses: List[str],
     ) -> str:
         """Returns a query string with double quotation mark formatting as required by Snowflake syntax."""
-        return f'SELECT {field_list} FROM {self._generate_table_name()} WHERE ({" OR ".join(clauses)})'
+        return f'SELECT {field_list} FROM {self.generate_table_name()} WHERE ({" OR ".join(clauses)})'
 
     def format_key_map_for_update_stmt(self, param_map: Dict[str, Any]) -> List[str]:
         """Adds the appropriate formatting for update statements in this datastore."""
@@ -69,4 +69,4 @@ class SnowflakeQueryConfig(SQLQueryConfig):
         where_clauses: List[str],
     ) -> str:
         """Returns a parameterized update statement in Snowflake dialect."""
-        return f'UPDATE {self._generate_table_name()} SET {", ".join(update_clauses)} WHERE {" AND ".join(where_clauses)}'
+        return f'UPDATE {self.generate_table_name()} SET {", ".join(update_clauses)} WHERE {" AND ".join(where_clauses)}'

fides/api/service/connectors/rds_mysql_connector.py

@@ -158,6 +158,7 @@ class RDSMySQLConnector(RDSConnectorMixin, SQLConnector):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """DSR execution not yet supported for RDS MySQL"""
         return 0

fides/api/service/connectors/rds_postgres_connector.py

@@ -147,6 +147,7 @@ class RDSPostgresConnector(RDSConnectorMixin, SQLConnector):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """DSR execution not yet supported for RDS Postgres"""
         return 0

fides/api/service/connectors/s3_connector.py

@@ -66,6 +66,7 @@ class S3Connector(BaseConnector):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """DSR execution not yet supported for S3"""
         return 0

fides/api/service/connectors/saas_connector.py

@@ -517,6 +517,7 @@ class SaaSConnector(BaseConnector[AuthenticatedClient], Contextualizable):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """Execute a masking request. Return the number of rows that have been updated."""
         self.set_privacy_request_state(privacy_request, node, request_task)

fides/api/service/connectors/scylla_connector.py

@@ -150,6 +150,7 @@ class ScyllaConnector(BaseConnector[Cluster]):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """Execute a masking request"""
         query_config = self.query_config(node)

fides/api/service/connectors/snowflake_connector.py

@@ -3,11 +3,11 @@ from typing import Any, Dict, Union
 from cryptography.hazmat.backends import default_backend
 from cryptography.hazmat.primitives import serialization
 from snowflake.sqlalchemy import URL as Snowflake_URL
+from sqlalchemy import text
 from sqlalchemy.orm import Session
 
 from fides.api.graph.execution import ExecutionNode
 from fides.api.schemas.connection_configuration import SnowflakeSchema
-from fides.api.service.connectors.query_configs.query_config import SQLQueryConfig
 from fides.api.service.connectors.query_configs.snowflake_query_config import (
     SnowflakeQueryConfig,
 )
@@ -69,10 +69,63 @@ class SnowflakeConnector(SQLConnector):
                 connect_args["private_key"] = private_key
         return connect_args
 
-    def query_config(self, node: ExecutionNode) -> SQLQueryConfig:
+    def query_config(self, node: ExecutionNode) -> SnowflakeQueryConfig:
         """Query wrapper corresponding to the input execution_node."""
 
         db: Session = Session.object_session(self.configuration)
         return SnowflakeQueryConfig(
             node, SQLConnector.get_namespace_meta(db, node.address.dataset)
         )
+
+    def get_qualified_table_name(self, node: ExecutionNode) -> str:
+        """Get fully qualified Snowflake table name using existing query config logic"""
+        query_config = self.query_config(node)
+        return query_config.generate_table_name()
+
+    def table_exists(self, qualified_table_name: str) -> bool:
+        """
+        Check if table exists in Snowflake using the proper three-part naming convention.
+
+        Snowflake supports database.schema.table naming, and the generic SQLConnector
+        table_exists method doesn't handle quoted identifiers properly.
+        """
+        try:
+            client = self.create_client()
+            with client.connect() as connection:
+                # Remove quotes and split the parts
+                clean_name = qualified_table_name.replace('"', "")
+                parts = clean_name.split(".")
+
+                if len(parts) == 1:
+                    # Simple table name - use current schema
+                    table_name = parts[0]
+                    result = connection.execute(text(f'DESC TABLE "{table_name}"'))
+                elif len(parts) == 2:
+                    # schema.table format
+                    schema_name, table_name = parts
+                    result = connection.execute(
+                        text(f'DESC TABLE "{schema_name}"."{table_name}"')
+                    )
+                elif len(parts) >= 3:
+                    # database.schema.table format
+                    database_name, schema_name, table_name = (
+                        parts[-3],
+                        parts[-2],
+                        parts[-1],
+                    )
+                    # Use the database.schema.table format
+                    result = connection.execute(
+                        text(
+                            f'DESC TABLE "{database_name}"."{schema_name}"."{table_name}"'
+                        )
+                    )
+                else:
+                    return False
+
+                # If we get here without an exception, the table exists
+                result.close()
+                return True
+
+        except Exception:
+            # Table doesn't exist or other error
+            return False

fides/api/service/connectors/sql_connector.py

@@ -6,7 +6,7 @@ import paramiko
 import sshtunnel  # type: ignore
 from aiohttp.client_exceptions import ClientResponseError
 from loguru import logger
-from sqlalchemy import Column, select
+from sqlalchemy import Column, inspect, select
 from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.engine import (  # type: ignore
     Connection,
@@ -22,16 +22,19 @@ from sqlalchemy.sql.elements import TextClause
 from fides.api.common_exceptions import (
     ConnectionException,
     SSHTunnelConfigNotFoundException,
+    TableNotFound,
 )
 from fides.api.graph.execution import ExecutionNode
 from fides.api.models.connectionconfig import ConnectionConfig, ConnectionTestStatus
 from fides.api.models.policy import Policy
 from fides.api.models.privacy_request import PrivacyRequest, RequestTask
+from fides.api.schemas.application_config import SqlDryRunMode
 from fides.api.schemas.connection_configuration import ConnectionConfigSecretsSchema
 from fides.api.service.connectors.base_connector import BaseConnector
 from fides.api.service.connectors.query_configs.query_config import SQLQueryConfig
 from fides.api.util.collection_util import Row
 from fides.config import get_config
+from fides.config.config_proxy import ConfigProxy
 
 from fides.api.models.sql_models import (  # type: ignore[attr-defined] # isort: skip
     Dataset as CtlDataset,
@@ -58,6 +61,23 @@ class SQLConnector(BaseConnector[Engine]):
             )
         self.ssh_server: sshtunnel._ForwardServer = None
 
+    def should_dry_run(self, mode_to_check: SqlDryRunMode) -> bool:
+        """
+        Check if SQL dry run is enabled for the specified mode.
+
+        Args:
+            mode_to_check: The SqlDryRunMode to check for
+
+        Returns:
+            bool: True if the current mode matches the mode to check
+        """
+        from fides.api.api.deps import get_autoclose_db_session as get_db
+
+        with get_db() as db:
+            config_proxy = ConfigProxy(db)
+            current_mode = getattr(config_proxy.execution, "sql_dry_run", None)
+            return current_mode == mode_to_check
+
     @staticmethod
     def cursor_result_to_rows(results: CursorResult) -> List[Row]:
         """Convert SQLAlchemy results to a list of dictionaries"""
@@ -140,6 +160,10 @@ class SQLConnector(BaseConnector[Engine]):
         if query is None:
             return []
 
+        if self.should_dry_run(SqlDryRunMode.access):
+            logger.warning(f"SQL DRY RUN - Would execute SQL: {query}")
+            return []
+
         with client.connect() as connection:
             self.set_schema(connection)
             results = connection.execute(query)
@@ -160,16 +184,34 @@ class SQLConnector(BaseConnector[Engine]):
         if stmt is None:
             return []
 
+        if self.should_dry_run(SqlDryRunMode.access):
+            logger.warning(f"SQL DRY RUN - Would execute SQL: {stmt}")
+            return []
+
         logger.info("Starting data retrieval for {}", node.address)
         with client.connect() as connection:
-            self.set_schema(connection)
-            if (
-                query_config.partitioning
-            ):  # only BigQuery supports partitioning, for now
-                return self.partitioned_retrieval(query_config, connection, stmt)
-
-            results = connection.execute(stmt)
-            return self.cursor_result_to_rows(results)
+            try:
+                self.set_schema(connection)
+                if (
+                    query_config.partitioning
+                ):  # only BigQuery supports partitioning, for now
+                    return self.partitioned_retrieval(query_config, connection, stmt)
+
+                results = connection.execute(stmt)
+                return self.cursor_result_to_rows(results)
+            except Exception as exc:
+                # Check if table exists using qualified table name
+                qualified_table_name = self.get_qualified_table_name(node)
+                if not self.table_exists(qualified_table_name):
+                    # Central decision point - will raise TableNotFound or ConnectionException
+                    self.handle_table_not_found(
+                        node=node,
+                        table_name=qualified_table_name,
+                        operation_context="data retrieval",
+                        original_exception=exc,
+                    )
+                # Table exists or can't check - re-raise original exception
+                raise
 
     def mask_data(
         self,
@@ -178,20 +220,41 @@ class SQLConnector(BaseConnector[Engine]):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """Execute a masking request. Returns the number of records masked"""
         query_config = self.query_config(node)
         update_ct = 0
         client = self.client()
+
         for row in rows:
             update_stmt: Optional[TextClause] = query_config.generate_update_stmt(
                 row, policy, privacy_request
             )
             if update_stmt is not None:
-                with client.connect() as connection:
-                    self.set_schema(connection)
-                    results: LegacyCursorResult = connection.execute(update_stmt)
-                    update_ct = update_ct + results.rowcount
+                if self.should_dry_run(SqlDryRunMode.erasure):
+                    logger.warning(f"SQL DRY RUN - Would execute SQL: {update_stmt}")
+                else:
+                    with client.connect() as connection:
+                        try:
+                            self.set_schema(connection)
+                            results: LegacyCursorResult = connection.execute(
+                                update_stmt
+                            )
+                            update_ct = update_ct + results.rowcount
+                        except Exception as exc:
+                            # Check if table exists using qualified table name
+                            qualified_table_name = self.get_qualified_table_name(node)
+                            if not self.table_exists(qualified_table_name):
+                                # Central decision point - will raise TableNotFound or ConnectionException
+                                self.handle_table_not_found(
+                                    node=node,
+                                    table_name=qualified_table_name,
+                                    operation_context="data erasure",
+                                    original_exception=exc,
+                                )
+                            # Table exists or can't check - re-raise original exception
+                            raise
         return update_ct
 
     def close(self) -> None:
@@ -258,3 +321,86 @@ class SQLConnector(BaseConnector[Engine]):
         raise NotImplementedError(
             "Partitioned retrieval is only supported for BigQuery currently!"
         )
+
+    def get_qualified_table_name(self, node: ExecutionNode) -> str:
+        """
+        Get the fully qualified table name for this database.
+
+        Default: Returns the simple collection name
+        Override: Database-specific connectors can implement namespace resolution
+        """
+        return node.collection.name
+
+    def table_exists(self, qualified_table_name: str) -> bool:
+        """
+        Check if table exists using SQLAlchemy introspection.
+
+        This is a generic implementation that should work for most SQL databases.
+        Override: Connectors can implement database-specific table existence checking
+        """
+        try:
+            client = self.create_client()
+            with client.connect() as connection:
+                inspector = inspect(connection)
+
+                # For simple table names
+                if "." not in qualified_table_name:
+                    return inspector.has_table(qualified_table_name)
+
+                # For qualified names like schema.table or database.schema.table
+                parts = qualified_table_name.split(".")
+
+                if len(parts) == 2:
+                    # schema.table format
+                    schema_name, table_name = parts
+                    return inspector.has_table(table_name, schema=schema_name)
+
+                if len(parts) >= 3:
+                    # database.schema.table format (use schema.table)
+                    schema_name, table_name = parts[-2], parts[-1]
+                    return inspector.has_table(table_name, schema=schema_name)
+
+                # Fallback for unexpected format
+                return inspector.has_table(qualified_table_name)
+
+        except Exception as exc:
+            # Graceful fallback - if we can't check, assume table exists
+            # to preserve existing behavior for connectors that don't implement this
+            logger.error("Unable to check if table exists, assuming it does: {}", exc)
+            return True
+
+    def handle_table_not_found(
+        self,
+        node: ExecutionNode,
+        table_name: str,
+        operation_context: str,
+        original_exception: Optional[Exception] = None,
+    ) -> None:
+        """
+        Central decision point for table-not-found scenarios.
+
+        Raises TableNotFound (for collection skipping) or ConnectionException (for hard errors).
+        The raised exception will be caught by the @retry decorator in graph_task.py.
+
+        Args:
+            node: The ExecutionNode being processed
+            table_name: Name of the missing table
+            operation_context: Context like "data retrieval" or "data masking"
+            original_exception: The original exception that triggered this check
+        """
+        if node.has_outgoing_dependencies():
+            # Collection has dependencies - cannot skip safely
+            error_msg = (
+                f"Table '{table_name}' did not exist during {operation_context}. "
+                f"Cannot skip collection '{node.address}' because other collections depend on it."
+            )
+            if original_exception:
+                raise ConnectionException(error_msg) from original_exception
+            raise ConnectionException(error_msg)
+
+        # Safe to skip - raise TableNotFound for @retry decorator to catch
+        skip_msg = f"Table '{table_name}' did not exist during {operation_context}."
+        if original_exception:
+            raise TableNotFound(skip_msg) from original_exception
+
+        raise TableNotFound(skip_msg)

fides/api/service/connectors/website_connector.py

@@ -75,6 +75,7 @@ class WebsiteConnector(BaseConnector):
         privacy_request: PrivacyRequest,
         request_task: RequestTask,
         rows: List[Row],
+        input_data: Optional[Dict[str, List[Any]]] = None,
     ) -> int:
         """DSR execution not supported for website connector"""
         return 0

fides/api/service/privacy_request/dsr_package/templates/welcome.html

@@ -48,8 +48,8 @@
          <div class="dsr-information-text">
             <p>
                This web link contains all data requested as part of your Data Subject Request (DSR). Your
-               information has been compiled from the following areas. Click on each section to open a file and view
-               your data in PDF format.
+               information has been compiled from the following areas. Click on each section to open and view
+               your data.
             </p>
          </div>
          <div class="dataset-list">