class singer_sdk.SQLSink#

SQL-type sink type.

__init__(target: Target, stream_name: str, schema: dict, key_properties: list[str] | None, connector: SQLConnector | None = None) None#

Initialize SQL Sink.

  • target – The target object.

  • stream_name – The source tap’s stream name.

  • schema – The JSON Schema definition.

  • key_properties – The primary key columns.

  • connector – Optional connector to reuse.

activate_version(new_version: int) None#

Bump the active version of the target table.


new_version – The version number to activate.

bulk_insert_records(full_table_name: str, schema: dict, records: Iterable[dict[str, Any]]) int | None#

Bulk insert records to an existing destination table.

The default implementation uses a generic SQLAlchemy bulk insert operation. This method may optionally be overridden by developers in order to provide faster, native bulk uploads.

  • full_table_name – the target table name.

  • schema – the JSON schema for the new table, to be used when inferring column names.

  • records – the input records.


True if table exists, False if not, None if unsure or undetectable.

conform_name(name: str, object_type: str | None = None) str#

Conform a stream property name to one suitable for the target system.

Transforms names to snake case by default, applicable to most common DBMSs’. Developers may override this method to apply custom transformations to database/schema/table/column names.

  • name – Property name.

  • object_type – One of database, schema, table or column.


The name transformed to snake case.

conform_record(record: dict) dict#

Return record dictionary with property names conformed.


record – Dictionary representing a single record.


New record dictionary with conformed column names.

conform_schema(schema: dict) dict#

Return schema dictionary with property names conformed.


schema – JSON schema dictionary.


A schema dictionary with the property names conformed.

property connection: Connection#

Get or set the SQLAlchemy connection for this sink.


A connection object.

property connector: SQLConnector#

The connector object.


The connector object.

property database_name: str | None#

Return the DB name or None if using names with no database part.

property full_schema_name: str#

Return the fully qualified schema name.


The fully qualified schema name.

property full_table_name: str#

Return the fully qualified table name.


The fully qualified table name.

generate_insert_statement(full_table_name: str, schema: dict) str | Executable#

Generate an insert statement for the given records.

  • full_table_name – the target table name.

  • schema – the JSON schema for the new table.


An insert statement.

property key_properties: list[str]#

Return key properties, conformed to target system naming requirements.


A list of key properties, conformed with self.conform_name()

merge_upsert_from_table(target_table_name: str, from_table_name: str, join_keys: list[str]) int | None#

Merge upsert data from one table to another.

  • target_table_name – The destination table name.

  • from_table_name – The source table name.

  • join_keys – The merge upsert keys, or None to append.


The number of records copied, if detectable, or None if the API does not report number of records affected/inserted.


NotImplementedError – if the merge upsert capability does not exist or is undefined.

process_batch(context: dict) None#

Process a batch with the given batch context.

Writes a batch to the SQL target. Developers may override this method in order to provide a more efficient upload/upsert process.


context – Stream partition or context dictionary.

property schema_name: str | None#

Return the schema name or None if using names with no schema part.


The target schema name.

setup() None#

Set up Sink.

This method is called on Sink creation, and creates the required Schema and Table entities in the target database.

property table_name: str#

Return the table name, with no schema or database part.


The target table name.