drb-iceberg

Purpose: Turn a product topic's iceberg: ontological configuration into a declarative IcebergMapping that describes table name, partitions, column extractors, and derived sub-tables. The add-on performs planning only — no Iceberg catalog I/O.

Contract

IcebergAddon.apply(node, topic=None, **kwargs) returns an IcebergMapping object containing:

• table: main table name (string)
• ingestible: ingestion enabled (boolean)
• ingest_priority: optional priority level (int or None)
• identifier: unique key extractor (XQuery-based, not executed)
• partitions: list of partition field definitions
• columns: column schema with row extractors (delegated to drb-metadata)
• derived: list of derived (sub-)table definitions

The mapping is declarative: extractors are not executed by the add-on. A separate consumer processes the plan and executes row extraction/table writes.

Dependencies

• drb>=2.2,<3
• drb-metadata>=1.4,<2 — provides column extraction via extract_for_class(class_uri, graph)

The iceberg: vocabulary

Attached to a product topic at namespace http://www.gael.fr/iceberg#:

PartitionStrategy

• partitionField (string) — field name
• partitionTransform (string: identity, year, month, day, etc.)
• partitionOrder (xs:int) — sort order

DerivedTable

• tableName (string) — suffix appended to main table with a dot
• entityListExtractor (XQuery) — yields the list of row nodes
• entityClass (resource) OR entityMetadataExtractor (XQuery) — column source (entityClass wins if both present)
• entityIdentifier (XQuery) — unique key per row