transform

sqllocks_spindle.transform

Transform layer — post-processing for star schema and CDM output.

Classes

DimSpec dataclass

Specification for building one dimension table.

FactSpec dataclass

Specification for building one fact table.

StarSchemaMap

Describes how to transform a domain result into a star schema.

StarSchemaResult dataclass

Result of a star schema transform.

Methods:

all_tables()

Return all tables (dimensions + date_dim + facts) as a flat dict.

StarSchemaTransform

Transform a 3NF GenerationResult into a star schema using a StarSchemaMap.

Methods:

transform(tables, schema_map)

Apply the star schema transform.

Parameters:

Name	Type	Description	Default
tables	dict[str, DataFrame]	Dict of table_name → DataFrame (from GenerationResult.tables).	required
schema_map	StarSchemaMap	Mapping spec defining dims and facts.	required

Returns:

Type	Description
StarSchemaResult	StarSchemaResult with dimensions, facts, and date_dim.

CdmMapper

Export tables as a Microsoft CDM folder (model.json + data files).

The output folder structure is compatible with Fabric CDM connectors, Dataverse, Power Platform, and Azure Data Lake Storage CDM folders.

Usage::

from sqllocks_spindle.transform import CdmMapper, CdmEntityMap

mapper = CdmMapper()
mapper.write_cdm_folder(
    tables=result.tables,
    output_dir="./cdm",
    domain_name="SpindleRetail",
    entity_map=CdmEntityMap({"customer": "Contact", "order": "SalesOrder"}),
)

Methods:

write_cdm_folder(tables, output_dir, domain_name='SpindleOutput', entity_map=None, fmt='csv')

Write a CDM folder to disk.

Parameters:

Name	Type	Description	Default
tables	dict[str, DataFrame]	Dict of table_name → DataFrame.	required
output_dir	str | Path	Root directory for CDM folder output.	required
domain_name	str	The CDM model name (appears in model.json).	'SpindleOutput'
entity_map	CdmEntityMap | None	Optional mapping of table names to CDM entity names.	None
fmt	str	Data file format — "csv" (default) or "parquet".	'csv'

Returns:

Type	Description
list[Path]	List of written file paths.

to_model_json(tables, domain_name='SpindleOutput', entity_map=None, fmt='csv')

Generate a model.json manifest dict without writing to disk.

Useful for in-memory CDM metadata generation or Fabric notebook use.

CdmEntityMap

Optional mapping from source table names to CDM entity names.

If a table is not in the map, it defaults to PascalCase of the table name.