delta_writer

`sqllocks_spindle.output.delta_writer` ¶

Delta Lake writer for Spindle output.

Write generated tables as Delta Lake tables.

Uses the deltalake (delta-rs) package — works both locally and inside Microsoft Fabric Notebooks without requiring Spark or JVM.

Install the required extra::

pip install sqllocks-spindle[fabric]

Parameters:

Name	Type	Description	Default
`output_dir`	`str \| Path \| None`	Root directory for Delta tables. Each table is written to a subdirectory `{output_dir}/{table_name}/`. If None and running inside a Fabric Notebook, defaults to `/lakehouse/default/Tables/`.	`None`
`partition_by`	`dict[str, list[str]] \| None`	Per-table partition specs from the schema output config. Keys are table names, values are lists of specs like `["order_date:year", "order_date:month"]`.	`None`
`mode`	`str`	Write mode — `"overwrite"` (default) or `"append"`.	`'overwrite'`

Example::

from sqllocks_spindle.output import DeltaWriter

writer = DeltaWriter(output_dir="./delta_output")
paths = writer.write_all(result.tables)

Write all tables as Delta Lake tables in parallel.

delta-rs releases the GIL during Parquet/Delta writes, so ThreadPoolExecutor gives genuine parallelism here.

Parameters:

Name	Type	Description	Default
`tables`	`dict[str, DataFrame]`	Mapping of table name to DataFrame (the standard Spindle output format from `GenerationResult.tables`).	required
`max_workers`	`int`	Maximum parallel write threads (default 4).	`4`

Returns:

Type	Description
`list[Path]`	List of paths to the written Delta table directories.

Write a single table as a Delta Lake table.

Parameters:

Name	Type	Description	Default
`table_name`	`str`	Name of the table (becomes the subdirectory name).	required
`df`	`DataFrame`	DataFrame to write.	required

Returns:

Type	Description
`Path`	Path to the written Delta table directory.