Skip to content

warehouse

sqllocks_spindle.engine.sinks.warehouse

Classes

WarehouseSink

Delegate sink that writes chunks to a Fabric Warehouse via COPY INTO.

Thin wrapper around :class:~sqllocks_spindle.fabric.WarehouseBulkWriter. Accumulates numpy array chunks as DataFrames per table, then flushes via write_tables() on close.

Requires pip install sqllocks-spindle[fabric-sql,parquet].

Parameters:

Name Type Description Default
connection_string str

pyodbc connection string for the Warehouse endpoint.

required
staging_lakehouse_path str

abfss:// path to a Lakehouse Files area for staging Parquet files before COPY INTO.

required
auth_method str

"cli", "msi", or "spn" (default "cli").

'cli'
schema_name str

SQL schema for target tables (default "dbo").

'dbo'
client_id str | None

Service principal client ID (required when auth_method="spn").

None
client_secret str | None

Service principal secret (required when auth_method="spn").

None
tenant_id str | None

Azure AD tenant ID (required when auth_method="spn").

None
chunk_size int

Rows per staged Parquet file (default 1 000 000).

1000000