Set Delta liquid-clustering keys on an existing table. Metadata-only and idempotent: re-running just re-asserts the keys. Subsequent

Aggregate enriched orders by the report dimensions. Mirrors GenerateOrdersAgg.aggregate_orders exactly. total_value sums item_total — th

Four-way join: order_item ⨝ order ⨝ customer ⨝ product. Mirrors GenerateOrders.enrich_order exactly. product_name is read from the product

Resolve a SQL warehouse to use for DDL. Picking warehouses[0] is fragile — the API returns warehouses in an implementation-defined order, so

The product ids selected for a rename on a given day, by a day-anchored rotation. Deterministic so reruns of the same date are idempotent.

Integration test: setup → run (job1 batch) + run_sdp (pipeline) in parallel → validate. A single validate task waits for both run and run_sdp to

Prod integration job: seed_sources → run (job1 batch) + run_sdp (pipeline) in parallel. Mirrors the staging integration test structure but uses s

Read the DQX version from pyproject.toml so it stays in sync with the runtime dep.

DROP statements to try in order. The kind-specific form first (when there is one), then plain DROP TABLE as a fallback for warehouses whose parser

Stable category attribute of a product, mirroring the Spark expression in _seed_initial: category = (product_id - 1) % 10 + 1. Used to build the n

Static unit_price of a product, mirroring the Spark expression in _seed_initial. unit_price never changes after the initial load; it is carried on

Read version from pyproject.toml so the bundle pins to the exact built wheel.

Compute the target catalog at generation time, mirroring Config.__init__. - dev → dev_<sanitized_user> (per-developer sandbox) - stagin

Target-level run_as and root_path pinned to the SP for staging/prod. Two independent gates stack here: 1. Target-level run_as makes mode: p

Analyze job costs for a specific date using system tables. Args: workspace: Databricks WorkspaceClient instance target_date:

Fallback method using Jobs API when system tables are unavailable. Note: This doesn't provide cost data, only run information.

Get today's date in YYYY-MM-DD format.

Get the first available SQL warehouse ID.

Main entry point for the script.

Pre-joined enriched orders matching the output of enrich_order().

One aggregated row per distinct customer name.

Aggregated qty and value must match hand-calculated totals.