Staging Tables

The epm_staging database contains intermediate views used by the dbt staging layer, and write-back tables for user-submitted data.

Staging Views

Staging models are materialized as views in the epm_staging schema. They perform field renames, joins, and JSON parsing without persisting data.

Staging models sit between Bronze (raw) and Silver (cleaned) layers:

epm_bronze (Airbyte tables) → epm_staging (views) → epm_silver (tables)

Each staging view typically: 1. Selects from the corresponding Bronze table 2. Renames fields from PascalCase to snake_case 3. Casts types using db_adapter.sql macros 4. Joins related reference tables (e.g., journal headers to line items)

Staging views are not directly queried by the API or Excel — they're internal to the dbt pipeline.

Write-Back Tables (Planned)

The epm_staging database is also reserved for future write-back scenarios:

Table	Purpose	Status
staging_budget_submissions	Budget entries submitted from Excel	Planned
staging_consolidation_adjustments	Top-side journals submitted via UI	Planned

The planned workflow: 1. User submits data from Excel or Frappe UI 2. Data lands in epm_staging table 3. Approval workflow in Frappe 4. Approved data is picked up by dbt on next build

ClickHouse Databases

Created by clickhouse/init-db.sql:

CREATE DATABASE IF NOT EXISTS epm_bronze;
CREATE DATABASE IF NOT EXISTS epm_staging;
CREATE DATABASE IF NOT EXISTS epm_silver;
CREATE DATABASE IF NOT EXISTS epm_gold;

Database	Managed By	Contents
epm_bronze	Airbyte	Raw D365 OData tables
epm_staging	dbt	Staging views + write-back tables
epm_silver	dbt	Cleaned, standardized tables
epm_gold	dbt	Business logic + seed tables