merica-dlh/merica-pipelines
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
c64fe92543f88c7d2145ff27d85c9b7a855e7d5d
merica-pipelines/metadata/e2e_orders_master_medallion_flow.json
T

481 lines
18 KiB
JSON
Raw Normal View History

feat(pipeline): add e2e orders master medallion flow and dbt models
2026-10-19 01:53:47 +07:00
{
"id": "e2e_orders_master_medallion_flow",
"description": "Dynamic Data Lakehouse Pipeline via DLH Studio",
"schedule": "0 */1 * * *",
"owner_id": "local-admin",
"owner_name": "Administrator Lokal (Break-Glass)",
"nodes": [
{
"id": "node-9826",
"type": "customNode",
"position": {
"x": 80,
"y": 80
},
"data": {
"label": "Extract Master Facilities",
"type": "source_db",
"subtitle": "Postgres / MySQL / Oracle",
"isSource": true,
"configured": true,
"config": {
"conn": "Postgres",
"conn_id": "conn_postgres_source_db",
"table": "public.ref_waste_facilities",
"table_name": "public.ref_waste_facilities",
"target_type": "minio_s3"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-0827",
"type": "customNode",
"position": {
"x": 270,
"y": 80
},
"data": {
"label": "DuckDB Master Facilities to Iceberg Bronze",
"type": "duckdb_transform",
"subtitle": "In-process SQL (<512MB)",
"configured": true,
"config": {
"layer": "silver",
"source_parquet_path": "s3://dlh-bronze/staging/ref_facilities/*.parquet",
"source_path": "s3://dlh-bronze/staging/ref_facilities/*.parquet",
"target_table": "bronze_ref_facilities",
"table_name": "bronze_ref_facilities",
"primary_keys": "facility_id",
"query": "SELECT facility_id, facility_name, operator_name, license_number, region_zone, facility_type, max_daily_capacity_ton, facility_status, compliance_rating, created_at FROM read_parquet('s3://dlh-bronze/staging/ref_facilities/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY facility_id ORDER BY created_at DESC) = 1",
"sql_query": "SELECT facility_id, facility_name, operator_name, license_number, region_zone, facility_type, max_daily_capacity_ton, facility_status, compliance_rating, created_at FROM read_parquet('s3://dlh-bronze/staging/ref_facilities/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY facility_id ORDER BY created_at DESC) = 1",
"transformation_sql": "SELECT facility_id, facility_name, operator_name, license_number, region_zone, facility_type, max_daily_capacity_ton, facility_status, compliance_rating, created_at FROM read_parquet('s3://dlh-bronze/staging/ref_facilities/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY facility_id ORDER BY created_at DESC) = 1"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-1812",
"type": "customNode",
"position": {
"x": 460,
"y": 80
},
"data": {
"label": "Extract Raw Waste Orders",
"type": "source_db",
"subtitle": "Postgres / MySQL / Oracle",
"isSource": true,
"configured": true,
"config": {
"conn": "Postgres",
"conn_id": "conn_postgres_source_db",
"table": "public.raw_waste_orders",
"table_name": "public.raw_waste_orders",
"target_type": "minio_s3"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-2810",
"type": "customNode",
"position": {
"x": 650,
"y": 80
},
"data": {
"label": "GE Orders Staging Quality Gate",
"type": "ge_quality",
"subtitle": "Great Expectations",
"configured": true,
"config": {
"ge_scope_level": "all",
"ge_table_name": "staging_waste_orders",
"test_table_row_count": true,
"test_table_schema_match": true,
"test_table_table_not_empty": true,
"table_min_rows": 1,
"table_max_rows": 1000000,
"ge_col_table": "stasiun_spku",
"ge_target_column": "stasiun_id",
"ge_column_test_type": "not_null",
"sql_assertion": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"sql": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"expected": 0,
"error_message": "Unexpected rows detected in table",
"suite": "bronze",
"fail": "Stop DAG"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-3793",
"type": "customNode",
"position": {
"x": 840,
"y": 80
},
"data": {
"label": "DuckDB Orders to Iceberg Bronze",
"type": "duckdb_transform",
"subtitle": "In-process SQL (<512MB)",
"configured": true,
"config": {
"layer": "silver",
"source_parquet_path": "s3://dlh-bronze/staging/waste_orders/*.parquet",
"source_path": "s3://dlh-bronze/staging/waste_orders/*.parquet",
"target_table": "bronze_waste_orders",
"table_name": "bronze_waste_orders",
"primary_keys": "order_id",
"query": "SELECT order_id, manifest_code, facility_id, generator_company, waste_category, volume_ton, tariff_per_ton, transport_fee, order_status, order_date, created_at FROM read_parquet('s3://dlh-bronze/staging/waste_orders/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY order_id ORDER BY created_at DESC) = 1",
"sql_query": "SELECT order_id, manifest_code, facility_id, generator_company, waste_category, volume_ton, tariff_per_ton, transport_fee, order_status, order_date, created_at FROM read_parquet('s3://dlh-bronze/staging/waste_orders/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY order_id ORDER BY created_at DESC) = 1",
"transformation_sql": "SELECT order_id, manifest_code, facility_id, generator_company, waste_category, volume_ton, tariff_per_ton, transport_fee, order_status, order_date, created_at FROM read_parquet('s3://dlh-bronze/staging/waste_orders/*.parquet') QUALIFY ROW_NUMBER() OVER (PARTITION BY order_id ORDER BY created_at DESC) = 1"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-4793",
"type": "customNode",
"position": {
"x": 1030,
"y": 80
},
"data": {
"label": "GE Orders Bronze Iceberg Quality",
"type": "ge_quality",
"subtitle": "Great Expectations",
"configured": true,
"config": {
"ge_scope_level": "all",
"ge_table_name": "bronze_waste_orders",
"test_table_row_count": true,
"test_table_schema_match": true,
"test_table_table_not_empty": true,
"table_min_rows": 1,
"table_max_rows": 1000000,
"ge_col_table": "stasiun_spku",
"ge_target_column": "stasiun_id",
"ge_column_test_type": "not_null",
"sql_assertion": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"sql": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"expected": 0,
"error_message": "Unexpected rows detected in table",
"suite": "bronze",
"fail": "Stop DAG"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-5809",
"type": "customNode",
"position": {
"x": 1220,
"y": 80
},
"data": {
"label": "dbt Silver Waste Manifest Enriched",
"type": "dbt_transform",
"subtitle": "dbt-postgres",
"configured": true,
"config": {
"dbt_layer": "silver",
"dbt_model": "silver_waste_manifest_enriched",
"sel": "silver_waste_manifest_enriched",
"tgt": "postgres",
"trino_catalog": "iceberg",
"sql_query_preview": "-- dbt Silver Cleansed Manifest Orders Enriched with Master Facilities\nSELECT \n o.order_id,\n o.manifest_code,\n o.order_date,\n o.generator_company,\n o.waste_category,\n o.volume_ton,\n o.tariff_per_ton,\n o.transport_fee,\n ROUND((o.volume_ton * o.tariff_per_ton) + o.transport_fee, 2) AS total_order_amount,\n o.order_status,\n f.facility_id,\n f.facility_name,\n f.region_zone,\n f.facility_type,\n f.compliance_rating,\n f.facility_status,\n CASE \n WHEN f.facility_status = 'ACTIVE' AND o.volume_ton <= f.max_daily_capacity_ton THEN 'COMPLIANT'\n WHEN f.facility_status != 'ACTIVE' THEN 'UNLICENSED_FACILITY_WARNING'\n ELSE 'CAPACITY_EXCEEDED_WARNING'\n END AS environmental_compliance_status\nFROM {{ ref('bronze_waste_orders') }} o\nLEFT JOIN {{ ref('bronze_ref_facilities') }} f\n ON o.facility_id = f.facility_id\nWHERE o.order_status != 'CANCELLED'",
"query": "-- dbt Silver Cleansed Manifest Orders Enriched with Master Facilities\nSELECT \n o.order_id,\n o.manifest_code,\n o.order_date,\n o.generator_company,\n o.waste_category,\n o.volume_ton,\n o.tariff_per_ton,\n o.transport_fee,\n ROUND((o.volume_ton * o.tariff_per_ton) + o.transport_fee, 2) AS total_order_amount,\n o.order_status,\n f.facility_id,\n f.facility_name,\n f.region_zone,\n f.facility_type,\n f.compliance_rating,\n f.facility_status,\n CASE \n WHEN f.facility_status = 'ACTIVE' AND o.volume_ton <= f.max_daily_capacity_ton THEN 'COMPLIANT'\n WHEN f.facility_status != 'ACTIVE' THEN 'UNLICENSED_FACILITY_WARNING'\n ELSE 'CAPACITY_EXCEEDED_WARNING'\n END AS environmental_compliance_status\nFROM {{ ref('bronze_waste_orders') }} o\nLEFT JOIN {{ ref('bronze_ref_facilities') }} f\n ON o.facility_id = f.facility_id\nWHERE o.order_status != 'CANCELLED'"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-6793",
"type": "customNode",
"position": {
"x": 1410,
"y": 80
},
"data": {
"label": "GE Silver Compliance Quality Gate",
"type": "ge_quality",
"subtitle": "Great Expectations",
"configured": true,
"config": {
"ge_scope_level": "all",
"ge_table_name": "silver_waste_manifest_enriched",
"test_table_row_count": true,
"test_table_schema_match": true,
"test_table_table_not_empty": true,
"table_min_rows": 1,
"table_max_rows": 1000000,
"ge_col_table": "stasiun_spku",
"ge_target_column": "stasiun_id",
"ge_column_test_type": "not_null",
"sql_assertion": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"sql": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"expected": 0,
"error_message": "Unexpected rows detected in table",
"suite": "bronze",
"fail": "Stop DAG"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-7793",
"type": "customNode",
"position": {
"x": 1600,
"y": 80
},
"data": {
"label": "dbt Gold Regional Waste Revenue Mart",
"type": "dbt_transform",
"subtitle": "dbt-postgres",
"configured": true,
"config": {
"dbt_layer": "gold",
"dbt_model": "gold_regional_waste_metrics",
"sel": "gold_regional_waste_metrics",
"tgt": "postgres",
"trino_catalog": "iceberg",
"sql_query_preview": "-- dbt Gold Regional Waste Revenue & Compliance Mart\nSELECT \n region_zone,\n facility_type,\n waste_category,\n COUNT(order_id) AS total_orders,\n COUNT(DISTINCT facility_id) AS active_facilities_count,\n COUNT(DISTINCT generator_company) AS active_generators_count,\n ROUND(SUM(volume_ton), 2) AS total_waste_volume_ton,\n ROUND(SUM(total_order_amount), 2) AS total_gross_revenue,\n ROUND(AVG(total_order_amount / NULLIF(volume_ton, 0)), 2) AS avg_revenue_per_ton,\n COUNT(CASE WHEN environmental_compliance_status = 'COMPLIANT' THEN 1 END) AS compliant_orders_count,\n CURRENT_TIMESTAMP AS mart_refreshed_at\nFROM {{ ref('silver_waste_manifest_enriched') }}\nGROUP BY region_zone, facility_type, waste_category\nORDER BY total_waste_volume_ton DESC",
"query": "-- dbt Gold Regional Waste Revenue & Compliance Mart\nSELECT \n region_zone,\n facility_type,\n waste_category,\n COUNT(order_id) AS total_orders,\n COUNT(DISTINCT facility_id) AS active_facilities_count,\n COUNT(DISTINCT generator_company) AS active_generators_count,\n ROUND(SUM(volume_ton), 2) AS total_waste_volume_ton,\n ROUND(SUM(total_order_amount), 2) AS total_gross_revenue,\n ROUND(AVG(total_order_amount / NULLIF(volume_ton, 0)), 2) AS avg_revenue_per_ton,\n COUNT(CASE WHEN environmental_compliance_status = 'COMPLIANT' THEN 1 END) AS compliant_orders_count,\n CURRENT_TIMESTAMP AS mart_refreshed_at\nFROM {{ ref('silver_waste_manifest_enriched') }}\nGROUP BY region_zone, facility_type, waste_category\nORDER BY total_waste_volume_ton DESC"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-8871",
"type": "customNode",
"position": {
"x": 1790,
"y": 80
},
"data": {
"label": "GE Gold Mart Integrity Gate",
"type": "ge_quality",
"subtitle": "Great Expectations",
"configured": true,
"config": {
"ge_scope_level": "all",
"ge_table_name": "gold_regional_waste_metrics",
"test_table_row_count": true,
"test_table_schema_match": true,
"test_table_table_not_empty": true,
"table_min_rows": 1,
"table_max_rows": 1000000,
"ge_col_table": "stasiun_spku",
"ge_target_column": "stasiun_id",
"ge_column_test_type": "not_null",
"sql_assertion": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"sql": "SELECT count(*) FROM silver.table WHERE status NOT IN ('APPROVED', 'IN_REVIEW', 'REJECTED')",
"expected": 0,
"error_message": "Unexpected rows detected in table",
"suite": "bronze",
"fail": "Stop DAG"
}
},
"width": 96,
"height": 96,
"selected": false,
"dragging": false
},
{
"id": "node-9855",
"type": "customNode",
"position": {
"x": 1980,
"y": 80
},
"data": {
"label": "SITOMAT Governance Callback",
"type": "sitomat_callback",
"subtitle": "Status callback",
"configured": true,
"config": {
"layer": "gold",
"artifact_ref": "lakehouse.gold.gold_regional_waste_metrics",
"checksum": "sha256-verified-waste-orders-pass",
"sitomat_base_url": "http://localhost:3000",
"openmetadata_fqn": "dlh_airflow_service.pipeline.task",
"ge_status": "PASSED",
"url": "/sitomat/callback",
"stage": "bronze"
}
},
"width": 96,
"height": 96,
"selected": true,
"dragging": false
}
],
"edges": [
{
"source": "node-9826",
"sourceHandle": null,
"target": "node-0827",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-9826-node-0827"
},
{
"source": "node-0827",
"sourceHandle": null,
"target": "node-5809",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-0827-node-5809"
},
{
"source": "node-1812",
"sourceHandle": null,
"target": "node-2810",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-1812-node-2810"
},
{
"source": "node-2810",
"sourceHandle": null,
"target": "node-3793",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-2810-node-3793"
},
{
"source": "node-3793",
"sourceHandle": null,
"target": "node-4793",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-3793-node-4793"
},
{
"source": "node-4793",
"sourceHandle": null,
"target": "node-5809",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-4793-node-5809"
},
{
"source": "node-5809",
"sourceHandle": null,
"target": "node-6793",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-5809-node-6793"
},
{
"source": "node-6793",
"sourceHandle": null,
"target": "node-7793",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-6793-node-7793"
},
{
"source": "node-7793",
"sourceHandle": null,
"target": "node-8871",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-7793-node-8871"
},
{
"source": "node-8871",
"sourceHandle": null,
"target": "node-9855",
"targetHandle": null,
"type": "removableEdge",
"style": {
"stroke": "var(--muted)",
"strokeWidth": 2
},
"data": {},
"id": "reactflow__edge-node-8871-node-9855"
}
],
"notes": [],
"is_draft": false,
"status": "deployed",
"updated_at": "2026-10-18T18:28:07.599927Z"
}
Reference in New Issue Copy Permalink