Snowflake의 dbt

dbt/nyc_taxi_dbt/
├── dbt_project.yml          # Project config: name, folder layout, materialization defaults
├── profiles.yml.example     # Connection config template (copy to ~/.dbt/profiles.yml)
├── packages.yml             # Third-party dbt packages
│
├── macros/
│   ├── generate_schema_name.sql   # Overrides dbt's default schema naming logic
│   └── generate_surrogate_key.sql # Wrapper for consistent surrogate key generation
│
└── models/
    ├── sources.yml          # Declares RAW.TRIPS_RAW as an external source
    │
    ├── staging/             # Layer 1: clean and rename raw columns
    │   ├── schema.yml       # Column-level tests for staging models
    │   ├── stg_trips.sql
    │   └── stg_taxi_zones.sql
    │
    ├── intermediate/        # Layer 2: joins and enrichment (no physical table)
    │   └── int_trips_enriched.sql
    │
    └── analytics/           # Layer 3: final tables consumed by dashboards
        ├── schema.yml
        ├── fact_trips.sql
        ├── agg_hourly_zone_trips.sql
        ├── dim_date.sql
        ├── dim_payment_type.sql
        ├── dim_taxi_zones.sql
        └── dim_vendor.sql

-- Snowflake: colon-path notation
TRIP_METADATA:driver.rating::FLOAT     AS driver_rating,
TRIP_METADATA:app.surge_multiplier::FLOAT AS surge_multiplier

-- dbt_project.yml
intermediate:
  +materialized: ephemeral   # compiled inline, no CREATE TABLE

{% if is_incremental() %}
  WHERE pickup_at > (SELECT MAX(pickup_at) FROM {{ this }})
{% endif %}

{% if is_incremental() %}
  WHERE pickup_at >= DATEADD('hour', -2, CURRENT_TIMESTAMP())
{% endif %}

dbt run --full-refresh

MERGE INTO ANALYTICS.FACT_TRIPS AS target
USING (SELECT ...) AS source
ON target.trip_id = source.trip_id
WHEN MATCHED THEN UPDATE SET ...
WHEN NOT MATCHED THEN INSERT ...;

target schema = STAGING  +  custom schema = ANALYTICS  →  STAGING_ANALYTICS  (wrong)

{% macro generate_schema_name(custom_schema_name, node) -%}
  {%- if custom_schema_name is none -%}
    {{ target.schema | upper }}      -- no custom schema → use target schema
  {%- else -%}
    {{ custom_schema_name | upper }}  -- custom schema → use it directly
  {%- endif -%}
{%- endmacro %}

# dbt_project.yml
profile: 'nyc_taxi'

# ~/.dbt/profiles.yml
nyc_taxi:
  target: dev
  outputs:
    dev:
      type: snowflake
      account: "{{ env_var('SNOWFLAKE_ORG') }}-{{ env_var('SNOWFLAKE_ACCOUNT') }}"
      role: DBT_ROLE
      database: NYC_TAXI_DB
      warehouse: TRANSFORM_WH
      schema: STAGING       # ← this is the "target schema" / default schema
      threads: 4

- name: trip_id
  tests:
    - not_null
    - unique
- name: total_amount_usd
  tests:
    - dbt_expectations.expect_column_values_to_be_between:
        min_value: 0
        max_value: 1000

-- tests/assert_revenue_positive.sql
-- A passing test returns 0 rows
SELECT trip_id, total_amount_usd
FROM {{ ref('fact_trips') }}
WHERE total_amount_usd < 0

dbt test

packages:
  - package: dbt-labs/dbt_utils
    version: [">=1.0.0", "<2.0.0"]
  - package: calogica/dbt_expectations
    version: [">=0.10.0", "<1.0.0"]

dbt deps

RAW.TRIPS_RAW (source — not managed by dbt)
    └── stg_trips (view)
            └── int_trips_enriched (ephemeral)
                    ├── fact_trips (incremental table)
                    └── agg_hourly_zone_trips (incremental table)

ANALYTICS.DIM_TAXI_ZONES (seeded by SQL script)
    └── stg_taxi_zones (view)
            ├── int_trips_enriched
            └── dim_taxi_zones (table)

dbt_utils.date_spine
    └── dim_date (table)

terraform apply          → creates warehouses, database, schemas, roles
scripts/01_create_tables.sql → creates raw tables, seeds dimension data
scripts/02_seed_data.sql → loads 50M synthetic trip rows
dbt deps && dbt build    → transforms raw data into analytics-ready tables
scripts/03_create_streams_tasks.sql → creates CDC stream and scheduled task