Merge pull request #3675 from cal-itp/add-labels-to-models

Add labels to DBT models
cal-itp · Feb 5, 2025 · 36b9127 · 36b9127
2 parents d6a16c9 + 30595dc
commit 36b9127
Show file tree

Hide file tree

Showing 2 changed files with 179 additions and 19 deletions.
diff --git a/warehouse/dbt_project.yml b/warehouse/dbt_project.yml
@@ -39,42 +39,154 @@ models:
     staging:
       +materialized: view
       schema: staging
+      amplitude:
+        +labels:
+          domain: staging
+          dataset: benefits
+      audit:
+        +labels:
+          domain: staging
+          dataset: audit
+      gtfs:
+        +labels:
+          domain: staging
+          dataset: gtfs
+      gtfs_quality:
+        +labels:
+          domain: staging
+          dataset: gtfs_quality
+      ntd:
+        +labels:
+          domain: staging
+          dataset: ntd
+      ntd_annual_reporting:
+        +labels:
+          domain: staging
+          dataset: ntd_annual_reporting
+      ntd_ridership:
+        +labels:
+          domain: staging
+          dataset: ntd_ridership
+      ntd_safety_and_security:
+        +labels:
+          domain: staging
+          dataset: ntd_safety_and_security
+      ntd_validation:
+        +labels:
+          domain: staging
+          dataset: ntd_validation
+      payments:
+        +labels:
+          domain: staging
+          dataset: payments
+      rt:
+        +labels:
+          domain: staging
+          dataset: gtfs
+      state_geoportal:
+        +labels:
+          domain: staging
+          dataset: state_geoportal
+      transit_database:
+        +labels:
+          domain: staging
+          dataset: transit_database
 
     intermediate:
+      gtfs:
+        +labels:
+          domain: intermediate
+          dataset: gtfs
       gtfs_quality:
+        +labels:
+          domain: intermediate
+          dataset: gtfs_quality
         guidelines_checks:
           +materialized: table
+      ntd:
+        +labels:
+          domain: intermediate
+          dataset: ntd
+      ntd_validation:
+        +labels:
+          domain: intermediate
+          dataset: ntd_validation
+      payments:
+        +labels:
+          domain: intermediate
+          dataset: payments
+      transit_database:
+        +labels:
+          domain: intermediate
+          dataset: transit_database
 
     mart:
-      transit_database:
-        schema: mart_transit_database
-      transit_database_latest:
-        schema: mart_transit_database_latest
-      gtfs_quality:
-        schema: mart_gtfs_quality
+      audit:
+        +labels:
+          domain: mart
+          dataset: audit
+        schema: mart_audit
+      benefits:
+        +labels:
+          domain: mart
+          dataset: benefits
+        schema: mart_benefits
       gtfs:
+        +labels:
+          domain: mart
+          dataset: gtfs
         schema: mart_gtfs
+      gtfs_quality:
+        +labels:
+          domain: mart
+          dataset: gtfs_quality
+        schema: mart_gtfs_quality
       gtfs_schedule_latest:
+        +labels:
+          domain: mart
+          dataset: gtfs_schedule_latest
         schema: mart_gtfs_schedule_latest
-      ad_hoc:
-        schema: mart_ad_hoc
-      audit:
-        schema: mart_audit
       ntd:
+        +labels:
+          domain: mart
+          dataset: ntd
         schema: mart_ntd
-      payments:
-        +materialized: table
-        schema: mart_payments
-      benefits:
-        schema: mart_benefits
-      ntd_validation:
-        schema: mart_ntd_validation
       ntd_annual_reporting:
         +materialized: table
+        +labels:
+          domain: mart
+          dataset: ntd_annual_reporting
         schema: mart_ntd_annual_reporting
+      ntd_ridership:
+        +materialized: table
+        +labels:
+          domain: mart
+          dataset: ntd_ridership
+        schema: mart_ntd_ridership
       ntd_safety_and_security:
         +materialized: table
+        +labels:
+          domain: mart
+          dataset: ntd_safety_and_security
         schema: mart_ntd_safety_and_security
-      ntd_ridership:
+      ntd_validation:
+        +labels:
+          domain: mart
+          dataset: ntd_validation
+        schema: mart_ntd_validation
+      payments:
         +materialized: table
-        schema: mart_ntd_ridership
+        +labels:
+          domain: mart
+          dataset: payments
+        schema: mart_payments
+      transit_database:
+        +labels:
+          domain: mart
+          dataset: transit_database
+        schema: mart_transit_database
+      transit_database_latest:
+        +labels:
+          domain: mart
+          dataset: transit_database_latest
+        schema: mart_transit_database_latest
diff --git a/warehouse/seeds/_seeds.yml b/warehouse/seeds/_seeds.yml
@@ -4,6 +4,10 @@ seeds:
   - name: miles_traveled
     description: |
       A matrix of the distance between origin/destination pairs in miles.
+    config:
+      labels:
+        domain: seeds
+        dataset: payments
     columns:
       - name: location_name
         description: On location in the O/D pair
@@ -19,6 +23,10 @@ seeds:
   - name: ntd_modes_to_full_names
     description: |
       A list of ntd 2 letter mode codes and their full names
+    config:
+      labels:
+        domain: seeds
+        dataset: ntd
     columns:
       - name: ntd_mode_abbreviation
         description: The two letter abbreviation mode
@@ -28,6 +36,10 @@ seeds:
         description: The mode's full name
 
   - name: payments_entity_mapping
+    config:
+      labels:
+        domain: seeds
+        dataset: payments
     columns:
       - name: gtfs_dataset_source_record_id
         description: Unversioned key to dim_gtfs_datasets natural key from Airtable.
@@ -45,6 +57,10 @@ seeds:
     description: |
       A list of validation codes output by the GTFS RT validator, and their severities.
       Originally sourced from https://docs.google.com/spreadsheets/d/1GDDaDlsBPCYn3dtYPSABnce9ns3ekJ8Jzfgyy56lZz4/edit#gid=617612870.
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -56,6 +72,10 @@ seeds:
     description: |
       A list of validation codes output by the GTFS Schedule validator, and their severities.
       Originally sourced from https://docs.google.com/spreadsheets/d/1GDDaDlsBPCYn3dtYPSABnce9ns3ekJ8Jzfgyy56lZz4/edit#gid=0.
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: name
         tests:
@@ -67,6 +87,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v2.0.0 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/archive/refs/tags/v2.0.0.zip
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -87,6 +111,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v3.1.1 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/archive/refs/tags/v3.1.1.zip
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -107,6 +135,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v4.0.0 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/archive/refs/tags/v4.0.0.zip
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -127,6 +159,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v4.1.0 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/archive/refs/tags/v4.1.0.zip
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -147,6 +183,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v4.2.0 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/archive/refs/tags/v4.2.0.zip
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -167,6 +207,10 @@ seeds:
       A list of validation codes output by the GTFS Schedule validator, and their severities and descriptions.
       This data was manually parsed from the contents of the RULES.md file in the v5.0.0 release of the validator,
       sourced from: https://github.com/MobilityData/gtfs-validator/releases/tag/v5.0.0
+    config:
+      labels:
+        domain: seeds
+        dataset: gtfs_quality
     columns:
       - name: code
         tests:
@@ -200,6 +244,10 @@ seeds:
 
       There are also a few records in here that were manually added after looking at the 2021 NTD
       data and comparing that to see if any records were missing.
+    config:
+      labels:
+        domain: seeds
+        dataset: transit_database
     tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns: