Add COLP usetypes mapping as dbt seed

.github/workflows/colp_build.yml

@@ -47,6 +47,10 @@ jobs:
         AWS_SECRET_ACCESS_KEY: "op://Data Engineering/DO_keys/AWS_SECRET_ACCESS_KEY"
         AWS_ACCESS_KEY_ID: "op://Data Engineering/DO_keys/AWS_ACCESS_KEY_ID"
         BUILD_ENGINE_SERVER: "op://Data Engineering/EDM_DATA/server_url"
+        BUILD_ENGINE_HOST: "op://Data Engineering/EDM_DATA/server"
+        BUILD_ENGINE_USER: "op://Data Engineering/EDM_DATA/username"
+        BUILD_ENGINE_PASSWORD: "op://Data Engineering/EDM_DATA/password"
+        BUILD_ENGINE_PORT: "op://Data Engineering/EDM_DATA/port"
 
     - name: Finish container setup ...
       working-directory: ./

products/colp/bash/02a_load_data.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+source bash/config.sh
+
+# Load seeds via dbt
+dbt seed --select usetype_mappings
+run_sql_file sql/load_modifications.sql
+run_sql_file sql/geo_inputs.sql

products/colp/bash/02b_python.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+source bash/config.sh
+
+python3 -m python.geocode
+python3 -m python.geo_qaqc

products/colp/bash/02_build.sh → products/colp/bash/02c_sql_build.sh

@@ -1,13 +1,6 @@
 #!/bin/bash
 source bash/config.sh
 
-run_sql_file sql/load_modifications.sql  
-
-run_sql_file sql/geo_inputs.sql
-
-python3 -m python.geocode
-python3 -m python.geo_qaqc
-
 run_sql_file sql/_procedures.sql
 run_sql_file sql/clean_parcelname.sql
 run_sql_file sql/create_colp.sql

products/colp/colp.sh

@@ -2,10 +2,17 @@
 
 case $1 in 
     dataloading ) ./bash/01_dataloading.sh ;;
-    build ) ./bash/02_build.sh ;;
+    build ) 
+        ./bash/02a_load_data.sh
+        ./bash/02b_python.sh
+        ./bash/02c_sql_build.sh
+        ;;
+    load_data ) ./bash/02a_load_data.sh ;;
+    python ) ./bash/02b_python.sh ;;
+    sql_build ) ./bash/02c_sql_build.sh ;;
     qaqc ) ./bash/03_qaqc.sh ;;
     export ) ./bash/04_export.sh ;;
     upload ) python3 -m dcpy.connectors.edm.publishing upload -p db-colp -a public-read ;;
     sql) sql $@ ;;
-    * ) echo "COMMAND \"$1\" is not found. (valid commands: dataloading|build|export|upload)" ;;
+    * ) echo "COMMAND \"$1\" is not found. (valid commands: dataloading|build|load_data|python|sql_build|qaqc|export|upload)" ;;
 esac

products/colp/dbt_project.yml

@@ -0,0 +1,5 @@
+name: "colp"
+
+profile: "dcp-de-postgres"
+
+seed-paths: ["seeds"]

products/colp/profiles.yml

@@ -0,0 +1,11 @@
+dcp-de-postgres:
+  target: dev
+  outputs:
+    dev:
+      type: postgres
+      host: "{{ env_var('BUILD_ENGINE_HOST') }}"
+      user: "{{ env_var('BUILD_ENGINE_USER') }}"
+      password: "{{ env_var('BUILD_ENGINE_PASSWORD') }}"
+      port: "{{ env_var('BUILD_ENGINE_PORT') | as_number }}"
+      dbname: "{{ env_var('BUILD_ENGINE_DB') }}"
+      schema: "{{ env_var('BUILD_ENGINE_SCHEMA') }}"

products/colp/seeds/_seeds.yml

@@ -0,0 +1,11 @@
+version: 2
+
+seeds:
+  - name: usetype_mappings
+    description: |
+      Mapping table for cleaning IPIS use type values.
+      Maps source use type names to standardized cleaned versions.
+    config:
+      column_types:
+        source: text
+        cleaned: text

products/colp/seeds/usetype_mappings.csv

@@ -0,0 +1,82 @@
+source,cleaned
+NATURAL AREA/ETC,NATURAL AREA/WETLAND/WILDLIFE REFUGE
+IN USE-RES STRUCTURE,IN USE-RESIDENTIAL STRUCTURE
+PLAYGRND/SPORTS AREA,PLAYGROUND/SPORTS AREA
+AGCY EXECUTVE OFFICE,AGENCY EXECUTIVE OFFICE
+AGCY OFFICE,AGENCY OFFICE
+NO USE-WFT (ANY AGY),NO USE-WATERFRONT (ANY AGENCY)
+HIGHWAY STRIP/ETC,MALL/TRIANGLE/HIGHWAY STRIP/PARK STRIP
+OTHER TRANS FACILITY,OTHER TRANSPORTATION FACILITY
+NO USE-NON RESIDENTIAL STRUCTURES,NO USE-NON RESIDENTIAL STRUCTURE
+NO USE-NON RES STRC,NO USE-NON RESIDENTIAL STRUCTURE
+COMBINED MAINT/STRG,COMBINED MAINTENANCE/STORAGE FACILITY
+MUNICPL PKNG-OUTDR,MUNICIPAL PARKING - OUTDOOR PUBLIC
+INDOOR PKNG GARAGE,INDOOR PARKING GARAGE
+UNDEVELOP OPEN SPACE,UNDEVELOPED OPEN SPACE
+OUTDOOR PKNG-AGENCY,OUTDOOR PARKING - AGENCY VEHICLES
+TRANSL HSG - FAMILY,TRANSITIONAL HOUSING - FAMILY
+RECREATIONL FACILITY,RECREATIONAL FACILITY
+NO USE-RES STRUCTURE,NO USE-RESIDENTIAL STRUCTURE
+WASTEWATER PUMPING,WASTEWATER PUMPING STATION
+OUTDR PKNG-EMPLOYEES,OUTDOOR PARKING - EMPLOYEES
+SPECIAL EDUC SCHOOL,SPECIAL EDUCATION SCHOOL
+OTHER PUB SAFETY FAC,OTHER PUBLIC SAFETY FACILITY
+OTHER CULTURAL FAC,OTHER CULTURAL FACILITY
+AMBUL HEALTH FACIL,AMBULATORY HEALTH FACILITY
+INDOOR MAINT/STORAGE,INDOOR MAINTENANCE/STORAGE/GARAGE FACILITY
+EARLY CHILDHD CENTER,EARLY CHILDHOOD CENTER
+INDOOR STORAGE(WHSE),INDOOR STORAGE (WAREHOUSE)
+POLLUTION CONTRL FAC,POLLUTION CONTROL FACILITY
+TRANSL HSG - ADULT,TRANSITIONAL HOUSING - ADULT
+VOCATION HIGH SCHOOL,VOCATIONAL HIGH SCHOOL
+WATER POLL CNTRL PLT,WATER POLLUTION CONTROL PLANT
+OUTDOOR STRG-EQUIP,OUTDOOR STORAGE - EQUIPMENT
+OUTDOOR PKNG-CLIENTS,OUTDOOR PARKING - CLIENTS
+OTHER HEALTH FACILTY,OTHER HEALTH FACILITY
+RECREATION CTR/GYM,RECREATION CENTER/GYMNASIUM
+OTH WATER SUPPLY FAC,OTHER WATER SUPPLY FACILITY
+OFC LEASED BY EDUC,OFFICE LEASED BY EDUC
+INDOOR STORAGE-EQUIP,INDOOR STORAGE - EQUIPMENT
+SECTION STATION,SANITATION SECTION STATION
+FACILITIES MAINT,FACILITIES MAINTENANCE
+OUTDOOR MAINT/STORGE,OUTDOOR MAINTENANCE/STORAGE FACILITY
+OTHER COLLEGE/UNIV,OTHER COLLEGE/UNIVERSITY
+OUTDOOR STORAGE-BULK,OUTDOOR STORAGE - BULK MATERIAL
+ALTERNTV HIGH SCHOOL,ALTERNATIVE HIGH SCHOOL
+OTHR SPC REC-OUTDOOR,OTHER SPECIAL RECREATIONAL FACILITY - OUTDOOR
+SOLID WASTE TSFR STN,SOLID WASTE TRANSFER STATION
+SPECIAL REC-OUTDOOR,SPECIAL RECREATIONAL FACILITY - OUTDOOR
+MUNICIPL PKNG-INDR,MUNICIPAL PARKING - INDOOR PUBLIC
+WATER SUPPLY FACIL,WATER SUPPLY FACILITY
+NEIGHBORHD SIT AREA,NEIGHBORHOOD SITTING AREA
+SPEC RECREATN-INDOOR,SPECIAL RECREATIONAL FACILITY - INDOOR
+SLUDGE DE-WATERING,SLUDGE DE-WATERING FACILITY
+PUB SAFETY FACILITY,PUBLIC SAFETY FACILITY
+OTH OUTDOOR STORAGE,OTHER OUTDOOR STORAGE
+TRANS/PUBLIC PARKING,TRANSPORTATION/PUBLIC PARKING FACILITY
+INDOOR STRG-PRINTED,INDOOR STORAGE - PRINTED MATERIAL/SUPPLIES
+EDUC SKILLS CENTER,EDUCATIONAL SKILLS CENTER
+OTHER POLL CNTRL FAC,OTHER POLLUTION CONTROL FACILITY
+OTH DETENTION FACIL,OTHER DETENTION FACILITY
+OTH OUTDR MAINT/STG,OTHER OUTDOOR MAINTENANCE/STORAGE FACILITY
+TRANSL HSG - MEN,TRANSITIONAL HOUSING - ADULT MEN
+NON-RES SOC SERV FAC,NON-RESIDENTIAL SOCIAL SERVICE FACILITY
+O/D STRG-AGY VEHICLE,OUTDOOR STORAGE - AGENCY VEHICLES
+INDOOR STRG-SUPPLIES,INDOOR STORAGE - SUPPLIES
+OTHER REC FACILITY,OTHER RECREATIONAL FACILITY
+OTH INDOOR MAINT/STG,OTHER INDOOR MAINTENANCE/STORAGE FACILITY
+INDOOR STORAGE-BULK,INDOOR STORAGE - BULK MATERIAL
+RES HEALTH CARE FAC,RESIDENTIAL HEALTH CARE FACILITY
+OTH SECURE DETENTION,OTHER SECURE DETENTION FACILITY
+SOCIAL SERVICE FACIL,SOCIAL SERVICE FACILITY
+TRAIN FACIL-OUTDOOR,TRAINING FACILITY - OUTDOOR
+TRAIN FAC-INDOOR,TRAINING FACILITY - INDOOR
+TRANSL HSG - WOMEN,TRANSITIONAL HOUSING - ADULT WOMEN
+STORMWATER PUMPING,STORMWATER PUMPING STATION
+RES FACIL - CHILDREN,RESIDENTIAL FACILITY FOR CHILDREN
+OTHER TESTING FACIL,OTHER TESTING FACILITY
+OTH WATER TRANS FAC,OTHER WATERFRONT TRANSPORTATION FACILITY
+CRIMINAL JUSTICE FAC,CRIMINAL JUSTICE FACILITY
+OTH SOLID WASTE FAC,OTHER SOLID WASTE FACILITY
+OUTDOOR TOW POUND,OUTDOOR PARKING - TOW POUND
+OTHER SPC REC-INDOOR,OTHER SPECIAL RECREATIONAL FACILITY - INDOOR

products/colp/sql/create_colp.sql

@@ -205,6 +205,39 @@ normed_name_merge AS (
         ON a._parcelname = b.old_name
 ),
 
+usetype_mapped AS (
+    SELECT
+        a.uid,
+        a.borough,
+        a.block,
+        a.lot,
+        a.bbl,
+        a.geo_bbl,
+        a.mapbbl,
+        a.hnum,
+        a._sname,
+        a._parcelname,
+        a.parcelname,
+        a.agency,
+        a._usecode,
+        -- Apply usetype mappings, keeping original if no mapping exists
+        coalesce(b.cleaned, a._usetype) AS _usetype,
+        a.ownership,
+        a.leased,
+        a.finalcom,
+        a.agreement,
+        a.xcoord,
+        a.ycoord,
+        a.latitude,
+        a.longitude,
+        a.geom,
+        a.cd,
+        a.sname
+    FROM normed_name_merge AS a
+    LEFT JOIN usetype_mappings AS b
+        ON a._usetype = b.source
+),
+
 categorized AS (
     SELECT
         a.*,
@@ -309,7 +342,7 @@ categorized AS (
                 a._usecode = '1410'
                 OR a._usecode = '1400' THEN 'PROPERTY WITH RESIDENTIAL USE'
         END) AS excatdesc
-    FROM normed_name_merge AS a
+    FROM usetype_mapped AS a
 )
 
 -- Reorder columns for output