fix to use latest dca_operatingbusinesses columns

NYCPlanning · Nov 7, 2024 · c1aa3cd · c1aa3cd
1 parent 663deb8
commit c1aa3cd
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 20 deletions.
diff --git a/products/facilities/facdb/pipelines.py b/products/facilities/facdb/pipelines.py
@@ -52,12 +52,10 @@ def dca_operatingbusinesses(df: pd.DataFrame):
     ]
     today = datetime.datetime.today()
     covid_freeze = datetime.datetime.strptime("03/12/2020", "%m/%d/%Y")
-    df.license_expiration_date = pd.to_datetime(
-        df["license_expiration_date"], format="%m/%d/%Y"
-    )
+    df.expiration_date = pd.to_datetime(df["expiration_date"], format="%m/%d/%Y")
     # fmt:off
-    df = df.loc[((df.license_expiration_date >= today) & (df.industry == "Scrap Metal Processor"))|((df.license_expiration_date >= covid_freeze) & (df.industry != "Scrap Metal Processor")), :]\
-        .loc[df.industry.isin(industry), :]
+    df = df.loc[((df.expiration_date >= today) & (df.business_category == "Scrap Metal Processor"))|((df.expiration_date >= covid_freeze) & (df.business_category != "Scrap Metal Processor")), :]\
+        .loc[df.business_category.isin(industry), :]
     # fmt:on
     df = sanitize_df(df)
     df = FunctionBN(bin_field="bin").geocode_a_dataframe(df)

diff --git a/products/facilities/facdb/sql/pipelines/dca_operatingbusinesses.sql b/products/facilities/facdb/sql/pipelines/dca_operatingbusinesses.sql
@@ -4,29 +4,29 @@ SELECT
     uid,
     source,
     initcap(business_name) AS facname,
-    address_building AS addressnum,
-    address_street_name AS streetname,
-    address_building || ' ' || address_street_name AS address,
-    address_city AS city,
-    address_zip AS zipcode,
-    address_borough AS boro,
-    borough_code AS borocode,
+    building_number AS addressnum,
+    street1 AS streetname,
+    building_number || ' ' || street1 AS address,
+    city,
+    zip_code AS zipcode,
+    borough AS boro,
+    NULL AS borocode,
     bin,
     bbl,
     (
         CASE
-            WHEN industry LIKE '%Scrap Metal%' THEN 'Scrap Metal Processing'
-            WHEN industry LIKE '%Tow%' THEN 'Tow Truck Company'
-            ELSE concat('Commercial ', industry)
+            WHEN business_category LIKE '%Scrap Metal%' THEN 'Scrap Metal Processing'
+            WHEN business_category LIKE '%Tow%' THEN 'Tow Truck Company'
+            ELSE concat('Commercial ', business_category)
         END
     ) AS factype,
     (
         CASE
-            WHEN industry = 'Scrap Metal Processor' THEN 'Solid Waste Processing'
-            WHEN industry = 'Parking Lot' THEN 'Parking Lots and Garages'
-            WHEN industry = 'Garage' THEN 'Parking Lots and Garages'
-            WHEN industry = 'Garage and Parking Lot' THEN 'Parking Lots and Garages'
-            WHEN industry = 'Tow Truck Company' THEN 'Parking Lots and Garages'
+            WHEN business_category = 'Scrap Metal Processor' THEN 'Solid Waste Processing'
+            WHEN business_category = 'Parking Lot' THEN 'Parking Lots and Garages'
+            WHEN business_category = 'Garage' THEN 'Parking Lots and Garages'
+            WHEN business_category = 'Garage and Parking Lot' THEN 'Parking Lots and Garages'
+            WHEN business_category = 'Tow Truck Company' THEN 'Parking Lots and Garages'
         END
     ) AS facsubgrp,
     initcap(business_name) AS opname,