Merge pull request #155 from GeoDaCenter/update-herop-ids

remove year suffix from HEROP_ID
GeoDaCenter · Mar 26, 2024 · 6d1eeb6 · 6d1eeb6
2 parents 65da78b + 15372c6
commit 6d1eeb6
Show file tree

Hide file tree

Showing 25 changed files with 399,275 additions and 399,278 deletions.
diff --git a/code/SpatialIds/add_csv_herop_ids.py b/code/SpatialIds/add_csv_herop_ids.py
@@ -71,8 +71,6 @@
 
     rows = []
     new_geoid_field = 'HEROP_ID'
-    # for the "latest" CSVs the year is 2018, for all older ones it's 2010
-    year_suffix = "2018" if "Latest" in str(path) else "2010"
     with open(path, "r") as r:
         reader = csv.DictReader(r)
 
@@ -82,7 +80,7 @@
 
         for r in reader:
             str_geoid = str(r[f_lookup[path.name]]).zfill(sl_lookup[geo]['id_length'])
-            new_geoid = f"{sl_lookup[geo]['code']}US{str_geoid}-{year_suffix}"
+            new_geoid = f"{sl_lookup[geo]['code']}US{str_geoid}"
             r[new_geoid_field] = new_geoid
             rows.append(r)
 

diff --git a/code/SpatialIds/add_shp_herop_ids.py b/code/SpatialIds/add_shp_herop_ids.py
@@ -40,9 +40,9 @@
     "county/counties2018.shp": "GEOID",
     "tract/tracts2018.shp": "GEOID",
     "zcta/zctas2018.shp": "GEOID10",
-    "tl_2010_state/states2010.shp": "STATEFP",
-    "tl_2010_county/counties2010.shp": "GEOID",
-    "tl_2010_tract/tracts2010.shp": "GEOID",
+    "state/states2010.shp": "STATEFP",
+    "county/counties2010.shp": "GEOID",
+    "tract/tracts2010.shp": "GEOID",
 }
 
 new_suffix = "bq"
@@ -68,11 +68,6 @@
         raise Exception("unexpected input shapefile")
 
     out_path = out_dir / Path(path).name
-    year = ''
-    if '2010' in str(out_path):
-        year = '2010'
-    if '2018' in str(out_path):
-        year = '2018'
     with fiona.open(shp_path) as src:
         dst_schema = src.schema
         dst_schema['properties']['HEROP_ID'] = 'str'
@@ -86,9 +81,13 @@
         ) as dst:
             for feat in src:
                 geo_id = str(feat.properties[f_lookup[path]]).zfill(sl['id_length'])
-                herop_id =  f"{sl['code']}US{geo_id}-{year}"
-                props = Properties.from_dict(
-                    **feat.properties,
-                    HEROP_ID=herop_id,
-                )
+                herop_id =  f"{sl['code']}US{geo_id}"
+                if "HEROP_ID" in feat.properties:
+                    feat.properties['HEROP_ID'] = herop_id
+                    props = feat.properties
+                else:
+                    props = Properties.from_dict(
+                        **feat.properties,
+                        HEROP_ID=herop_id,
+                    )
                 dst.write(Feature(geometry=feat.geometry, properties=props))
diff --git a/data_final/full_tables/C_1980.csv b/data_final/full_tables/C_1980.csv
diff --git a/data_final/full_tables/C_1990.csv b/data_final/full_tables/C_1990.csv
diff --git a/data_final/full_tables/C_2000.csv b/data_final/full_tables/C_2000.csv
diff --git a/data_final/full_tables/C_2010.csv b/data_final/full_tables/C_2010.csv
diff --git a/data_final/full_tables/C_Latest.csv b/data_final/full_tables/C_Latest.csv
diff --git a/data_final/full_tables/S_1980.csv b/data_final/full_tables/S_1980.csv
diff --git a/data_final/full_tables/S_1990.csv b/data_final/full_tables/S_1990.csv
diff --git a/data_final/full_tables/S_2000.csv b/data_final/full_tables/S_2000.csv
diff --git a/data_final/full_tables/S_2010.csv b/data_final/full_tables/S_2010.csv
diff --git a/data_final/full_tables/S_Latest.csv b/data_final/full_tables/S_Latest.csv
diff --git a/data_final/full_tables/T_1980.csv b/data_final/full_tables/T_1980.csv
diff --git a/data_final/full_tables/T_1990.csv b/data_final/full_tables/T_1990.csv
diff --git a/data_final/full_tables/T_2000.csv b/data_final/full_tables/T_2000.csv