more vera matching, skip enrichment when it's only vera

Signed-off-by: John Seekins <john@robot-house.us>

Author: johnseekins

enrichers/general.py

@@ -36,6 +36,9 @@ def _enrich_facility(facility_data: tuple) -> tuple:
     """enrich a single facility"""
     facility_id, facility = facility_data
     facility_name = facility["name"]
+    if len(facility["source_urls"]) == 1 and "vera-institute/ice-detention-trends" in facility["source_urls"][0]:
+        logger.debug("  Skipping enrichment of facility with only vera.org data: %s", facility["name"])
+        return facility_id, facility
     logger.info("Enriching facility %s...", facility_name)
     enriched_facility = copy.deepcopy(facility)
 

ice_scrapers/general.py

@@ -14,8 +14,8 @@ def facilities_scrape_wrapper(keep_sheet: bool = True, force_download: bool = Tr
     facilities_data = copy.deepcopy(facilities_schema)
     facilities = load_sheet(keep_sheet, force_download)
     facilities_data["facilities"] = copy.deepcopy(facilities)
-    facilities_data = collect_vera_facility_data(facilities_data, keep_sheet, force_download)
     facilities_data = scrape_facilities(facilities_data)
+    facilities_data = collect_vera_facility_data(facilities_data, keep_sheet, force_download)
     field_offices = scrape_field_offices()
     facilities_data = merge_field_offices(facilities_data, field_offices)
     facilities_data = insert_additional_facilities(facilities_data)

ice_scrapers/utils.py

@@ -195,6 +195,8 @@ def repair_name(name: str, locality: str) -> Tuple[str, bool]:
             "locality": "HURRICANE",
         },
         {"match": "ETOWAH COUNTY JAIL (ALABAMA)", "replace": "ETOWAH COUNTY JAIL", "locality": "GADSDEN"},
+        {"match": "BURLEIGH COUNTY", "replace": "BURLEIGH COUNTY JAIL", "locality": "BISMARCK"},
+        {"match": "NELSON COLEMAN CORRECTION", "replace": "NELSON COLEMAN CORRECTIONS CENTER", "locality": "KILLONA"},
     ]
     cleaned = False
     for m in matches:

ice_scrapers/vera_data.py

@@ -178,6 +178,12 @@ def _vera_name_fixes(name: str, city: str) -> Tuple[str, bool]:
             "city": "Bay St. Louis",
         },
         {"match": "Brooks County Jail (Contract)", "replace": "Brooks County Jail", "city": "Falfurrias"},
+        {"match": "Burleigh Co. Jail, ND", "replace": "Burleigh County Jail", "city": "Bismarck"},
+        {"match": "Lubbock County Jail", "replace": "Lubbock County Detention Center", "city": "Lubbock"},
+        {"match": "Montgomery County Jail", "replace": "Montgomery Ice Processing Center", "city": "Conroe"},
+        {"match": "Sebastian County Det Cnt", "replace": "Sebastian County Detention Center", "city": "Smith"},
+        {"match": "Atlanta U.S. Pen.", "replace": "FCI Atlanta", "city": "Atlanta"},
+        {"match": "Clinton County Corr. Fac.", "replace": "Clinton County Correctional Facility", "city": "McElhattan"},
     ]
     fixed = False
     for m in matches: