wrapped up some conflicts in patches

Author: HongPong

ice_scrapers/__init__.py

@@ -39,7 +39,7 @@
         "description": "A publicly-owned facility operated by state/local government(s), or private contractors, in which ICE contracts to use all bed space via a Dedicated Intergovernmental Service Agreement; or facilities used by ICE pursuant to Intergovernmental Service Agreements, which house only ICE detainees – typically these are operated by private contractors pursuant to their agreements with local governments.",
     },
     "DOD": {
-        "expanded_name": "Department of Defence",
+        "expanded_name": "Department of Defense",
         "description": "Department of Defence facilities - Often Army bases",
     },
     "FAMILY": {
@@ -66,14 +66,14 @@
         "expanded_name": "Other",
         "description": "Facilities including but not limited to transportation-related facilities, hotels, and/or other facilities",
     },
-    "Unknown": {
-        "expanded_name": "Unknown",
-        "description": "A facility whose type could not be identified",
-    },
     "SPC": {
         "expanded_name": "Service Processing Center",
         "description": "A facility owned by the government and staffed by a combination of federal and contract employees.",
     },
+    "Unknown": {
+        "expanded_name": "Unknown",
+        "description": "A facility whose type could not be identified",
+    },
     "USMS": {
         "expanded_name": "United States Marshals Service",
         "description": "A facility primarily contracted with the USMS for housing of USMS detainees, in which ICE contracts with the USMS for bed space.",
@@ -134,9 +134,9 @@
 from .utils import (  # noqa: E402
     get_ice_scrape_pages,  # noqa: F401
     repair_locality,  # noqa: F401
-    repair_name,  # noqa: F401
     repair_street,  # noqa: F401
     repair_zip,  # noqa: F401
+    repair_name,  # noqa: F401
     special_facilities,  # noqa: F401
     update_facility,  # noqa: F401
 )

ice_scrapers/facilities_scraper.py

@@ -7,6 +7,7 @@
     repair_locality,
     repair_street,
     repair_zip,
+    repair_name,
     special_facilities,
     update_facility,
 )
@@ -28,6 +29,7 @@ def scrape_facilities(facilities_data: dict) -> dict:
     logger.info("Starting to scrape ICE.gov detention facilities...")
     facilities_data["scraped_date"] = datetime.datetime.now(datetime.UTC)
     urls = get_ice_scrape_pages(base_scrape_url)
+
     scraped_count = 0
     for page_num, url in enumerate(urls):
         logger.info("Scraping page %s/%s...", page_num + 1, len(urls))
@@ -53,7 +55,7 @@ def scrape_facilities(facilities_data: dict) -> dict:
             if cleaned:
                 addr["locality"] = locality
                 facility["_repaired_record"] = True
-            name, cleaned = repair_locality(facility["name"], addr["locality"])
+            name, cleaned = repair_name(facility["name"], addr["locality"])
             if cleaned:
                 facility["name"] = name
                 facility["_repaired_record"] = True
@@ -181,6 +183,7 @@ def _scrape_page(page_url: str) -> list:
             facilities.append(facility_data)
 
     logger.info("  Extracted %s facilities from page", len(facilities))
+
     return facilities
 
 
@@ -193,6 +196,7 @@ def _find_facility_patterns(container):
         r"([A-Z][^|]+(?:\|[^|]+)?)\s*([A-Z][^A-Z]*Field Office)",
         r"([^-]+)\s*-\s*([A-Z][^A-Z]*Field Office)",
     ]
+
     text_content = container.get_text()
 
     for pattern in facility_patterns:

ice_scrapers/spreadsheet_load.py

@@ -26,7 +26,6 @@
 SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__))
 base_xlsx_url = "https://www.ice.gov/detain/detention-management"
 filename = f"{SCRIPT_DIR}{os.sep}detentionstats.xlsx"
-
 # extracted ADP sheet header list 2025-09-07
 facility_sheet_header = [
     "Name",
@@ -198,5 +197,5 @@ def load_sheet(keep_sheet: bool = True, force_download: bool = True) -> dict:
         details["field_office"]["id"] = row["AOR"]
         details["address_str"] = full_address
         results[full_address] = details
-    logger.info("  Loaded %s facilties", len(results.keys()))
+    logger.info("  Loaded %s facilities", len(results.keys()))
     return results

ice_scrapers/utils.py

@@ -35,6 +35,67 @@ def special_facilities(facility: dict) -> dict:
     return facility
 
 
+def repair_name(name: str, locality: str) -> Tuple[str, bool]:
+    """Even facility names are occasionally bad"""
+    matches = [
+        {"match": "ALEXANDRIA STAGING FACILI", "replace": "Alexandria Staging Facility", "locality": "ALEXANDRIA"},
+        {"match": "ORANGE COUNTY JAIL (NY)", "replace": "ORANGE COUNTY JAIL", "locality": "GOSHEN"},
+        {"match": "NORTH LAKE CORRECTIONAL F", "replace": "NORTH LAKE CORRECTIONAL FACILITY", "locality": "BALDWIN"},
+        {"match": "PHELPS COUNTY JAIL (MO)", "replace": "Phelps County Jail", "locality": "ROLLA"},
+        {
+            "match": "PENNINGTON COUNTY JAIL (SOUTH DAKOTA)",
+            "replace": "PENNINGTON COUNTY JAIL",
+            "locality": "RAPID CITY",
+        },
+        {
+            "match": "CORR. CTR OF NORTHWEST OHIO",
+            "replace": "CORRECTIONS CENTER OF NORTHWEST OHIO",
+            "locality": "STRYKER",
+        },
+        {
+            "match": "FOLKSTON D RAY ICE PROCES",
+            "replace": "D. RAY JAMES CORRECTIONAL INSTITUTION",
+            "locality": "FOLKSTON",
+        },
+        {"match": "COLLIER COUNTY NAPLES JAIL CENTER", "replace": "COLLIER COUNTY JAIL", "locality": "NAPLES"},
+        {
+            "match": "IAH SECURE ADULT DETENTION FACILITY (POLK)",
+            "replace": "IAM SECURE ADULT DET. FACILITY",
+            "locality": "LIVINGSTON",
+        },
+        {"match": "CIMMARRON CORR FACILITY", "replace": "CIMMARRON CORRECTIONAL FACILITY", "locality": "CUSHING"},
+        {"match": "ORANGE COUNTY JAIL (FL)", "replace": "ORANGE COUNTY JAIL", "locality": "ORLANDO"},
+        {"match": "CLARK COUNTY JAIL (IN)", "replace": "CLARK COUNTY JAIL", "locality": "JEFFERSONVILLE"},
+        {"match": "PRINCE EDWARD COUNTY (FARMVILLE)", "replace": "ICA - FARMVILLE", "locality": "FARMVILLE"},
+        {"match": "PHELPS COUNTY JAIL (NE)", "replace": "PHELPS COUNTY JAIL", "locality": "HOLDREGE"},
+        {
+            "match": "WASHINGTON COUNTY JAIL (PURGATORY CORRECTIONAL FAC",
+            "replace": "WASHINGTON COUNTY JAIL",
+            "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"},
+        {
+            "match": "CIMMARRON CORR FACILITY",
+            "replace": "CIMARRON CORRECTIONAL FACILITY",
+            "locality": "CUSHING",
+        },
+        {
+            "match": "IAM SECURE ADULT DET. FACILITY",
+            "replace": "IAH SECURE ADULT DET. FACILITY",
+            "locality": "LIVINGSTON",
+        },
+    ]
+    cleaned = False
+    for m in matches:
+        if m["match"] == name and m["locality"] == locality:
+            name = m["replace"]
+            cleaned = True
+            break
+    return name, cleaned
+
+
 def repair_street(street: str, locality: str = "") -> Tuple[str, bool]:
     """Generally, we'll let the spreadsheet win arguments just to be consistent"""
     street_filters = [
@@ -156,67 +217,6 @@ def repair_street(street: str, locality: str = "") -> Tuple[str, bool]:
     return street, cleaned
 
 
-def repair_name(name: str, locality: str) -> Tuple[str, bool]:
-    """Even facility names are occasionally bad"""
-    matches = [
-        {"match": "ALEXANDRIA STAGING FACILI", "replace": "Alexandria Staging Facility", "locality": "ALEXANDRIA"},
-        {"match": "ORANGE COUNTY JAIL (NY)", "replace": "ORANGE COUNTY JAIL", "locality": "GOSHEN"},
-        {"match": "NORTH LAKE CORRECTIONAL F", "replace": "NORTH LAKE CORRECTIONAL FACILITY", "locality": "BALDWIN"},
-        {"match": "PHELPS COUNTY JAIL (MO)", "replace": "Phelps County Jail", "locality": "ROLLA"},
-        {
-            "match": "PENNINGTON COUNTY JAIL (SOUTH DAKOTA)",
-            "replace": "PENNINGTON COUNTY JAIL",
-            "locality": "RAPID CITY",
-        },
-        {
-            "match": "CORR. CTR OF NORTHWEST OHIO",
-            "replace": "CORRECTIONS CENTER OF NORTHWEST OHIO",
-            "locality": "STRYKER",
-        },
-        {
-            "match": "FOLKSTON D RAY ICE PROCES",
-            "replace": "D. RAY JAMES CORRECTIONAL INSTITUTION",
-            "locality": "FOLKSTON",
-        },
-        {"match": "COLLIER COUNTY NAPLES JAIL CENTER", "replace": "COLLIER COUNTY JAIL", "locality": "NAPLES"},
-        {
-            "match": "IAH SECURE ADULT DETENTION FACILITY (POLK)",
-            "replace": "IAM SECURE ADULT DET. FACILITY",
-            "locality": "LIVINGSTON",
-        },
-        {"match": "CIMMARRON CORR FACILITY", "replace": "CIMMARRON CORRECTIONAL FACILITY", "locality": "CUSHING"},
-        {"match": "ORANGE COUNTY JAIL (FL)", "replace": "ORANGE COUNTY JAIL", "locality": "ORLANDO"},
-        {"match": "CLARK COUNTY JAIL (IN)", "replace": "CLARK COUNTY JAIL", "locality": "JEFFERSONVILLE"},
-        {"match": "PRINCE EDWARD COUNTY (FARMVILLE)", "replace": "ICA - FARMVILLE", "locality": "FARMVILLE"},
-        {"match": "PHELPS COUNTY JAIL (NE)", "replace": "PHELPS COUNTY JAIL", "locality": "HOLDREGE"},
-        {
-            "match": "WASHINGTON COUNTY JAIL (PURGATORY CORRECTIONAL FAC",
-            "replace": "WASHINGTON COUNTY JAIL",
-            "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"},
-        {
-            "match": "CIMMARRON CORR FACILITY",
-            "replace": "CIMARRON CORRECTIONAL FACILITY",
-            "locality": "CUSHING",
-        },
-        {
-            "match": "IAM SECURE ADULT DET. FACILITY",
-            "replace": "IAH SECURE ADULT DET. FACILITY",
-            "locality": "LIVINGSTON",
-        },
-    ]
-    cleaned = False
-    for m in matches:
-        if m["match"] == name and m["locality"] == locality:
-            name = m["replace"]
-            cleaned = True
-            break
-    return name, cleaned
-
-
 def repair_zip(zip_code: int, locality: str) -> Tuple[str, bool]:
     """
     Excel does a cool thing where it strips leading 0s

schemas.py

@@ -118,10 +118,10 @@
 }
 
 # enrichment print details
-enrichment_print_schema = {
+enrichment_print_schema: dict = {
     "osm_found": 0,
     "wiki_found": 0,
     "wikidata_found": 0,
 }
 
-supported_output_types = ["csv", "json", "parquet", "xlsx"]
+supported_output_types: list = ["csv", "json", "parquet", "xlsx"]