match two JTF records

Signed-off-by: John Seekins <[email protected]>

Author: johnseekins

ice_scrapers/__init__.py

@@ -53,6 +53,10 @@
         "expanded_name": "Dedicated Intergovernmental Service Agreement",
         "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 Defense",
+        "description": "Military facility",
+    },
     "IGSA": {
         "expanded_name": "Intergovernmental Service Agreement",
         "description": "A publicly-owned facility operated by state/local government(s), or private contractors, in which ICE contracts for bed space via an Intergovernmental Service Agreement; or local jails used by ICE pursuant to Intergovernmental Service Agreements, which house both ICE and non-ICE detainees, typically county prisoners awaiting trial or serving short sentences, but sometimes also USMS prisoners.",

ice_scrapers/facilities_scraper.py

@@ -21,6 +21,19 @@
 base_scrape_url = "https://www.ice.gov/detention-facilities"
 
 
+def _special_facilities(facility: dict) -> dict:
+    match facility["name"]:
+        case "Naval Station Guantanamo Bay (JTF Camp Six and Migrant Ops Center Main A)":
+            facility["address"]["country"] = "Cuba"
+            facility["address"]["administrative_area"] = "FPO"
+            facility["address"]["locality"] = "FPO"
+            facility["address"]["postal_code"] = "34009"
+            facility["address"]["street"] = "AVENUE C PSC 1005 BOX 55"
+        case _:
+            pass
+    return facility
+
+
 def scrape_facilities(facilities_data: dict) -> dict:
     """Scrape all ICE detention facility data from all discovered pages"""
     start_time = time.time()
@@ -37,6 +50,7 @@ def scrape_facilities(facilities_data: dict) -> dict:
         logger.debug("Found %s facilities on page %s", len(facilities), page_num + 1)
         time.sleep(1)  # Be respectful to the server
         for facility in facilities:
+            facility = _special_facilities(facility)
             addr = facility["address"]
             street, cleaned = repair_street(addr["street"], addr["locality"])
             if cleaned:

ice_scrapers/spreadsheet_load.py

@@ -72,6 +72,17 @@ def _download_sheet(keep_sheet: bool = True) -> Tuple[polars.DataFrame, str]:
     return df, actual_link
 
 
+def _special_rows(row: dict) -> dict:
+    match row["name"]:
+        case "JTF CAMP SIX":
+            row["address"]["country"] = "Cuba"
+            row["address"]["administrative_area"] = "FPO"
+            row["name"] = "Naval Station Guantanamo Bay (JTF Camp Six and Migrant Ops Center Main A)"
+        case _:
+            pass
+    return row
+
+
 def load_sheet(keep_sheet: bool = True) -> dict:
     df, sheet_url = _download_sheet(keep_sheet)
     """Convert the detentionstats sheet data into something we can update our facilities with"""
@@ -94,12 +105,20 @@ def load_sheet(keep_sheet: bool = True) -> dict:
         locality, cleaned = repair_locality(row["City"], row["State"])
         if cleaned:
             details["_repaired_record"] = True
-        full_address = ",".join([street, locality, row["State"], zcode]).upper()
         details["address"]["administrative_area"] = row["State"]
         details["address"]["locality"] = locality
         details["address"]["postal_code"] = zcode
         details["address"]["street"] = street
         details["name"] = row["Name"]
+        details = _special_rows(details)
+        full_address = ",".join(
+            [
+                details["address"]["street"],
+                details["address"]["locality"],
+                details["address"]["administrative_area"],
+                details["address"]["postal_code"],
+            ]
+        ).upper()
 
         """
         population statistics