Merge pull request #53 from johnseekins/jtf-fix match two JTF records

match two JTF records

Author: HongPong

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.",
@@ -127,6 +131,7 @@
     repair_locality,  # noqa: F401
     repair_street,  # noqa: F401
     repair_zip,  # noqa: F401
+    special_facilities,  # noqa: F401
     update_facility,  # noqa: F401
 )
 from .facilities_scraper import scrape_facilities  # noqa: F401,E402

ice_scrapers/facilities_scraper.py

@@ -7,6 +7,7 @@
     repair_locality,
     repair_street,
     repair_zip,
+    special_facilities,
     update_facility,
 )
 from schemas import facility_schema
@@ -37,6 +38,7 @@ 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

@@ -15,6 +15,7 @@
     repair_locality,
     repair_street,
     repair_zip,
+    special_facilities,
 )
 from typing import Tuple
 from utils import (
@@ -94,12 +95,20 @@ 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_facilities(details)
+        full_address = ",".join(
+            [
+                details["address"]["street"],
+                details["address"]["locality"],
+                details["address"]["administrative_area"],
+                details["address"]["postal_code"],
+            ]
+        ).upper()
 
         """
         population statistics

ice_scrapers/utils.py

@@ -7,6 +7,34 @@
 )
 
 
+def special_facilities(facility: dict) -> dict:
+    """
+    Some very specific facilities have unique fixes
+    that are hard to fit into our normal repair_* pattern.
+
+    Please don't expand this function unless it's necessary
+    """
+    match facility["name"]:
+        case "Naval Station Guantanamo Bay (JTF Camp Six and Migrant Ops Center Main A)":
+            """
+            First special case? JTF Camp Six is purely a mess.
+            While we work on getting a consistent address for this facility,
+            we'll need to make the two records converge.
+            """
+            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 "JTF CAMP SIX":
+            facility["address"]["country"] = "Cuba"
+            facility["address"]["administrative_area"] = "FPO"
+            facility["name"] = "Naval Station Guantanamo Bay (JTF Camp Six and Migrant Ops Center Main A)"
+        case _:
+            pass
+    return facility
+
+
 def repair_street(street: str, locality: str = "") -> Tuple[str, bool]:
     """Generally, we'll let the spreadsheet win arguments just to be consistent"""
     street_filters = [