GTC-3137 Fix extract_level_gid to deal with misformats in GADM 4.1

app/routes/political/id_lookup.py

@@ -165,4 +165,14 @@ def form_admin_id_lookup_response(
 
 def extract_level_gid(gid_level: int, match):
     gid_level_name = f"gid_{gid_level}"
-    return (match[gid_level_name].rsplit("_")[0]).split(".")[gid_level]
+    gid_str = match[gid_level_name]
+
+    # Exception because of bad formatting of GHA gids in gadm_administrative_boundaries/v4.1
+    if gid_str.startswith("GHA") and not gid_str.startswith("GHA."):
+        gid_str = "GHA." + gid_str[3:]
+    # Exception because bad ids IDN.35.4, IDN.35.8, IDN.35.9, IDN.35.13, IDN.35.14
+    # (they are missing final '_1') in gadm_administrative_boundaries/v4.1
+    if gid_str.startswith("IDN") and not gid_str.endswith("_1"):
+        gid_str += "_1"
+
+    return (gid_str.rsplit("_")[0]).split(".")[gid_level]

tests_v2/unit/app/routes/political/id_lookup/test_id_lookup.py

@@ -6,7 +6,7 @@
 
 from app.models.pydantic.geostore import GeostoreCommon
 from app.routes.political import id_lookup
-from app.routes.political.id_lookup import _admin_boundary_lookup_sql, normalize_names
+from app.routes.political.id_lookup import _admin_boundary_lookup_sql, normalize_names, extract_level_gid
 
 ENDPOINT_UNDER_TEST = "/political/id-lookup"
 
@@ -279,6 +279,27 @@ async def test_id_lookup_matches_hide_extraneous(
     assert resp.status_code == 200
 
 
+@pytest.mark.asyncio
+async def test_extract_level_gid() -> None:
+    # Normal gid values
+    match1 = {"gid_0": "USA", "gid_1": "USA.5_1", "gid_2": "USA.5.10_1"}
+    assert extract_level_gid(0, match1) == "USA"
+    assert extract_level_gid(1, match1) == "5"
+    assert extract_level_gid(2, match1) == "10"
+
+    # Ghana values with bad formatting (missing dot after GHA in gadm 4.1)
+    match2 = {"gid_0": "GHA", "gid_1": "GHA7_2", "gid_2": "GHA7.1_2"}
+    assert extract_level_gid(0, match2) == "GHA"
+    assert extract_level_gid(1, match2) == "7"
+    assert extract_level_gid(2, match2) == "1"
+
+    # Indonesia values with bad formatting (missing suffix _1 in gadm 4.1)
+    match3 = {"gid_0": "IDN", "gid_1": "IDN.35_1", "gid_2": "IDN.35.4"}
+    assert extract_level_gid(0, match3) == "IDN"
+    assert extract_level_gid(1, match3) == "35"
+    assert extract_level_gid(2, match3) == "4"
+
+
 async def _query_dataset_json_mocked_no_results(
     dataset: str,
     version: str,