fix: generate Data360 verification URLs for non-WDI databases to prevent broken classic API links

Author: reichaves

lib/wb_client/wdi_client.py

@@ -450,12 +450,21 @@ def _clean(text: str | None) -> str | None:
                     f"?DATABASE_ID={_wgi_db_id}&INDICATOR={wgi_ind_id}"
                     f"&REF_AREA=WLD&COMP_BREAKDOWN_1=WGI_EST"
                 )
-            elif indicator_code.startswith("WB_") or indicator_code.startswith("IMF_"):
-                parts = indicator_code.split("_")
-                classic_code = ".".join(parts[2:])
+            elif indicator_code in _GHG_CODES:
                 verifier_url = (
-                    f"https://api.worldbank.org/v2/indicator/{classic_code}?format=json"
+                    f"{DATA360_BASE_URL}/data360/data"
+                    f"?DATABASE_ID={meta_db_id}&INDICATOR={full_indicator_id}&REF_AREA=WLD"
                 )
+            elif indicator_code.startswith("WB_") or indicator_code.startswith("IMF_"):
+                parts = indicator_code.split("_")
+                if len(parts) >= 2 and parts[1] == "WDI":
+                    classic_code = ".".join(parts[2:])
+                    verifier_url = f"https://api.worldbank.org/v2/indicator/{classic_code}?format=json"
+                else:
+                    verifier_url = (
+                        f"{DATA360_BASE_URL}/data360/data"
+                        f"?DATABASE_ID={meta_db_id}&INDICATOR={full_indicator_id}&REF_AREA=WLD"
+                    )
             else:
                 verifier_url = f"https://api.worldbank.org/v2/indicator/{indicator_code}?format=json"
 
@@ -807,19 +816,31 @@ def fetch_time_series(
                 display_code = indicator_code  # keep CC.EST as the label
             elif indicator_code.startswith("WB_") or indicator_code.startswith("IMF_"):
                 parts = indicator_code.split("_")
-                seal_code = ".".join(parts[2:])
-                countries_str = ";".join(countries)
-                verifiable_url = (
-                    f"{CLASSIC_BASE_URL}/country/{countries_str}/indicator/{seal_code}"
-                    f"?format=json&per_page=500"
-                )
-                if start_year and end_year:
-                    verifiable_url += f"&date={start_year}:{end_year}"
-                elif start_year:
-                    verifiable_url += f"&date={start_year}:2030"
-                elif end_year:
-                    verifiable_url += f"&date=1960:{end_year}"
-                display_code = seal_code
+                if len(parts) >= 2 and parts[1] == "WDI":
+                    seal_code = ".".join(parts[2:])
+                    countries_str = ";".join(countries)
+                    verifiable_url = (
+                        f"{CLASSIC_BASE_URL}/country/{countries_str}/indicator/{seal_code}"
+                        f"?format=json&per_page=500"
+                    )
+                    if start_year and end_year:
+                        verifiable_url += f"&date={start_year}:{end_year}"
+                    elif start_year:
+                        verifiable_url += f"&date={start_year}:2030"
+                    elif end_year:
+                        verifiable_url += f"&date=1960:{end_year}"
+                    display_code = seal_code
+                else:
+                    # Non-WDI database (like WB_ESG, IMF_IFS, etc.) -> use Data360 GET URL
+                    verifiable_url = (
+                        f"{DATA360_BASE_URL}/data360/data?DATABASE_ID={database_id}"
+                        f"&INDICATOR={full_indicator_id}&REF_AREA={','.join(countries)}"
+                    )
+                    if start_year:
+                        verifiable_url += f"&timePeriodFrom={start_year}"
+                    if end_year:
+                        verifiable_url += f"&timePeriodTo={end_year}"
+                    display_code = indicator_code
             elif indicator_code in _GHG_CODES:
                 # GHG indicators: Classic WB API removed these in 2023; Data360 is authoritative.
                 # Use the already-computed database_id and full_indicator_id.

tests/test_data_api.py

@@ -438,6 +438,32 @@ def test_ghg_total_seal_url_uses_data360(self, responses):
         assert "data360api.worldbank.org" in result["retrieval_url"]
         assert "api.worldbank.org/v2" not in result["retrieval_url"]
 
+    def test_esg_seal_url_uses_data360(self, responses):
+        """Indicators in non-WDI databases like WB_ESG must use Data360 URL, not Classic API."""
+        responses.add(
+            responses_lib.GET,
+            "https://data360api.worldbank.org/data360/data",
+            json={
+                "count": 1,
+                "value": [
+                    {
+                        "REF_AREA": "IND",
+                        "TIME_PERIOD": "2020",
+                        "OBS_VALUE": "1.57",
+                        "INDICATOR": "WB_ESG_EN_ATM_CO2E_PC",
+                        "DATABASE_ID": "WB_ESG",
+                    }
+                ],
+            },
+            status=200,
+        )
+
+        result = get_time_series("WB_ESG_EN_ATM_CO2E_PC", ["IND"])
+
+        assert result["data_available"] is True
+        assert "data360api.worldbank.org" in result["retrieval_url"]
+        assert "api.worldbank.org/v2" not in result["retrieval_url"]
+
 
 class TestDataCoverageNote:
     """Bug 3 fix: data_coverage_note fires when max_data_year < current_year - 1."""