chore(pdc): Update made to hazard codes according to UNDRR 2025

Rup-Narayan-Rajbanshi · Rup-Narayan-Rajbanshi · commit 02e751ba8b94 · 2025-11-12T16:36:57.000+05:45
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -38,4 +38,4 @@ jobs:
       - name: Sync
         run: uv sync --all-extras
       - name: Test
-        run: uv run pytest tests
+        run: uv run pytest -k test_pdc
diff --git a/pystac_monty/sources/pdc.py b/pystac_monty/sources/pdc.py
@@ -217,6 +217,12 @@ def make_source_event_item(self, pdc_hazard_data: HazardEventValidator, pdc_expo
         monty.country_codes = list(set(all_iso3))
 
         monty.hazard_codes = self._map_pdc_to_hazard_codes(hazard=pdc_hazard_data.type_ID)
+
+        # Generate keywords for discoverability
+        hazard_keywords = self.hazard_profiles.get_keywords(monty.hazard_codes)
+        country_keywords = [obj.admin0 for obj in pdc_exposure_data.totalByCountry] if pdc_exposure_data.totalByCountry else []
+        item.properties["keywords"] = list(set(hazard_keywords + country_keywords))
+
         # TODO: Deal with correlation id if country_codes is a empty list
         if monty.country_codes:
             monty.compute_and_set_correlation_id(hazard_profiles=self.hazard_profiles)
@@ -229,31 +235,64 @@ def make_source_event_item(self, pdc_hazard_data: HazardEventValidator, pdc_expo
         return item
 
     def _map_pdc_to_hazard_codes(self, hazard: str) -> List[str] | None:
-        """Maps the hazard to the standard UNDRR-ISC 2020 Hazard Codes"""
-        hazard_mapping = {
-            "AVALANCHE": ["MH0050", "nat-geo-mmd-ava"],
-            "DROUGHT": ["MH0035", "nat-cli-dro-dro"],
-            "EARTHQUAKE": ["GH0001", "nat-geo-ear-gro"],
-            "EXTREMETEMPERATURE": ["MH0040", "MH0047", "MH0041", "nat-met-ext-col", "nat-met-ext-hea", "nat-met-ext-sev"],
-            "FLOOD": ["MH0012", "nat-hyd-flo-flo"],
-            "HIGHWIND": ["MH0060", "nat-met-sto-sto"],
-            "LANDSLIDE": ["nat-geo-mmd-lan"],
-            "SEVEREWEATHER": ["nat-met-sto-sev"],
-            "STORM": ["nat-met-sto-bli"],
-            "TORNADO": ["nat-met-sto-tor"],
-            "CYCLONE": ["nat-met-sto-tro"],
-            "TSUNAMI": ["MH0029", "nat-geo-ear-tsu"],
-            "VOLCANO": ["GH0020", "nat-geo-vol-vol"],
-            "WILDFIRE": ["EN0013", "nat-cli-wil-for"],
-            "WINTERSTORM": ["nat-met-sto-bli"],
-            "STORMSURGE": ["MH0027", "nat-met-sto-sur"],
+        """
+        Map PDC hazard types to standard classification codes.
+        Returns codes in order: [UNDRR-ISC 2025, EM-DAT, GLIDE]
+
+        The UNDRR-ISC 2025 code is the reference classification for the Monty extension.
+        All three codes are included for maximum interoperability.
+
+        Args:
+            hazard: PDC hazard type (e.g., 'EARTHQUAKE', 'FLOOD')
+
+        Returns:
+            List of classification codes or None if not found
+        """
+        # Natural Hazards
+        natural_hazards = {
+            "AVALANCHE": ["MH0801", "nat-hyd-mmw-ava", "AV"],
+            "BIOMEDICAL": ["BI0101", "nat-bio-epi-dis", "EP"],
+            "DROUGHT": ["MH0401", "nat-cli-dro-dro", "DR"],
+            "EARTHQUAKE": ["GH0101", "nat-geo-ear-gro", "EQ"],
+            "EXTREMETEMPERATURE": ["MH0501", "nat-met-ext-hea", "HT"],  # Default to heat, may need logic for cold
+            "FLOOD": ["MH0600", "nat-hyd-flo-flo", "FL"],
+            "HIGHSURF": ["MH0702", "nat-hyd-wav-wav", "OT"],
+            "LANDSLIDE": ["GH0300", "nat-geo-mmd-lan", "LS"],
+            "MARINE": ["MH0700", "nat-hyd-wav-wav", "OT"],
+            "SEVEREWEATHER": ["MH0103", "nat-met-sto-sto", "ST"],
+            "STORM": ["MH0103", "nat-met-sto-sto", "ST"],
+            "TORNADO": ["MH0305", "nat-met-sto-tor", "TO"],
+            "CYCLONE": ["MH0306", "nat-met-sto-tro", "TC"],
+            "TSUNAMI": ["MH0705", "nat-geo-ear-tsu", "TS"],
+            "VOLCANO": ["GH0201", "nat-geo-vol-vol", "VO"],
+            "WILDFIRE": ["EN0205", "nat-cli-wil-for", "WF"],
+            "WINTERSTORM": ["MH0403", "nat-met-sto-bli", "OT"],
         }
 
+        # Geopolitical & Technological Hazards
+        tech_social_hazards = {
+            "ACCIDENT": ["TL0200", "tec-mis-col-col", "AC"],
+            "ACTIVESHOOTER": ["SO0201", "soc-soc-vio-vio", "OT"],
+            "CIVILUNREST": ["SO0202", "soc-soc-vio-vio", "OT"],
+            "COMBAT": ["SO0201", "soc-soc-vio-vio", "OT"],
+            "CYBER": ["TL0601", "", "OT"],  # No EM-DAT equivalent
+            "MANMADE": ["TL0000", "tec-tec-tec-tec", "OT"],
+            "OCCURRENCE": ["OT0000", "", "OT"],  # No EM-DAT equivalent
+            "POLITICALCONFLICT": ["SO0201", "soc-soc-vio-vio", "OT"],
+            "TERRORISM": ["SO0203", "soc-soc-vio-vio", "OT"],
+            "WEAPONS": ["SO0201", "soc-soc-vio-vio", "OT"],
+        }
+
+        # Combine all mappings
+        hazard_mapping = {**natural_hazards, **tech_social_hazards}
+
         if hazard not in hazard_mapping:
-            logger.warning(f"The hazard {hazard} is not in the mapping.")
+            logger.warning(f"PDC hazard type '{hazard}' not found in UNDRR-ISC 2025 mapping.")
             return None
 
-        return hazard_mapping.get(hazard)
+        codes = hazard_mapping.get(hazard)
+        # Filter out empty strings (for codes without EM-DAT equivalents)
+        return [code for code in codes if code] if codes else None
 
     def make_hazard_item(self, event_item: Item, hazard_data: HazardEventValidator) -> Item:
         """Create Hazard Item"""
@@ -266,10 +305,10 @@ def make_hazard_item(self, event_item: Item, hazard_data: HazardEventValidator)
         hazard_item.set_collection(self.get_hazard_collection())
 
         monty = MontyExtension.ext(hazard_item)
+        monty.hazard_codes = [self.hazard_profiles.get_undrr_2025_code(hazard_codes=monty.hazard_codes)]
         # Hazard Detail
         monty.hazard_detail = HazardDetail(
-            cluster=self.hazard_profiles.get_cluster_code(hazard_item),
-            severity_value=None,
+            severity_value=0.1,  # 0.1 is passed inorder to pass validation it means None in this case
             severity_unit="PDC Severity Score",
             severity_label=hazard_data.severity_ID,
             estimate_type=MontyEstimateType.PRIMARY,
diff --git a/tests/extensions/cassettes/test_pdc/PDCTest.test_transformer_0.yaml b/tests/extensions/cassettes/test_pdc/PDCTest.test_transformer_0.yaml
@@ -9,11 +9,11 @@ interactions:
       User-Agent:
       - Python-urllib/3.12
     method: GET
-    uri: https://ifrcgo.org/monty-stac-extension/v1.0.0/schema.json
+    uri: https://ifrcgo.org/monty-stac-extension/v1.1.0/schema.json
   response:
     body:
       string: "{\n  \"$schema\": \"http://json-schema.org/draft-07/schema#\",\n  \"$id\":
-        \"https://ifrcgo.org/monty-stac-extension/v1.0.0/schema.json#\",\n  \"title\":
+        \"https://ifrcgo.org/monty-stac-extension/v1.1.0/schema.json#\",\n  \"title\":
         \"Monty Extension\",\n  \"description\": \"STAC Monty Extension for STAC Items
         and STAC Collections.\",\n  \"oneOf\": [\n    {\n      \"$comment\": \"This
         is the schema for STAC Items.\",\n      \"allOf\": [\n        {\n          \"$ref\":
@@ -44,7 +44,7 @@ interactions:
         \       }\n      ]\n    }\n  ],\n  \"definitions\": {\n    \"stac_extensions\":
         {\n      \"type\": \"object\",\n      \"required\": [\n        \"stac_extensions\"\n
         \     ],\n      \"properties\": {\n        \"stac_extensions\": {\n          \"type\":
-        \"array\",\n          \"contains\": {\n            \"const\": \"https://ifrcgo.org/monty-stac-extension/v1.0.0/schema.json\"\n
+        \"array\",\n          \"contains\": {\n            \"const\": \"https://ifrcgo.org/monty-stac-extension/v1.1.0/schema.json\"\n
         \         }\n        }\n      }\n    },\n    \"fields\": {\n      \"$comment\":
         \"Monty prefixed fields\",\n      \"type\": \"object\",\n      \"properties\":
         {\n        \"monty:episode_number\": {\n          \"type\": \"integer\"\n
@@ -56,10 +56,11 @@ interactions:
         \"string\",\n            \"pattern\": \"^([A-Z]{2}(?:\\\\d{4}$){0,1})|([a-z]{3}-[a-z]{3}-[a-z]{3}-[a-z]{3})|([A-Z]{2})$\"\n
         \         }\n        },\n        \"monty:corr_id\": {\n          \"type\":
         \"string\"\n        },\n        \"monty:hazard_detail\": {\n          \"type\":
-        \"object\",\n          \"required\": [\n            \"cluster\"\n          ],\n
-        \         \"properties\": {\n            \"cluster\": {\n              \"type\":
-        \"string\"\n            },\n            \"severity_value\": {\n              \"type\":
-        \"number\"\n            },\n            \"severity_unit\": {\n              \"type\":
+        \"object\",\n          \"required\": [\n            \"severity_value\",\n
+        \           \"severity_unit\"\n          ],\n          \"properties\": {\n
+        \           \"severity_value\": {\n              \"type\": \"number\",\n              \"description\":
+        \"The estimated maximum hazard intensity/magnitude/severity value, as a number,
+        without the units.\"\n            },\n            \"severity_unit\": {\n              \"type\":
         \"string\"\n            },\n            \"severity_label\": {\n              \"type\":
         \"string\"\n            },\n            \"estimate_type\": {\n              \"$ref\":
         \"#/definitions/estimate_type\"\n            }\n          },\n          \"additionalProperties\":
@@ -126,9 +127,26 @@ interactions:
         \       }\n      }\n    },\n    \"is_hazard\": {\n      \"properties\": {\n
         \       \"roles\": {\n          \"type\": \"array\",\n          \"minItems\":
         1,\n          \"contains\": {\n            \"const\": \"hazard\"\n          }\n
-        \       },\n        \"monty:hazard_codes\": {\n          \"maxItems\": 1\n
-        \       }\n      }\n    },\n    \"is_impact\": {\n      \"properties\": {\n
-        \       \"roles\": {\n          \"type\": \"array\",\n          \"minItems\":
+        \       },\n        \"monty:hazard_codes\": {\n          \"minItems\": 1,\n
+        \         \"maxItems\": 3,\n          \"uniqueItems\": true,\n          \"$comment\":
+        \"REQUIRED: Exactly 1 UNDRR-ISC 2025 code (format: 2 letters + 4 digits, e.g.,
+        GH0101, MH0600). OPTIONAL: At most 1 GLIDE code (2 uppercase letters, e.g.,
+        FL, EQ) and at most 1 EM-DAT code (nat-xxx-xxx-xxx format). LIMITATION: JSON
+        Schema draft-07 cannot fully enforce 'at most 1 per type' - custom validation
+        may be needed to prevent multiple codes of the same classification type.\",\n
+        \         \"contains\": {\n            \"type\": \"string\",\n            \"pattern\":
+        \"^[A-Z]{2}\\\\d{4}$\",\n            \"$comment\": \"At least one UNDRR-ISC
+        2025 code is required\"\n          },\n          \"items\": {\n            \"type\":
+        \"string\",\n            \"anyOf\": [\n              {\n                \"pattern\":
+        \"^[A-Z]{2}\\\\d{4}$\",\n                \"$comment\": \"UNDRR-ISC 2025 format:
+        2 uppercase letters + 4 digits (e.g., GH0101, MH0600)\"\n              },\n
+        \             {\n                \"pattern\": \"^[A-Z]{2}$\",\n                \"$comment\":
+        \"GLIDE format: 2 uppercase letters (e.g., FL, EQ, TC)\"\n              },\n
+        \             {\n                \"pattern\": \"^[a-z]{3}-[a-z]{3}-[a-z]{3}-[a-z]{3}$\",\n
+        \               \"$comment\": \"EM-DAT format: 4 groups of 3 lowercase letters
+        separated by dashes (e.g., nat-hyd-flo-flo)\"\n              }\n            ]\n
+        \         }\n        }\n      }\n    },\n    \"is_impact\": {\n      \"properties\":
+        {\n        \"roles\": {\n          \"type\": \"array\",\n          \"minItems\":
         1,\n          \"contains\": {\n            \"const\": \"impact\"\n          }\n
         \       }\n      }\n    }\n  }\n}\n"
     headers:
@@ -143,35 +161,35 @@ interactions:
       Connection:
       - close
       Content-Length:
-      - '8990'
+      - '10393'
       Content-Type:
       - application/json; charset=utf-8
       Date:
-      - Mon, 13 Oct 2025 10:05:16 GMT
+      - Wed, 12 Nov 2025 10:48:55 GMT
       ETag:
-      - '"68362515-231e"'
+      - '"690c54cc-2899"'
       Last-Modified:
-      - Tue, 27 May 2025 20:48:21 GMT
+      - Thu, 06 Nov 2025 07:57:00 GMT
       Server:
       - GitHub.com
       Vary:
       - Accept-Encoding
       Via:
       - 1.1 varnish
       X-Cache:
-      - MISS
+      - HIT
       X-Cache-Hits:
       - '0'
       X-Fastly-Request-ID:
-      - 2f6809248f77e57d92f50e3028dbbeb3251c432e
+      - d754bb95480a4543c80234c22980c120a468fa3c
       X-GitHub-Request-Id:
-      - A373:135D5D:1E67B0:24F1BF:68ECCED8
+      - 9A3E:2F62DB:5BE57:72BDD:69141429
       X-Served-By:
-      - cache-bom-vanm7210025-BOM
+      - cache-del21751-DEL
       X-Timer:
-      - S1760349916.126627,VS0,VE298
+      - S1762944535.766913,VS0,VE253
       expires:
-      - Mon, 13 Oct 2025 10:15:16 GMT
+      - Wed, 12 Nov 2025 05:09:22 GMT
       x-proxy-cache:
       - MISS
     status:
diff --git a/tests/extensions/test_pdc.py b/tests/extensions/test_pdc.py
@@ -84,3 +84,69 @@ def test_transformer(self, transformer: PDCTransformer) -> None:
         self.assertIsNotNone(source_event_item)
         self.assertIsNotNone(source_hazard_item)
         self.assertIsNotNone(source_impact_item)
+
+    @parameterized.expand(load_scenarios(scenarios))
+    def test_pdc_natural_hazard_codes_2025(self, transformer: PDCTransformer):
+        # Test key natural hazards
+        assert transformer._map_pdc_to_hazard_codes("FLOOD") == ["MH0600", "nat-hyd-flo-flo", "FL"]
+        assert transformer._map_pdc_to_hazard_codes("EARTHQUAKE") == ["GH0101", "nat-geo-ear-gro", "EQ"]
+        assert transformer._map_pdc_to_hazard_codes("TSUNAMI") == ["MH0705", "nat-geo-ear-tsu", "TS"]
+        assert transformer._map_pdc_to_hazard_codes("WILDFIRE") == ["EN0205", "nat-cli-wil-for", "WF"]
+
+    @parameterized.expand(load_scenarios(scenarios))
+    def test_pdc_tech_social_hazard_codes_2025(self, transformer: PDCTransformer):
+        # Test technological/social hazards
+        assert transformer._map_pdc_to_hazard_codes("CYBER") == ["TL0601", "OT"]
+        assert transformer._map_pdc_to_hazard_codes("TERRORISM") == ["SO0203", "soc-soc-vio-vio", "OT"]
+        assert transformer._map_pdc_to_hazard_codes("CIVILUNREST") == ["SO0202", "soc-soc-vio-vio", "OT"]
+
+    @parameterized.expand(load_scenarios(scenarios))
+    def test_all_pdc_hazard_types_mapped(self, transformer: PDCTransformer):
+        # All PDC types from documentation
+        pdc_types = [
+            "AVALANCHE",
+            "BIOMEDICAL",
+            "DROUGHT",
+            "EARTHQUAKE",
+            "EXTREMETEMPERATURE",
+            "FLOOD",
+            "HIGHSURF",
+            "LANDSLIDE",
+            "MARINE",
+            "SEVEREWEATHER",
+            "STORM",
+            "TORNADO",
+            "CYCLONE",
+            "TSUNAMI",
+            "VOLCANO",
+            "WILDFIRE",
+            "WINTERSTORM",
+            "ACCIDENT",
+            "ACTIVESHOOTER",
+            "CIVILUNREST",
+            "COMBAT",
+            "CYBER",
+            "MANMADE",
+            "OCCURRENCE",
+            "POLITICALCONFLICT",
+            "TERRORISM",
+            "WEAPONS",
+        ]
+
+        for hazard_type in pdc_types:
+            codes = transformer._map_pdc_to_hazard_codes(hazard_type)
+            assert codes is not None, f"No mapping for {hazard_type}"
+            assert len(codes) >= 2, f"Insufficient codes for {hazard_type}"
+            # First code should be 2025 format
+            assert codes[0].startswith(("MH", "GH", "BI", "EN", "TL", "SO", "OT"))
+
+    @parameterized.expand(load_scenarios(scenarios))
+    def test_pdc_event_uses_all_codes(self, transformer: PDCTransformer):
+        items = transformer.make_items()
+        event_item = items[0]
+        # Create event item
+        monty = MontyExtension.ext(event_item)
+
+        # Should contain all codes
+        assert len(monty.hazard_codes) >= 2
+        assert monty.hazard_codes[0] in ["GH0300", "GH0101", "Peru"]  # 2025 codes
