Merge pull request #36 from IFRCGo/feat/src-validators

Feat/src validators

Author: tnagorra

Diff for: .pre-commit-config.yaml

@@ -50,6 +50,12 @@ repos:
       - id: ruff-format
         types_or: [python, pyi, jupyter, toml]
 
+  - repo: https://github.com/astral-sh/uv-pre-commit
+    rev: 0.6.9
+    hooks:
+      - id: uv-lock
+        args: ["--locked", "--offline"]
+
   - repo: https://github.com/RobertCraigie/pyright-python
     rev: v1.1.396
     hooks:

Diff for: pyproject.toml

@@ -33,7 +33,7 @@ dependencies = [
     "pytz>=2021.1",
     "pandas>=2.2.0",
     "lxml>=5.3.0",
-    "pydantic",
+    "pydantic>=2.10.6",
 ]
 dynamic = ["version"]
 

Diff for: pystac_monty/geocoding.py

@@ -501,7 +501,8 @@ def get_geometry_by_country_name(self, country_name: str) -> Optional[Dict[str,
             return None
 
     def get_iso3_from_geometry(self, geometry: Dict[str, Any]) -> Optional[str]:
-        raise NotImplementedError("Method not implemented")
+        # FIXME: Implement this later
+        return "UNK"
 
     def get_geometry_from_iso3(self, iso3: str) -> Optional[Dict[str, Any]]:
         raise NotImplementedError("Method not implemented")

Diff for: pystac_monty/sources/desinventar.py

@@ -24,107 +24,19 @@
 )
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataTransformer
-
-STAC_EVENT_ID_PREFIX = "desinventar-event-"
-STAC_HAZARD_ID_PREFIX = "desinventar-hazard-"
-STAC_IMPACT_ID_PREFIX = "desinventar-impact-"
+from pystac_monty.validators.desinventar import (
+    STAC_EVENT_ID_PREFIX,
+    STAC_HAZARD_ID_PREFIX,
+    STAC_IMPACT_ID_PREFIX,
+    DataRow,
+    GeoDataEntry,
+)
 
 logger = logging.getLogger(__name__)
 
 T = typing.TypeVar("T")
 
 
-class GeoDataEntry(TypedDict):
-    level: Optional[str]
-    property_code: Optional[str]
-    shapefile_data: Optional[gpd.GeoDataFrame]
-
-
-# Properties extracted from desinventar
-class DataRow(pydantic.BaseModel):
-    serial: str
-    comment: str | None
-    # source: str | None
-
-    deaths: float | None
-    injured: float | None
-    missing: float | None
-    houses_destroyed: float | None
-    houses_damaged: float | None
-    directly_affected: float | None
-    indirectly_affected: float | None
-    relocated: float | None
-    evacuated: float | None
-    losses_in_dollar: float | None
-    losses_local_currency: float | None
-    # education_centers: str | None
-    # hospitals: str | None
-    damages_in_crops_ha: float | None
-    lost_cattle: float | None
-    damages_in_roads_mts: float | None
-
-    level0: str | None
-    level1: str | None
-    level2: str | None
-    # name0: str | None
-    # name1: str | None
-    # name2: str | None
-    # latitude: str | None
-    # longitude: str | None
-
-    # haz_maxvalue: str | None
-    event: str | None
-    # glide: str | None
-    location: str | None
-
-    # duration: str | None
-    year: int
-    month: int | None
-    day: int | None
-
-    # Added fields
-
-    iso3: str
-    data_source_url: str | None
-
-    @property
-    def event_stac_id(self):
-        return f"{STAC_EVENT_ID_PREFIX}{self.iso3}-{self.serial}"
-
-    @property
-    def event_title(self):
-        return f"{self.event} in {self.location} on {self.event_start_date}"
-
-    @property
-    def event_description(self):
-        return f"{self.event} in {self.location}: {self.comment}"
-
-    @property
-    def event_start_date(self):
-        if self.year is None:
-            return
-
-        start_year = self.year
-        start_month = self.month or 1
-        start_day = self.day or 1
-
-        try:
-            start_dt = datetime(start_year, start_month, start_day)
-            return pytz.utc.localize(start_dt)
-        except Exception:
-            return None
-
-    @property
-    def lowest_level(self):
-        if self.level2 is not None:
-            return 'level2'
-        if self.level1 is not None:
-            return 'level1'
-        if self.level0 is not None:
-            return 'level0'
-        return None
-
-
 # TODO: move to common utils
 def get_list_item_safe(lst: list[T], index: int, default_value: T | None = None) -> T | None:
     try:
@@ -686,6 +598,8 @@ def get_stac_items(self) -> typing.Generator[Item, None, None]:
                         if event_item := self._create_event_item_from_row(row_data):
                             yield event_item
                             yield from self._create_impact_items_from_row(row_data, event_item)
+                        else:
+                            failed_items_count += 1
                 except Exception:
                     failed_items_count += 1
                     logger.error('Failed to process desinventar', exc_info=True)

Diff for: pystac_monty/sources/emdat.py

@@ -19,6 +19,7 @@
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataSource, MontyDataTransformer
 from pystac_monty.utils import rename_columns
+from pystac_monty.validators.em_dat import EmdatDataValidator
 
 STAC_EVENT_ID_PREFIX = "emdat-event-"
 STAC_HAZARD_ID_PREFIX = "emdat-hazard-"
@@ -40,12 +41,24 @@ def __init__(self, source_url: str, data: Union[str, pd.DataFrame]):
             self.df = data
         elif isinstance(data, dict):
             # If data is a dict, assume it's Json content
-            data = data["data"]["public_emdat"]["data"]
+            # data = data["data"]["public_emdat"]["data"]
+            data = self.source_data_validator(data["data"]["public_emdat"]["data"])
             df = pd.DataFrame(data)
             self.df = rename_columns(df)
         else:
             raise ValueError("Data must be either Excel content (str) or pandas DataFrame or Json")
 
+    def source_data_validator(self, data):
+        valid_list = []
+        error_list = []
+        for item in data:
+            if EmdatDataValidator.validate_event(item):
+                valid_list.append(item)
+            else:
+                error_list.append(item)
+
+        return valid_list
+
     def get_data(self) -> pd.DataFrame:
         return self.df
 

Diff for: pystac_monty/sources/gdacs.py

@@ -22,6 +22,8 @@
 )
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataSource, MontyDataTransformer
+from pystac_monty.validators.gdacs_events import GdacsDataValidatorEvents
+from pystac_monty.validators.gdacs_geometry import GdacsDataValidatorGeometry
 
 # Constants
 
@@ -46,7 +48,24 @@ def __init__(self, source_url: str, data: Any, type: GDACSDataSourceType):
         super().__init__(source_url, data)
         self.type = type
         # all gdacs data are json
-        self.data = json.loads(data)
+        self.data = self.source_data_validator(json.loads(data))
+
+    def source_data_validator(self, data: dict):
+        # Debug print
+        if self.type == GDACSDataSourceType.EVENT:
+            result = GdacsDataValidatorEvents.validate_event(data)
+            if result:
+                return data
+        elif self.type == GDACSDataSourceType.GEOMETRY:
+            new_data = {}  # Store the filtered dictionary
+            for key, value in data.items():
+                if key == "features" and isinstance(value, list):
+                    # Validate each feature in the list and skip the ones with 'Figure cause' = 'Conflict'
+                    new_data[key] = [feature for feature in value if GdacsDataValidatorGeometry.validate_event(feature)]
+                else:
+                    # Keep normal key-value pairs unchanged
+                    new_data[key] = value
+            return new_data
 
     def get_type(self) -> GDACSDataSourceType:
         return self.type

Diff for: pystac_monty/sources/gfd.py

@@ -15,6 +15,7 @@
 )
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataSource, MontyDataTransformer
+from pystac_monty.validators.gfd import GFDSourceValidator
 
 # Constants
 
@@ -28,7 +29,20 @@ class GFDDataSource(MontyDataSource):
 
     def __init__(self, source_url: str, data: Any):
         super().__init__(source_url, data)
-        self.data = json.loads(data)
+        self.data = self.source_data_validator(json.loads(data))
+
+    def source_data_validator(self, data: list[dict]):
+        """Validate the source data and collect only the success items"""
+        # TODO Handle the failed_items
+        failed_items = []
+        success_items = []
+        for item in data:
+            is_valid = GFDSourceValidator.validate_event(item)
+            if is_valid:
+                success_items.append(item)
+            else:
+                failed_items.append(item)
+        return success_items
 
 
 class GFDTransformer(MontyDataTransformer[GFDDataSource]):
@@ -85,7 +99,7 @@ def make_source_event_item(self, data: dict) -> Item:
         enddate = pytz.utc.localize(datetime.fromtimestamp(data["system:time_end"] / 1000))
 
         item = Item(
-            id=f'{STAC_EVENT_ID_PREFIX}{data["id"]}',
+            id=f"{STAC_EVENT_ID_PREFIX}{data['id']}",
             geometry=geometry,
             bbox=bbox,
             datetime=startdate,

Diff for: pystac_monty/sources/gidd.py

@@ -17,6 +17,7 @@
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataSource, MontyDataTransformer
 from pystac_monty.sources.utils import IDMCUtils
+from pystac_monty.validators.gidd import GiddValidator
 
 logger = logging.getLogger(__name__)
 
@@ -31,6 +32,21 @@ class GIDDDataSource(MontyDataSource):
     def __init__(self, source_url: str, data: Any):
         super().__init__(source_url, data)
         self.data = json.loads(data)
+        self.data = self.source_data_validator(json.loads(data))
+
+    def source_data_validator(self, data: dict):
+        """Validate only the items inside 'features' while keeping other keys unchanged."""
+
+        new_data = {}  # Store the filtered dictionary
+
+        for key, value in data.items():
+            if key == "features" and isinstance(value, list):
+                # Validate each feature in the list and skip the ones with 'Figure cause' = 'Conflict'
+                new_data[key] = [feature for feature in value if GiddValidator.validate_event(feature)]
+            else:
+                # Keep normal key-value pairs unchanged
+                new_data[key] = value
+        return new_data
 
 
 class GIDDTransformer(MontyDataTransformer[GIDDDataSource]):

Diff for: pystac_monty/sources/glide.py

@@ -9,6 +9,7 @@
 from pystac_monty.extension import HazardDetail, MontyEstimateType, MontyExtension
 from pystac_monty.hazard_profiles import MontyHazardProfiles
 from pystac_monty.sources.common import MontyDataSource, MontyDataTransformer
+from pystac_monty.validators.glide import GlideSetValidator
 
 STAC_EVENT_ID_PREFIX = "glide-event-"
 STAC_HAZARD_ID_PREFIX = "glide-hazard-"
@@ -17,7 +18,19 @@
 class GlideDataSource(MontyDataSource):
     def __init__(self, source_url: str, data: Any):
         super().__init__(source_url, data)
-        self.data = json.loads(data)
+        self.data = self.source_data_validator(json.loads(data))
+
+    def source_data_validator(self, data: dict[dict]):
+        """Validate the source data and collect only the success items"""
+        failed_items = []
+        success_items = []
+        for item in data["glideset"]:
+            is_valid = GlideSetValidator.validate_event(item)
+            if is_valid:
+                success_items.append(item)
+            else:
+                failed_items.append(item)
+        return {"glideset": success_items}
 
 
 class GlideTransformer(MontyDataTransformer[GlideDataSource]):