warn: preventing subset on stereographic data (#217)

Add a error if trying to subset on latitude or longitude a dataset with part originalGrid (ie stereographic projection)

copernicusmarine/__init__.py

@@ -34,9 +34,9 @@
     ServiceNotSupported,
     VariableDoesNotExistInTheDataset,
     WrongDatetimeFormat,
+    GeospatialSubsetNotAvailableForNonLatLon,
 )
 from copernicusmarine.core_functions.models import (
-    DatasetCoordinatesExtent,
     FileGet,
     FileStatus,
     GeographicalExtent,
@@ -78,7 +78,6 @@
     "CoperniusMarineServiceShortNames",
     "CouldNotConnectToAuthenticationSystem",
     "CredentialsCannotBeNone",
-    "DatasetCoordinatesExtent",
     "DatasetNotFound",
     "DatasetVersionNotFound",
     "DatasetVersionPartNotFound",
@@ -96,6 +95,7 @@
     "ServiceNotAvailable",
     "ServiceNotHandled",
     "ServiceNotSupported",
+    "GeospatialSubsetNotAvailableForNonLatLon",
     "StatusCode",
     "FileStatus",
     "StatusMessage",

copernicusmarine/catalogue_parser/fields_query_builder.py

@@ -1,4 +1,12 @@
-from typing import Optional, Type, get_args, get_origin, get_type_hints
+from typing import (
+    Literal,
+    Optional,
+    Type,
+    Union,
+    get_args,
+    get_origin,
+    get_type_hints,
+)
 
 from pydantic import BaseModel
 
@@ -55,29 +63,50 @@ def build_query(
         ) in get_type_hints(type_to_check).items():
             if field_name in self.fields_to_include_or_exclude:
                 query[field_name] = True
-            elif get_origin(field_type) is list:
+                continue
+            all_base_models = self._get_base_models_in_type(field_type)
+            for base_model, in_an_iterable in (all_base_models or {}).items():
                 if field_name not in query:
-                    query[field_name] = {"__all__": {}}
+                    query[field_name] = (
+                        {"__all__": {}} if in_an_iterable else {}
+                    )
                 result = self.build_query(
-                    get_args(field_type)[0],
-                    query[field_name]["__all__"],
+                    base_model,
+                    (
+                        query[field_name]["__all__"]
+                        if in_an_iterable
+                        else query[field_name]
+                    ),
                 )
                 if not result:
                     del query[field_name]
-            elif get_origin(field_type) is dict:
-                if field_name not in query:
-                    query[field_name] = {"__all__": {}}
-                result = self.build_query(
-                    get_args(field_type)[1],
-                    query[field_name]["__all__"],
-                )
-                if not result:
-                    del query[field_name]
-            elif check_type_is_base_model(field_type):
-                if field_name not in query:
-                    query[field_name] = {}
-                result = self.build_query(field_type, query[field_name])
-                if not result:
-                    del query[field_name]
-
         return query
+
+    def _get_base_models_in_type(
+        self,
+        type_to_check: Type,
+        in_an_iterable: bool = False,
+    ) -> Optional[dict[Type, Literal["__all__", None]]]:
+        models = {}
+        if check_type_is_base_model(type_to_check):
+            return {type_to_check: "__all__" if in_an_iterable else None}
+        elif get_origin(type_to_check) is list:
+            result = self._get_base_models_in_type(
+                get_args(type_to_check)[0], in_an_iterable=True
+            )
+            if result:
+                models.update(result)
+        elif get_origin(type_to_check) is dict:
+            result = self._get_base_models_in_type(
+                get_args(type_to_check)[1], in_an_iterable=True
+            )
+            if result:
+                models.update(result)
+        elif get_origin(type_to_check) is Union:
+            for union_type in get_args(type_to_check):
+                result = self._get_base_models_in_type(
+                    union_type, in_an_iterable=in_an_iterable
+                )
+                if result:
+                    models.update(result)
+        return models

copernicusmarine/core_functions/exceptions.py

@@ -41,6 +41,24 @@ def __init__(self, message: str):
         self.__setattr__("custom_exception_message", message)
 
 
+class GeospatialSubsetNotAvailableForNonLatLon(Exception):
+    """
+    The data you are requesting is using a projection that is not on the
+    normalised latitude and longitude grid. The geospatial subset of such
+    datasets is not yet available.
+
+    Please check other parts of the dataset to subset it. The geospatial subset
+    of the datasets with different gridding will be fully available soon.
+    """
+
+    def __init__(self):
+        super().__init__(
+            "The geospatial subset of datasets in a projection that is not in "
+            "latitude and longitude is not yet available. "
+            "We are developing such feature and will be supported in future versions."
+        )
+
+
 class NetCDFCompressionNotAvailable(Exception):
     """
     Exception raised when the NetCDF compression is not available.

copernicusmarine/core_functions/models.py

@@ -1,6 +1,6 @@
 import pathlib
 from enum import Enum
-from typing import Literal, Optional, get_args
+from typing import Literal, Optional, Union, get_args
 
 from pydantic import BaseModel, ConfigDict
 
@@ -153,32 +153,19 @@ def add_s3_file(self, s3_file: S3FileInfo):
 class GeographicalExtent(BaseModel):
     """Interval for geographical coordinates."""
 
-    minimum: Optional[float]
-    maximum: Optional[float]
+    minimum: float
+    maximum: float
     unit: Optional[str]
+    coordinate_id: str
 
 
 class TimeExtent(BaseModel):
     """Interval for time coordinates."""
 
-    minimum: Optional[str]
-    maximum: Optional[str]
-    unit: Optional[str]
-
-
-class DatasetCoordinatesExtent(BaseModel):
-    #: Longitude interval of the subsetted data.
-    longitude: Optional[GeographicalExtent]
-    #: Latitude interval of the subsetted data.
-    latitude: Optional[GeographicalExtent]
-    #: Time interval of the subsetted data in iso8601 string.
-    time: Optional[TimeExtent]
-    #: Depth interval of the subsetted data.
-    depth: Optional[GeographicalExtent] = None
-    #: Elevation interval of the subsetted data.
-    #: Is relevant if data are requested for elevation
-    #: instead of depth.
-    elevation: Optional[GeographicalExtent] = None
+    minimum: str
+    maximum: str
+    unit: str
+    coordinate_id: str
 
 
 class ResponseSubset(BaseModel):
@@ -200,7 +187,7 @@ class ResponseSubset(BaseModel):
     #: Variables of the subsetted dataset.
     variables: list[str]
     #: The bounds of the subsetted dataset.
-    coordinates_extent: DatasetCoordinatesExtent
+    coordinates_extent: list[Union[GeographicalExtent, TimeExtent]]
     #: Status of the request.
     status: StatusCode
     #: Message explaning the status.

copernicusmarine/core_functions/services_utils.py

@@ -263,6 +263,7 @@ class RetrievalService:
     uri: str
     dataset_valid_start_date: Optional[Union[str, int, float]]
     service: CopernicusMarineService
+    is_original_grid: bool = False
 
 
 def get_retrieval_service(
@@ -408,6 +409,7 @@ def _get_retrieval_service_from_dataset_version(
         dataset_valid_start_date=dataset_start_date,
         service_format=service.service_format,
         service=service,
+        is_original_grid=dataset_part.name == "originalGrid",
     )
 
 

copernicusmarine/core_functions/subset.py

@@ -170,6 +170,7 @@ def subset_function(
         dataset_subset=subset_request.get_time_and_space_subset(),
         coordinates_selection_method=subset_request.coordinates_selection_method,
         dataset_valid_date=retrieval_service.dataset_valid_start_date,
+        is_original_grid=retrieval_service.is_original_grid,
     )
     logger.info(
         "Downloading using service " f"{retrieval_service.service_name}..."

copernicusmarine/download_functions/subset_xarray.py

@@ -16,6 +16,7 @@
 from copernicusmarine.core_functions import custom_open_zarr
 from copernicusmarine.core_functions.exceptions import (
     CoordinatesOutOfDatasetBounds,
+    GeospatialSubsetNotAvailableForNonLatLon,
     MinimumLongitudeGreaterThanMaximumLongitude,
     ServiceNotSupported,
     VariableDoesNotExistInTheDataset,
@@ -584,6 +585,7 @@ def check_dataset_subset_bounds(
     dataset_subset: DatasetTimeAndSpaceSubset,
     coordinates_selection_method: CoordinatesSelectionMethod,
     dataset_valid_date: Optional[Union[str, int, float]],
+    is_original_grid: bool,
 ) -> None:
     if service_name in [
         CopernicusMarineServiceNames.GEOSERIES,
@@ -597,6 +599,16 @@ def check_dataset_subset_bounds(
         dataset_coordinates = dataset.coords
     else:
         raise ServiceNotSupported(service_name)
+    if is_original_grid:
+        logger.debug("Dataset part has the non lat lon projection.")
+        if (
+            dataset_subset.minimum_latitude is not None
+            or dataset_subset.maximum_latitude is not None
+            or dataset_subset.minimum_longitude is not None
+            or dataset_subset.maximum_longitude is not None
+        ):
+            raise GeospatialSubsetNotAvailableForNonLatLon()
+
     for coordinate_label in COORDINATES_LABEL["latitude"]:
         if coordinate_label in dataset.sizes:
             latitudes = dataset_coordinates[coordinate_label].values

copernicusmarine/download_functions/utils.py

@@ -13,7 +13,6 @@
 )
 from copernicusmarine.core_functions.models import (
     DEFAULT_FILE_EXTENSIONS,
-    DatasetCoordinatesExtent,
     FileFormat,
     GeographicalExtent,
     TimeExtent,
@@ -190,17 +189,12 @@ def _format_datetimes(
 
 def get_dataset_coordinates_extent(
     dataset: xarray.Dataset,
-) -> DatasetCoordinatesExtent:
-    coordinates_extent = DatasetCoordinatesExtent(
-        longitude=_get_coordinate_extent(dataset, "longitude"),  # type: ignore
-        latitude=_get_coordinate_extent(dataset, "latitude"),  # type: ignore
-        time=_get_coordinate_extent(dataset, "time"),  # type: ignore
-    )
-    depth_or_elevation_extent = _get_coordinate_extent(dataset, "depth")
-    if "depth" in dataset.sizes:
-        coordinates_extent.depth = depth_or_elevation_extent  # type: ignore
-    elif "elevation" in dataset.sizes:
-        coordinates_extent.elevation = depth_or_elevation_extent  # type: ignore
+) -> list[Union[GeographicalExtent, TimeExtent]]:
+    coordinates_extent = []
+    for coord_label in ["longitude", "latitude", "time", "depth"]:
+        if coordinate_extent := _get_coordinate_extent(dataset, coord_label):
+            coordinates_extent.append(coordinate_extent)
+
     return coordinates_extent
 
 
@@ -225,11 +219,13 @@ def _get_coordinate_extent(
                     minimum=minimum,
                     maximum=maximum,
                     unit=unit,
+                    coordinate_id=coord_label,
                 )
             return GeographicalExtent(
                 minimum=minimum,
                 maximum=maximum,
                 unit=unit,
+                coordinate_id=coord_label,
             )
     return None
 

copernicusmarine/python_interface/load_utils.py

@@ -52,6 +52,7 @@ def load_data_object_from_load_request(
         dataset_subset=load_request.get_time_and_space_subset(),
         coordinates_selection_method=load_request.coordinates_selection_method,
         dataset_valid_date=retrieval_service.dataset_valid_start_date,
+        is_original_grid=retrieval_service.is_original_grid,
     )
     if retrieval_service.service_name in [
         CopernicusMarineServiceNames.GEOSERIES,

doc/response-types.rst

@@ -52,13 +52,6 @@ Subtypes
     :exclude-members: model_computed_fields, model_config, model_fields
     :member-order: bysource
 
-.. autoclass:: copernicusmarine.DatasetCoordinatesExtent()
-    :members:
-    :undoc-members:
-    :exclude-members: model_computed_fields, model_config, model_fields
-    :member-order: bysource
-
-
 .. autoclass:: copernicusmarine.GeographicalExtent()
     :members:
     :undoc-members:

tests/test_describe_released_date.py

@@ -36,7 +36,9 @@ def then_I_dont_get_the_not_released_products_version_and_datasets(
         assert 1 == len(describe_result.products)
         assert (
             describe_result.model_dump(
-                exclude_none=True, exclude_unset=True, exclude=exclude_query
+                exclude_none=True,
+                exclude_unset=True,
+                exclude=exclude_query,
             )
             == snapshot
         )
@@ -55,7 +57,9 @@ def then_I_get_all_products_versions_and_datasets(
         assert 2 == len(describe_result.products)
         assert (
             describe_result.model_dump(
-                exclude_none=True, exclude_unset=True, exclude=exclude_query
+                exclude_none=True,
+                exclude_unset=True,
+                exclude=exclude_query,
             )
             == snapshot
         )