ocf-data-sampler 0.2.23__tar.gz → 0.2.25__tar.gz

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocf-data-sampler
-Version: 0.2.23
+Version: 0.2.25
 Author: James Fulton, Peter Dudfield
 Author-email: Open Climate Fix team <info@openclimatefix.org>
 License: MIT License

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/ocf_data_sampler/config/model.py

@@ -291,6 +291,8 @@ class GSP(TimeWindowMixin, DropoutMixin):
         description="Version of the GSP boundaries to use. Options are '20220314' or '20250109'.",
     )
 
+    public: bool = Field(False, description="Whether the NWP data is public or private")
+
 
 class Site(TimeWindowMixin, DropoutMixin):
     """Site configuration model."""

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/ocf_data_sampler/load/gsp.py

@@ -26,13 +26,17 @@ def get_gsp_boundaries(version: str) -> pd.DataFrame:
     )
 
 
-def open_gsp(zarr_path: str, boundaries_version: str = "20220314") -> xr.DataArray:
+def open_gsp(zarr_path: str,
+             boundaries_version: str = "20220314",
+             public: bool = False,
+    ) -> xr.DataArray:
     """Open the GSP data.
 
     Args:
         zarr_path: Path to the GSP zarr data
         boundaries_version: Version of the GSP boundaries to use. Options are "20220314" or
         "20250109".
+        public: Whether the data is public or private.
 
     Returns:
         xr.DataArray: The opened GSP data
@@ -40,12 +44,18 @@ def open_gsp(zarr_path: str, boundaries_version: str = "20220314") -> xr.DataArr
     # Load UK GSP locations
     df_gsp_loc = get_gsp_boundaries(boundaries_version)
 
+    backend_kwargs ={}
     # Open the GSP generation data
+    if public:
+        backend_kwargs ={"storage_options":{"anon": True}}
+        # Currently only compatible with S3 bucket.
+
     ds = (
-        xr.open_zarr(zarr_path)
+        xr.open_dataset(zarr_path,engine="zarr",backend_kwargs=backend_kwargs)
         .rename({"datetime_gmt": "time_utc"})
     )
 
+
     if not (ds.gsp_id.isin(df_gsp_loc.index)).all():
         raise ValueError(
             "Some GSP IDs in the GSP generation data are not available in the locations file.",

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/ocf_data_sampler/load/load_dataset.py

@@ -24,6 +24,7 @@ def get_dataset_dict(
         da_gsp = open_gsp(
             zarr_path=input_config.gsp.zarr_path,
             boundaries_version=input_config.gsp.boundaries_version,
+            public=input_config.gsp.public,
         ).compute()
 
         if gsp_ids is None:

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/ocf_data_sampler/select/select_spatial_slice.py

@@ -239,8 +239,19 @@ def select_spatial_slice_pixels(
         if allow_partial_slice:
             da = _select_padded_slice(da, left_idx, right_idx, bottom_idx, top_idx, x_dim, y_dim)
         else:
+            issues = []
+            if left_idx < 0:
+                issues.append(f"left_idx ({left_idx}) < 0")
+            if right_idx > data_width_pixels:
+                issues.append(f"right_idx ({right_idx}) > data_width_pixels ({data_width_pixels})")
+            if bottom_idx < 0:
+                issues.append(f"bottom_idx ({bottom_idx}) < 0")
+            if top_idx > data_height_pixels:
+                issues.append(f"top_idx ({top_idx}) > data_height_pixels ({data_height_pixels})")
+            issue_details = "\n".join(issues)
             raise ValueError(
-                f"Window for location {location} not available.  Padding required. "
+                f"Window for location {location} not available.  Padding required due to: \n"
+                f"{issue_details}\n"
                 "You may wish to set `allow_partial_slice=True`",
             )
     else:

{ocf_data_sampler-0.2.23 → ocf_data_sampler-0.2.25}/ocf_data_sampler.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocf-data-sampler
-Version: 0.2.23
+Version: 0.2.25
 Author: James Fulton, Peter Dudfield
 Author-email: Open Climate Fix team <info@openclimatefix.org>
 License: MIT License