cecil 0.0.22__tar.gz → 0.0.24__tar.gz

{cecil-0.0.22 → cecil-0.0.24}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cecil
-Version: 0.0.22
+Version: 0.0.24
 Summary: Python SDK for Cecil Earth
 License-Expression: MIT
 License-File: LICENSE.txt

{cecil-0.0.22 → cecil-0.0.24}/src/cecil/client.py

@@ -1,12 +1,14 @@
 import os
-from typing import Dict, List, Optional
 
 import pandas as pd
 import requests
 import snowflake.connector
+import xarray
+
 from pydantic import BaseModel
 from requests import auth
 from cryptography.hazmat.primitives import serialization
+from typing import Dict, List, Optional
 
 from .errors import (
     Error,
@@ -31,8 +33,10 @@ from .models import (
     TransformationCreate,
     User,
     UserCreate,
+    DataRequestMetadata,
 )
 from .version import __version__
+from .xarray import load_xarray
 
 
 class Client:
@@ -78,6 +82,11 @@ class Client:
         res = self._get(url="/v0/data-requests")
         return [DataRequest(**record) for record in res["records"]]
 
+    def load_xarray(self, data_request_id: str) -> xarray.Dataset:
+        res = self._get(url=f"/v0/data-requests/{data_request_id}/metadata")
+        metadata = DataRequestMetadata(**res)
+        return load_xarray(metadata)
+
     def create_transformation(
         self, data_request_id: str, crs: str, spatial_resolution: float
     ) -> Transformation:

{cecil-0.0.22 → cecil-0.0.24}/src/cecil/models.py

@@ -1,5 +1,5 @@
 import datetime
-from typing import Dict, Optional
+from typing import Dict, Optional, List
 
 from pydantic import BaseModel, ConfigDict, SecretStr
 from pydantic.alias_generators import to_camel
@@ -110,3 +110,28 @@ class UserCreate(BaseModel):
     first_name: str
     last_name: str
     email: str
+
+
+class Band(BaseModel):
+    model_config = ConfigDict(alias_generator=to_camel, populate_by_name=True)
+    variable_name: str
+    time: str
+    time_pattern: str
+    number: int
+
+
+class File(BaseModel):
+    model_config = ConfigDict(alias_generator=to_camel, populate_by_name=True)
+    url: str
+    bands: List[Band]
+
+
+class DataRequestMetadata(BaseModel):
+    model_config = ConfigDict(alias_generator=to_camel, populate_by_name=True)
+    provider_name: str
+    dataset_id: str
+    dataset_name: str
+    dataset_crs: str
+    aoi_id: str
+    data_request_id: str
+    files: List[File]

cecil-0.0.24/src/cecil/version.py

@@ -0,0 +1 @@
+__version__ = "0.0.24"

cecil-0.0.24/src/cecil/xarray.py

@@ -0,0 +1,74 @@
+import os
+import rioxarray
+import xarray
+
+from datetime import datetime
+
+from .errors import Error
+from .models import DataRequestMetadata
+
+os.environ["GDAL_NUM_THREADS"] = "1"
+os.environ["GDAL_DISABLE_READDIR_ON_OPEN"] = "FALSE"
+
+
+def align_pixel_grids(time_series):
+    # Use the first timestep as reference
+    reference_da = time_series[0]
+    aligned_series = [reference_da]
+
+    # Align all other timesteps to the reference grid
+    for i, da in enumerate(time_series[1:], 1):
+        try:
+            aligned_da = da.rio.reproject_match(reference_da)
+            aligned_series.append(aligned_da)
+        except Exception as e:
+            raise Error
+            continue
+
+    return aligned_series
+
+
+def load_xarray(metadata: DataRequestMetadata) -> xarray.Dataset:
+    data_vars = {}
+
+    for f in metadata.files:
+        dataset = rioxarray.open_rasterio(
+            f.url,
+            chunks={"x": 2000, "y": 2000},
+        )
+
+        for b in f.bands:
+            band = dataset.sel(band=b.number, drop=True)
+
+            if b.time and b.time_pattern:
+                time = datetime.strptime(b.time, b.time_pattern)
+                band = band.expand_dims("time")
+                band = band.assign_coords(time=[time])
+
+            band.name = b.variable_name
+
+            if b.variable_name not in data_vars:
+                data_vars[b.variable_name] = []
+
+            data_vars[b.variable_name].append(band)
+
+    for variable_name, time_series in data_vars.items():
+        if "time" in time_series[0].dims:
+            # time_series = align_pixel_grids(time_series)
+            data_vars[variable_name] = xarray.concat(
+                time_series, dim="time", join="exact"
+            )
+        else:
+            data_vars[variable_name] = time_series[0]
+
+    return xarray.Dataset(
+        data_vars=data_vars,
+        attrs={
+            "provider_name": metadata.provider_name,
+            "dataset_id": metadata.dataset_id,
+            "dataset_name": metadata.dataset_name,
+            "dataset_crs": metadata.dataset_crs,
+            "aoi_id": metadata.aoi_id,
+            "data_request_id": metadata.data_request_id,
+        },
+    )

{cecil-0.0.22 → cecil-0.0.24}/tests/test_client.py

@@ -20,6 +20,7 @@ def test_client_create_data_request():
             "id": "id",
             "aoiId": "aoi_id",
             "datasetId": "dataset_id",
+            "externalRef": "external_ref",
             "created_at": FROZEN_TIME,
             "created_by": "user_id",
         },
@@ -33,6 +34,7 @@ def test_client_create_data_request():
         id="id",
         aoiId="aoi_id",
         datasetId="dataset_id",
+        externalRef="external_ref",
         created_at="2024-01-01T00:00:00.000Z",
         created_by="user_id",
     )
@@ -49,6 +51,7 @@ def test_client_list_data_requests():
                     "id": "data_request_id_1",
                     "aoiId": "aoi_id",
                     "datasetId": "dataset_id",
+                    "externalRef": "external_ref",
                     "created_at": "2024-09-19T04:45:57.561Z",
                     "created_by": "user_id",
                 },
@@ -56,6 +59,7 @@ def test_client_list_data_requests():
                     "id": "data_request_id_2",
                     "aoiId": "aoi_id",
                     "datasetId": "dataset_id",
+                    "externalRef": "",
                     "created_at": "2024-09-19T04:54:38.252Z",
                     "created_by": "user_id",
                 },
@@ -71,6 +75,7 @@ def test_client_list_data_requests():
             id="data_request_id_1",
             aoiId="aoi_id",
             datasetId="dataset_id",
+            externalRef="external_ref",
             created_at="2024-09-19T04:45:57.561Z",
             created_by="user_id",
         ),
@@ -78,6 +83,7 @@ def test_client_list_data_requests():
             id="data_request_id_2",
             aoiId="aoi_id",
             datasetId="dataset_id",
+            externalRef="",
             created_at="2024-09-19T04:54:38.252Z",
             created_by="user_id",
         ),

cecil-0.0.22/src/cecil/version.py

@@ -1 +0,0 @@
-__version__ = "0.0.22"