Merge branch 'issue666-load_stac-temporal-metadata'

Author: soxofaan

CHANGELOG.md

@@ -17,6 +17,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `load_stac`: use fallback temporal dimension when no "cube:dimensions" in STAC Collection ([#666](https://github.com/Open-EO/openeo-python-client/issues/666))
 
 ## [0.35.0] - 2024-11-19
 

openeo/metadata.py

@@ -11,7 +11,7 @@
 import pystac.extensions.item_assets
 
 from openeo.internal.jupyter import render_component
-from openeo.util import deep_get
+from openeo.util import Rfc3339, deep_get
 
 _log = logging.getLogger(__name__)
 
@@ -691,6 +691,11 @@ def get_temporal_dimension(self, stac_obj: pystac.STACObject) -> Union[TemporalD
                 if len(temporal_dims) == 1:
                     name, extent = temporal_dims[0]
                     return TemporalDimension(name=name, extent=extent)
+            elif isinstance(stac_obj, pystac.Collection) and stac_obj.extent.temporal:
+                # No explicit "cube:dimensions": build fallback from "extent.temporal",
+                # with dimension name "t" (openEO API recommendation).
+                extent = [Rfc3339(propagate_none=True).normalize(d) for d in stac_obj.extent.temporal.intervals[0]]
+                return TemporalDimension(name="t", extent=extent)
         else:
             if isinstance(stac_obj, pystac.Item):
                 cube_dimensions = stac_obj.properties.get("cube:dimensions", {})

openeo/rest/datacube.py

@@ -203,6 +203,7 @@ def load_collection(
                 metadata = None
             if metadata:
                 bands = [b if isinstance(b, str) else metadata.band_dimension.band_name(b) for b in bands]
+                # TODO: also apply spatial/temporal filters to metadata?
                 metadata = metadata.filter_bands(bands)
             arguments['bands'] = bands
 
@@ -385,6 +386,9 @@ def load_stac(
         graph = PGNode("load_stac", arguments=arguments)
         try:
             metadata = metadata_from_stac(url)
+            if bands:
+                # TODO: also apply spatial/temporal filters to metadata?
+                metadata = metadata.filter_bands(band_names=bands)
         except Exception:
             log.warning(f"Failed to extract cube metadata from STAC URL {url}", exc_info=True)
             metadata = None

openeo/util.py

@@ -172,7 +172,7 @@ def parse_date_or_datetime(
     @classmethod
     def _format_datetime(cls, d: dt.datetime) -> str:
         """Format given datetime as RFC-3339 date-time string."""
-        if d.tzinfo not in {None, dt.timezone.utc}:
+        if not (d.tzinfo is None or d.tzinfo.tzname(d) == "UTC"):
             # TODO: add support for non-UTC timezones?
             raise ValueError(f"No support for non-UTC timezone {d.tzinfo}")
         return d.strftime(cls._FMT_DATETIME)

setup.py

@@ -30,6 +30,7 @@
     "pyproj>=3.2.0",  # Pyproj is an optional, best-effort runtime dependency
     "dirty_equals>=0.8.0",
     "pyarrow>=10.0.1",  # For Parquet read/write support in pandas
+    "python-dateutil>=2.7.0",
 ]
 
 docs_require = [

tests/rest/test_connection.py

@@ -18,6 +18,7 @@
 import openeo
 from openeo.capabilities import ApiVersionException
 from openeo.internal.graph_building import FlatGraphableMixin, PGNode
+from openeo.metadata import _PYSTAC_1_9_EXTENSION_INTERFACE, TemporalDimension
 from openeo.rest import (
     CapabilitiesException,
     OpenEoApiError,
@@ -40,7 +41,7 @@
 )
 from openeo.rest.vectorcube import VectorCube
 from openeo.testing.stac import StacDummyBuilder
-from openeo.util import ContextTimer, dict_no_none
+from openeo.util import ContextTimer, deep_get, dict_no_none
 
 from .auth.test_cli import auth_config, refresh_token_store
 
@@ -2622,6 +2623,51 @@ def test_load_stac_reduce_temporal(self, con120, tmp_path, temporal_dim):
             },
         }
 
+    @pytest.mark.skipif(
+        not _PYSTAC_1_9_EXTENSION_INTERFACE,
+        reason="No backport of implementation/test below PySTAC 1.9 extension interface",
+    )
+    @pytest.mark.parametrize(
+        ["collection_extent", "dim_extent"],
+        [
+            (
+                {"spatial": {"bbox": [[3, 4, 5, 6]]}, "temporal": {"interval": [["2024-01-01", "2024-05-05"]]}},
+                ["2024-01-01T00:00:00Z", "2024-05-05T00:00:00Z"],
+            ),
+            (
+                {"spatial": {"bbox": [[3, 4, 5, 6]]}, "temporal": {"interval": [[None, "2024-05-05"]]}},
+                [None, "2024-05-05T00:00:00Z"],
+            ),
+        ],
+    )
+    def test_load_stac_no_cube_extension_temporal_dimension(self, con120, tmp_path, collection_extent, dim_extent):
+        """
+        Metadata detection when STAC metadata does not use "cube" extension
+        https://github.com/Open-EO/openeo-python-client/issues/666
+        """
+        stac_path = tmp_path / "stac.json"
+        stac_data = StacDummyBuilder.collection(extent=collection_extent)
+        # No cube:dimensions, but at least "temporal" extent is set as indicator for having a temporal dimension
+        assert "cube:dimensions" not in stac_data
+        assert deep_get(stac_data, "extent", "temporal")
+        stac_path.write_text(json.dumps(stac_data))
+
+        cube = con120.load_stac(str(stac_path))
+        assert cube.metadata.temporal_dimension == TemporalDimension(name="t", extent=dim_extent)
+
+    def test_load_stac_band_filtering(self, con120, tmp_path):
+        stac_path = tmp_path / "stac.json"
+        stac_data = StacDummyBuilder.collection(
+            summaries={"eo:bands": [{"name": "B01"}, {"name": "B02"}, {"name": "B03"}]}
+        )
+        stac_path.write_text(json.dumps(stac_data))
+
+        cube = con120.load_stac(str(stac_path))
+        assert cube.metadata.band_names == ["B01", "B02", "B03"]
+
+        cube = con120.load_stac(str(stac_path), bands=["B03", "B02"])
+        assert cube.metadata.band_names == ["B03", "B02"]
+
 
 @pytest.mark.parametrize(
     "data",

tests/test_metadata.py

@@ -849,6 +849,10 @@ def test_metadata_from_stac_collection_bands_from_item_assets(test_data, tmp_pat
     assert warn_count == (0 if eo_extension_is_declared else 1)
 
 
+@pytest.mark.skipif(
+    not _PYSTAC_1_9_EXTENSION_INTERFACE,
+    reason="No backport of implementation/test below PySTAC 1.9 extension interface",
+)
 @pytest.mark.parametrize(
     ["stac_dict", "expected"],
     [
@@ -868,7 +872,7 @@ def test_metadata_from_stac_collection_bands_from_item_assets(test_data, tmp_pat
         ),
         (
             StacDummyBuilder.collection(),
-            None,
+            ("t", ["2024-01-01T00:00:00Z", "2024-05-05T00:00:00Z"]),
         ),
         (
             StacDummyBuilder.collection(

tests/test_util.py

@@ -7,6 +7,7 @@
 import unittest.mock as mock
 from typing import List, Union
 
+import dateutil.tz
 import pyproj
 import pytest
 import shapely.geometry
@@ -87,21 +88,12 @@ def test_datetime(self):
         assert "2020-03-17T12:34:56Z" == rfc3339.datetime([2020, 3, 17, 12, 34, 56])
         assert "2020-03-17T12:34:56Z" == rfc3339.datetime(2020, 3, 17, 12, 34, 56)
         assert "2020-03-17T12:34:00Z" == rfc3339.datetime(2020, 3, 17, 12, 34)
-        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(
-            (2020, "3", 17, "12", "34", 56)
-        )
-        assert "2020-09-17T12:34:56Z" == rfc3339.datetime(
-            [2020, "09", 17, "12", "34", 56]
-        )
-        assert "2020-09-17T12:34:56Z" == rfc3339.datetime(
-            2020, "09", "17", "12", "34", 56
-        )
-        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(
-            dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=None)
-        )
-        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(
-            dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dt.timezone.utc)
-        )
+        assert "2020-03-17T12:34:56Z" == rfc3339.datetime((2020, "3", 17, "12", "34", 56))
+        assert "2020-09-17T12:34:56Z" == rfc3339.datetime([2020, "09", 17, "12", "34", 56])
+        assert "2020-09-17T12:34:56Z" == rfc3339.datetime(2020, "09", "17", "12", "34", 56)
+        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=None))
+        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dt.timezone.utc))
+        assert "2020-03-17T12:34:56Z" == rfc3339.datetime(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dateutil.tz.UTC))
         assert "2020-03-17T12:34:56Z" == rfc3339.datetime(
             dt.datetime(
                 *(2020, 3, 17, 12, 34, 56),
@@ -125,15 +117,10 @@ def test_normalize(self):
             "2020-03-17T12:34:56.44546546Z"
         )
         assert "2020-03-17" == rfc3339.normalize(dt.date(2020, 3, 17))
-        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(
-            dt.datetime(2020, 3, 17, 12, 34, 56)
-        )
-        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(
-            dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=None)
-        )
-        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(
-            dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dt.timezone.utc)
-        )
+        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(dt.datetime(2020, 3, 17, 12, 34, 56))
+        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=None))
+        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dt.timezone.utc))
+        assert "2020-03-17T12:34:56Z" == rfc3339.normalize(dt.datetime(2020, 3, 17, 12, 34, 56, tzinfo=dateutil.tz.UTC))
         assert "2020-03-17T12:34:56Z" == rfc3339.normalize(
             dt.datetime(
                 *(2020, 3, 17, 12, 34, 56),