return STAC Items with valid date/time for time series job results

* be explicit about which metadata Open-EO/openeo-geopyspark-driver#852 * expose start/end_datetime from asset metadata Open-EO/openeo-geopyspark-driver#852 * update CHANGELOG Open-EO/openeo-geopyspark-driver#852 * update version and CHANGELOG Open-EO/openeo-geopyspark-driver#852
Open-EO · Sep 13, 2024 · adf218f · adf218f
1 parent 911cfff
commit adf218f
Show file tree

Hide file tree

Showing 6 changed files with 120 additions and 13 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -21,6 +21,10 @@ and start a new "In Progress" section above it.
 
 ## In progress
 
+## 0.107.7
+
+- return STAC Items with valid date/time for time series job results ([Open-EO/openeo-geopyspark-driver#852](https://github.com/Open-EO/openeo-geopyspark-driver/issues/852))
+
 ## 0.107.6
 
 - support passing the output of `raster_to_vector` to `aggregate_spatial` during dry run ([EU-GRASSLAND-WATCH/EUGW#7](https://github.com/EU-GRASSLAND-WATCH/EUGW/issues/7))

diff --git a/openeo_driver/_version.py b/openeo_driver/_version.py
@@ -1 +1 @@
-__version__ = "0.107.6a1"
+__version__ = "0.107.7a1"
diff --git a/openeo_driver/dummy/dummy_backend.py b/openeo_driver/dummy/dummy_backend.py
@@ -732,6 +732,30 @@ def get_result_assets(self, job_id: str, user_id: str) -> Dict[str, dict]:
                 },
             }
 
+        if job_id == "j-24083059866540a38cab32b028be0ab5":
+            return {
+                "timeseries.parquet": {
+                    "roles": ["data"],
+                    "type": "application/parquet; profile=geo",
+                    "href": "/data/projects/OpenEO/j-24083059866540a38cab32b028be0ab5/timeseries.parquet",
+                    "bands": [Band(name="S1-SIGMA0-VV"), Band(name="S1-SIGMA0-VH"), Band(name="S2-L2A-B01")],
+                    "geometry": {
+                        "type": "Polygon",
+                        "coordinates": [
+                            [
+                                [34.456156, -0.910085],
+                                [34.456156, -0.345477],
+                                [34.796396, -0.345477],
+                                [34.796396, -0.910085],
+                                [34.456156, -0.910085],
+                            ]
+                        ],
+                    },
+                    "start_datetime": "2016-10-30T00:00:00+00:00",
+                    "end_datetime": "2018-05-03T00:00:00+00:00",
+                },
+            }
+
         return {
             "output.tiff": {
                 "output_dir": f"{self._output_root()}/{job_id}",

diff --git a/openeo_driver/save_result.py b/openeo_driver/save_result.py
@@ -314,9 +314,15 @@ def write_assets(self, directory: Union[str, Path]) -> Dict[str, StacAsset]:
             with open(filename, 'w') as f:
                 json.dump(self.prepare_for_json(), f)
         asset["href"] = filename
-        if self._metadata is not None and self._metadata.has_band_dimension():
-            bands = [b._asdict() for b in self._metadata.bands]
-            asset["bands"] = bands
+
+        if self._metadata is not None:
+            if self._metadata.has_band_dimension():
+                bands = [b._asdict() for b in self._metadata.bands]
+                asset["bands"] = bands
+            if self._metadata.has_temporal_dimension():
+                start_datetime, end_datetime = self._metadata.temporal_dimension.extent
+                asset["start_datetime"] = start_datetime
+                asset["end_datetime"] = end_datetime
 
         the_file = Path(filename)
         if the_file.exists():

diff --git a/openeo_driver/views.py b/openeo_driver/views.py
@@ -1297,19 +1297,19 @@ def _get_job_result_item(job_id, item_id, user_id):
         if len(assets_for_item_id) != 1:
             raise AssertionError(f"expected exactly 1 asset with file name {item_id}")
 
-        asset_filename, metadata = next(iter(assets_for_item_id.items()))
+        asset_filename, asset_metadata = next(iter(assets_for_item_id.items()))
 
         job_info = backend_implementation.batch_jobs.get_job_info(job_id, user_id)
 
-        geometry = metadata.get("geometry", job_info.geometry)
-        bbox = metadata.get("bbox", job_info.bbox)
+        geometry = asset_metadata.get("geometry", job_info.geometry)
+        bbox = asset_metadata.get("bbox", job_info.bbox)
 
-        properties = {"datetime": metadata.get("datetime")}
+        properties = {"datetime": asset_metadata.get("datetime")}
         if properties["datetime"] is None:
             to_datetime = Rfc3339(propagate_none=True).datetime
 
-            start_datetime = to_datetime(job_info.start_datetime)
-            end_datetime = to_datetime(job_info.end_datetime)
+            start_datetime = asset_metadata.get("start_datetime") or to_datetime(job_info.start_datetime)
+            end_datetime = asset_metadata.get("end_datetime") or to_datetime(job_info.end_datetime)
 
             if start_datetime == end_datetime:
                 properties["datetime"] = start_datetime
@@ -1358,7 +1358,7 @@ def _get_job_result_item(job_id, item_id, user_id):
                     "type": "application/json",
                 },
             ],
-            "assets": {asset_filename: _asset_object(job_id, user_id, asset_filename, metadata,job_info)},
+            "assets": {asset_filename: _asset_object(job_id, user_id, asset_filename, asset_metadata, job_info)},
             "collection": job_id,
         }
         # Add optional items, if they are present.

diff --git a/tests/test_views.py b/tests/test_views.py
@@ -1198,6 +1198,14 @@ def _fresh_job_registry(next_job_id="job-1234", output_root: Optional[Path] = No
                     start_datetime=datetime(2020, 1, 1, 0, 0, 0),
                     end_datetime=datetime(2020, 3, 15, 0, 0, 0),
                 ),
+                (TEST_USER, "j-24083059866540a38cab32b028be0ab5"): BatchJobMetadata(
+                    id="j-24083059866540a38cab32b028be0ab5",
+                    status='finished',
+                    created=datetime(2024, 8, 30, 12, 16, 34),
+                    bbox=[34.456156, -0.910085, 34.796396, -0.345477],
+                    start_datetime=None,
+                    end_datetime=None,
+                ),
             }
             dummy_backend.DummyBatchJobs._job_result_registry = {}
 
@@ -1383,6 +1391,12 @@ def test_list_user_jobs_100(self, api100):
                     'progress': 100,
                     'created': "2024-06-04T14:20:23Z",
                 },
+                {
+                    "id": "j-24083059866540a38cab32b028be0ab5",
+                    "status": "finished",
+                    "progress": 100,
+                    "created": "2024-08-30T12:16:34Z",
+                }
             ],
             "links": []
         }
@@ -2684,8 +2698,7 @@ def test_get_job_result_item(self, flask_app, api110, backend_config_overrides):
         ("timeseries.csv", "text/csv"),
         ("timeseries.parquet", "application/parquet; profile=geo"),
     ])
-    def test_get_vector_cube_job_result_item(self, flask_app, api110, backend_config_overrides, vector_item_id,
-                                             vector_asset_media_type):
+    def test_get_vector_cube_job_result_item(self, flask_app, api110, vector_item_id, vector_asset_media_type):
         vector_asset_filename = vector_item_id
 
         with self._fresh_job_registry():
@@ -2747,6 +2760,66 @@ def test_get_vector_cube_job_result_item(self, flask_app, api110, backend_config
             extensions=resp_data.get("stac_extensions", []),
         )
 
+    def test_get_job_result_item_with_temporal_extent_on_asset(self, flask_app, api110):
+        with self._fresh_job_registry():
+            resp = api110.get(f"/jobs/j-24083059866540a38cab32b028be0ab5/results/items/timeseries.parquet",
+                              headers=self.AUTH_HEADER)
+
+        resp_data = resp.assert_status_code(200).json
+
+        assert resp_data == DictSubSet({
+            "type": "Feature",
+            "stac_version": "1.0.0",
+            "id": "timeseries.parquet",
+            "geometry": {"type": "Polygon",
+                                 "coordinates": [[[34.456156, -0.910085],
+                                                  [34.456156, -0.345477],
+                                                  [34.796396, -0.345477],
+                                                  [34.796396, -0.910085],
+                                                  [34.456156, -0.910085]]]},
+            "bbox": [34.456156, -0.910085, 34.796396, -0.345477],
+            "properties": {
+                "datetime": None,
+                "start_datetime": "2016-10-30T00:00:00+00:00",
+                "end_datetime": "2018-05-03T00:00:00+00:00",
+            },
+            "collection": "j-24083059866540a38cab32b028be0ab5",
+            "links": [
+                {
+                    "href": f"http://oeo.net/openeo/1.1.0/jobs/j-24083059866540a38cab32b028be0ab5/results/items/timeseries.parquet",
+                    "rel": "self",
+                    "type": "application/geo+json",
+                },
+                {
+                    "href": "http://oeo.net/openeo/1.1.0/jobs/j-24083059866540a38cab32b028be0ab5/results",
+                    "rel": "collection",
+                    "type": "application/json",
+                }
+            ],
+            "assets": {
+                "timeseries.parquet": {
+                    "href": f"http://oeo.net/openeo/1.1.0/jobs/j-24083059866540a38cab32b028be0ab5/results/assets/timeseries.parquet",
+                    "type": "application/parquet; profile=geo",
+                    "roles": ["data"],
+                    "title": "timeseries.parquet",
+                    "eo:bands": [
+                        {"name": "S1-SIGMA0-VV"},
+                        {"name": "S1-SIGMA0-VH"},
+                        {"name": "S2-L2A-B01"}
+                    ],
+                }
+            }
+        })
+
+        assert resp.headers["Content-Type"] == "application/geo+json"
+
+        pystac.validation.stac_validator.JsonSchemaSTACValidator().validate(
+            stac_dict=resp_data,
+            stac_object_type=pystac.STACObjectType.ITEM,
+            stac_version=resp_data.get("stac_version", "0.9.0"),
+            extensions=resp_data.get("stac_extensions", []),
+        )
+
     @mock.patch("time.time", mock.MagicMock(return_value=1234))
     @pytest.mark.parametrize("backend_config_overrides", [{"url_signer": UrlSigner(secret="123&@#", expiration=1000)}])
     def test_download_ml_model_metadata_signed(self, flask_app, api110, backend_config_overrides):