Issue #699/#780 skip "no band source" warnings when walking STAC constructs

no need to trigger low level warnings when traversing and merging larger STAC hierarchies

Author: soxofaan

openeo/metadata.py

@@ -701,6 +701,10 @@ def merge(cls, band_lists: Iterable[_BandList]) -> _BandList:
         return cls(unique(all_bands, key=lambda b: b.name))
 
 
+_ON_EMPTY_WARN = "warn"
+_ON_EMPTY_IGNORE = "ignore"
+
+
 class _StacMetadataParser:
     """
     Helper to extract openEO metadata from STAC metadata resource
@@ -791,7 +795,7 @@ def bands_from_stac_object(self, obj: Union[pystac.STACObject, pystac.Asset]) ->
         else:
             raise ValueError(f"Unsupported STAC object: {obj!r}")
 
-    def bands_from_stac_catalog(self, catalog: pystac.Catalog) -> _BandList:
+    def bands_from_stac_catalog(self, catalog: pystac.Catalog, *, on_empty: str = _ON_EMPTY_WARN) -> _BandList:
         # TODO: "eo:bands" vs "bands" priority based on STAC and EO extension version information
         summaries = catalog.extra_fields.get("summaries", {})
         if "eo:bands" in summaries:
@@ -801,12 +805,17 @@ def bands_from_stac_catalog(self, catalog: pystac.Catalog) -> _BandList:
         elif "bands" in summaries:
             return _BandList(self._band_from_common_bands_metadata(b) for b in summaries["bands"])
 
-        # TODO: instead of warning: exception, or return None?
-        self._warn("bands_from_stac_catalog: no band name source found")
+        if on_empty == _ON_EMPTY_WARN:
+            self._warn("bands_from_stac_catalog: no band name source found")
         return _BandList([])
 
     def bands_from_stac_collection(
-        self, collection: pystac.Collection, *, consult_items: bool = True, consult_assets: bool = True
+        self,
+        collection: pystac.Collection,
+        *,
+        consult_items: bool = True,
+        consult_assets: bool = True,
+        on_empty: str = _ON_EMPTY_WARN,
     ) -> _BandList:
         # TODO: "eo:bands" vs "bands" priority based on STAC and EO extension version information
         self._log(f"bands_from_stac_collection with {collection.summaries.lists.keys()=}")
@@ -836,18 +845,25 @@ def bands_from_stac_collection(
         # If no band metadata so far: traverse items in collection
         elif consult_items:
             bands = _BandList.merge(
-                self.bands_from_stac_item(item=i, consult_collection=False, consult_assets=consult_assets)
+                self.bands_from_stac_item(
+                    item=i, consult_collection=False, consult_assets=consult_assets, on_empty=_ON_EMPTY_IGNORE
+                )
                 for i in collection.get_items()
             )
             if bands:
                 return bands
 
-        # TODO: instead of warning: exception, or return None?
-        self._warn("bands_from_stac_collection: no band name source found")
+        if on_empty == _ON_EMPTY_WARN:
+            self._warn("bands_from_stac_collection: no band name source found")
         return _BandList([])
 
     def bands_from_stac_item(
-        self, item: pystac.Item, *, consult_collection: bool = True, consult_assets: bool = True
+        self,
+        item: pystac.Item,
+        *,
+        consult_collection: bool = True,
+        consult_assets: bool = True,
+        on_empty: str = _ON_EMPTY_WARN,
     ) -> _BandList:
         # TODO: "eo:bands" vs "bands" priority based on STAC and EO extension version information
         self._log(f"bands_from_stac_item with {item.properties.keys()=}")
@@ -859,19 +875,21 @@ def bands_from_stac_item(
             return self.bands_from_stac_collection(collection=parent_collection)
         elif consult_assets:
             # TODO: filter on asset roles?
-            bands = _BandList.merge(self.bands_from_stac_asset(asset=a) for a in item.get_assets().values())
+            bands = _BandList.merge(
+                self.bands_from_stac_asset(asset=a, on_empty=_ON_EMPTY_IGNORE) for a in item.get_assets().values()
+            )
             if bands:
                 return bands
 
-        # TODO: instead of warning: exception, or return None?
-        self._warn("bands_from_stac_item: no band name source found")
+        if on_empty == _ON_EMPTY_WARN:
+            self._warn("bands_from_stac_item: no band name source found")
         return _BandList([])
 
     def _warn_undeclared_metadata(self, *, field: str, ext: str):
         """Helper to warn about using metadata from undeclared STAC extension"""
         self._warn(f"Using {field!r} metadata, but STAC extension {ext} was not declared.")
 
-    def bands_from_stac_asset(self, asset: pystac.Asset) -> _BandList:
+    def bands_from_stac_asset(self, asset: pystac.Asset, *, on_empty: str = _ON_EMPTY_WARN) -> _BandList:
         # TODO: "eo:bands" vs "bands" priority based on STAC and EO extension version information
         if _PYSTAC_1_9_EXTENSION_INTERFACE and asset.owner and asset.ext.has("eo") and asset.ext.eo.bands is not None:
             return _BandList(self._band_from_eo_bands_metadata(b) for b in asset.ext.eo.bands)
@@ -883,8 +901,8 @@ def bands_from_stac_asset(self, asset: pystac.Asset) -> _BandList:
             # TODO: avoid `extra_fields`, but built-in "bands" support seems to be scheduled for pystac V2
             return _BandList(self._band_from_common_bands_metadata(b) for b in asset.extra_fields["bands"])
 
-        # TODO: instead of warning: exception, or return None?
-        self._warn("bands_from_stac_asset: no band name source found")
+        if on_empty == _ON_EMPTY_WARN:
+            self._warn("bands_from_stac_asset: no band name source found")
         return _BandList([])
 
     def _bands_from_item_asset_definition(

tests/test_metadata.py

@@ -1354,9 +1354,10 @@ def test_bands_from_stac_collection(self, data, expected, caplog, expected_warni
             assert caplog.messages == expected_warnings
 
     @pytest.mark.parametrize(
-        ["entities", "kwargs", "expected"],
+        ["entities", "kwargs", "expected", "expected_warnings"],
         [
             (
+                # Basic: Collection with one item with eo:bands
                 {
                     "collection.json": StacDummyBuilder.collection(
                         stac_version="1.0.0",
@@ -1374,8 +1375,10 @@ def test_bands_from_stac_collection(self, data, expected, caplog, expected_warni
                 },
                 {},
                 [Band("B02"), Band("B03")],
+                [],
             ),
             (
+                # Collection with two items with different bands (eo:bands metadata)
                 {
                     "collection.json": StacDummyBuilder.collection(
                         stac_version="1.0.0",
@@ -1401,8 +1404,10 @@ def test_bands_from_stac_collection(self, data, expected, caplog, expected_warni
                 },
                 {},
                 [Band("B02"), Band("B03")],
+                [],
             ),
             (
+                # Collection with two items with different bands in different metadata format
                 {
                     "collection.json": StacDummyBuilder.collection(
                         stac_version="1.0.0",
@@ -1427,8 +1432,10 @@ def test_bands_from_stac_collection(self, data, expected, caplog, expected_warni
                 },
                 {},
                 [Band("B02"), Band("B03")],
+                [],
             ),
             (
+                # Collection with one item, with band metadata in asset
                 {
                     "collection.json": StacDummyBuilder.collection(
                         stac_version="1.0.0",
@@ -1450,15 +1457,51 @@ def test_bands_from_stac_collection(self, data, expected, caplog, expected_warni
                 },
                 {},
                 [Band("B02")],
+                [],
+            ),
+            (
+                # Collection with multiple items and assets, only partially with band metadata
+                {
+                    "collection.json": StacDummyBuilder.collection(
+                        stac_version="1.0.0",
+                        links=[
+                            {"rel": "item", "href": "item1.json"},
+                            {"rel": "item", "href": "item2.json"},
+                        ],
+                    ),
+                    "item1.json": StacDummyBuilder.item(
+                        stac_version="1.0.0",
+                        stac_extensions=["https://stac-extensions.github.io/eo/v1.1.0/schema.json"],
+                        assets={
+                            "asset11": {
+                                "href": "https://stac.test/asset11.tif",
+                                "roles": ["data"],
+                                "eo:bands": [{"name": "B02"}],
+                            },
+                            "asset12": {
+                                "href": "https://stac.test/asset12.tif",
+                                "roles": ["data"],
+                            },
+                        },
+                    ),
+                    "item2.json": StacDummyBuilder.item(),
+                },
+                {},
+                [Band("B02")],
+                [],
             ),
         ],
     )
-    def test_bands_from_stac_collection_consult_items(self, tmp_path, entities, kwargs, expected):
+    def test_bands_from_stac_collection_consult_items(
+        self, tmp_path, entities, kwargs, expected, caplog, expected_warnings
+    ):
         for name, content in entities.items():
             (tmp_path / name).write_text(json.dumps(content))
         collection = pystac.Collection.from_file(tmp_path / "collection.json")
         assert _StacMetadataParser().bands_from_stac_collection(collection=collection) == expected
 
+        assert caplog.messages == expected_warnings
+
     @pytest.mark.parametrize(
         ["data", "expected"],
         [