feat: external enhanced product types metadata

eodag/api/core.py

@@ -288,6 +288,7 @@ def build_index(self) -> None:
                 missionStartDate=fields.ID,
                 missionEndDate=fields.ID,
                 keywords=fields.KEYWORD(analyzer=kw_analyzer),
+                stacCollection=fields.STORED,
             )
             self._product_types_index = create_in(index_dir, product_types_schema)
             ix_writer = self._product_types_index.writer()

eodag/rest/core.py

@@ -689,6 +689,7 @@ def crunch_products(
 def eodag_api_init() -> None:
     """Init EODataAccessGateway server instance, pre-running all time consuming tasks"""
     eodag_api.fetch_product_types_list()
+    StacCollection.fetch_external_stac_collections(eodag_api)
 
     # pre-build search plugins
     for provider in eodag_api.available_providers():

eodag/rest/stac.py

@@ -36,6 +36,7 @@
 
 from eodag.api.product.metadata_mapping import (
     DEFAULT_METADATA_MAPPING,
+    NOT_AVAILABLE,
     format_metadata,
     get_metadata_path,
 )
@@ -53,8 +54,11 @@
 from eodag.utils.exceptions import (
     NoMatchingProductType,
     NotAvailableError,
+    RequestError,
+    TimeOutError,
     ValidationError,
 )
+from eodag.utils.requests import fetch_json
 
 if TYPE_CHECKING:
     from eodag.api.core import EODataAccessGateway
@@ -610,6 +614,34 @@ class StacCollection(StacCommon):
     :type root: str
     """
 
+    # External STAC collections
+    ext_stac_collections: Dict[str, Dict[str, Any]] = dict()
+
+    @classmethod
+    def fetch_external_stac_collections(cls, eodag_api: EODataAccessGateway) -> None:
+        """Load external STAC collections
+
+        :param eodag_api: EODAG python API instance
+        :type eodag_api: :class:`eodag.api.core.EODataAccessGateway`
+        """
+        list_product_types = eodag_api.list_product_types(fetch_providers=False)
+        for product_type in list_product_types:
+            ext_stac_collection_path = product_type.get("stacCollection")
+            if not ext_stac_collection_path:
+                continue
+            logger.info(f"Fetching external STAC collection for {product_type['ID']}")
+
+            try:
+                ext_stac_collection = fetch_json(ext_stac_collection_path)
+            except (RequestError, TimeOutError) as e:
+                logger.debug(e)
+                logger.warning(
+                    f"Could not read remote external STAC collection from {ext_stac_collection_path}",
+                )
+                ext_stac_collection = {}
+
+            cls.ext_stac_collections[product_type["ID"]] = ext_stac_collection
+
     def __init__(
         self,
         url: str,
@@ -717,6 +749,19 @@ def __get_collection_list(
                     "providers": providers_models,
                 },
             )
+            # override EODAG's collection with the external collection
+            product_type_id = product_type.get("_id", None) or product_type["ID"]
+            ext_stac_collection = self.ext_stac_collections.get(product_type_id, {})
+            # merge "keywords" list
+            merged_keywords = product_type_collection.get("keywords", [])
+            if "keywords" in ext_stac_collection:
+                new_keywords = ext_stac_collection["keywords"]
+                for v in new_keywords:
+                    if v != NOT_AVAILABLE and v not in merged_keywords:
+                        merged_keywords.append(v)
+            product_type_collection.update(ext_stac_collection)
+            if merged_keywords:
+                product_type_collection["keywords"] = merged_keywords
             # parse f-strings
             format_args = deepcopy(self.stac_config)
             format_args["collection"] = dict(

tests/units/test_stac_utils.py

@@ -27,6 +27,7 @@
 from pygeofilter.values import Geometry
 
 import eodag.rest.utils.rfc3339 as rfc3339
+from eodag.rest.stac import StacCollection
 from eodag.rest.types.stac_search import SearchPostRequest
 from eodag.rest.utils.cql_evaluate import EodagEvaluator
 from eodag.utils.exceptions import ValidationError
@@ -54,6 +55,12 @@ def setUpClass(cls):
 
         cls.rest_utils = rest_utils
 
+        import eodag.rest.core as rest_core
+
+        importlib.reload(rest_core)
+
+        cls.rest_core = rest_core
+
         search_results_file = os.path.join(
             TEST_RESOURCES_PATH, "eodag_search_result_peps.geojson"
         )
@@ -246,6 +253,42 @@ def test_get_datetime(self):
         self.assertEqual(dtstart, start)
         self.assertEqual(dtend, end)
 
+    def test_fetch_external_stac_collections(self):
+        """Load external STAC collections"""
+        external_json = """{
+            "new_field":"New Value",
+            "title":"A different title for Sentinel 2 MSI Level 1C",
+            "keywords":["New Keyword"]
+        }"""
+        product_type_conf = self.rest_core.eodag_api.product_types_config["S2_MSI_L1C"]
+        ext_stac_collection_path = "/path/to/external/stac/collections/S2_MSI_L1C.json"
+        product_type_conf["stacCollection"] = ext_stac_collection_path
+
+        with mock.patch(
+            "eodag.rest.stac.fetch_json",
+            autospec=True,
+            return_value=json.loads(external_json),
+        ) as mock_fetch_json:
+            # Check if the returned STAC collection contains updated data
+            StacCollection.fetch_external_stac_collections(self.rest_core.eodag_api)
+            stac_coll = self.rest_core.get_stac_collection_by_id(
+                url="", root="", collection_id="S2_MSI_L1C"
+            )
+            mock_fetch_json.assert_called_with(ext_stac_collection_path)
+            # New field
+            self.assertIn("new_field", stac_coll)
+            # Merge keywords
+            self.assertListEqual(
+                ["MSI", "SENTINEL2", "S2A,S2B", "L1", "OPTICAL", "New Keyword"],
+                stac_coll["keywords"],
+            )
+            # Override existing fields
+            self.assertEqual(
+                "A different title for Sentinel 2 MSI Level 1C", stac_coll["title"]
+            )
+            # Restore previous state
+            StacCollection.ext_stac_collections.clear()
+
 
 class TestEodagCql2jsonEvaluator(unittest.TestCase):
     def setUp(self):