COPDS-2654: schema.org & isPartOf & distribution (#112)

* chore: schema.org

* chore: distribution metadata

Author: alfredotoselli

cads_catalogue_api_service/models/schema_org.py

@@ -35,7 +35,7 @@ class Organization(pydantic.BaseModel):
 class DataDownload(pydantic.BaseModel):
     type: str = pydantic.Field("DataDownload", const=True, alias="@type")
     encodingFormat: str
-    contentUrl: str
+    url: str | None = None
 
 
 class GeoShape(pydantic.BaseModel):

cads_catalogue_api_service/schema_org.py

@@ -34,6 +34,18 @@
 )
 
 
+# List subject to change in case of new portals
+CADS_SITE_TO_LONG_NAME = {
+    "cds": "Climate Data Store",
+    "ads": "Atmosphere Data Store",
+    "cems": "CEMS Early Warning Data Store",
+    "ecds": "ECMWF Data Store",
+    "xds": "ECMWF Cross Data Store",
+    "eds": "Energy Thematic Hub",
+    "hds": "Health Thematic Hub",
+}
+
+
 def query_collection(
     session: sa.orm.Session,
     collection_id: str,
@@ -80,12 +92,14 @@ def schema_org_json_ld(
         url = get_url_link(collection, "self")
         license = get_url_link(collection, "license")
         distribution = get_url_link(collection, "layout")
+        retrieve_url = get_url_link(collection, "retrieve")
     temporal_coverage = (
         collection.get("extent", {}).get("temporal", {}).get("interval", [])
     )
     temporal_coverage = (
         list(filter(None, temporal_coverage[0])) if temporal_coverage else []
     )
+    download_url = f"{os.getenv(f'{site.upper()}_PROJECT_URL', None)}/datasets/{collection_id}?tab=download"
 
     box = collection.get("extent", {}).get("spatial", {}).get("bbox", [])
 
@@ -117,13 +131,21 @@ def schema_org_json_ld(
             (
                 models.schema_org.DataDownload(
                     encodingFormat=collection.get("file_format")
-                    # Sometimes the file_format is not defined on the input data
                     or "application/octet-stream",
-                    contentUrl=f"{url}?tab=download",
+                    url=f"{retrieve_url}",
                 )
                 if distribution
                 else ""
-            )
+            ),
+            (
+                models.schema_org.DataDownload(
+                    encodingFormat=collection.get("file_format")
+                    or "application/octet-stream",
+                    url=download_url,
+                )
+                if distribution
+                else ""
+            ),
         ],
         temporalCoverage="/".join(temporal_coverage) if temporal_coverage else None,
         spatialCoverage=models.schema_org.Place(
@@ -141,7 +163,7 @@ def schema_org_json_ld(
             {
                 "@type": "DataCatalog",
                 "identifier": site,
-                "name": "ECMWF Data Store",
+                "name": CADS_SITE_TO_LONG_NAME.get(site, "ECMWF Data Store"),
                 "url": f"{os.getenv(f'{site.upper()}_PROJECT_URL', None)}/datasets"
                 if site
                 else None,

tests/test_40_schema_org.py

@@ -81,6 +81,11 @@ def static_collection_query(
                 "href": "http://localhost:8080/api/catalogue/v1/collections/era5-something/layout",
                 "title": "Distribution",
             },
+            {
+                "rel": "retrieve",
+                "href": "http://localhost:8080/api/retrieve/v1/processes/era5-something",
+                "title": "Retrieve",
+            },
         ],
         "assets": {
             "thumbnail": {
@@ -139,8 +144,13 @@ def test_schema_org_jsonId(monkeypatch) -> None:
             {
                 "@type": "DataDownload",
                 "encodingFormat": "application/octet-stream",
-                "contentUrl": "http://localhost:8080/api/catalogue/v1/collections/era5-something?tab=download",
-            }
+                "url": "http://localhost:8080/api/retrieve/v1/processes/era5-something",
+            },
+            {
+                "@type": "DataDownload",
+                "encodingFormat": "application/octet-stream",
+                "url": "https://cds.climate.copernicus.eu/datasets/era5-something?tab=download",
+            },
         ],
         "temporalCoverage": "2019-11-05T00:00:00Z/2023-06-22T00:00:00Z",
         "spatialCoverage": {
@@ -155,7 +165,7 @@ def test_schema_org_jsonId(monkeypatch) -> None:
             {
                 "@type": "DataCatalog",
                 "identifier": "cds",
-                "name": "ECMWF Data Store",
+                "name": "Climate Data Store",
                 "url": "https://cds.climate.copernicus.eu/datasets",
             }
         ],