codemeta: Workaround pyld issue for compact operation

pyld is not correctly loading json-ld content from http*://schema.org
(see digitalbazaar/pyld#154) and raises an
exception when attempting to compact a codemeta document having
schema.org in its @context list.

As a workaround, remove schema.org from the @context list of a codemeta
document before compacting it.

Author: anlambert

swh/indexer/codemeta.py

@@ -146,6 +146,13 @@ def compact(doc, forgefed: bool):
           This is typically used for extrinsic metadata documents, which frequently
           use properties from these namespaces.
     """
+    if "@context" in doc and isinstance(doc["@context"], list):
+        # workaround pyld issue (https://github.com/digitalbazaar/pyld/issues/154)
+        doc["@context"] = [
+            context
+            for context in doc["@context"]
+            if not context.endswith("/schema.org")
+        ]
     contexts: List[Any] = [CODEMETA_CONTEXT_URL]
     if forgefed:
         contexts.append(

swh/indexer/tests/test_bibtex.py

@@ -299,6 +299,34 @@ def test_invalid_date():
     )
 
 
+def test_context_contains_schema_org():
+    assert codemeta_to_bibtex(
+        {
+            "@context": [
+                "https://doi.org/10.5063/schema/codemeta-2.0",
+                "http://schema.org",
+            ],
+            "author": {"name": "Jane Doe"},
+            "name": "Example Software",
+            "url": "http://example.org/",
+            "datePublished": "2023-10-10",
+            "license": "https://spdx.org/licenses/Apache-2.0",
+        }
+    ) == textwrap.dedent(
+        """\
+        @software{REPLACEME,
+            author = "Doe, Jane",
+            license = "Apache-2.0",
+            date = "2023-10-10",
+            year = "2023",
+            month = oct,
+            title = "Example Software",
+            url = "http://example.org/"
+        }
+        """
+    )
+
+
 def test_cff_empty():
     assert cff_to_bibtex("") == textwrap.dedent(
         """\