invenio-vocabularies 6.3.1__tar.gz → 6.4.1__tar.gz

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/CHANGES.rst

@@ -8,6 +8,16 @@
 Changes
 =======
 
+Version v6.4.1 (released 2024-10-15)
+
+- fix: exclude unknown fields when updating awards with subjects
+- fix: revert generic writer and define OpenAIRE awards writer logic
+
+Version v6.4.0 (released 2024-10-15)
+
+- jobs: add import awards OpenAIRE; Update CORDIS
+- awards: rollback to use the 2nd part of funding stream as program
+
 Version v6.3.1 (released 2024-10-11)
 
 - jobs: pass since as string to task

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.3.1
+Version: 6.4.1
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -50,6 +50,16 @@ Description: ..
         Changes
         =======
         
+        Version v6.4.1 (released 2024-10-15)
+        
+        - fix: exclude unknown fields when updating awards with subjects
+        - fix: revert generic writer and define OpenAIRE awards writer logic
+        
+        Version v6.4.0 (released 2024-10-15)
+        
+        - jobs: add import awards OpenAIRE; Update CORDIS
+        - awards: rollback to use the 2nd part of funding stream as program
+        
         Version v6.3.1 (released 2024-10-11)
         
         - jobs: pass since as string to task

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "6.3.1"
+__version__ = "6.4.1"
 
 __all__ = ("__version__", "InvenioVocabularies")

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/contrib/affiliations/datastreams.py

@@ -9,8 +9,11 @@
 
 """Affiliations datastreams, transformers, writers and readers."""
 
+from copy import deepcopy
+
 from flask import current_app
 
+from ...datastreams import StreamEntry
 from ...datastreams.errors import TransformerError, WriterError
 from ...datastreams.transformers import BaseTransformer
 from ...datastreams.writers import ServiceWriter
@@ -96,15 +99,24 @@ class OpenAIREAffiliationsServiceWriter(ServiceWriter):
         """Get the id from an entry."""
         return entry["id"]
 
-    def write(self, stream_entry, *args, **kwargs):
-        """Writes the input entry using a given service."""
-        entry = stream_entry.entry
-
-        return super().write(stream_entry, *args, **kwargs)
-
-    def write_many(self, stream_entries, *args, **kwargs):
-        """Writes the input entries using a given service."""
-        return super().write_many(stream_entries, *args, **kwargs)
+    def _do_update(self, entry):
+        vocab_id = self._entry_id(entry)
+        current = self._resolve(vocab_id)
+        updated = deepcopy(current.to_dict())
+
+        if "identifiers" in entry:
+            # For each new identifier
+            for new_identifier in entry["identifiers"]:
+                # Either find an existing identifier with the same scheme and update the "identifier" value
+                for existing_identifier in updated["identifiers"]:
+                    if existing_identifier["scheme"] == new_identifier["scheme"]:
+                        existing_identifier["identifier"] = new_identifier["identifier"]
+                        break
+                # Or add the new identifier to the list of identifiers
+                else:
+                    updated["identifiers"].append(new_identifier)
+
+        return StreamEntry(self._service.update(self._identity, vocab_id, updated))
 
 
 VOCABULARIES_DATASTREAM_READERS = {}

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/contrib/awards/datastreams.py

@@ -74,7 +74,10 @@ class OpenAIREProjectTransformer(BaseTransformer):
 
         funding = next(iter(record.get("funding", [])), None)
         if funding:
-            program = funding.get("fundingStream", {}).get("id", "")
+            funding_stream_id = funding.get("fundingStream", {}).get("id", "")
+            # Example funding stream ID: `EC::HE::HORIZON-AG-UN`. We need the `HE`
+            # string, i.e. the second "part" of the identifier.
+            program = next(iter(funding_stream_id.split("::")[1:2]), "")
             if program:
                 award["program"] = program
 

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/contrib/subjects/schema.py

@@ -13,7 +13,7 @@
 from functools import partial
 
 from invenio_i18n import get_locale
-from marshmallow import Schema, fields, pre_load
+from marshmallow import EXCLUDE, Schema, fields, pre_load
 from marshmallow_utils.fields import IdentifierSet, SanitizedUnicode
 from marshmallow_utils.schemas import IdentifierSchema
 
@@ -59,6 +59,14 @@ class SubjectSchema(BaseVocabularySchema):
 class SubjectRelationSchema(ContribVocabularyRelationSchema):
     """Schema to define an optional subject relation in another schema."""
 
+    # If re-running an OpenAIRE awards update on existing awards which already have subjects,
+    # the subject entries will contains `scheme` and `props`, which are unknown since they are `dump_only`.
+    # This makes the update exclude unknown field and go through with the update.
+    class Meta:
+        """Metadata class."""
+
+        unknown = EXCLUDE
+
     ftf_name = "subject"
     parent_field_name = "subjects"
     subject = SanitizedUnicode()

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/datastreams/writers.py

@@ -87,23 +87,8 @@ class ServiceWriter(BaseWriter):
     def _do_update(self, entry):
         vocab_id = self._entry_id(entry)
         current = self._resolve(vocab_id)
-        combined_dict = current.to_dict()
-
-        # Update fields from entry
-        for key, value in entry.items():
-            if key in combined_dict:
-                if isinstance(combined_dict[key], list) and isinstance(value, list):
-                    combined_dict[key].extend(
-                        item for item in value if item not in combined_dict[key]
-                    )
-                else:
-                    combined_dict[key] = value
-            else:
-                combined_dict[key] = value
-
-        return StreamEntry(
-            self._service.update(self._identity, vocab_id, combined_dict)
-        )
+        updated = dict(current.to_dict(), **entry)
+        return StreamEntry(self._service.update(self._identity, vocab_id, updated))
 
     def write(self, stream_entry, *args, **kwargs):
         """Writes the input entry using a given service."""

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies/jobs.py

@@ -131,3 +131,66 @@ class ProcessRORFundersJob(ProcessDataStreamJob):
                 "transformers": [{"type": "ror-funders"}],
             }
         }
+
+
+class ImportAwardsOpenAIREJob(ProcessDataStreamJob):
+    """Import awards from OpenAIRE registered task."""
+
+    description = "Import awards from OpenAIRE"
+    title = "Import Awards OpenAIRE"
+    id = "import_awards_openaire"
+
+    @classmethod
+    def default_args(cls, job_obj, **kwargs):
+        """Generate default job arguments."""
+        return {
+            "config": {
+                "readers": [
+                    {
+                        "type": "openaire-http",
+                        "args": {"origin": "diff", "tar_href": "/project.tar"},
+                    },
+                    {
+                        "type": "tar",
+                        "args": {
+                            "mode": "r",
+                            "regex": "\\.json.gz$",
+                        },
+                    },
+                    {"type": "gzip"},
+                    {"type": "jsonl"},
+                ],
+                "transformers": [{"type": "openaire-award"}],
+                "writers": [
+                    {"args": {"writer": {"type": "awards-service"}}, "type": "async"}
+                ],
+            }
+        }
+
+
+class UpdateAwardsCordisJob(ProcessDataStreamJob):
+    """Update awards from CORDIS registered task."""
+
+    description = "Update awards from CORDIS"
+    title = "Update Awards CORDIS"
+    id = "update_awards_cordis"
+
+    @classmethod
+    def default_args(cls, job_obj, **kwargs):
+        """Generate default job arguments."""
+        return {
+            "config": {
+                "readers": [
+                    {"args": {"origin": "HE"}, "type": "cordis-project-http"},
+                    {"args": {"mode": "r", "regex": "\\.xml$"}, "type": "zip"},
+                    {"args": {"root_element": "project"}, "type": "xml"},
+                ],
+                "transformers": [{"type": "cordis-award"}],
+                "writers": [
+                    {
+                        "args": {"writer": {"type": "cordis-awards-service"}},
+                        "type": "async",
+                    }
+                ],
+            }
+        }

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.3.1
+Version: 6.4.1
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -50,6 +50,16 @@ Description: ..
         Changes
         =======
         
+        Version v6.4.1 (released 2024-10-15)
+        
+        - fix: exclude unknown fields when updating awards with subjects
+        - fix: revert generic writer and define OpenAIRE awards writer logic
+        
+        Version v6.4.0 (released 2024-10-15)
+        
+        - jobs: add import awards OpenAIRE; Update CORDIS
+        - awards: rollback to use the 2nd part of funding stream as program
+        
         Version v6.3.1 (released 2024-10-11)
         
         - jobs: pass since as string to task

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/invenio_vocabularies.egg-info/entry_points.txt

@@ -50,8 +50,10 @@ vocabulary_model = invenio_vocabularies.records.models
 invenio_vocabularies = invenio_vocabularies
 
 [invenio_jobs.jobs]
+import_awards_openaire = invenio_vocabularies.jobs:ImportAwardsOpenAIREJob
 process_ror_affiliations = invenio_vocabularies.jobs:ProcessRORAffiliationsJob
 process_ror_funders = invenio_vocabularies.jobs:ProcessRORFundersJob
+update_awards_cordis = invenio_vocabularies.jobs:UpdateAwardsCordisJob
 
 [invenio_jsonschemas.schemas]
 affiliations = invenio_vocabularies.contrib.affiliations.jsonschemas

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/setup.cfg

@@ -105,6 +105,8 @@ invenio_celery.tasks =
 invenio_jobs.jobs = 
 	process_ror_affiliations = invenio_vocabularies.jobs:ProcessRORAffiliationsJob
 	process_ror_funders = invenio_vocabularies.jobs:ProcessRORFundersJob
+	import_awards_openaire = invenio_vocabularies.jobs:ImportAwardsOpenAIREJob
+	update_awards_cordis = invenio_vocabularies.jobs:UpdateAwardsCordisJob
 
 [build_sphinx]
 source-dir = docs/

{invenio-vocabularies-6.3.1 → invenio-vocabularies-6.4.1}/tests/contrib/awards/test_awards_datastreams.py

@@ -92,7 +92,7 @@ def expected_from_award_json():
         "id": "021nxhr62::0751743",
         "identifiers": [{"identifier": "https://test.com", "scheme": "url"}],
         "number": "0751743",
-        "program": "NSF::GEO/OAD::GEO/OCE",
+        "program": "GEO/OAD",
         "title": {"en": "Test title"},
         "funder": {"id": "021nxhr62"},
         "acronym": "TA",
@@ -110,7 +110,7 @@ def expected_from_award_json_ec():
         "title": {"en": "Test title"},
         "funder": {"id": "00k4n6c32"},
         "acronym": "TS",
-        "program": "TST::test::test",
+        "program": "test",
     }