invenio-vocabularies 6.3.0__tar.gz → 6.4.0__tar.gz

{invenio-vocabularies-6.3.0 → invenio-vocabularies-6.4.0}/CHANGES.rst

@@ -8,6 +8,15 @@
 Changes
 =======
 
+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
+
 Version v6.3.0 (released 2024-10-11)
 
 - awards: get program from CORDIS

{invenio-vocabularies-6.3.0 → invenio-vocabularies-6.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.3.0
+Version: 6.4.0
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -50,6 +50,15 @@ Description: ..
         Changes
         =======
         
+        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
+        
         Version v6.3.0 (released 2024-10-11)
         
         - awards: get program from CORDIS

{invenio-vocabularies-6.3.0 → invenio-vocabularies-6.4.0}/invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "6.3.0"
+__version__ = "6.4.0"
 
 __all__ = ("__version__", "InvenioVocabularies")

{invenio-vocabularies-6.3.0 → invenio-vocabularies-6.4.0}/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-6.3.0 → invenio-vocabularies-6.4.0}/invenio_vocabularies/jobs.py

@@ -68,7 +68,7 @@ class ProcessRORAffiliationsJob(ProcessDataStreamJob):
             "config": {
                 "readers": [
                     {
-                        "args": {"since": since},
+                        "args": {"since": str(since)},
                         "type": "ror-http",
                     },
                     {"args": {"regex": "_schema_v2\\.json$"}, "type": "zip"},
@@ -111,7 +111,7 @@ class ProcessRORFundersJob(ProcessDataStreamJob):
             "config": {
                 "readers": [
                     {
-                        "args": {"since": since},
+                        "args": {"since": str(since)},
                         "type": "ror-http",
                     },
                     {"args": {"regex": "_schema_v2\\.json$"}, "type": "zip"},
@@ -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-6.3.0 → invenio-vocabularies-6.4.0}/invenio_vocabularies.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.3.0
+Version: 6.4.0
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -50,6 +50,15 @@ Description: ..
         Changes
         =======
         
+        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
+        
         Version v6.3.0 (released 2024-10-11)
         
         - awards: get program from CORDIS

{invenio-vocabularies-6.3.0 → invenio-vocabularies-6.4.0}/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-6.3.0 → invenio-vocabularies-6.4.0}/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-6.3.0 → invenio-vocabularies-6.4.0}/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",
     }