invenio-vocabularies 6.3.1__py2.py3-none-any.whl → 6.4.0__py2.py3-none-any.whl

invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "6.3.1"
+__version__ = "6.4.0"
 
 __all__ = ("__version__", "InvenioVocabularies")

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/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.dist-info → invenio_vocabularies-6.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.3.1
+Version: 6.4.0
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -83,6 +83,11 @@ https://invenio-vocabularies.readthedocs.io/
 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

{invenio_vocabularies-6.3.1.dist-info → invenio_vocabularies-6.4.0.dist-info}/RECORD

@@ -1,10 +1,10 @@
-invenio_vocabularies/__init__.py,sha256=hDg_LXDMIoa_G123kgz3K8g9J3LghXoNWpYs4e3YSE4,377
+invenio_vocabularies/__init__.py,sha256=jky3oxkIimkLgucO1IDyWkzGZqBqrjPrgFGLkFtnx8s,377
 invenio_vocabularies/cli.py,sha256=S3lBsLxsSYa83sCDaGZr5mP7TwPvmmwxzzbB13h8VBI,5856
 invenio_vocabularies/config.py,sha256=bpNKVgwfb7bgkP5zbmoawnAMD6bHWXIJV-6CpEhi-M8,5752
 invenio_vocabularies/ext.py,sha256=GujJ4UARd4Fxf4z7zznRk9JAgHamZuYCOdrKU5czg00,5987
 invenio_vocabularies/factories.py,sha256=mVg4yGKe58e4uS8rYe0DmIO6oMpmtNTcK3wH9eM5jVU,4380
 invenio_vocabularies/fixtures.py,sha256=nNWwH04HFASjfj1oy5kMdcQGKmVjzUuA5wSw-ER1QAg,1585
-invenio_vocabularies/jobs.py,sha256=LxLt3zmfgulDdf5z7RGB70QF8GMPzgdD4vAGf8ZQENY,4285
+invenio_vocabularies/jobs.py,sha256=OoQOBQoWOLHZ16EMlVM54OezgO58u_aMDemAjbCuUk8,6371
 invenio_vocabularies/proxies.py,sha256=k7cTUgWfnCoYIuNqAj_VFi1zBN33KNNclRSVnBkObEM,711
 invenio_vocabularies/views.py,sha256=PNJ5nvc3O7ASwNe56xmqy5YaU9n3UYF3W2JwvtE_kYs,1561
 invenio_vocabularies/webpack.py,sha256=hzTM0qx6iiRHkmjti53yuZ5ebfPMR5mpO9uNewBat74,1891
@@ -68,7 +68,7 @@ invenio_vocabularies/contrib/awards/__init__.py,sha256=KwCmwFalz-3pDs9iTa5TKUidB
 invenio_vocabularies/contrib/awards/api.py,sha256=OXukE7PLXs45BTtqVrhvGBNqLmQaI-CgXmHTCi36LZk,303
 invenio_vocabularies/contrib/awards/awards.py,sha256=tOLvcvTPiN1gn1QAl-hSh1bwcclg8Kx2ZMmJMEhI7vk,2959
 invenio_vocabularies/contrib/awards/config.py,sha256=PlDHabkWDUzwa1Fvk_U2hG83kQYBqM1IyChg8Yg_VlY,1630
-invenio_vocabularies/contrib/awards/datastreams.py,sha256=k7GtQ0pmY9EehYUtus45G6opJb8v2Kc63YIn5zO4SU8,14004
+invenio_vocabularies/contrib/awards/datastreams.py,sha256=4J6JFCqYosDO8y4FRq-LTWgRNPbeH4Tj4Tur6ziChtY,14234
 invenio_vocabularies/contrib/awards/models.py,sha256=mM-kSNf7kDH3oIbV8epxxbUi7muYqi4JreXxgWXlVzw,318
 invenio_vocabularies/contrib/awards/resources.py,sha256=_9YTqbhz8axFXGhG5y4WyjE27p9n-7e3c6HoBRditPA,411
 invenio_vocabularies/contrib/awards/schema.py,sha256=P_k9EONMMx0eWpALVuhGBzZlDeh4599elLlmMis-Vko,3302
@@ -297,10 +297,10 @@ invenio_vocabularies/translations/zh_CN/LC_MESSAGES/messages.mo,sha256=g1I5aNO8r
 invenio_vocabularies/translations/zh_CN/LC_MESSAGES/messages.po,sha256=vg8qC8ofpAdJ3mQz7mWM1ylKDpiNWXFs7rlMdSPkgKk,4629
 invenio_vocabularies/translations/zh_TW/LC_MESSAGES/messages.mo,sha256=cqSm8NtMAwrP9O6qbmtkDtRT1e9D93qpsJN5X9_PPVw,600
 invenio_vocabularies/translations/zh_TW/LC_MESSAGES/messages.po,sha256=9ACePz_EpB-LfcIJajZ2kp8Q04tcdrQLOtug162ZUss,4115
-invenio_vocabularies-6.3.1.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
-invenio_vocabularies-6.3.1.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
-invenio_vocabularies-6.3.1.dist-info/METADATA,sha256=i448NUrQmoezzIp3sY16Lxd2aQ4OKNUOdoOWCrtWUIo,10103
-invenio_vocabularies-6.3.1.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
-invenio_vocabularies-6.3.1.dist-info/entry_points.txt,sha256=P9wLzJDDXGd4BEzRX4ylmnWge8wg_Q9jI0NBxNRVA2Q,2972
-invenio_vocabularies-6.3.1.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
-invenio_vocabularies-6.3.1.dist-info/RECORD,,
+invenio_vocabularies-6.4.0.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
+invenio_vocabularies-6.4.0.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
+invenio_vocabularies-6.4.0.dist-info/METADATA,sha256=bex12nauN9gFRDmQpgZPHPfhUM3Hrdd7K41hGPwogsI,10260
+invenio_vocabularies-6.4.0.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
+invenio_vocabularies-6.4.0.dist-info/entry_points.txt,sha256=xskfdwhOV074XHL1g8x8BZ6BNsdZ6D7XPVqcU5JqB2g,3118
+invenio_vocabularies-6.4.0.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
+invenio_vocabularies-6.4.0.dist-info/RECORD,,

{invenio_vocabularies-6.3.1.dist-info → invenio_vocabularies-6.4.0.dist-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