invenio-vocabularies 6.3.1__py2.py3-none-any.whl → 6.4.1__py2.py3-none-any.whl

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/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/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/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/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/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-6.4.1.dist-info}/METADATA

@@ -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
@@ -83,6 +83,16 @@ https://invenio-vocabularies.readthedocs.io/
 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.dist-info → invenio_vocabularies-6.4.1.dist-info}/RECORD

@@ -1,10 +1,10 @@
-invenio_vocabularies/__init__.py,sha256=hDg_LXDMIoa_G123kgz3K8g9J3LghXoNWpYs4e3YSE4,377
+invenio_vocabularies/__init__.py,sha256=cnTvl-gNQL8JCB5A30kCWNNoQi84lT9ksWrLQbmr8eY,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
@@ -47,7 +47,7 @@ invenio_vocabularies/contrib/affiliations/__init__.py,sha256=rV8YAzBRoSKsBYcVjCN
 invenio_vocabularies/contrib/affiliations/affiliations.py,sha256=BSEjzAjLq17GGKHyPrgC5VLluvZqVnKMepIiqqLQKzo,2077
 invenio_vocabularies/contrib/affiliations/api.py,sha256=5nIOvpfcseuAAg2XgblHc8jb7TAdfU79XOBRpL-p398,326
 invenio_vocabularies/contrib/affiliations/config.py,sha256=kk-zAl7Uyi6N4qVOiCkBD9cN1KKOvDcILDHJ-zoIzJA,2084
-invenio_vocabularies/contrib/affiliations/datastreams.py,sha256=vEH6wj951LLa9pSsNb0rS8IVJ1zWbEQDDe0Cu770J9k,5479
+invenio_vocabularies/contrib/affiliations/datastreams.py,sha256=M99qM0R2_Tpa1IR_WC9xI2jHQDaVjvEHSn7Mh17_AHI,6060
 invenio_vocabularies/contrib/affiliations/facets.py,sha256=w316MGvtdyTpRCPOpCEmMxxLraRkbFFb1VvLkFlEc9o,1229
 invenio_vocabularies/contrib/affiliations/models.py,sha256=JUcj-1ydc2Cw2Rsc24JwXE3TFBJ_6fivhUYhGq4rT8A,329
 invenio_vocabularies/contrib/affiliations/resources.py,sha256=DBEbRxQmp-o-PeZlgFG588Q4sGcruuwIL8L9O-SzCes,435
@@ -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
@@ -139,7 +139,7 @@ invenio_vocabularies/contrib/subjects/datastreams.py,sha256=CiLSTm9PukzLKm34VDmm
 invenio_vocabularies/contrib/subjects/facets.py,sha256=qQ7_rppFBzsmrlZu4-MvOIdUcjeOmDA9gOHAcs0lWwI,695
 invenio_vocabularies/contrib/subjects/models.py,sha256=8XgbVRxDDvhWPjMWsoCriNlOKdmV_113a14yLRtlvM4,363
 invenio_vocabularies/contrib/subjects/resources.py,sha256=0KRfUMizwgIziZybk4HnIjiSsXbrCv_XmguNPwnxoo8,506
-invenio_vocabularies/contrib/subjects/schema.py,sha256=5pM9FFVdeb1nUuTwr2SqkUJh-oHIjYN-EXXI8_SSjCQ,2446
+invenio_vocabularies/contrib/subjects/schema.py,sha256=MtJ-YijukyK77MlCayGHJbjEMf7YTiRAS9nfFJ3YAyI,2813
 invenio_vocabularies/contrib/subjects/services.py,sha256=s1U6HMmpjuz7rrgR0DtT9C28TC6sZEeDTsa4Jh1TXQk,864
 invenio_vocabularies/contrib/subjects/subjects.py,sha256=NwZycExLyV8l7ikGStH4GOecVuDSxFT70KoNv6qC78I,1877
 invenio_vocabularies/contrib/subjects/euroscivoc/__init__.py,sha256=e5L9E4l5JHqVzijAX8tn2DIa2n01vJ5wOAZdN62RnIo,247
@@ -162,7 +162,7 @@ invenio_vocabularies/datastreams/factories.py,sha256=H8a2gAy7KNImtdCdtqpVKC5gIvE
 invenio_vocabularies/datastreams/readers.py,sha256=Gr_KbN-tZtoQ8R4TJG4dTBTE8028eLRlcxSv2-IqI9c,11386
 invenio_vocabularies/datastreams/tasks.py,sha256=0fuH_PRt9Ncv6WHM4pkYmfheRVGDKkERZiMPvgV4bZU,1129
 invenio_vocabularies/datastreams/transformers.py,sha256=0ymZiHtNtgfYxt2MIjthtSzikRH1jAqhIAZ1yScoAHs,1733
-invenio_vocabularies/datastreams/writers.py,sha256=PO9UPQ8rT4pstfoCiB-zCfuvCNOHRgFbysq6-spXm34,7218
+invenio_vocabularies/datastreams/writers.py,sha256=FMTQdGavRgq6Qk21UcifYeDVH3jUA3rmwdyICr1ywxU,6719
 invenio_vocabularies/datastreams/xml.py,sha256=HFa-lfxj7kFrr2IjeN1jxSLDfcvpBwO9nZLZF2-BryE,997
 invenio_vocabularies/records/__init__.py,sha256=Uj7O6fYdAtLOkLXUGSAYPADBB7aqP4yVs9b6OAjA158,243
 invenio_vocabularies/records/api.py,sha256=Lynt6Sz4BVN1orh0zgJ5ljhnUobEtcq8c22PmSeUo2U,1494
@@ -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-6.4.1.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
+invenio_vocabularies-6.4.1.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
+invenio_vocabularies-6.4.1.dist-info/METADATA,sha256=54qe6nCLmGJy9rNAImi2olrld8Cl2p3BGn2jp6u9ADY,10433
+invenio_vocabularies-6.4.1.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
+invenio_vocabularies-6.4.1.dist-info/entry_points.txt,sha256=xskfdwhOV074XHL1g8x8BZ6BNsdZ6D7XPVqcU5JqB2g,3118
+invenio_vocabularies-6.4.1.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
+invenio_vocabularies-6.4.1.dist-info/RECORD,,

{invenio_vocabularies-6.3.1.dist-info → invenio_vocabularies-6.4.1.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