invenio-vocabularies 6.4.0__py2.py3-none-any.whl → 6.5.0__py2.py3-none-any.whl

invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "6.4.0"
+__version__ = "6.5.0"
 
 __all__ = ("__version__", "InvenioVocabularies")

invenio_vocabularies/config.py

@@ -172,7 +172,7 @@ VOCABULARIES_TYPES_SEARCH = {
 }
 """Vocabulary type search configuration."""
 
-SUBJECTS_EUROSCIVOC_FILE_URL = "https://op.europa.eu/o/opportal-service/euvoc-download-handler?cellarURI=http%3A%2F%2Fpublications.europa.eu%2Fresource%2Fdistribution%2Feuroscivoc%2F20231115-0%2Frdf%2Fskos_ap_eu%2FEuroSciVoc-skos-ap-eu.rdf&fileName=EuroSciVoc-skos-ap-eu.rdf"
+SUBJECTS_EUROSCIVOC_FILE_URL = "https://publications.europa.eu/resource/distribution/euroscivoc/rdf/skos_ap_eu/EuroSciVoc-skos-ap-eu.rdf"
 """Subject EuroSciVoc file download link."""
 
 VOCABULARIES_ORCID_ACCESS_KEY = "TODO"

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/funders/mappings/os-v1/funders/funder-v2.0.0.json

@@ -14,8 +14,8 @@
           "type": "custom",
           "char_filter": ["strip_special_chars"],
           "filter": [
-            "lowercase",
-            "asciifolding",
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal",
             "edgegrams"
           ]
         },
@@ -24,8 +24,8 @@
           "type": "custom",
           "char_filter": ["strip_special_chars"],
           "filter": [
-            "lowercase",
-            "asciifolding"
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal"
           ]
         }
       },
@@ -40,11 +40,11 @@
         }
       },
       "filter": {
-        "lowercase": {
+        "lowercasepreserveoriginal": {
           "type": "lowercase",
           "preserve_original": true
         },
-        "asciifolding": {
+        "asciifoldingpreserveoriginal": {
           "type": "asciifolding",
           "preserve_original": true
         },

invenio_vocabularies/contrib/funders/mappings/os-v2/funders/funder-v2.0.0.json

@@ -14,8 +14,8 @@
           "type": "custom",
           "char_filter": ["strip_special_chars"],
           "filter": [
-            "lowercase",
-            "asciifolding",
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal",
             "edgegrams"
           ]
         },
@@ -24,8 +24,8 @@
           "type": "custom",
           "char_filter": ["strip_special_chars"],
           "filter": [
-            "lowercase",
-            "asciifolding"
+            "lowercasepreserveoriginal",
+            "asciifoldingpreserveoriginal"
           ]
         }
       },
@@ -40,11 +40,11 @@
         }
       },
       "filter": {
-        "lowercase": {
+        "lowercasepreserveoriginal": {
           "type": "lowercase",
           "preserve_original": true
         },
-        "asciifolding": {
+        "asciifoldingpreserveoriginal": {
           "type": "asciifolding",
           "preserve_original": true
         },

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-6.4.0.dist-info → invenio_vocabularies-6.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.4.0
+Version: 6.5.0
 Summary: Invenio module for managing vocabularies.
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -83,6 +83,17 @@ https://invenio-vocabularies.readthedocs.io/
 Changes
 =======
 
+Version v6.5.0 (released 2024-10-31)
+
+- subjects: euroscivoc: change default to latest version-less URL
+- Rename patched filters so the normalizer uses the default ones (#409)
+    * rename patched filters so the normalizer uses the default ones
+
+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

{invenio_vocabularies-6.4.0.dist-info → invenio_vocabularies-6.5.0.dist-info}/RECORD

@@ -1,6 +1,6 @@
-invenio_vocabularies/__init__.py,sha256=jky3oxkIimkLgucO1IDyWkzGZqBqrjPrgFGLkFtnx8s,377
+invenio_vocabularies/__init__.py,sha256=RojgHMHpQAzjFxZhRH31r4GvA4kZAkhfR-T-HZvyrGQ,377
 invenio_vocabularies/cli.py,sha256=S3lBsLxsSYa83sCDaGZr5mP7TwPvmmwxzzbB13h8VBI,5856
-invenio_vocabularies/config.py,sha256=bpNKVgwfb7bgkP5zbmoawnAMD6bHWXIJV-6CpEhi-M8,5752
+invenio_vocabularies/config.py,sha256=l-mKnoZaIO0p-9K2gnlBvExdjzpQw5aU1lFI16ZI4Mk,5614
 invenio_vocabularies/ext.py,sha256=GujJ4UARd4Fxf4z7zznRk9JAgHamZuYCOdrKU5czg00,5987
 invenio_vocabularies/factories.py,sha256=mVg4yGKe58e4uS8rYe0DmIO6oMpmtNTcK3wH9eM5jVU,4380
 invenio_vocabularies/fixtures.py,sha256=nNWwH04HFASjfj1oy5kMdcQGKmVjzUuA5wSw-ER1QAg,1585
@@ -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
@@ -104,10 +104,10 @@ invenio_vocabularies/contrib/funders/jsonschemas/funders/funder-v1.0.0.json,sha2
 invenio_vocabularies/contrib/funders/mappings/__init__.py,sha256=aSr-tZd9rsjet6leeS336gdSdZHXwZKdaPStNtVNQVk,244
 invenio_vocabularies/contrib/funders/mappings/os-v1/__init__.py,sha256=xXEX3tacmXp0I1KFtDw7ohIahozd2oIGp1UN40IhFic,251
 invenio_vocabularies/contrib/funders/mappings/os-v1/funders/funder-v1.0.0.json,sha256=E7Zp4IHsQGdaxVrksr-SaQtieV7tV0W6-LgGe231G1w,1646
-invenio_vocabularies/contrib/funders/mappings/os-v1/funders/funder-v2.0.0.json,sha256=31shkTYZDvx7tSCnOA4OlqRmQ0IEv2qhePJsIStReJo,3385
+invenio_vocabularies/contrib/funders/mappings/os-v1/funders/funder-v2.0.0.json,sha256=gx2CkNWNcTxP94iCU9remCrWxR-AmNqloYTh47gwNHQ,3481
 invenio_vocabularies/contrib/funders/mappings/os-v2/__init__.py,sha256=YvMRlKYTnEmyTzI9smZp_lO3w-zcK-8IpqT-jGUXEEY,251
 invenio_vocabularies/contrib/funders/mappings/os-v2/funders/funder-v1.0.0.json,sha256=E7Zp4IHsQGdaxVrksr-SaQtieV7tV0W6-LgGe231G1w,1646
-invenio_vocabularies/contrib/funders/mappings/os-v2/funders/funder-v2.0.0.json,sha256=31shkTYZDvx7tSCnOA4OlqRmQ0IEv2qhePJsIStReJo,3385
+invenio_vocabularies/contrib/funders/mappings/os-v2/funders/funder-v2.0.0.json,sha256=gx2CkNWNcTxP94iCU9remCrWxR-AmNqloYTh47gwNHQ,3481
 invenio_vocabularies/contrib/funders/mappings/v7/__init__.py,sha256=yFHmi3QYD65YKzLU5vMEtwAZn0gwkFYa6Db_tSbHjKE,254
 invenio_vocabularies/contrib/funders/mappings/v7/funders/funder-v1.0.0.json,sha256=E7Zp4IHsQGdaxVrksr-SaQtieV7tV0W6-LgGe231G1w,1646
 invenio_vocabularies/contrib/names/__init__.py,sha256=DBfsM7JMETZGaV5QmXEwE7zhCaAXvc2SZN6uXnW_V-c,451
@@ -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-6.4.0.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-6.5.0.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
+invenio_vocabularies-6.5.0.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
+invenio_vocabularies-6.5.0.dist-info/METADATA,sha256=VRg_aDoS0NrNaNMJTEiX-k-kMCuSMDKo7pQBHgOjP9I,10679
+invenio_vocabularies-6.5.0.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
+invenio_vocabularies-6.5.0.dist-info/entry_points.txt,sha256=xskfdwhOV074XHL1g8x8BZ6BNsdZ6D7XPVqcU5JqB2g,3118
+invenio_vocabularies-6.5.0.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
+invenio_vocabularies-6.5.0.dist-info/RECORD,,