invenio-vocabularies 6.2.0__py2.py3-none-any.whl → 6.3.0__py2.py3-none-any.whl

invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "6.2.0"
+__version__ = "6.3.0"
 
 __all__ = ("__version__", "InvenioVocabularies")

invenio_vocabularies/contrib/awards/datastreams.py

@@ -211,6 +211,56 @@ class CORDISProjectTransformer(BaseTransformer):
 
                 award["organizations"].append(organization_data)
 
+        programmes = (
+            record.get("relations", {}).get("associations", {}).get("programme", {})
+        )
+        if programmes:
+            # Projects with a single programme (this is the case of some projects in FP7) are not wrapped in a list,
+            # so we do this here to be able to iterate over it.
+            programmes = programmes if isinstance(programmes, list) else [programmes]
+
+            programmes_related_legal_basis = [
+                {
+                    "code": programme["code"],
+                    "uniqueprogrammepart": programme.get("@uniqueprogrammepart"),
+                }
+                for programme in programmes
+                if programme.get("@type") == "relatedLegalBasis"
+            ]
+
+            if len(programmes_related_legal_basis) == 0:
+                raise TransformerError(
+                    _(
+                        "No related legal basis programme found for project {project_id}".format(
+                            project_id=record["id"]
+                        )
+                    )
+                )
+            elif len(programmes_related_legal_basis) == 1:
+                # FP7 projects have only one related legal basis programme and do not have a 'uniqueprogrammepart' field.
+                unique_programme_related_legal_basis = programmes_related_legal_basis[0]
+            elif len(programmes_related_legal_basis) >= 1:
+                # The entry with the field 'uniqueprogrammepart' == 'true' is the high level programme code,
+                # while the other entry is a more specific sub-programme.
+                unique_programme_related_legal_basis = [
+                    programme_related_legal_basis
+                    # A few H2020 projects have more than one 'uniqueprogrammepart' == 'true',
+                    # for instance https://cordis.europa.eu/project/id/825673 (showing as "main programme" in the page)
+                    # which has one entry with the code 'H2020-EU.1.2.',
+                    # and one with the code 'H2020-EU.1.2.3.'.
+                    # We sort them from the shortest code to the longest code, and take the first item,
+                    # so that it conforms more with other projects which all have the shortest code as the main one.
+                    for programme_related_legal_basis in sorted(
+                        programmes_related_legal_basis, key=lambda d: len(d["code"])
+                    )
+                    if programme_related_legal_basis["uniqueprogrammepart"] == "true"
+                ][0]
+
+            # Store the code of the programme.
+            # For instance the code "HORIZON.1.2" which means "Marie Skłodowska-Curie Actions (MSCA)"
+            # See https://cordis.europa.eu/programme/id/HORIZON.1.2
+            award["program"] = unique_programme_related_legal_basis["code"]
+
         stream_entry.entry = award
         return stream_entry
 
@@ -222,7 +272,7 @@ class CORDISAwardsServiceWriter(ServiceWriter):
         """Constructor."""
         service_or_name = kwargs.pop("service_or_name", "awards")
         # Here we only update and we do not insert, since CORDIS data is used to augment existing awards
-        # (with subjects and organizations information) and is not used to create new awards.
+        # (with subjects, organizations, and program information) and is not used to create new awards.
         super().__init__(
             service_or_name=service_or_name, insert=False, update=True, *args, **kwargs
         )

invenio_vocabularies/contrib/common/ror/datastreams.py

@@ -160,6 +160,8 @@ class RORTransformer(BaseTransformer):
                     acronym = name["value"]
                 else:
                     aliases.add(name["value"])
+        if "en" not in ror["title"]:
+            ror["title"]["en"] = ror["name"]
         if acronym:
             ror["acronym"] = acronym
         if aliases:

invenio_vocabularies/jobs.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2021-2022 CERN.
+# Copyright (C) 2021-2024 CERN.
 #
 # Invenio-Vocabularies is free software; you can redistribute it and/or
 # modify it under the terms of the MIT License; see LICENSE file for more
@@ -60,8 +60,10 @@ class ProcessRORAffiliationsJob(ProcessDataStreamJob):
         if since is None and job_obj.last_runs["success"]:
             since = job_obj.last_runs["success"].started_at
         else:
-            since = datetime.datetime.now()
+            since = since or datetime.datetime.now()
 
+        # NOTE: Update is set to False for now given we don't have the logic to re-index dependent records yet.
+        # Since jobs support custom args, update true can be passed via that.
         return {
             "config": {
                 "readers": [
@@ -77,7 +79,7 @@ class ProcessRORAffiliationsJob(ProcessDataStreamJob):
                         "args": {
                             "writer": {
                                 "type": "affiliations-service",
-                                "args": {"update": True},
+                                "args": {"update": False},
                             }
                         },
                         "type": "async",
@@ -86,3 +88,46 @@ class ProcessRORAffiliationsJob(ProcessDataStreamJob):
                 "transformers": [{"type": "ror-affiliations"}],
             }
         }
+
+
+class ProcessRORFundersJob(ProcessDataStreamJob):
+    """Process ROR funders datastream registered task."""
+
+    description = "Process ROR funders"
+    title = "Load ROR funders"
+    id = "process_ror_funders"
+
+    @classmethod
+    def default_args(cls, job_obj, since=None, **kwargs):
+        """Generate default job arguments here."""
+        if since is None and job_obj.last_runs["success"]:
+            since = job_obj.last_runs["success"].started_at
+        else:
+            since = since or datetime.datetime.now()
+
+        # NOTE: Update is set to False for now given we don't have the logic to re-index dependent records yet.
+        # Since jobs support custom args, update true can be passed via that.
+        return {
+            "config": {
+                "readers": [
+                    {
+                        "args": {"since": since},
+                        "type": "ror-http",
+                    },
+                    {"args": {"regex": "_schema_v2\\.json$"}, "type": "zip"},
+                    {"type": "json"},
+                ],
+                "writers": [
+                    {
+                        "args": {
+                            "writer": {
+                                "type": "funders-service",
+                                "args": {"update": False},
+                            }
+                        },
+                        "type": "async",
+                    }
+                ],
+                "transformers": [{"type": "ror-funders"}],
+            }
+        }

invenio_vocabularies/webpack.py

@@ -35,7 +35,7 @@ theme = WebpackThemeBundle(
                 "react-dropzone": "^11.0.0",
                 "react-i18next": "^11.11.0",
                 "react-invenio-forms": "^4.0.0",
-                "react-searchkit": "^2.0.0",
+                "react-searchkit": "^3.0.0",
                 "yup": "^0.32.0",
             },
             aliases={

{invenio_vocabularies-6.2.0.dist-info → invenio_vocabularies-6.3.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 6.2.0
+Version: 6.3.0
 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 v6.3.0 (released 2024-10-11)
+
+- awards: get program from CORDIS
+- fix: add 'en' title if missing ROR
+- fix: since not passed to args
+- jobs: add process funders job
+
+Version v6.2.1 (released 2024-10-10)
+
+- webpack: bump react-searchkit due to axios upgrade
+
 Version v6.2.0 (released 2024-10-10)
 
 - tests: update axios version (needed only for local js tests)

{invenio_vocabularies-6.2.0.dist-info → invenio_vocabularies-6.3.0.dist-info}/RECORD

@@ -1,13 +1,13 @@
-invenio_vocabularies/__init__.py,sha256=3owW-TfH2DDgOkGa0ObjwTskfWvOzTJH9ZqJap3QH2A,377
+invenio_vocabularies/__init__.py,sha256=uT7gvYe4pp5fuXqR4Kh8SNLYVi0ZRPcdav187WrWiyk,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=QBWgScrUlKJagpOvwXN12G1OsvW6NM-IFKRgkbW_edg,2549
+invenio_vocabularies/jobs.py,sha256=yPK4HDnJI9vP7rAnRUXhZPRKfiznOUaO-FJU7RMjGQ0,4275
 invenio_vocabularies/proxies.py,sha256=k7cTUgWfnCoYIuNqAj_VFi1zBN33KNNclRSVnBkObEM,711
 invenio_vocabularies/views.py,sha256=PNJ5nvc3O7ASwNe56xmqy5YaU9n3UYF3W2JwvtE_kYs,1561
-invenio_vocabularies/webpack.py,sha256=bp2vz3O8QdZwrznsV-SjAmS0g1nV9WGWqc-7ZJQY5PQ,1891
+invenio_vocabularies/webpack.py,sha256=hzTM0qx6iiRHkmjti53yuZ5ebfPMR5mpO9uNewBat74,1891
 invenio_vocabularies/administration/__init__.py,sha256=0bDp2Aw8aZth7C-q9Xn9rxeCUQQRoIUxoWWWwPvKbXA,308
 invenio_vocabularies/administration/views/__init__.py,sha256=31DP4jLG6q4HQlzSRiGLPxUjHPUCCl4N34y4XMuPP6g,313
 invenio_vocabularies/administration/views/vocabularies.py,sha256=JyKr1OYF9DO89zvWCNOtNfClO_QptdHdjvgY1kkImnw,1290
@@ -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=j6dtimniWg9faHzV1J49QpAgUJDDW3rjQR0_Mhcfqm4,11147
+invenio_vocabularies/contrib/awards/datastreams.py,sha256=k7GtQ0pmY9EehYUtus45G6opJb8v2Kc63YIn5zO4SU8,14004
 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
@@ -87,7 +87,7 @@ invenio_vocabularies/contrib/common/__init__.py,sha256=DdbGYRthEpQtObhY_YK4vOT9Z
 invenio_vocabularies/contrib/common/openaire/__init__.py,sha256=L7UtSimFJ3NI6j53bHzYKsWpFti1uo4fPb9OaTl7leI,244
 invenio_vocabularies/contrib/common/openaire/datastreams.py,sha256=BV6NtBCPFuii6KbTHGkgNQO5tt_3Hn9T_219bz8AINg,3514
 invenio_vocabularies/contrib/common/ror/__init__.py,sha256=3u2-fre1SQ-4nz3Ay0nxj3ntmMZ8Ujh_4eV-fyxfmtc,239
-invenio_vocabularies/contrib/common/ror/datastreams.py,sha256=arhB2OKu3Hl0Sn6urTxBQh_kim4oc9TTMRAwzi4pbXE,8057
+invenio_vocabularies/contrib/common/ror/datastreams.py,sha256=wHpJy5mRKgJ4nQ7IniLyjxoxjTCZKHaEKKKr_OKrqFM,8139
 invenio_vocabularies/contrib/funders/__init__.py,sha256=YxFXBDnT7NM8rFwxT_Ge3xXR2n17EM0alknQq7r_Bt8,478
 invenio_vocabularies/contrib/funders/api.py,sha256=QKGGeSnPHSoBfucvpaVruXT_txYidofZ080G3IxFkIo,306
 invenio_vocabularies/contrib/funders/config.py,sha256=EU7UrwLOkr2Bem9Skz_HJIxyBQRkXEdPT8zIuV8vbzI,2217
@@ -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-6.2.0.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
-invenio_vocabularies-6.2.0.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
-invenio_vocabularies-6.2.0.dist-info/METADATA,sha256=nmPiV7XQY2p8bV7PHfPV6oIWo5IHEKhrHKIijWiWfMk,9761
-invenio_vocabularies-6.2.0.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
-invenio_vocabularies-6.2.0.dist-info/entry_points.txt,sha256=qm4ydo2p2KWN3937X8argeeSHw-BX_IJ7gS11pkADEU,2903
-invenio_vocabularies-6.2.0.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
-invenio_vocabularies-6.2.0.dist-info/RECORD,,
+invenio_vocabularies-6.3.0.dist-info/AUTHORS.rst,sha256=8d0p_WWE1r9DavvzMDi2D4YIGBHiMYcN3LYxqQOj8sY,291
+invenio_vocabularies-6.3.0.dist-info/LICENSE,sha256=UvI8pR8jGWqe0sTkb_hRG6eIrozzWwWzyCGEpuXX4KE,1062
+invenio_vocabularies-6.3.0.dist-info/METADATA,sha256=9NYJRAerai4L9KSUjbnUV1O4CBk9ewHXpShhXi-PMTM,10027
+invenio_vocabularies-6.3.0.dist-info/WHEEL,sha256=-G_t0oGuE7UD0DrSpVZnq1hHMBV9DD2XkS5v7XpmTnk,110
+invenio_vocabularies-6.3.0.dist-info/entry_points.txt,sha256=P9wLzJDDXGd4BEzRX4ylmnWge8wg_Q9jI0NBxNRVA2Q,2972
+invenio_vocabularies-6.3.0.dist-info/top_level.txt,sha256=x1gRNbaODF_bCD0SBLM3nVOFPGi06cmGX5X94WKrFKk,21
+invenio_vocabularies-6.3.0.dist-info/RECORD,,

{invenio_vocabularies-6.2.0.dist-info → invenio_vocabularies-6.3.0.dist-info}/entry_points.txt

@@ -51,6 +51,7 @@ invenio_vocabularies = invenio_vocabularies
 
 [invenio_jobs.jobs]
 process_ror_affiliations = invenio_vocabularies.jobs:ProcessRORAffiliationsJob
+process_ror_funders = invenio_vocabularies.jobs:ProcessRORFundersJob
 
 [invenio_jsonschemas.schemas]
 affiliations = invenio_vocabularies.contrib.affiliations.jsonschemas