invenio-vocabularies 3.4.2__tar.gz → 4.1.1__tar.gz

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/.github/workflows/tests.yml

@@ -45,7 +45,7 @@ jobs:
     env:
       DB: ${{ matrix.db-service }}
       SEARCH: ${{ matrix.search-service }}
-      EXTRAS: tests,${{ matrix.SEARCH_EXTRAS }}
+      EXTRAS: tests,oaipmh,${{ matrix.SEARCH_EXTRAS }}
 
     steps:
       - name: Checkout

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/CHANGES.rst

@@ -8,14 +8,35 @@
 Changes
 =======
 
-Version v3.4.2 (released 2024-07-15)
+Version v4.1.1 (released 2024-07-15)
 
-- schema: add administration UI attributes
-
-Version 3.4.1 (released 2024-04-19)
+- installation: use invenio-oaipmh-scythe from PyPI
 
-- funders: exclude ROR schema v2 json (#309)
+Version v4.1.0 (released 2024-07-15)
 
+- readers: make OAI-PMH an optional extra
+- schema: add administration UI attributes
+- ror: fix duplicate acronymns and aliases
+- affiliations: fix title search
+- datastreams: have yaml writer output utf8
+- datastreams: add configs for funders and affiliations
+- affiliations: add datastreams
+- datastreams: move ror transformer to common
+- vocabulary-types: services, resources, and administration UI (#310)
+- config: add OpenAIRE mapping for "Latvian Council of Science"
+- funders: fix country name display (#343)
+- Initial implementation of OAIPMHReader (#329)
+- global: add "tags" field to all vocabularies
+
+Version 4.0.0 (released 2024-06-04)
+
+- datastreams: implement factories for generating vocabulary configurations
+- datastreams: added ROR HTTP reader
+- funders: use ROR v2 dump instead of v1
+- datastreams: added celery task for funders using ROR HTTP reader
+- datastreams: add OpenAIRE Project HTTP Reader
+- datastreams: fix OpenAIRE graph dataset parsing
+- installation: upgrade invenio-records-resources
 
 Version 3.4.0 (released 2024-04-19)
 

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: invenio-vocabularies
-Version: 3.4.2
+Version: 4.1.1
 Summary: "Invenio module for managing vocabularies."
 Home-page: https://github.com/inveniosoftware/invenio-vocabularies
 Author: CERN
@@ -50,14 +50,35 @@ Description: ..
         Changes
         =======
         
-        Version v3.4.2 (released 2024-07-15)
+        Version v4.1.1 (released 2024-07-15)
         
-        - schema: add administration UI attributes
-        
-        Version 3.4.1 (released 2024-04-19)
+        - installation: use invenio-oaipmh-scythe from PyPI
         
-        - funders: exclude ROR schema v2 json (#309)
+        Version v4.1.0 (released 2024-07-15)
         
+        - readers: make OAI-PMH an optional extra
+        - schema: add administration UI attributes
+        - ror: fix duplicate acronymns and aliases
+        - affiliations: fix title search
+        - datastreams: have yaml writer output utf8
+        - datastreams: add configs for funders and affiliations
+        - affiliations: add datastreams
+        - datastreams: move ror transformer to common
+        - vocabulary-types: services, resources, and administration UI (#310)
+        - config: add OpenAIRE mapping for "Latvian Council of Science"
+        - funders: fix country name display (#343)
+        - Initial implementation of OAIPMHReader (#329)
+        - global: add "tags" field to all vocabularies
+        
+        Version 4.0.0 (released 2024-06-04)
+        
+        - datastreams: implement factories for generating vocabulary configurations
+        - datastreams: added ROR HTTP reader
+        - funders: use ROR v2 dump instead of v1
+        - datastreams: added celery task for funders using ROR HTTP reader
+        - datastreams: add OpenAIRE Project HTTP Reader
+        - datastreams: fix OpenAIRE graph dataset parsing
+        - installation: upgrade invenio-records-resources
         
         Version 3.4.0 (released 2024-04-19)
         
@@ -221,6 +242,7 @@ Keywords: invenio vocabulary management
 Platform: any
 Classifier: Development Status :: 5 - Production/Stable
 Requires-Python: >=3.7
+Provides-Extra: oaipmh
 Provides-Extra: tests
 Provides-Extra: elasticsearch7
 Provides-Extra: opensearch1

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/__init__.py

@@ -10,6 +10,6 @@
 
 from .ext import InvenioVocabularies
 
-__version__ = "3.4.2"
+__version__ = "4.1.1"
 
 __all__ = ("__version__", "InvenioVocabularies")

invenio-vocabularies-4.1.1/invenio_vocabularies/administration/__init__.py

@@ -0,0 +1,10 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+# Copyright (C) 2024 Uni Münster.
+#
+# 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
+# details.
+
+"""Invenio administration views module for vocabularies."""

invenio-vocabularies-4.1.1/invenio_vocabularies/administration/views/__init__.py

@@ -0,0 +1,10 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2020-2024 CERN.
+# Copyright (C) 2024 Uni Münster.
+#
+# 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
+# details.
+
+"""Invenio administration views module for Vocabularies."""

invenio-vocabularies-4.1.1/invenio_vocabularies/administration/views/vocabularies.py

@@ -0,0 +1,44 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2020-2024 CERN.
+# Copyright (C) 2024 Uni Münster.
+#
+# 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
+# details.
+
+"""Vocabularies admin interface."""
+from invenio_administration.views.base import (
+    AdminResourceEditView,
+    AdminResourceListView,
+)
+
+
+class VocabulariesListView(AdminResourceListView):
+    """Configuration for vocabularies list view."""
+
+    api_endpoint = "/vocabularies/"
+    name = "vocabulary-types"
+    menu_label = "Vocabulary Types"
+    resource_config = "vocabulary_admin_resource"
+    search_request_headers = {"Accept": "application/json"}
+    title = "Vocabulary Types"
+    category = "Site management"
+
+    pid_path = "id"
+    icon = "exchange"
+    template = "invenio_administration/search.html"
+
+    display_search = True
+    display_delete = False
+    display_edit = False
+    display_create = False
+
+    item_field_list = {
+        "id": {"text": "Name", "order": 1},
+        "count": {"text": "Number of entries", "order": 2},
+    }
+
+    search_config_name = "VOCABULARIES_TYPES_SEARCH"
+    search_facets_config_name = "VOCABULARIES_TYPES_FACETS"
+    search_sort_config_name = "VOCABULARIES_TYPES_SORT_OPTIONS"

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/assets/semantic-ui/js/invenio_vocabularies/src/contrib/forms/Funding/CustomAwardForm.js

@@ -1,5 +1,5 @@
 // This file is part of InvenioVocabularies
-// Copyright (C) 2021-2023 CERN.
+// Copyright (C) 2021-2024 CERN.
 // Copyright (C) 2021 Northwestern University.
 //
 // Invenio is free software; you can redistribute it and/or modify it
@@ -16,21 +16,9 @@ import Overridable from "react-overridable";
 
 function CustomAwardForm({ deserializeFunder, selectedFunding }) {
   function deserializeFunderToDropdown(funderItem) {
-    let funderName = null;
-    let funderPID = null;
-    let funderCountry = null;
-
-    if (funderItem.name) {
-      funderName = funderItem.name;
-    }
-
-    if (funderItem.id) {
-      funderPID = funderItem.id;
-    }
-
-    if (funderItem.country) {
-      funderCountry = funderItem.country;
-    }
+    const funderName = funderItem?.name;
+    const funderPID = funderItem?.id;
+    const funderCountry = funderItem?.country_name ?? funderItem?.country;
 
     if (!funderName && !funderPID) {
       return {};
@@ -100,7 +88,7 @@ function CustomAwardForm({ deserializeFunder, selectedFunding }) {
       </Overridable>
       <Overridable id="InvenioVocabularies.CustomAwardForm.AwardInformationHeader.Container">
         <Header as="h3" size="small">
-          {i18next.t("Award information")} ({i18next.t("optional")})
+          {i18next.t("Additional information")} ({i18next.t("optional")})
         </Header>
       </Overridable>
       <Form.Group widths="equal">
@@ -110,7 +98,7 @@ function CustomAwardForm({ deserializeFunder, selectedFunding }) {
         >
           <TextField
             label={i18next.t("Number")}
-            placeholder={i18next.t("Award number")}
+            placeholder={i18next.t("Award/Grant number")}
             fieldPath="selectedFunding.award.number"
           />
         </Overridable>
@@ -120,7 +108,7 @@ function CustomAwardForm({ deserializeFunder, selectedFunding }) {
         >
           <TextField
             label={i18next.t("Title")}
-            placeholder={i18next.t("Award Title")}
+            placeholder={i18next.t("Award/Grant Title")}
             fieldPath="selectedFunding.award.title"
           />
         </Overridable>
@@ -130,7 +118,7 @@ function CustomAwardForm({ deserializeFunder, selectedFunding }) {
         >
           <TextField
             label={i18next.t("URL")}
-            placeholder={i18next.t("Award URL")}
+            placeholder={i18next.t("Award/Grant URL")}
             fieldPath="selectedFunding.award.url"
           />
         </Overridable>

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/assets/semantic-ui/js/invenio_vocabularies/src/contrib/forms/Funding/FundingField.js

@@ -1,5 +1,5 @@
 // This file is part of InvenioVocabularies
-// Copyright (C) 2021-2023 CERN.
+// Copyright (C) 2021-2024 CERN.
 // Copyright (C) 2021 Northwestern University.
 //
 // Invenio is free software; you can redistribute it and/or modify it
@@ -123,7 +123,7 @@ function FundingFieldForm(props) {
                 className="mb-5"
               >
                 <Icon name="add" />
-                {i18next.t("Add award")}
+                {i18next.t("Add")}
               </Button>
             }
             onAwardChange={(selectedFunding) => {

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/assets/semantic-ui/js/invenio_vocabularies/src/contrib/forms/Funding/FundingModal.js

@@ -1,5 +1,5 @@
 // This file is part of InvenioVocabularies
-// Copyright (C) 2021-2023 CERN.
+// Copyright (C) 2021-2024 CERN.
 // Copyright (C) 2021 Northwestern University.
 //
 // Invenio is free software; you can redistribute it and/or modify it
@@ -157,8 +157,8 @@ function FundingModal({
         >
           <Modal.Header as="h2" className="pt-10 pb-10">
             {mode === "standard"
-              ? i18next.t("Add standard award")
-              : i18next.t("Add custom award")}
+              ? i18next.t("Add standard award/grant")
+              : i18next.t("Add custom funding")}
           </Modal.Header>
           <Modal.Content>
             {mode === ModalTypes.STANDARD && (
@@ -172,7 +172,7 @@ function FundingModal({
                   <Grid.Row>
                     <Grid.Column width={11} floated="left" verticalAlign="middle">
                       <SearchBar
-                        placeholder={i18next.t("Search for awards")}
+                        placeholder={i18next.t("Search for awards/grants")}
                         autofocus
                         actionProps={{
                           icon: "search",
@@ -234,9 +234,7 @@ function FundingModal({
               primary
               icon="checkmark"
               content={
-                action === ModalActions.ADD
-                  ? i18next.t("Add award")
-                  : i18next.t("Change award")
+                action === ModalActions.ADD ? i18next.t("Add") : i18next.t("Change")
               }
             />
           </Modal.Actions>

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/assets/semantic-ui/js/invenio_vocabularies/src/contrib/forms/Funding/NoAwardResults.js

@@ -1,5 +1,5 @@
 // This file is part of InvenioVocabularies
-// Copyright (C) 2021-2023 CERN.
+// Copyright (C) 2021-2024 CERN.
 // Copyright (C) 2021 Northwestern University.
 //
 // Invenio is free software; you can redistribute it and/or modify it
@@ -16,7 +16,7 @@ export function NoAwardResults({ switchToCustom }) {
       basic
       content={
         <p>
-          {i18next.t("Did not find your award? ")}
+          {i18next.t("Did not find your award/grant? ")}
           <a
             href="/"
             onClick={(e) => {
@@ -24,7 +24,7 @@ export function NoAwardResults({ switchToCustom }) {
               switchToCustom();
             }}
           >
-            {i18next.t("Add a custom award.")}
+            {i18next.t("Add a custom award/grant.")}
           </a>
         </p>
       }

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/cli.py

@@ -9,46 +9,14 @@
 
 """Commands to create and manage vocabularies."""
 
-from copy import deepcopy
 
 import click
-import yaml
 from flask.cli import with_appcontext
 from invenio_access.permissions import system_identity
 from invenio_pidstore.errors import PIDDeletedError, PIDDoesNotExistError
-from invenio_records_resources.proxies import current_service_registry
 
-from .contrib.awards.datastreams import DATASTREAM_CONFIG as awards_ds_config
-from .contrib.funders.datastreams import DATASTREAM_CONFIG as funders_ds_config
-from .contrib.names.datastreams import DATASTREAM_CONFIG as names_ds_config
 from .datastreams import DataStreamFactory
-
-
-def get_config_for_ds(vocabulary, filepath=None, origin=None):
-    """Calculates the configuration for a Data Stream."""
-    config = None
-    if vocabulary == "names":  # FIXME: turn into a proper factory
-        config = deepcopy(names_ds_config)
-    elif vocabulary == "funders":
-        config = deepcopy(funders_ds_config)
-    elif vocabulary == "awards":
-        config = deepcopy(awards_ds_config)
-
-    if config:
-        if filepath:
-            with open(filepath) as f:
-                config = yaml.safe_load(f).get(vocabulary)
-        if origin:
-            config["readers"][0].setdefault("args", {})
-            config["readers"][0]["args"]["origin"] = origin
-
-    return config
-
-
-def get_service_for_vocabulary(vocabulary):
-    """Calculates the configuration for a Data Stream."""
-    if vocabulary == "names":  # FIXME: turn into a proper factory
-        return current_service_registry.get("names")
+from .factories import get_vocabulary_config
 
 
 @click.group()
@@ -111,7 +79,9 @@ def import_vocab(vocabulary, filepath=None, origin=None, num_samples=None):
         click.secho("One of --filepath or --origin must be present.", fg="red")
         exit(1)
 
-    config = get_config_for_ds(vocabulary, filepath, origin)
+    vc = get_vocabulary_config(vocabulary)
+    config = vc.get_config(filepath, origin)
+
     success, errored, filtered = _process_vocab(config, num_samples)
 
     _output_process(vocabulary, "imported", success, errored, filtered)
@@ -127,8 +97,8 @@ def update(vocabulary, filepath=None, origin=None):
     if not filepath and not origin:
         click.secho("One of --filepath or --origin must be present.", fg="red")
         exit(1)
-
-    config = get_config_for_ds(vocabulary, filepath, origin)
+    vc = get_vocabulary_config(vocabulary)
+    config = vc.get_config(filepath, origin)
 
     for w_conf in config["writers"]:
         w_conf["args"]["update"] = True
@@ -153,7 +123,8 @@ def convert(vocabulary, filepath=None, origin=None, target=None, num_samples=Non
         )
         exit(1)
 
-    config = get_config_for_ds(vocabulary, filepath, origin)
+    vc = get_vocabulary_config(vocabulary)
+    config = vc.get_config(filepath, origin)
     if not filepath:
         config["writers"] = [{"type": "yaml", "args": {"filepath": target}}]
 
@@ -176,8 +147,8 @@ def delete(vocabulary, identifier, all):
     if not id and not all:
         click.secho("An identifier or the --all flag must be present.", fg="red")
         exit(1)
-
-    service = get_service_for_vocabulary(vocabulary)
+    vc = get_vocabulary_config(vocabulary)
+    service = vc.get_service()
     if identifier:
         try:
             if service.delete(identifier, system_identity):

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/config.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2020 CERN.
+# Copyright (C) 2020-2024 CERN.
 # Copyright (C) 2021 Northwestern University.
 #
 # Invenio-Vocabularies is free software; you can redistribute it and/or
@@ -17,6 +17,7 @@ from .datastreams.readers import (
     GzipReader,
     JsonLinesReader,
     JsonReader,
+    OAIPMHReader,
     TarReader,
     XMLReader,
     YamlReader,
@@ -24,8 +25,8 @@ from .datastreams.readers import (
 )
 from .datastreams.transformers import XMLTransformer
 from .datastreams.writers import ServiceWriter, YamlWriter
-from .resources.resource import VocabulariesResourceConfig
-from .services.service import VocabulariesServiceConfig
+from .resources import VocabulariesResourceConfig
+from .services.config import VocabulariesServiceConfig
 
 VOCABULARIES_RESOURCE_CONFIG = VocabulariesResourceConfig
 """Configure the resource."""
@@ -75,6 +76,7 @@ VOCABULARIES_AWARDS_OPENAIRE_FUNDERS = {
     "fwf_________": "013tf3c58",
     "inca________": "03m8vkq32",
     "irb_hr______": "03n51vw80",
+    "lcs_________": "02ar66p97",
     "mestd_______": "01znas443",
     "nhmrc_______": "011kf5r70",
     "nih_________": "01cwqze88",
@@ -102,6 +104,15 @@ VOCABULARIES_NAMES_SCHEMES = {
 }
 """Names allowed identifier schemes."""
 
+# configure CUSTOM_VOCABULARY_TYPES to differentiate output. Is used in VocabulariesServiceConfig
+VOCABULARIES_CUSTOM_VOCABULARY_TYPES = [
+    "names",
+    "affiliations",
+    "awards",
+    "funders",
+    "subjects",
+]
+
 VOCABULARIES_DATASTREAM_READERS = {
     "csv": CSVReader,
     "json": JsonReader,
@@ -111,6 +122,7 @@ VOCABULARIES_DATASTREAM_READERS = {
     "yaml": YamlReader,
     "zip": ZipReader,
     "xml": XMLReader,
+    "oai-pmh": OAIPMHReader,
 }
 """Data Streams readers."""
 
@@ -124,3 +136,21 @@ VOCABULARIES_DATASTREAM_WRITERS = {
     "yaml": YamlWriter,
 }
 """Data Streams writers."""
+
+VOCABULARIES_TYPES_SORT_OPTIONS = {
+    "name": dict(
+        title=_("Name"),
+        fields=["id"],
+    ),
+    "count": dict(
+        title=_("Number of entries"),
+        fields=["count"],
+    ),
+}
+"""Definitions of available Vocabulary types sort options. """
+
+VOCABULARIES_TYPES_SEARCH = {
+    "facets": [],
+    "sort": ["name", "count"],
+}
+"""Vocabulary type search configuration."""

{invenio-vocabularies-3.4.2 → invenio-vocabularies-4.1.1}/invenio_vocabularies/contrib/affiliations/config.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2021 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
@@ -26,7 +26,7 @@ class AffiliationsSearchOptions(SearchOptions):
     """Search options."""
 
     suggest_parser_cls = SuggestQueryParser.factory(
-        fields=["name^100", "acronym^20", "id^20"],
+        fields=["name^100", "acronym^50", "title.*^20", "id^20", "aliases^20"],
     )
 
     sort_default = "bestmatch"

invenio-vocabularies-4.1.1/invenio_vocabularies/contrib/affiliations/datastreams.py

@@ -0,0 +1,67 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2022-2024 CERN.
+# Copyright (C) 2024 California Institute of Technology.
+#
+# 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
+# details.
+
+"""Affiliations datastreams, transformers, writers and readers."""
+
+from invenio_access.permissions import system_identity
+from invenio_i18n import lazy_gettext as _
+
+from ...datastreams.writers import ServiceWriter
+from .config import affiliation_schemes
+
+
+class AffiliationsServiceWriter(ServiceWriter):
+    """Affiliations service writer."""
+
+    def __init__(self, *args, **kwargs):
+        """Constructor."""
+        service_or_name = kwargs.pop("service_or_name", "affiliations")
+        super().__init__(service_or_name=service_or_name, *args, **kwargs)
+
+    def _entry_id(self, entry):
+        """Get the id from an entry."""
+        return entry["id"]
+
+
+VOCABULARIES_DATASTREAM_WRITERS = {
+    "affiliations-service": AffiliationsServiceWriter,
+}
+
+
+DATASTREAM_CONFIG = {
+    "readers": [
+        {
+            "type": "zip",
+            "args": {
+                "regex": "_schema_v2\\.json$",
+            },
+        },
+        {"type": "json"},
+    ],
+    "transformers": [
+        {
+            "type": "ror",
+            "args": {
+                "vocab_schemes": affiliation_schemes,
+            },
+        },
+    ],
+    "writers": [
+        {
+            "type": "affiliations-service",
+            "args": {
+                "identity": system_identity,
+            },
+        }
+    ],
+}
+"""Data Stream configuration.
+
+An origin is required for the reader.
+"""

invenio-vocabularies-4.1.1/invenio_vocabularies/contrib/affiliations/jsonschemas/affiliations/affiliation-v1.0.0.json

@@ -0,0 +1,63 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "local://affiliations/affiliation-v1.0.0.json",
+  "type": "object",
+  "additionalProperties": false,
+  "properties": {
+    "$schema": {
+      "$ref": "local://definitions-v1.0.0.json#/$schema"
+    },
+    "tags": {
+      "$ref": "local://vocabularies/definitions-v1.0.0.json#/tags"
+    },
+    "country": {
+      "type": "string",
+      "description": "Represents a affiliation's country as a country code."
+    },
+    "country_name": {
+      "type": "string",
+      "description": "Represents a affiliation's country as a full name."
+    },
+    "location_name": {
+      "type": "string",
+      "description": "Represents a affiliation's location name (usually a city)."
+    },
+    "acronym": {
+      "type": "string"
+    },
+    "identifiers": {
+      "description": "Alternate identifiers for the record.",
+      "type": "array",
+      "items": {
+        "$ref": "local://definitions-v1.0.0.json#/identifiers_with_scheme"
+      },
+      "uniqueItems": true
+    },
+    "status": {
+        "type": "string",
+        "description": "Status of the affiliation organization."
+    },
+    "aliases": {
+      "description": "Alternate names for the affiliation.",
+      "type": "array",
+      "items": {
+        "type": "string"
+      },
+      "uniqueItems": true
+    },
+    "types": {
+      "description": "Types of affiliation organization.",
+      "type": "array",
+      "items": {
+        "type": "string"
+      },
+      "uniqueItems": true
+    },
+    "name": {
+      "type": "string"
+    },
+    "title": {
+      "$ref": "local://vocabularies/definitions-v1.0.0.json#/title"
+    }
+  }
+}