record: added swh field and serializers

Pipfile

@@ -17,7 +17,7 @@ sentry-sdk = ">=1.45,<2.0.0"
 zenodo_rdm = {editable="True", path="./site"}
 zenodo_legacy = {editable="True", path="./legacy"}
 # TODO: Remove once we fix PyPI package issues
-invenio-swh = {git = "https://github.com/inveniosoftware/invenio-swh", ref = "v0.10.3"}
+invenio-swh = {git = "https://github.com/inveniosoftware/invenio-swh", ref = "v0.11.0"}
 jsonschema = ">=4.17.0,<4.18.0" # due to compatibility issues with alpha
 ipython = "!=8.1.0"
 uwsgi = ">=2.0"

invenio.cfg

@@ -1064,3 +1064,49 @@ COMMUNITIES_SHOW_BROWSE_MENU_ENTRY = True
 
 JOBS_ADMINISTRATION_ENABLED = True
 """Enable Jobs administration view."""
+
+from invenio_app_rdm.config import APP_RDM_RECORD_EXPORTERS as default_exporters
+
+APP_RDM_RECORD_EXPORTERS = {
+    **default_exporters, 
+    "bibtex": {
+        "name": _("BibTeX"),
+        "serializer": ("zenodo_rdm.serializers:ZenodoBibtexSerializer"),
+        "params": {},
+        "content-type": "application/x-bibtex",
+        "filename": "{id}.bib"
+    },
+    "codemeta": {
+        "name": _("Codemeta"),
+        "serializer": "zenodo_rdm.serializers:ZenodoCodemetaSerializer",
+        "params": {},
+        "content-type": "application/ld+json",
+        "filename": "{id}.json",
+    },
+    "datacite-json": {
+        "name": _("DataCite JSON"),
+        "serializer": (
+            "zenodo_rdm.serializers:ZenodoDataciteJSONSerializer"
+        ),
+        "params": {"options": {"indent": 2, "sort_keys": True}},
+        "content-type": "application/vnd.datacite.datacite+json",
+        "filename": "{id}.json",
+    },
+    "datacite-xml": {
+        "name": _("DataCite XML"),
+        "serializer": (
+            "zenodo_rdm.serializers:ZenodoDataciteXMLSerializer"
+        ),
+        "params": {},
+        "content-type": "application/vnd.datacite.datacite+xml",
+        "filename": "{id}.xml",
+    },
+     "cff": {
+        "name": _("Citation File Format"),
+        "serializer": "zenodo_rdm.serializers:ZenodoCFFSerializer",
+        "params": {},
+        "content-type": "application/x-yaml",
+        "filename": "{id}.yaml",
+    },
+
+}

site/zenodo_rdm/api.py

@@ -11,6 +11,7 @@
 from invenio_pidstore.providers.recordid import RecordIdProvider
 from invenio_rdm_records.records.api import RDMDraft, RDMParent, RDMRecord
 from invenio_records_resources.records.systemfields import PIDField
+from invenio_swh.records.systemfields import SWHSysField
 
 
 class DraftRecordIdProvider(RecordIdProvider):
@@ -50,6 +51,8 @@ class ZenodoRDMRecord(RDMRecord):
         ZenodoRDMParent, create=False, soft_delete=False, hard_delete=False
     )
 
+    swh = SWHSysField("swh")
+
 
 class ZenodoRDMDraft(RDMDraft):
     """Zenodo RDMDraft API class."""

site/zenodo_rdm/config.py

@@ -6,6 +6,7 @@
 # under the terms of the MIT License; see LICENSE file for more details.
 """Custom code config."""
 
+
 from .params import ZenodoArgsSchema, ZenodoSearchOptions
 from .redirector import (
     communities_detail_view_function,
@@ -27,6 +28,7 @@
     redirect_records_search_slash,
     search_view_function,
 )
+from .schema import ZenodoRecordSchema
 
 # I18N_TRANSLATIONS_PATHS = [os.path.abspath("./site/zenodo_rdm/translations")]
 
@@ -370,3 +372,7 @@ def lock_edit_record_published_files(service, identity, record=None, draft=None)
     "//cdnjs.cloudflare.com/ajax/libs/mathjax/3.2.2/es5/tex-mml-chtml.js"
     "?config=TeX-AMS-MML_HTMLorMML"
 )
+
+
+RDM_RECORD_SCHEMA = ZenodoRecordSchema
+"""Base record schema."""

site/zenodo_rdm/legacy/serializers/schemas/zenodojson.py

@@ -177,6 +177,8 @@ class ZenodoSchema(common.LegacySchema):
     files = fields.Method("dump_files", dump_only=True)
     metadata = fields.Nested(MetadataSchema)
 
+    swh = fields.Dict(dump_only=True)
+
     owners = fields.Method("dump_owners")
 
     def dump_owners(self, obj):

site/zenodo_rdm/schema.py

@@ -0,0 +1,26 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# Zenodo-RDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo-RDM service schema."""
+
+from invenio_rdm_records.services.schemas import RDMRecordSchema
+from marshmallow import Schema, fields
+
+
+class SWHSchema(Schema):
+    """Software Heritage schema."""
+
+    swhid = fields.Str()
+
+
+class ZenodoRecordSchema(RDMRecordSchema):
+    """Zenodo service schema.
+
+    This schema subclasses the base schema and extends it with Zenodo-specific
+    fields.
+    """
+
+    swh = fields.Nested(SWHSchema, dump_only=True)

site/zenodo_rdm/serializers/__init__.py

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo serializers."""
+
+from .bibtex import ZenodoBibtexSerializer
+from .cff import ZenodoCFFSerializer
+from .codemeta import ZenodoCodemetaSerializer
+from .datacite import ZenodoDataciteJSONSerializer, ZenodoDataciteXMLSerializer
+
+__all__ = (
+    "ZenodoBibtexSerializer",
+    "ZenodoCodemetaSerializer",
+    "ZenodoDataciteJSONSerializer",
+    "ZenodoDataciteXMLSerializer",
+    "ZenodoCFFSerializer",
+)

site/zenodo_rdm/serializers/bibtex.py

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo bibtex serializer."""
+
+from flask_resources import BaseListSchema, MarshmallowSerializer
+from flask_resources.serializers import SimpleSerializer
+from invenio_rdm_records.resources.serializers import BibtexSerializer
+from invenio_rdm_records.resources.serializers.bibtex.schema import BibTexSchema
+from marshmallow import fields, missing
+
+
+class ZenodoBibtexSchema(BibTexSchema):
+    """Zenodo bibtex schema."""
+
+    swhid = fields.Method("get_swhid")
+
+    def get_swhid(self, obj):
+        """Get swhid."""
+        return obj.get("swh", {}).get("swhid") or missing
+
+
+class ZenodoBibtexSerializer(MarshmallowSerializer):
+    """Zenodo bibtex serializer."""
+
+    def __init__(self, **options):
+        """Initialize serializer."""
+        super().__init__(
+            format_serializer_cls=SimpleSerializer,
+            object_schema_cls=ZenodoBibtexSchema,
+            list_schema_cls=BaseListSchema,
+            encoder=self.bibtex_tostring,
+        )
+
+    @classmethod
+    def bibtex_tostring(cls, record):
+        """Stringify a BibTex record."""
+        return record

site/zenodo_rdm/serializers/cff.py

@@ -0,0 +1,40 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo CFF serializer."""
+
+import yaml
+from flask_resources import BaseListSchema, MarshmallowSerializer
+from flask_resources.serializers import SimpleSerializer
+from invenio_rdm_records.resources.serializers.cff.schema import CFFSchema
+from marshmallow import missing
+
+
+class ZenodoCFFSchema(CFFSchema):
+    """Zenodo Codemeta schema."""
+
+    def get_identifiers(self, obj):
+        """Get identifiers."""
+        ret = super().get_identifiers(obj) or []
+        swhid = obj.get("swh", {}).get("swhid")
+        if swhid:
+            ret.append({"value": swhid, "type": "swh"})
+        return ret or missing
+
+
+class ZenodoCFFSerializer(MarshmallowSerializer):
+    """Zenodo Codemeta serializer."""
+
+    def __init__(self, **options):
+        """Initialize serializer."""
+        encoder = options.get("encoder", yaml.dump)
+        super().__init__(
+            format_serializer_cls=SimpleSerializer,
+            object_schema_cls=ZenodoCFFSchema,
+            list_schema_cls=BaseListSchema,
+            encoder=encoder,
+            **options,
+        )

site/zenodo_rdm/serializers/codemeta.py

@@ -0,0 +1,54 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo codemeta serializer."""
+
+from flask import current_app
+from flask_resources import BaseListSchema, MarshmallowSerializer
+from flask_resources.serializers import JSONSerializer
+from idutils import normalize_doi, to_url
+from invenio_rdm_records.contrib.codemeta.processors import CodemetaDumper
+from invenio_rdm_records.resources.serializers.codemeta.schema import CodemetaSchema
+from marshmallow import fields, missing
+
+
+class ZenodoCodemetaSchema(CodemetaSchema):
+    """Zenodo Codemeta schema."""
+
+    identifier = fields.Method("get_identifiers")
+
+    def get_identifiers(self, obj):
+        """Compute the "identifier".
+
+        It uses the DOI expressed as a URL and the Software Hash ID as `swhid`.
+        If only one identifier is present, it returns it as a single entry.
+        """
+        doi = obj.get("pids", {}).get("doi", {}).get("identifier")
+        ret = []
+        if doi:
+            doi_url = to_url(normalize_doi(doi), "doi")
+            ret.append({"@type": "doi", "value": doi, "propertyID": doi_url})
+        swhid = obj.get("swh", {}).get("swhid")
+        if swhid:
+            swh_url = f"{current_app.config['SWH_UI_BASE_URL']}/{swhid}"
+            ret.append({"@type": "swhid", "value": swhid, "propertyID": swh_url})
+        if len(ret) == 1:
+            return ret[0]
+        return ret or missing
+
+
+class ZenodoCodemetaSerializer(MarshmallowSerializer):
+    """Zenodo Codemeta serializer."""
+
+    def __init__(self, **options):
+        """Initialize serializer."""
+        super().__init__(
+            format_serializer_cls=JSONSerializer,
+            object_schema_cls=ZenodoCodemetaSchema,
+            list_schema_cls=BaseListSchema,
+            schema_kwargs={"dumpers": [CodemetaDumper()]},  # Order matters
+            **options,
+        )

site/zenodo_rdm/serializers/datacite.py

@@ -0,0 +1,63 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+"""Zenodo datacite serializer."""
+
+from datacite import schema43
+from flask import current_app
+from flask_resources import BaseListSchema, MarshmallowSerializer
+from flask_resources.serializers import JSONSerializer, SimpleSerializer
+from invenio_rdm_records.contrib.journal.processors import JournalDataciteDumper
+from invenio_rdm_records.resources.serializers.datacite.schema import DataCite43Schema
+from marshmallow import missing
+
+
+class ZenodoDataciteSchema(DataCite43Schema):
+    """Zenodo Datacite schema."""
+
+    def get_related_identifiers(self, obj):
+        """Get related identifiers."""
+        ret = super().get_related_identifiers(obj) or []
+        swhid = obj.get("swh", {}).get("swhid")
+        if swhid:
+            _url = f"{current_app.config['SWH_UI_BASE_URL']}/{swhid}"
+            ret.append(
+                {
+                    "relatedIdentifier": _url,
+                    "relatedIdentifierType": "URL",
+                    "relationType": "IsIdenticalTo",
+                }
+            )
+        return ret or missing
+
+
+class ZenodoDataciteJSONSerializer(MarshmallowSerializer):
+    """Zenodo Datacite serializer."""
+
+    def __init__(self, **options):
+        """Instantiate serializer."""
+        super().__init__(
+            format_serializer_cls=JSONSerializer,
+            object_schema_cls=ZenodoDataciteSchema,
+            list_schema_cls=BaseListSchema,
+            schema_kwargs={"dumpers": [JournalDataciteDumper()]},  # Order matters
+            **options,
+        )
+
+
+class ZenodoDataciteXMLSerializer(MarshmallowSerializer):
+    """Zenodo Datacite XML serializer."""
+
+    def __init__(self, **options):
+        """Instantiate serializer."""
+        encoder = options.get("encoder", schema43.tostring)
+        super().__init__(
+            format_serializer_cls=SimpleSerializer,
+            object_schema_cls=ZenodoDataciteSchema,
+            list_schema_cls=BaseListSchema,
+            schema_kwargs={"dumpers": [JournalDataciteDumper()]},  # Order matters
+            encoder=encoder,
+        )