feat: update datacite model to version 4.6

I am not sure what the best way to deal with schema versions is. The
easiest thing and the current strategy is to provide the latest versions
that are backward-compatible with prior versions (within reason) i.e.:

* `CodeMetaV3` can accept v2 fields and turn them into v3 equivalents
* `DataCiteV46` implements the latest changes to v4, and does not break
   compatibility with any other v4 version

Citation File Format provides jsonschema so in this case, it would be
easy to provide seperate models

Author: sgfost

codemeticulous/cff/models.py

@@ -1575,7 +1575,7 @@ class Reference(BaseModel):
     """
 
 
-class CitationFileFormat(ByAliasExcludeNoneMixin, BaseModel):
+class CitationFileFormatV120(ByAliasExcludeNoneMixin, BaseModel):
     model_config = ConfigDict(
         extra="forbid",
         populate_by_name=True,
@@ -1694,3 +1694,6 @@ class CitationFileFormat(ByAliasExcludeNoneMixin, BaseModel):
     """
     The version of the software or dataset.
     """
+
+
+CitationFileFormat = CitationFileFormatV120

codemeticulous/codemeta/models.py

@@ -40,7 +40,7 @@ class VersionedLanguage(ComputerLanguage):
 SoftwareListOrSingle = Software | SoftwareList
 
 
-class CodeMeta(ByAliasExcludeNoneMixin, BaseModel):
+class CodeMetaV3(ByAliasExcludeNoneMixin, BaseModel):
     """CodeMeta v3 schema (supports v2 fields aliased to v3)
     see: https://codemeta.github.io/terms/
     and: https://github.com/codemeta/codemeta-generator/blob/master/js/validation/
@@ -272,3 +272,6 @@ def validate_sub_type(cls, value, base_class):
 
     class Config:
         allow_population_by_field_name = True
+
+
+CodeMeta = CodeMetaV3

codemeticulous/datacite/convert.py

@@ -154,6 +154,7 @@
         "financial backer",
     },
     ContributorType.Supervisor: {"supervisor", "advisor", "overseer", "mentor"},
+    ContributorType.Translator: { "translator", "translation" },
     ContributorType.WorkPackageLeader: {
         "work package leader",
         "package leader",
@@ -222,9 +223,17 @@ def codemeta_actors_to_datacite(
             # match roles to contributor types
             matched_roles = set()
             for role in extractor.role_names:
-                normalized_role = role.lower().replace(" ", "").replace("-", "").replace("_", "")
-                for contributor_type, synonyms, in CONTRIBUTOR_TYPE_MAP.items():
-                    if normalized_role in {s.lower().replace(" ", "").replace("-", "").replace("_", "") for s in synonyms}:
+                normalized_role = (
+                    role.lower().replace(" ", "").replace("-", "").replace("_", "")
+                )
+                for (
+                    contributor_type,
+                    synonyms,
+                ) in CONTRIBUTOR_TYPE_MAP.items():
+                    if normalized_role in {
+                        s.lower().replace(" ", "").replace("-", "").replace("_", "")
+                        for s in synonyms
+                    }:
                         matched_roles.add(contributor_type)
             # if we have no matched roles, default to "Other"
             if not matched_roles:
@@ -235,13 +244,16 @@ def codemeta_actors_to_datacite(
                     Contributor(
                         name=extractor.name,
                         nameType=(
-                            "Organizational" if extractor.is_organization else "Personal"
+                            "Organizational"
+                            if extractor.is_organization
+                            else "Personal"
                         ),
                         givenName=extractor.given_names,
                         familyName=extractor.family_names,
                         nameIdentifiers=[NameIdentifier(**i) for i in name_identifiers]
                         or None,
-                        affiliation=[AffiliationItem(**a) for a in affiliations] or None,
+                        affiliation=[AffiliationItem(**a) for a in affiliations]
+                        or None,
                         contributorType=role_type.value,
                     )
                 )

codemeticulous/datacite/models.py

@@ -1,5 +1,5 @@
 """
-generated by datamodel-codegen (schema/datacite/schema45.json)
+generated by datamodel-codegen (schema/datacite/schema46.json)
 with options:
     --output-model-type pydantic_v2.BaseModel \
     --field-constraints \
@@ -13,7 +13,7 @@
 MANUAL CHANGES:
 - add ByAliasExcludeNoneMixin to DataciteV45
 - add identifier and identifierType to Container, this is included in examples
-  for schema 4.5 but not in the schema itself..
+  but not in the source jsonschema
 """
 
 from __future__ import annotations
@@ -163,6 +163,7 @@ class ContributorType(Enum):
     RightsHolder = "RightsHolder"
     Sponsor = "Sponsor"
     Supervisor = "Supervisor"
+    Translator = "Translator"
     WorkPackageLeader = "WorkPackageLeader"
     Other = "Other"
 
@@ -179,6 +180,7 @@ class DateType(Enum):
     Available = "Available"
     Copyrighted = "Copyrighted"
     Collected = "Collected"
+    Coverage = "Coverage"
     Created = "Created"
     Issued = "Issued"
     Submitted = "Submitted"
@@ -190,6 +192,7 @@ class DateType(Enum):
 
 class ResourceTypeGeneral(Enum):
     Audiovisual = "Audiovisual"
+    Award = "Award"
     Book = "Book"
     BookChapter = "BookChapter"
     Collection = "Collection"
@@ -210,6 +213,7 @@ class ResourceTypeGeneral(Enum):
     PeerReview = "PeerReview"
     PhysicalObject = "PhysicalObject"
     Preprint = "Preprint"
+    Project = "Project"
     Report = "Report"
     Service = "Service"
     Software = "Software"
@@ -225,6 +229,7 @@ class RelatedIdentifierType(Enum):
     ARK = "ARK"
     arXiv = "arXiv"
     bibcode = "bibcode"
+    CTSR = "CTSR"
     DOI = "DOI"
     EAN13 = "EAN13"
     EISSN = "EISSN"
@@ -237,6 +242,7 @@ class RelatedIdentifierType(Enum):
     LSID = "LSID"
     PMID = "PMID"
     PURL = "PURL"
+    RRID = "RRID"
     UPC = "UPC"
     URL = "URL"
     URN = "URN"
@@ -280,6 +286,8 @@ class RelationType(Enum):
     Requires = "Requires"
     IsObsoletedBy = "IsObsoletedBy"
     Obsoletes = "Obsoletes"
+    isTranslationOf = "isTranslationOf"
+    hasTranslation = "hasTranslation"
 
 
 class RelatedObject(BaseModel):
@@ -469,7 +477,7 @@ class RelatedItem(RelatedObject):
     relationType: RelationType
 
 
-class DataciteV45(ByAliasExcludeNoneMixin, BaseModel):
+class DataciteV46(ByAliasExcludeNoneMixin, BaseModel):
     model_config = ConfigDict(
         extra="forbid",
         populate_by_name=True,
@@ -502,3 +510,6 @@ class DataciteV45(ByAliasExcludeNoneMixin, BaseModel):
         "http://datacite.org/schema/kernel-4"
     )
     container: Optional[Container] = None
+
+
+DataCite = DataciteV46