Merge pull request #260 from dandi/enh-version

Add "version" to datacite record and bundle datacite json serializations

dandischema/datacite.py → dandischema/datacite/__init__.py

@@ -1,11 +1,20 @@
+"""
+Interfaces and data to interact with DataCite metadata
+"""
+
+# TODO: RF into submodules for some next "minor" taking care not to break
+
 from copy import deepcopy
+from functools import lru_cache
+import json
+from pathlib import Path
 import re
 from typing import Any, Dict, Union
 
 from jsonschema import Draft7Validator
 import requests
 
-from .models import NAME_PATTERN, Organization, Person, PublishedDandiset, RoleType
+from ..models import NAME_PATTERN, Organization, Person, PublishedDandiset, RoleType
 
 DATACITE_CONTRTYPE = {
     "ContactPerson",
@@ -83,6 +92,8 @@ def to_datacite(
     ]
 
     attributes["doi"] = meta.doi
+    if meta.version:
+        attributes["version"] = meta.version
     attributes["titles"] = [{"title": meta.name}]
     attributes["descriptions"] = [
         {"description": meta.description, "descriptionType": "Abstract"}
@@ -240,15 +251,11 @@ def to_datacite(
     return datacite_dict
 
 
-def _get_datacite_schema() -> Any:
-    sr = requests.get(
-        "https://raw.githubusercontent.com/datacite/schema/"
-        "732cc7ef29f4cad4d6adfac83544133cd57a2e5e/"
-        "source/json/kernel-4.3/datacite_4.3_schema.json"
-    )
-    sr.raise_for_status()
-    schema = sr.json()
-    return schema
+@lru_cache()
+def _get_datacite_schema(version_id: str = "datacite-4.3-17-gaa5db56") -> Any:
+    """Load datacite schema based on the version id provided."""
+    schema_folder = Path(__file__).parent / "schema"
+    return json.loads((schema_folder / f"{version_id}.json").read_text())
 
 
 def validate_datacite(datacite_dict: dict) -> None:

dandischema/datacite/schema/README.md

@@ -0,0 +1,48 @@
+Folder contains copies of jsonschema serializations of datacite which were
+initially kept within datacite repository,
+
+   https://github.com/datacite/schema/tree/master/source/json
+
+but then moved to the "origin" of their manufacturing -- inveniosoftware
+
+   https://github.com/inveniosoftware/datacite/tree/master/datacite/schemas
+
+Those serializations are not "scripted" and apparently produced manually.
+Related issues/inquiries:
+
+- https://github.com/datacite/schema/issues/149
+- https://github.com/inveniosoftware/datacite/issues/101
+
+Versions in the suffixes of the files here were produced based on
+original version tag  and output of git describe so we capture "order" within
+MAJOR.MINOR versions.  e.g.
+
+    ❯ git describe --tags --match 4.3 732cc7
+    4.3-72-g732cc7e
+    ❯ git describe --tags --match 4.3 aa5db5
+    4.3-17-gaa5db56
+
+for those from datacite, for the last one based on commit when was last
+modified (not current master).  In "inveniosoftware" there are no tags for
+versions of datacite, so we base ordering of 0.1.0 first tag there:
+
+	❯ git log datacite-v4.3.json
+	commit 24fc2ba3ded44512ce8569dc11c958da4a29f70a
+	Author: Thorge Petersen <[email protected]>
+	Date:   Fri Aug 12 09:47:45 2022 +0200
+
+		schema: change affiliation definition to match property name
+
+	commit dc8403fd8556858e8917b960b0721884c52a588e
+	Author: Tom Morrell <[email protected]>
+	Date:   Thu Aug 15 12:17:21 2019 -0700
+
+		schema: Add support for DataCite 4.3 metadata schema
+
+	❯ git describe --match v0.1.0 24fc2ba3ded44512ce8569dc11c958da4a29f70a
+	v0.1.0-66-g24fc2ba
+
+so we get
+
+    ❯ cp /home/yoh/proj/datacite/inveniosoftware-datacite/datacite/schemas/datacite-v4.3.json inveniosoftware-4.3-66-g24fc2ba.json
+    ❯ cp /home/yoh/proj/datacite/inveniosoftware-datacite/datacite/schemas/datacite-v4.5.json inveniosoftware-4.5-81-g160250d.json