From 6d1089cb25b62e521ee68af7369d908c89e70f4c Mon Sep 17 00:00:00 2001
From: Valentin Lorentz <vlorentz@softwareheritage.org>
Date: Tue, 5 Apr 2022 15:40:52 +0200
Subject: [PATCH] Add support for HAL-ID as identifier
---
swh/deposit/api/checks.py | 38 ++++++++++++++++++----------
swh/deposit/tests/api/test_checks.py | 32 +++++++++++++++++++++++
swh/deposit/xsd/codemeta.xsd | 14 +++++++++-
3 files changed, 70 insertions(+), 14 deletions(-)
diff --git a/swh/deposit/api/checks.py b/swh/deposit/api/checks.py
index 3fb990c4..1d82a28e 100644
--- a/swh/deposit/api/checks.py
+++ b/swh/deposit/api/checks.py
@@ -16,6 +16,7 @@ Suggested fields:
import dataclasses
import functools
+import re
from typing import Dict, Iterator, Optional, Tuple, cast
import urllib
from xml.etree import ElementTree
@@ -55,22 +56,33 @@ def extra_validator(
# </xsd:simpleType>
# However, this would give an unreadable error, so we implement it here
# in Python instead.
- try:
- url = urllib.parse.urlparse(element.text)
- except ValueError:
+ yield from absolute_uri_validator(element, xsd_element)
+ elif type_name == "{https://doi.org/10.5063/SCHEMA/CODEMETA-2.0}identifierType":
+ # Made-up type, that allows both absolute URIs and HAL-IDs
+ if not re.match("hal-[0-9]+", element.text or ""):
+ yield from absolute_uri_validator(element, xsd_element)
+
+
+def absolute_uri_validator(
+ element: ElementTree.Element,
+ xsd_element: xmlschema.validators.elements.Xsd11Element,
+) -> Iterator[xmlschema.XMLSchemaValidationError]:
+ try:
+ url = urllib.parse.urlparse(element.text)
+ except ValueError:
+ yield xmlschema.XMLSchemaValidationError(
+ xsd_element, element, f"{element.text!r} is not a valid URI",
+ )
+ else:
+ if not url.scheme or not url.netloc:
+ yield xmlschema.XMLSchemaValidationError(
+ xsd_element, element, f"{element.text!r} is not an absolute URI",
+ )
+ elif " " in url.netloc:
+ # urllib is a little too permissive...
yield xmlschema.XMLSchemaValidationError(
xsd_element, element, f"{element.text!r} is not a valid URI",
)
- else:
- if not url.scheme or not url.netloc:
- yield xmlschema.XMLSchemaValidationError(
- xsd_element, element, f"{element.text!r} is not an absolute URI",
- )
- elif " " in url.netloc:
- # urllib is a little too permissive...
- yield xmlschema.XMLSchemaValidationError(
- xsd_element, element, f"{element.text!r} is not a valid URI",
- )
@dataclasses.dataclass
diff --git a/swh/deposit/tests/api/test_checks.py b/swh/deposit/tests/api/test_checks.py
index df6d8961..e3e55704 100644
--- a/swh/deposit/tests/api/test_checks.py
+++ b/swh/deposit/tests/api/test_checks.py
@@ -146,6 +146,38 @@ _parameters1 = [
</entry>
""",
),
+ (
+ "identifier-is-halid",
+ f"""\
+ <entry {XMLNS}>
+ <url>some url</url>
+ <codemeta:name>bar</codemeta:name>
+ <codemeta:author>
+ <codemeta:name>The Author</codemeta:name>
+ </codemeta:author>
+ <codemeta:identifier>hal-12345</codemeta:identifier>
+ {PROVENANCE_XML}
+ </entry>
+ """,
+ ),
+ (
+ "identifier-is-propertyvalue",
+ f"""\
+ <entry {XMLNS}>
+ <url>some url</url>
+ <codemeta:name>bar</codemeta:name>
+ <codemeta:author>
+ <codemeta:name>The Author</codemeta:name>
+ </codemeta:author>
+ <schema:identifier>
+ <codemeta:type>schema:PropertyValue</codemeta:type>
+ <schema:propertyID>HAL-ID</schema:propertyID>
+ <schema:value>hal-02527911</schema:value>
+ </schema:identifier>
+ {PROVENANCE_XML}
+ </entry>
+ """,
+ ),
(
"codemeta-dates",
f"""\
diff --git a/swh/deposit/xsd/codemeta.xsd b/swh/deposit/xsd/codemeta.xsd
index 64e5187c..474e4c9a 100644
--- a/swh/deposit/xsd/codemeta.xsd
+++ b/swh/deposit/xsd/codemeta.xsd
@@ -52,11 +52,23 @@
Therefore, more custom checks are implemented in swh/deposit/api/checks.py
in order to allow either. -->
+
+ <xsd:simpleType name="halId">
+ <xsd:restriction base="xsd:string">
+ <xsd:pattern value="hal-[0-9]+"/>
+ </xsd:restriction>
+ </xsd:simpleType>
+ <xsd:simpleType name="identifierType">
+ <!-- CodeMeta only allows URIs, but we make an exception for HAL-IDs,
+ in order not to break backward-compatibility. -->
+ <xsd:union memberTypes="xsd:anyURI codemeta:halId"/>
+ </xsd:simpleType>
+ <xsd:element name="identifier" type="codemeta:identifierType" />
+
<xsd:element name="name" type="xsd:string" />
<xsd:element name="givenName" type="xsd:string" />
<xsd:element name="familyName" type="xsd:string" />
<xsd:element name="email" type="xsd:string" />
- <xsd:element name="identifier" type="xsd:anyURI" />
<xsd:element name="applicationCategory" type="xsd:string" />
<xsd:element name="applicationSubCategory" type="xsd:string" />
--
GitLab