Skip to content
Snippets Groups Projects
Commit 3cfaed6d authored by vlorentz's avatar vlorentz Committed by vlorentz
Browse files

Check all tag names in Atom and Codemeta namespaces are valid

parent 0bf9351a
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
swh/deposit/api/checks.py
+ 154
− 0
View file @ 3cfaed6d
......@@ -35,6 +35,115 @@ METADATA_PROVENANCE_KEY = "swh:metadata-provenance"
AFFILIATION_NO_NAME = "Reason: affiliation does not have a <codemeta:name> element"
# from https://datatracker.ietf.org/doc/html/rfc4287
ATOM_ELEMENTS = [
"name",
"uri",
"email",
# specifically not allowing this one, because clients are supposed to send one
# entry at a time:
# "feed",
"entry",
# ditto:
# "content",
"author",
"category",
"contributor",
"generator",
"icon",
"id",
"link",
"logo",
"published",
"rights",
"source",
"subtitle",
"summary",
"title",
"updated",
# we used to recommend this, so we still need to support it so we don't break
# existing clients
"external_identifier",
]
# from https://github.com/codemeta/codemeta/blob/2.0/codemeta.jsonld
CODEMETA2_CONTEXT = {
"type": "@type",
"id": "@id",
"schema": "http://schema.org/",
"codemeta": "https://codemeta.github.io/terms/",
"Organization": {"@id": "schema:Organization"},
"Person": {"@id": "schema:Person"},
"SoftwareSourceCode": {"@id": "schema:SoftwareSourceCode"},
"SoftwareApplication": {"@id": "schema:SoftwareApplication"},
"Text": {"@id": "schema:Text"},
"URL": {"@id": "schema:URL"},
"address": {"@id": "schema:address"},
"affiliation": {"@id": "schema:affiliation"},
"applicationCategory": {"@id": "schema:applicationCategory", "@type": "@id"},
"applicationSubCategory": {"@id": "schema:applicationSubCategory", "@type": "@id"},
"citation": {"@id": "schema:citation"},
"codeRepository": {"@id": "schema:codeRepository", "@type": "@id"},
"contributor": {"@id": "schema:contributor"},
"copyrightHolder": {"@id": "schema:copyrightHolder"},
"copyrightYear": {"@id": "schema:copyrightYear"},
"creator": {"@id": "schema:creator"},
"dateCreated": {"@id": "schema:dateCreated", "@type": "schema:Date"},
"dateModified": {"@id": "schema:dateModified", "@type": "schema:Date"},
"datePublished": {"@id": "schema:datePublished", "@type": "schema:Date"},
"description": {"@id": "schema:description"},
"downloadUrl": {"@id": "schema:downloadUrl", "@type": "@id"},
"email": {"@id": "schema:email"},
"editor": {"@id": "schema:editor"},
"encoding": {"@id": "schema:encoding"},
"familyName": {"@id": "schema:familyName"},
"fileFormat": {"@id": "schema:fileFormat", "@type": "@id"},
"fileSize": {"@id": "schema:fileSize"},
"funder": {"@id": "schema:funder"},
"givenName": {"@id": "schema:givenName"},
"hasPart": {"@id": "schema:hasPart"},
"identifier": {"@id": "schema:identifier", "@type": "@id"},
"installUrl": {"@id": "schema:installUrl", "@type": "@id"},
"isAccessibleForFree": {"@id": "schema:isAccessibleForFree"},
"isPartOf": {"@id": "schema:isPartOf"},
"keywords": {"@id": "schema:keywords"},
"license": {"@id": "schema:license", "@type": "@id"},
"memoryRequirements": {"@id": "schema:memoryRequirements", "@type": "@id"},
"name": {"@id": "schema:name"},
"operatingSystem": {"@id": "schema:operatingSystem"},
"permissions": {"@id": "schema:permissions"},
"position": {"@id": "schema:position"},
"processorRequirements": {"@id": "schema:processorRequirements"},
"producer": {"@id": "schema:producer"},
"programmingLanguage": {"@id": "schema:programmingLanguage"},
"provider": {"@id": "schema:provider"},
"publisher": {"@id": "schema:publisher"},
"relatedLink": {"@id": "schema:relatedLink", "@type": "@id"},
"releaseNotes": {"@id": "schema:releaseNotes", "@type": "@id"},
"runtimePlatform": {"@id": "schema:runtimePlatform"},
"sameAs": {"@id": "schema:sameAs", "@type": "@id"},
"softwareHelp": {"@id": "schema:softwareHelp"},
"softwareRequirements": {"@id": "schema:softwareRequirements", "@type": "@id"},
"softwareVersion": {"@id": "schema:softwareVersion"},
"sponsor": {"@id": "schema:sponsor"},
"storageRequirements": {"@id": "schema:storageRequirements", "@type": "@id"},
"supportingData": {"@id": "schema:supportingData"},
"targetProduct": {"@id": "schema:targetProduct"},
"url": {"@id": "schema:url", "@type": "@id"},
"version": {"@id": "schema:version"},
"author": {"@id": "schema:author", "@container": "@list"},
"softwareSuggestions": {"@id": "codemeta:softwareSuggestions", "@type": "@id"},
"contIntegration": {"@id": "codemeta:contIntegration", "@type": "@id"},
"buildInstructions": {"@id": "codemeta:buildInstructions", "@type": "@id"},
"developmentStatus": {"@id": "codemeta:developmentStatus", "@type": "@id"},
"embargoDate": {"@id": "codemeta:embargoDate", "@type": "schema:Date"},
"funding": {"@id": "codemeta:funding"},
"readme": {"@id": "codemeta:readme", "@type": "@id"},
"issueTracker": {"@id": "codemeta:issueTracker", "@type": "@id"},
"referencePublication": {"@id": "codemeta:referencePublication", "@type": "@id"},
"maintainer": {"@id": "codemeta:maintainer"},
}
def extra_validator(
element: ElementTree.Element,
......@@ -237,6 +346,51 @@ def check_metadata(metadata: ElementTree.Element) -> Tuple[bool, Optional[Dict]]
)
break
for element in metadata.iter():
if element.tag.startswith("{http://www.w3.org/2005/Atom}"):
_, local_name = element.tag.split("}", 1)
if local_name not in ATOM_ELEMENTS:
if local_name in CODEMETA2_CONTEXT:
# Probably confused the two namespaces, display a nicer error
detail.append(
{
"fields": [local_name],
"summary": f"{local_name} is not a valid Atom element. "
"However, it would be a valid a Codemeta term; make sure "
"namespaces are not swapped",
}
)
else:
detail.append(
{
"fields": [local_name],
"summary": f"{local_name} is not a valid Atom element, "
"see https://datatracker.ietf.org/doc/html/rfc4287",
}
)
elif element.tag.startswith("{https://doi.org/10.5063/SCHEMA/CODEMETA-2.0}"):
_, local_name = element.tag.split("}", 1)
if local_name not in CODEMETA2_CONTEXT:
if local_name in CODEMETA2_CONTEXT:
# Probably confused the two namespaces, display a nicer error
detail.append(
{
"fields": [local_name],
"summary": f"{local_name} is not a valid Codemeta 2.0 term. "
"However, it would be a valid Atom element; make sure "
"namespaces are not swapped",
}
)
else:
detail.append(
{
"fields": [local_name],
"summary": f"{local_name} is not a valid Codemeta 2.0 term, "
"see "
"https://github.com/codemeta/codemeta/blob/2.0/codemeta.jsonld",
}
)
if detail:
return False, {"metadata": detail + suggested_fields}
......
swh/deposit/tests/api/test_checks.py
+ 76
− 14
View file @ 3cfaed6d
......@@ -89,20 +89,6 @@ _parameters1 = [
</entry>
""",
),
(
"unknown-codemeta-inner-element-after",
f"""\
<entry {XMLNS}>
<uri>some url</uri>
<codemeta:name>bar</codemeta:name>
<codemeta:author>
<codemeta:name>someone</codemeta:name>
<codemeta:unknown-tag>should allow anything here</codemeta:unknown-tag>
</codemeta:author>
{PROVENANCE_XML}
</entry>
""",
),
(
"unknown-schema-inner-element-after",
f"""\
......@@ -592,6 +578,82 @@ _parameters2 = [
"fields": ["atom:author or codemeta:author"],
},
),
(
"unknown-atom",
f"""\
<entry {XMLNS}>
<uri>some url</uri>
<unknown-tag>but in known namespace</unknown-tag>
<codemeta:name>bar</codemeta:name>
<codemeta:author>
<codemeta:name>someone</codemeta:name>
</codemeta:author>
{PROVENANCE_XML}
</entry>
""",
{
"summary": "unknown-tag is not a valid Atom element, see "
"https://datatracker.ietf.org/doc/html/rfc4287",
"fields": ["unknown-tag"],
},
),
(
"unknown-codemeta",
f"""\
<entry {XMLNS}>
<uri>some url</uri>
<codemeta:name>bar</codemeta:name>
<codemeta:unknown-tag>but in known namespace</codemeta:unknown-tag>
<codemeta:author>
<codemeta:name>someone</codemeta:name>
</codemeta:author>
{PROVENANCE_XML}
</entry>
""",
{
"summary": "unknown-tag is not a valid Codemeta 2.0 term, see "
"https://github.com/codemeta/codemeta/blob/2.0/codemeta.jsonld",
"fields": ["unknown-tag"],
},
),
(
"unknown-atom-in-codemeta",
f"""\
<entry {XMLNS}>
<uri>some url</uri>
<codemeta:name>bar</codemeta:name>
<codemeta:author>
<codemeta:name>someone</codemeta:name>
<unknown-tag>but in known namespace</unknown-tag>
</codemeta:author>
{PROVENANCE_XML}
</entry>
""",
{
"summary": "unknown-tag is not a valid Atom element, see "
"https://datatracker.ietf.org/doc/html/rfc4287",
"fields": ["unknown-tag"],
},
),
(
"unknown-codemeta-in-codemeta",
f"""\
<entry {XMLNS}>
<uri>some url</uri>
<codemeta:name>bar</codemeta:name>
<codemeta:author>
<codemeta:name>someone</codemeta:name>
<codemeta:unknown-tag>but in known namespace</codemeta:unknown-tag>
</codemeta:author>
{PROVENANCE_XML}
</entry>
""",
{
"summary": "unknown-tag is not a valid Codemeta 2.0 term, see "
"https://github.com/codemeta/codemeta/blob/2.0/codemeta.jsonld",
"fields": ["unknown-tag"],
},
),
]
]
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment