chore: serialized-metadata tests

CenterForOpenScience · May 26, 2023 · b61c00f · b61c00f
1 parent 7946749
commit b61c00f
Show file tree

Hide file tree

Showing 33 changed files with 2,006 additions and 674 deletions.
diff --git a/conftest.py b/conftest.py
@@ -56,6 +56,8 @@ def override_settings():
     website_settings.BCRYPT_LOG_ROUNDS = 1
     # Make sure we don't accidentally send any emails
     website_settings.SENDGRID_API_KEY = None
+    # or try to contact a SHARE
+    website_settings.SHARE_ENABLED = False
     # Set this here instead of in SILENT_LOGGERS, in case developers
     # call setLevel in local.py
     logging.getLogger('website.mails.mails').setLevel(logging.CRITICAL)

diff --git a/osf/metadata/osf_gathering.py b/osf/metadata/osf_gathering.py
@@ -430,16 +430,8 @@ def gather_versions(focus):
     if hasattr(focus.dbmodel, 'versions'):  # quacks like BaseFileNode
         last_fileversion = focus.dbmodel.versions.last()  # just the last version, for now
         if last_fileversion is not None:  # quacks like OsfStorageFileNode
-            from api.base.utils import absolute_reverse as apiv2_absolute_reverse
             fileversion_iri = rdflib.URIRef(
-                apiv2_absolute_reverse(
-                    'files:version-detail',
-                    kwargs={
-                        'version': 'v2',  # api version
-                        'file_id': focus.dbmodel._id,
-                        'version_id': last_fileversion.identifier,
-                    },
-                ),
+                f'{focus.iri}?revision={last_fileversion.identifier}'
             )
             yield (OSF.hasFileVersion, fileversion_iri)
             yield from _gather_fileversion(last_fileversion, fileversion_iri)
@@ -661,6 +653,7 @@ def gather_agents(focus):
     # TODO: contributor roles
     for user in getattr(focus.dbmodel, 'visible_contributors', ()):
         yield (DCTERMS.creator, OsfFocus(user))
+    # TODO: preserve order via rdflib.Seq
 
 
 @gather.er(OSF.affiliatedInstitution)

diff --git a/osf/metadata/serializers/datacite/datacite_tree_walker.py b/osf/metadata/serializers/datacite/datacite_tree_walker.py
@@ -379,7 +379,7 @@ def _visit_related(self, parent_el):
                 relation_pairs.add((datacite_relation, related_iri))
         related_identifiers_el = self.visit(parent_el, 'relatedIdentifiers', is_list=True)
         related_items_el = self.visit(parent_el, 'relatedItems', is_list=True)
-        for datacite_relation, related_iri in relation_pairs:
+        for datacite_relation, related_iri in sorted(relation_pairs):
             self._visit_related_identifier_and_item(
                 related_identifiers_el,
                 related_items_el,
@@ -396,11 +396,11 @@ def _visit_name_identifiers(self, parent_el, agent_iri):
 
     def _visit_subjects(self, parent_el):
         subjects_el = self.visit(parent_el, 'subjects', is_list=True)
-        for subject in self.basket[DCTERMS.subject]:
+        for subject in sorted(self.basket[DCTERMS.subject]):
             self.visit(subjects_el, 'subject', text=subject, attrib={
                 'subjectScheme': BEPRESS_SUBJECT_SCHEME,
             })
-        for keyword in self.basket[OSF.keyword]:
+        for keyword in sorted(self.basket[OSF.keyword]):
             self.visit(subjects_el, 'subject', text=keyword)
 
     def _visit_resource_type(self, parent_el):
@@ -417,6 +417,7 @@ def _visit_resource_type(self, parent_el):
         })
 
     def _get_resource_type_general(self, focus_iri):
+        # return just the first recognized type, preferably from dcterms:type
         type_terms = itertools.chain(
             self.basket[focus_iri:DCTERMS.type],
             self.basket[focus_iri:RDF.type],

diff --git a/osf/metadata/tools.py b/osf/metadata/tools.py
@@ -37,7 +37,7 @@ def pls_gather_metadata_file(osf_item, format_key, serializer_config=None) -> Se
     basket = pls_get_magic_metadata_basket(osfguid.referent)
     serializer = get_metadata_serializer(format_key, basket, serializer_config)
     return SerializedMetadataFile(
-        serializer.mediatype,
-        serializer.filename_for_itemid(osfguid._id),
-        serializer.serialize(),
+        mediatype=serializer.mediatype,
+        filename=serializer.filename_for_itemid(osfguid._id),
+        serialized_metadata=serializer.serialize(),
     )
diff --git a/osf_tests/metadata/expected_metadata_files/file_basic.datacite.json b/osf_tests/metadata/expected_metadata_files/file_basic.datacite.json
@@ -0,0 +1,86 @@
+{
+  "alternateIdentifiers": [
+    {
+      "alternateIdentifier": "http://localhost:5000/w3ibb",
+      "alternateIdentifierType": "URL"
+    }
+  ],
+  "contributors": [
+    {
+      "contributorName": {
+        "contributorName": "Center for Open Science",
+        "nameType": "Organizational"
+      },
+      "contributorType": "HostingInstitution",
+      "nameIdentifier": {
+        "nameIdentifier": "https://cos.io/",
+        "nameIdentifierScheme": "URL"
+      }
+    }
+  ],
+  "creators": [
+    {
+      "creatorName": {
+        "creatorName": "Person McNamington",
+        "nameType": "Personal"
+      },
+      "nameIdentifier": {
+        "nameIdentifier": "http://localhost:5000/w1ibb",
+        "nameIdentifierScheme": "URL"
+      }
+    }
+  ],
+  "dates": [
+    {
+      "date": "2123-05-04",
+      "dateType": "Created"
+    },
+    {
+      "date": "2123-05-04",
+      "dateType": "Updated"
+    }
+  ],
+  "descriptions": [],
+  "fundingReferences": [],
+  "identifier": {
+    "identifier": "http://localhost:5000/w3ibb",
+    "identifierType": "URL"
+  },
+  "publicationYear": "2123",
+  "publisher": "OSF",
+  "relatedIdentifiers": [
+    {
+      "relatedIdentifier": "10.70102/FK2osf.io/w2ibb",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsPartOf"
+    }
+  ],
+  "relatedItems": [
+    {
+      "publicationYear": "2252",
+      "publisher": "OSF",
+      "relatedItemIdentifier": {
+        "relatedItemIdentifier": "10.70102/FK2osf.io/w2ibb",
+        "relatedItemIdentifierType": "DOI"
+      },
+      "relatedItemType": "Text",
+      "relationType": "IsPartOf",
+      "titles": [
+        {
+          "title": "this is a project title!"
+        }
+      ]
+    }
+  ],
+  "resourceType": {
+    "resourceType": "File",
+    "resourceTypeGeneral": "Text"
+  },
+  "rightsList": [],
+  "subjects": [],
+  "titles": [
+    {
+      "title": "my-file.blarg"
+    }
+  ]
+}
diff --git a/osf_tests/metadata/expected_metadata_files/file_basic.datacite.xml b/osf_tests/metadata/expected_metadata_files/file_basic.datacite.xml
@@ -0,0 +1,47 @@
+<?xml version='1.0' encoding='utf-8'?>
+<resource xmlns="http://datacite.org/schema/kernel-4" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4.4/metadata.xsd">
+  <identifier identifierType="URL">http://localhost:5000/w3ibb</identifier>
+  <creators>
+    <creator>
+      <creatorName nameType="Personal">Person McNamington</creatorName>
+      <nameIdentifier nameIdentifierScheme="URL">http://localhost:5000/w1ibb</nameIdentifier>
+    </creator>
+  </creators>
+  <titles>
+    <title>my-file.blarg</title>
+  </titles>
+  <publisher>OSF</publisher>
+  <publicationYear>2123</publicationYear>
+  <subjects/>
+  <contributors>
+    <contributor contributorType="HostingInstitution">
+      <contributorName nameType="Organizational">Center for Open Science</contributorName>
+      <nameIdentifier nameIdentifierScheme="ROR">https://ror.org/05d5mza29</nameIdentifier>
+      <nameIdentifier nameIdentifierScheme="URL">https://cos.io/</nameIdentifier>
+    </contributor>
+  </contributors>
+  <dates>
+    <date dateType="Created">2123-05-04</date>
+    <date dateType="Updated">2123-05-04</date>
+  </dates>
+  <resourceType resourceTypeGeneral="Text">File</resourceType>
+  <alternateIdentifiers>
+    <alternateIdentifier alternateIdentifierType="URL">http://localhost:5000/w3ibb</alternateIdentifier>
+  </alternateIdentifiers>
+  <rightsList/>
+  <descriptions/>
+  <fundingReferences/>
+  <relatedIdentifiers>
+    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsPartOf">10.70102/FK2osf.io/w2ibb</relatedIdentifier>
+  </relatedIdentifiers>
+  <relatedItems>
+    <relatedItem relationType="IsPartOf" relatedItemType="Text">
+      <relatedItemIdentifier relatedItemIdentifierType="DOI">10.70102/FK2osf.io/w2ibb</relatedItemIdentifier>
+      <titles>
+        <title>this is a project title!</title>
+      </titles>
+      <publicationYear>2252</publicationYear>
+      <publisher>OSF</publisher>
+    </relatedItem>
+  </relatedItems>
+</resource>
diff --git a/osf_tests/metadata/expected_metadata_files/file_basic.turtle b/osf_tests/metadata/expected_metadata_files/file_basic.turtle
@@ -0,0 +1,41 @@
+@prefix dcterms: <http://purl.org/dc/terms/> .
+@prefix foaf: <http://xmlns.com/foaf/0.1/> .
+@prefix osf: <https://osf.io/vocab/2022/> .
+@prefix owl: <http://www.w3.org/2002/07/owl#> .
+
+<http://localhost:5000/w3ibb> a osf:File ;
+    dcterms:created "2123-05-04" ;
+    dcterms:identifier "http://localhost:5000/w3ibb" ;
+    dcterms:modified "2123-05-04" ;
+    osf:fileName "my-file.blarg" ;
+    osf:filePath "/my-file.blarg" ;
+    osf:hasFileVersion <http://localhost:5000/w3ibb?revision=1> ;
+    osf:isContainedBy <http://localhost:5000/w2ibb> .
+
+<http://localhost:5000/w2ibb> a osf:Project ;
+    dcterms:created "2123-05-04" ;
+    dcterms:creator <http://localhost:5000/w1ibb> ;
+    dcterms:identifier "http://localhost:5000/w2ibb",
+        "https://doi.org/10.70102/FK2osf.io/w2ibb" ;
+    dcterms:publisher <http://localhost:5000> ;
+    dcterms:title "this is a project title!" ;
+    owl:sameAs <https://doi.org/10.70102/FK2osf.io/w2ibb> .
+
+<http://localhost:5000/w3ibb?revision=1> a osf:FileVersion ;
+    dcterms:created "2123-05-04" ;
+    dcterms:creator <http://localhost:5000/w1ibb> ;
+    dcterms:extent "0.000007 MB" ;
+    dcterms:format "img/png" ;
+    dcterms:modified "2123-05-04" ;
+    dcterms:requires <urn:checksum:sha-256::6ac3c336e4094835293a3fed8a4b5fedde1b5e2626d9838fed50693bba00af0e> ;
+    osf:versionNumber "1" .
+
+<http://localhost:5000/w1ibb> a osf:Agent ;
+    dcterms:identifier "http://localhost:5000/w1ibb" ;
+    dcterms:type foaf:Person ;
+    foaf:name "Person McNamington" .
+
+<http://localhost:5000> a osf:Agent ;
+    dcterms:identifier "http://localhost:5000" ;
+    dcterms:type foaf:Organization ;
+    foaf:name "OSF" .
diff --git a/osf_tests/metadata/expected_metadata_files/file_full.datacite.json b/osf_tests/metadata/expected_metadata_files/file_full.datacite.json
@@ -0,0 +1,87 @@
+{
+  "alternateIdentifiers": [
+    {
+      "alternateIdentifier": "http://localhost:5000/w3ibb",
+      "alternateIdentifierType": "URL"
+    }
+  ],
+  "contributors": [
+    {
+      "contributorName": {
+        "contributorName": "Center for Open Science",
+        "nameType": "Organizational"
+      },
+      "contributorType": "HostingInstitution",
+      "nameIdentifier": {
+        "nameIdentifier": "https://cos.io/",
+        "nameIdentifierScheme": "URL"
+      }
+    }
+  ],
+  "creators": [
+    {
+      "creatorName": {
+        "creatorName": "Person McNamington",
+        "nameType": "Personal"
+      },
+      "nameIdentifier": {
+        "nameIdentifier": "http://localhost:5000/w1ibb",
+        "nameIdentifierScheme": "URL"
+      }
+    }
+  ],
+  "dates": [
+    {
+      "date": "2123-05-04",
+      "dateType": "Created"
+    },
+    {
+      "date": "2123-05-04",
+      "dateType": "Updated"
+    }
+  ],
+  "descriptions": [],
+  "fundingReferences": [],
+  "identifier": {
+    "identifier": "http://localhost:5000/w3ibb",
+    "identifierType": "URL"
+  },
+  "publicationYear": "2123",
+  "publisher": "OSF",
+  "relatedIdentifiers": [
+    {
+      "relatedIdentifier": "10.70102/FK2osf.io/w2ibb",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsPartOf"
+    }
+  ],
+  "relatedItems": [
+    {
+      "publicationYear": "2123",
+      "publisher": "OSF",
+      "relatedItemIdentifier": {
+        "relatedItemIdentifier": "10.70102/FK2osf.io/w2ibb",
+        "relatedItemIdentifierType": "DOI"
+      },
+      "relatedItemType": "Dataset",
+      "relationType": "IsPartOf",
+      "titles": [
+        {
+          "lang": "en",
+          "title": "this is a project title!"
+        }
+      ]
+    }
+  ],
+  "resourceType": {
+    "resourceType": "File",
+    "resourceTypeGeneral": "Text"
+  },
+  "rightsList": [],
+  "subjects": [],
+  "titles": [
+    {
+      "title": "my-file.blarg"
+    }
+  ]
+}
diff --git a/osf_tests/metadata/expected_metadata_files/file_full.datacite.xml b/osf_tests/metadata/expected_metadata_files/file_full.datacite.xml
@@ -0,0 +1,47 @@
+<?xml version='1.0' encoding='utf-8'?>
+<resource xmlns="http://datacite.org/schema/kernel-4" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4.4/metadata.xsd">
+  <identifier identifierType="URL">http://localhost:5000/w3ibb</identifier>
+  <creators>
+    <creator>
+      <creatorName nameType="Personal">Person McNamington</creatorName>
+      <nameIdentifier nameIdentifierScheme="URL">http://localhost:5000/w1ibb</nameIdentifier>
+    </creator>
+  </creators>
+  <titles>
+    <title>my-file.blarg</title>
+  </titles>
+  <publisher>OSF</publisher>
+  <publicationYear>2123</publicationYear>
+  <subjects/>
+  <contributors>
+    <contributor contributorType="HostingInstitution">
+      <contributorName nameType="Organizational">Center for Open Science</contributorName>
+      <nameIdentifier nameIdentifierScheme="ROR">https://ror.org/05d5mza29</nameIdentifier>
+      <nameIdentifier nameIdentifierScheme="URL">https://cos.io/</nameIdentifier>
+    </contributor>
+  </contributors>
+  <dates>
+    <date dateType="Created">2123-05-04</date>
+    <date dateType="Updated">2123-05-04</date>
+  </dates>
+  <resourceType resourceTypeGeneral="Text">File</resourceType>
+  <alternateIdentifiers>
+    <alternateIdentifier alternateIdentifierType="URL">http://localhost:5000/w3ibb</alternateIdentifier>
+  </alternateIdentifiers>
+  <rightsList/>
+  <descriptions/>
+  <fundingReferences/>
+  <relatedIdentifiers>
+    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsPartOf">10.70102/FK2osf.io/w2ibb</relatedIdentifier>
+  </relatedIdentifiers>
+  <relatedItems>
+    <relatedItem relationType="IsPartOf" relatedItemType="Dataset">
+      <relatedItemIdentifier relatedItemIdentifierType="DOI">10.70102/FK2osf.io/w2ibb</relatedItemIdentifier>
+      <titles>
+        <title xml:lang="en">this is a project title!</title>
+      </titles>
+      <publicationYear>2123</publicationYear>
+      <publisher>OSF</publisher>
+    </relatedItem>
+  </relatedItems>
+</resource>