fix: handle names consistently when formatting

- if no `cited_as` and no `name`, make some cultural assumptions and try building a name from parts (TODO in the future: stop using name parts at all) - correct ordering contributors for oai_dc
CenterForOpenScience · Mar 31, 2021 · 2b498ac · 2b498ac
1 parent 122f953
commit 2b498ac
Show file tree

Hide file tree

Showing 6 changed files with 125 additions and 7 deletions.
diff --git a/share/metadata_formats/oai_dc.py b/share/metadata_formats/oai_dc.py
@@ -1,6 +1,7 @@
 from lxml import etree
 
 from share.util.graph import MutableGraph
+from share.util.names import get_related_agent_name
 
 from share.oaipmh.util import format_datetime, ns, nsmap, SubEl
 from share.metadata_formats.base import MetadataFormatter
@@ -87,7 +88,10 @@ def build_dublin_core(self, work_node):
 
     def _get_related_agent_names(self, work_node, relation_types):
         def sort_key(relation_node):
-            return relation_node['order_cited'] or 9999999  # those without order_cited go last
+            order_cited = relation_node['order_cited']
+            if order_cited is None:
+                return 9999999  # those without order_cited go last
+            return int(order_cited)
 
         relation_nodes = sorted(
             [
@@ -97,10 +101,12 @@ def sort_key(relation_node):
             ],
             key=sort_key,
         )
-        return [
-            relation['cited_as'] or relation['agent']['name']
+
+        # remove falsy values
+        return filter(None, [
+            get_related_agent_name(relation)
             for relation in relation_nodes
-        ]
+        ])
 
     def _get_related_uris(self, work_node):
         related_work_uris = set()

diff --git a/share/metadata_formats/sharev2_elastic.py b/share/metadata_formats/sharev2_elastic.py
@@ -4,6 +4,7 @@
 from django.conf import settings
 
 from share.util.graph import MutableGraph
+from share.util.names import get_related_agent_name
 from share.util import IDObfuscator
 
 from .base import MetadataFormatter
@@ -135,7 +136,7 @@ def format(self, normalized_datum):
 
     def _get_related_agent_names(self, work_node, relation_types):
         return [
-            relation_node['cited_as'] or relation_node['agent']['name']
+            get_related_agent_name(relation_node)
             for relation_node in work_node['agent_relations']
             if relation_node.type in relation_types
         ]
@@ -181,7 +182,7 @@ def _build_list_agent(self, relation_node):
         return {
             'type': format_node_type(agent_node),
             'types': format_node_type_lineage(agent_node),
-            'name': agent_node['name'] or relation_node['cited_as'],
+            'name': get_related_agent_name(relation_node),
             'given_name': agent_node['given_name'],
             'family_name': agent_node['family_name'],
             'additional_name': agent_node['additional_name'],

diff --git a/share/util/names.py b/share/util/names.py
@@ -0,0 +1,28 @@
+
+def build_name_from_parts(agent_node):
+    """construct some name from parts, making wild cultural assumptions
+
+    @param agent_node: share.util.graph.MutableNode with concrete type 'abstractagent'
+    @returns string (possibly empty)
+    """
+    # filter out falsy parts
+    name_parts = filter(None, [
+        agent_node['given_name'],
+        agent_node['additional_name'],
+        agent_node['family_name'],
+        agent_node['suffix'],
+    ])
+    return ' '.join(name_parts).strip()
+
+
+def get_related_agent_name(relation_node):
+    """get the name to refer to a related agent
+
+    @param relation_node: share.util.graph.MutableNode with concrete type 'abstractagentworkrelation'
+    @returns string (possibly empty)
+    """
+    return (
+        relation_node['cited_as']
+        or relation_node['agent']['name']
+        or build_name_from_parts(relation_node['agent'])
+    )
diff --git a/tests/share/metadata_formats/base.py b/tests/share/metadata_formats/base.py
@@ -26,6 +26,24 @@
             },
         },
     },
+    'no-names-only-name-parts': {
+        'suid_id': 7,
+        'source_name': 'SomeSource',
+        'normalized_datum_kwargs': {
+            'created_at': dateutil.parser.isoparse('2017-04-07T21:09:05.023090+00:00'),
+            'data': {
+                '@graph': [
+                    {'@id': '_:cfed87cc7294471eac2b67d9ce92f60b', '@type': 'person', 'given_name': 'Suzanne', 'family_name': 'Simard', 'identifiers': [], 'related_agents': []},
+                    {'@id': '_:c786ef414acb423f878522690453a6b8', '@type': 'creator', 'agent': {'@id': '_:cfed87cc7294471eac2b67d9ce92f60b', '@type': 'person'}, 'order_cited': 0, 'creative_work': {'@id': '_:de04f3a34eb047e98891662b5345afd9', '@type': 'creativework'}},
+                    {'@id': '_:2afb5767c79c47c9ab6b87c7d5b3aa0a', '@type': 'person', 'given_name': 'Mary', 'family_name': 'Austi', 'identifiers': [], 'related_agents': []},
+                    {'@id': '_:44ec4e74e8ae487cbd86abcde5c2a075', '@type': 'creator', 'agent': {'@id': '_:2afb5767c79c47c9ab6b87c7d5b3aa0a', '@type': 'person'}, 'order_cited': 1, 'creative_work': {'@id': '_:de04f3a34eb047e98891662b5345afd9', '@type': 'creativework'}},
+                    {'@id': '_:8ae1b46cd2f341cb968fbf76c9a7f345', 'uri': 'http://dx.doi.org/10.5772/9813', '@type': 'workidentifier', 'creative_work': {'@id': '_:de04f3a34eb047e98891662b5345afd9', '@type': 'creativework'}},
+                    {'@id': '_:de04f3a34eb047e98891662b5345afd9', 'tags': [], '@type': 'creativework', 'extra': {'type': 'book-chapter', 'member': 'http://id.crossref.org/member/3774', 'titles': ['The Role of Mycorrhizas in Forest Soil Stability with Climate Change'], 'date_created': '2012-03-29T07:53:20+00:00', 'date_published': {'date_parts': [[2010, 8, 17]]}, 'container_title': ['Climate Change and Variability'], 'published_online': {'date_parts': [[2010, 8, 17]]}}, 'title': 'The Role of Mycorrhizas in Forest Soil Stability with Climate Change', 'identifiers': [{'@id': '_:8ae1b46cd2f341cb968fbf76c9a7f345', '@type': 'workidentifier'}], 'date_updated': '2017-03-31T05:39:48+00:00', 'related_agents': [{'@id': '_:c786ef414acb423f878522690453a6b8', '@type': 'creator'}, {'@id': '_:44ec4e74e8ae487cbd86abcde5c2a075', '@type': 'creator'}, {'@id': '_:e0fdb4b7b6194b699078f26a799cd232', '@type': 'publisher'}]},
+                ],
+                '@context': {}
+            },
+        },
+    },
     'with-is_deleted': {
         'suid_id': 57,
         'source_name': 'foo',

diff --git a/tests/share/metadata_formats/test_oai_dc_formatter.py b/tests/share/metadata_formats/test_oai_dc_formatter.py
@@ -39,14 +39,29 @@ def test_formatter(self, normalized_datum, expected_output):
             xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd"
         >
             <dc:title>The Role of Mycorrhizas in Forest Soil Stability with Climate Change</dc:title>
-            <dc:creator>Mary Austi</dc:creator>
             <dc:creator>Suzanne Simard</dc:creator>
+            <dc:creator>Mary Austi</dc:creator>
             <dc:publisher>InTech</dc:publisher>
             <dc:date>2017-03-31T05:39:48Z</dc:date>
             <dc:type>creativework</dc:type>
             <dc:identifier>http://dx.doi.org/10.5772/9813</dc:identifier>
         </oai_dc:dc>
         ''',
+        'no-names-only-name-parts': '''
+        <oai_dc:dc
+            xmlns:dc="http://purl.org/dc/elements/1.1/"
+            xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/"
+            xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+            xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd"
+        >
+            <dc:title>The Role of Mycorrhizas in Forest Soil Stability with Climate Change</dc:title>
+            <dc:creator>Suzanne Simard</dc:creator>
+            <dc:creator>Mary Austi</dc:creator>
+            <dc:date>2017-03-31T05:39:48Z</dc:date>
+            <dc:type>creativework</dc:type>
+            <dc:identifier>http://dx.doi.org/10.5772/9813</dc:identifier>
+        </oai_dc:dc>
+        ''',
         'with-is_deleted': None,
         'with-subjects': '''
         <oai_dc:dc

diff --git a/tests/share/metadata_formats/test_sharev2_elastic_formatter.py b/tests/share/metadata_formats/test_sharev2_elastic_formatter.py
@@ -87,6 +87,56 @@ def test_formatter(self, encode_mock, normalized_datum, expected_output):
                 ],
             },
         },
+        'no-names-only-name-parts': {
+            'contributors': ['Suzanne Simard', 'Mary Austi'],
+            'date': '2017-03-31T05:39:48+00:00',
+            'date_created': '2017-04-07T21:09:05.023090+00:00',
+            'date_modified': '2017-04-07T21:09:05.023090+00:00',
+            'date_updated': '2017-03-31T05:39:48+00:00',
+            'id': 'encoded-7',
+            'identifiers': ['http://dx.doi.org/10.5772/9813'],
+            'publishers': [],
+            'retracted': False,
+            'sources': ['SomeSource'],
+            'title': 'The Role of Mycorrhizas in Forest Soil Stability with Climate Change',
+            'type': 'creative work',
+            'types': ['creative work'],
+            'affiliations': [],
+            'funders': [],
+            'hosts': [],
+            'subject_synonyms': [],
+            'subjects': [],
+            'tags': [],
+            'lists': {
+                'affiliations': [],
+                'contributors': [
+                    {
+                        'family_name': 'Simard',
+                        'given_name': 'Suzanne',
+                        'identifiers': [],
+                        'name': 'Suzanne Simard',
+                        'order_cited': 0,
+                        'relation': 'creator',
+                        'type': 'person',
+                        'types': ['person', 'agent'],
+                    },
+                    {
+                        'family_name': 'Austi',
+                        'given_name': 'Mary',
+                        'identifiers': [],
+                        'name': 'Mary Austi',
+                        'order_cited': 1,
+                        'relation': 'creator',
+                        'type': 'person',
+                        'types': ['person', 'agent'],
+                    },
+                ],
+                'funders': [],
+                'hosts': [],
+                'lineage': [],
+                'publishers': [],
+            },
+        },
         'with-is_deleted': {
             'id': 'encoded-57',
             'is_deleted': True,