Merge pull request #794 from aaxelb/feature/suid-guid

[no ticket] index suid for search
CenterForOpenScience · Aug 9, 2022 · 938ed3e · 938ed3e
2 parents d2c268f + b2d5b2a
commit 938ed3e
Show file tree

Hide file tree

Showing 4 changed files with 24 additions and 0 deletions.
diff --git a/share/metadata_formats/sharev2_elastic.py b/share/metadata_formats/sharev2_elastic.py
@@ -74,6 +74,8 @@ def format(self, normalized_datum):
         return json.dumps(strip_empty_values({
             'id': IDObfuscator.encode(suid),
             'sources': [source_name],
+            'source_config': suid.source_config.label,
+            'source_unique_id': suid.identifier,
 
             'type': format_node_type(central_work),
             'types': format_node_type_lineage(central_work),

diff --git a/share/search/index_setup/postrend_backcompat.py b/share/search/index_setup/postrend_backcompat.py
@@ -105,6 +105,8 @@ def index_mappings(self):
                     'publishers': {'type': 'text', 'fields': exact_field},
                     'registration_type': {'type': 'keyword', 'include_in_all': False},
                     'retracted': {'type': 'boolean', 'include_in_all': False},
+                    'source_config': {'type': 'keyword', 'include_in_all': False},
+                    'source_unique_id': {'type': 'keyword'},
                     'sources': {'type': 'keyword', 'include_in_all': False},
                     'subjects': {'type': 'text', 'include_in_all': False, 'analyzer': 'subject_analyzer', 'search_analyzer': 'subject_search_analyzer'},
                     'subject_synonyms': {'type': 'text', 'include_in_all': False, 'analyzer': 'subject_analyzer', 'search_analyzer': 'subject_search_analyzer', 'copy_to': 'subjects'},

diff --git a/tests/share/metadata_formats/base.py b/tests/share/metadata_formats/base.py
@@ -10,7 +10,9 @@
 FORMATTER_TEST_INPUTS = {
     'mycorrhizas': {
         'suid_id': 7,
+        'suid_value': 'oof',
         'source_name': 'SomeSource',
+        'source_config_label': 'foo',
         'raw_datum_kwargs': {
             'date_created': dateutil.parser.isoparse('2017-04-07T21:09:05.023090+00:00'),
         },
@@ -33,7 +35,9 @@
     },
     'no-names-only-name-parts': {
         'suid_id': 7,
+        'suid_value': 'rab',
         'source_name': 'SomeSource',
+        'source_config_label': 'bar',
         'raw_datum_kwargs': {
             'date_created': dateutil.parser.isoparse('2017-04-07T21:09:05.023090+00:00'),
         },
@@ -54,7 +58,9 @@
     },
     'with-is_deleted': {
         'suid_id': 57,
+        'suid_value': 'zab',
         'source_name': 'foo',
+        'source_config_label': 'baz',
         'raw_datum_kwargs': {},
         'normalized_datum_kwargs': {
             'data': {
@@ -67,7 +73,9 @@
     },
     'with-subjects': {
         'suid_id': 123,
+        'suid_value': 'xuq',
         'source_name': 'osf reg',
+        'source_config_label': 'subj',
         'raw_datum_kwargs': {
             'date_created': dateutil.parser.isoparse('2020-02-02T20:20:02.02+00:00'),
         },
@@ -325,7 +333,9 @@
     },
     'with-osf-extra': {
         'suid_id': 99,
+        'suid_value': 'guidz',
         'source_name': 'OsfProbably',
+        'source_config_label': 'osf.io.v2_push',
         'raw_datum_kwargs': {
             'date_created': dateutil.parser.isoparse('2017-04-07T21:09:05.023090+00:00'),
         },
@@ -416,6 +426,8 @@ def normalized_datum(self, formatter_test_input):
         return NormalizedDataFactory(
             raw=RawDatumFactory(
                 suid__id=formatter_test_input['suid_id'],
+                suid__identifier=formatter_test_input['suid_value'],
+                suid__source_config__label=formatter_test_input['source_config_label'],
                 suid__source_config__source__long_title=formatter_test_input['source_name'],
                 **formatter_test_input['raw_datum_kwargs'],
             ),

diff --git a/tests/share/metadata_formats/test_sharev2_elastic_formatter.py b/tests/share/metadata_formats/test_sharev2_elastic_formatter.py
@@ -42,6 +42,8 @@ def assert_formatter_outputs_equal(self, actual_output, expected_output):
             'identifiers': ['http://dx.doi.org/10.5772/9813'],
             'publishers': ['InTech'],
             'retracted': False,
+            'source_config': 'foo',
+            'source_unique_id': 'oof',
             'sources': ['SomeSource'],
             'title': 'The Role of Mycorrhizas in Forest Soil Stability with Climate Change',
             'type': 'creative work',
@@ -102,6 +104,8 @@ def assert_formatter_outputs_equal(self, actual_output, expected_output):
             'identifiers': ['http://dx.doi.org/10.5772/9813'],
             'publishers': [],
             'retracted': False,
+            'source_config': 'bar',
+            'source_unique_id': 'rab',
             'sources': ['SomeSource'],
             'title': 'The Role of Mycorrhizas in Forest Soil Stability with Climate Change',
             'type': 'creative work',
@@ -157,6 +161,8 @@ def assert_formatter_outputs_equal(self, actual_output, expected_output):
             'identifiers': ['http://staging.osf.io/chair/'],
             'registration_type': 'Open-Ended Registration',
             'retracted': False,
+            'source_config': 'subj',
+            'source_unique_id': 'xuq',
             'sources': ['osf reg'],
             'subject_synonyms': [
                 'bepress|Life Sciences|Biology',
@@ -223,6 +229,8 @@ def assert_formatter_outputs_equal(self, actual_output, expected_output):
             'date_updated': '2017-03-31T05:39:48+00:00',
             'id': 'encoded-99',
             'identifiers': ['https://example.com/open'],
+            'source_config': 'osf.io.v2_push',
+            'source_unique_id': 'guidz',
             'sources': ['OsfProbably'],
             'subject_synonyms': [],
             'subjects': [],