Implement a encoded search to json-ld wrapper.

An open question is encoded returns terms like "foo.bar" where
bar is an attribute of an object name for foo. However since
foo has no id associated with it, at best I would have a term with
a blank node.

For the moment I'm just deleting those terms. I'm not sure what a
better solution might be.

diff --git a/htsworkflow/submission/encoded.py b/htsworkflow/submission/encoded.py
index 05813fd550e7a352a6177ac7195c2c1586a167d9..1228d387f015a39a9da12c8ad5ca296d218c68de 100644 (file)
--- a/htsworkflow/submission/encoded.py
+++ b/htsworkflow/submission/encoded.py
@@ -277,6 +277,28 @@ class ENCODED:
         url = urlunparse(url.values())
         return url
 
+    def search_jsonld(self, term, **kwargs):
+        '''Send search request to ENCODED
+        '''
+        url = self.prepare_url('/search/')
+        result = self.get_json(url, searchTerm=term, **kwargs)
+        self.convert_search_to_jsonld(result)
+        return result
+
+    def convert_search_to_jsonld(self, result):
+        '''Add the context to search result
+
+        Also remove hard to handle nested attributes
+          e.g. remove object.term when we have no id
+        '''
+        graph = result['@graph']
+        for i, obj in enumerate(graph):
+            # suppress nested attributes
+            graph[i] = {k: v for k, v in obj.items() if '.' not in k}
+
+        self.add_jsonld_context(result, self.prepare_url(result['@id']))
+        return result
+
     def validate(self, obj):
         obj_type = self.get_object_type(obj)
         schema_url = self.get_schema_url(obj)

diff --git a/htsworkflow/submission/test/test_encoded.py b/htsworkflow/submission/test/test_encoded.py
index 5483f6f325dbcc03452279d202940141d8892f23..675d944a4bc0bb33e1e1b78be7ba4608da64d511 100644 (file)
--- a/htsworkflow/submission/test/test_encoded.py
+++ b/htsworkflow/submission/test/test_encoded.py
@@ -122,6 +122,51 @@ class TestEncoded(TestCase):
         self.assertEqual(obj['@context']['OBO'], 'http://purl.obolibrary.org/obo/')
 
 
+    def test_convert_search_to_jsonld(self):
+        example = {'count': {'biosamples': 2},
+                   'portal_title': 'ENCODE',
+                   'title': 'Search',
+                   'notification': 'Success',
+                   'filters': [],
+                   '@id': '/search/?searchTerm=wold',
+                   '@type': ['search'],
+                   'facets': [],
+                    '@graph': [{
+                    u'@id': u'/biosamples/ENCBS125ENC/',
+                    u'@type': [u'biosample', u'item'],
+                    u'accession': u'ENCBS125ENC',
+                    u'award.rfa': u'ENCODE2-Mouse',
+                    u'biosample_term_name': u'myocyte',
+                    u'biosample_type': u'in vitro differentiated cells',
+                    u'characterizations.length': [],
+                    u'constructs.length': [],
+                    u'lab.title': u'Barbara Wold, Caltech',
+                    u'life_stage': u'unknown',
+                    u'organism.name': u'mouse',
+                    u'source.title': u'Barbara Wold',
+                    u'status': u'CURRENT',
+                    u'treatments.length': []},
+                    {u'@id': u'/biosamples/ENCBS126ENC/',
+                    u'@type': [u'biosample', u'item'],
+                    u'accession': u'ENCBS126ENC',
+                    u'award.rfa': u'ENCODE2-Mouse',
+                    u'biosample_term_name': u'myocyte',
+                    u'biosample_type': u'in vitro differentiated cells',
+                    u'characterizations.length': [],
+                    u'constructs.length': [],
+                    u'lab.title': u'Barbara Wold, Caltech',
+                    u'life_stage': u'unknown',
+                    u'organism.name': u'mouse',
+                    u'source.title': u'Barbara Wold',
+                    u'status': u'CURRENT',
+                    u'treatments.length': []},
+                    ]}
+
+        encode = ENCODED('test.encodedcc.org')
+        result = encode.convert_search_to_jsonld(example)
+        for obj in result['@graph']:
+            self.assertNotIn('award.rfa', obj)
+
     def _verify_context(self, context, obj_type):
         for context_key in [None, obj_type]:
             for k in ENCODED_CONTEXT[context_key]: