SciX ID included in output

config.py

@@ -13,7 +13,7 @@
 REFERENCE_SERVICE_MAX_RECORDS_SOLR = 100
 
 REFERENCE_SERVICE_QUERY_FIELDS_SOLR = "author,[fields author=10]author_norm,[fields author_norm=10],first_author_norm," \
-                                      "year,title,pub,pub_raw,aff_raw,[fields aff_raw=1]," \
+                                      "year,title,pub,pub_raw,aff_raw,[fields aff_raw=1],scix_id," \
                                       "volume,issue,page,page_range,bibstem,bibcode,identifier,doi,doctype"
 
 # maximum references that can be resolved in one call

referencesrv/resolver/common.py

@@ -264,7 +264,7 @@ class Solution(object):
     * score
     * source_hypothesis (the hypothesis that eventually got it right)
     """
-    def __init__(self, cited_bibcode, score, source_hypothesis='not given', citing_bibcode=None):
+    def __init__(self, cited_bibcode, score, source_hypothesis='not given', cited_bibcode=None, citing_bibcode=None, scix_id=None):
         """
 
         :param cited_bibcode:
@@ -276,14 +276,15 @@ def __init__(self, cited_bibcode, score, source_hypothesis='not given', citing_b
         self.score = score
         self.citing_bibcode = str(citing_bibcode)
         self.source_hypothesis = source_hypothesis
+        self.scix_id = scix_id
 
     def __str__(self):
         """
 
         :return:
         """
         if isinstance(self.score, Evidences):
-            return '%.1f %s'%(self.score.avg(),self.cited_bibcode)
+            return '%.1f bibcode:%s scixid:%s'%(self.score.avg(),self.cited_bibcode, self.scix_id)
         raise NoSolution("NotResolved")
 
     def __repr__(self):
@@ -459,4 +460,4 @@ def predicate(x):
     non_numbers = filterfalse(predicate, t2)
     sorted_numbers = sorted(numbers)
     sorted_non_numbers = sorted(non_numbers, key=str)
-    return sorted_numbers + sorted_non_numbers
+    return sorted_numbers + sorted_non_numbers

referencesrv/resolver/solrtestdata.py

@@ -20,6 +20,7 @@ def get_test_data():
                  u'numFound': 2,
                  u'docs': [
                      {u'bibcode': u'2019AAS...23320704A',
+                      u'scix_id': u'scix:6ANE-YQXJ-KRH0',
                       u'author': [u'Accomazzi, Alberto'],
                       u'title': [u'The NASA Astrophysics Data System\xe2\u20ac\u2122s Decadal Plan for the 2020s'],
                       u'doctype': u'abstract',
@@ -32,6 +33,7 @@ def get_test_data():
                       u'identifier': [u'2019AAS...23320704A'],
                       u'page': [u'207.04']},
                      {u'bibcode': u'2019AAS...23338108A',
+                      u'scix_id': u'scix:AGA3-9D3P-Y7EF',
                       u'author': [u'Accomazzi, Alberto', u'Kurtz, Michael J.', u'Henneken, Edwin', u'Grant, Carolyn S.', u'Thompson, Donna M.', u'Chyla, Roman', u'McDonald, Stephen', u'Blanco-Cuaresma, Sergi', u'Shapurian, Golnaz', u'Hostetler, Timothy', u'Templeton, Matthew', u'Lockhart, Kelly'],
                       u'title': [u'Transitioning from ADS Classic to the new ADS search platform'],
                       u'doctype': u'abstract',
@@ -45,4 +47,4 @@ def get_test_data():
                       u'page': [u'381.08']}
                  ]
                  }
-            }
+            }

referencesrv/resolver/solve.py

@@ -288,11 +288,10 @@ def solve_for_fields(hypothesis):
 
             current_app.logger.debug("evidences from %s"%(hypothesis.name))
             for score, sol in sorted2(scored):
-                current_app.logger.debug("score %s %s %s"%(sol['bibcode'], score.get_score(), score))
+                current_app.logger.debug("score %s %s %s"%(sol.get('bibcode',None), score.get_score(), score))
 
             score, sol = choose_solution(scored, query_string, hypothesis)
-
-            return Solution(sol["bibcode"], score, hypothesis.name)
+            return Solution(sol.get("bibcode",None), score, hypothesis.name, scix_id=sol.get("scix_id",None))
 
     raise OverflowOrNone("Got either too many or no records from solr")
 

referencesrv/tests/unittests/test_referencesrv_parser.py

@@ -756,6 +756,7 @@ def test_01(self):
                                                                       u'year': u'2020',
                                                                       u'page': u'2',
                                                                       u'bibcode': u'2020JHEP...09..002P',
+                                                                      u'scix_id': u'scix:5KGH-MC98-7AYN',
                                                                       u'author': [u'Penington, Geoffrey'], u'issue': u'9',
                                                                       u'aff_raw': u'Stanford Institute for Theoretical Physics, Stanford University, 450 Jane Stanford Way, 94305, Stanford, CA, USA',
                                                                       u'pub': u'Journal of High Energy Physics',
@@ -769,7 +770,7 @@ def test_01(self):
                                                           }
                                             })
             r = self.client.post(path='/text', data=json.dumps({'reference': ['Penington, G, 2020, JHEP, 9']}))
-            self.assertEqual(r.data, b"1.0 2020JHEP...09..002P -- Penington, G, 2020, JHEP, 9")
+            self.assertEqual(r.data, b"1.0 bibcode:2020JHEP...09..002P scixid:scix:5KGH-MC98-7AYN -- Penington, G, 2020, JHEP, 9")
 
     def test_02(self):
         """ test text endpoint when request is to return in json format """
@@ -785,6 +786,7 @@ def test_02(self):
                                                                       u'year': u'2020',
                                                                       u'page': u'2',
                                                                       u'bibcode': u'2020JHEP...09..002P',
+                                                                      u'scix_id': u'scix:5KGH-MC98-7AYN',
                                                                       u'author': [u'Penington, Geoffrey'], u'issue': u'9',
                                                                       u'aff_raw': u'Stanford Institute for Theoretical Physics, Stanford University, 450 Jane Stanford Way, 94305, Stanford, CA, USA',
                                                                       u'pub': u'Journal of High Energy Physics',
@@ -802,6 +804,7 @@ def test_02(self):
                                  data=json.dumps({'reference': ['Penington, G, 2020, JHEP, 9']}),
                                  headers={'accept':'application/json'})
             self.assertEqual(json.loads(r.data), {"resolved": [{"refstring": "Penington, G, 2020, JHEP, 9",
+                                                                "scix_id":"scix:5KGH-MC98-7AYN",
                                                                 "score": "1.0",
                                                                 "bibcode": "2020JHEP...09..002P"}]})
 

referencesrv/tests/unittests/test_referencesrv_resolver.py

@@ -356,8 +356,8 @@ def test_Solution(self):
         """
         e = Evidences()
         e.add_evidence(1, 'bibcode')
-        s = Solution(cited_bibcode='2013SPIE.8004.2013Z', score=e)
-        self.assertEqual(str(s), '1.0 2013SPIE.8004.2013Z')
+        s = Solution(cited_bibcode='2013SPIE.8004.2013Z', scix_id='foo', score=e)
+        self.assertEqual(str(s), '1.0 bibcode:2013SPIE.8004.2013Z scixid:foo')
         self.assertEqual(repr(s), "'2013SPIE.8004.2013Z'")
 
 
@@ -417,22 +417,22 @@ def test_solve_reference(self):
                'volume': '233',
                'year': '2019',
                'page': '207.04'}
-        self.assertEqual(str(solve_reference(Hypotheses(ref))), '1.0 2019AAS...23320704A')
+        self.assertEqual(str(solve_reference(Hypotheses(ref))), '1.0 bibcode:2019AAS...23320704A scixid:scix:6ANE-YQXJ-KRH0')
         # testing with first author only and page
         # eventhough other authors are missing but because of page match is found
         ref = {'authors': 'Accomazzi, A.',
                'journal': 'AAS233 Meeting',
                'volume': '233',
                'year': '2019',
                'page': '381.08'}
-        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 2019AAS...23338108A')
+        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 bibcode:2019AAS...23338108A scixid:scix:AGA3-9D3P-Y7EF')
         # testing with first author only and no page, hence record with only the first author is returned
         ref = {'authors': 'Accomazzi, A.',
                'journal': 'AAS233 Meeting',
                'volume': '233',
                'year': '2019',
                'page': '0'}
-        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 2019AAS...23320704A')
+        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 bibcode:2019AAS...23320704A scixid:scix:6ANE-YQXJ-KRH0')
         # when we have multiple solutions and not enough reference information to decide which
         # page and author are the deciding factor between these two test records
         # here first author and page are wrong
@@ -454,7 +454,7 @@ def test_solve_reference(self):
         # however the first record is authored by one author only and
         # it is the same first author of the second record
         # verify that the first record is returned
-        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 2019AAS...23320704A')
+        self.assertEqual(str(solve_reference(Hypotheses(ref))), '0.8 bibcode:2019AAS...23320704A scixid:scix:6ANE-YQXJ-KRH0')
 
 
     def test_add_volume_evidence(self):
@@ -637,7 +637,7 @@ def test_Querier(self):
         self.assertEqual(solrquery.make_params('author:("Accomazzi, A") AND year:"2019" AND bibstem:(AAS)'),
                          {'q': 'author:("Accomazzi, A") AND year:"2019" AND bibstem:(AAS)',
                           'rows': '100',
-                          'fl': u'author,[fields author=10]author_norm,[fields author_norm=10],first_author_norm,year,title,pub,pub_raw,aff_raw,[fields aff_raw=1],volume,issue,page,page_range,bibstem,bibcode,identifier,doi,doctype'})
+                          'fl': u'author,[fields author=10]author_norm,[fields author_norm=10],first_author_norm,year,title,pub,pub_raw,aff_raw,[fields aff_raw=1],scix_id,volume,issue,page,page_range,bibstem,bibcode,identifier,doi,doctype'})
 
         # no author_norm
         solution = {u'bibcode': u'2013JARS....7.3461V',

referencesrv/views.py

@@ -123,7 +123,9 @@ def format_resolved_reference(returned_format, resolved, reference, id, cache=Tr
         cache_resolved_set(reference, resolved)
     if 'application/json' in returned_format:
         resolved = resolved.split()
-        result = {'refstring': reference, 'score': resolved[0], 'bibcode': resolved[1]}
+        bibcode = resolved[1].replace('bibcode:','').strip()
+        scix_id = resolved[2].replace('scixid:','').strip()
+        result = {'refstring': reference, 'score': resolved[0], 'bibcode': bibcode, 'scix_id':scix_id}
         if comment:
             result['comment'] = comment
         if id: