Merge pull request #490 from edx/boost_orgs_typeahead

Boost orgs autocomplete in typeahead

course_discovery/apps/api/serializers.py

@@ -974,29 +974,23 @@ class AggregateSearchSerializer(HaystackSerializer):
         }
 
 
-class TypeaheadCourseRunSearchSerializer(serializers.Serializer):
-    org = serializers.CharField()
-    title = serializers.CharField()
-    key = serializers.CharField()
-    marketing_url = serializers.CharField()
-
-    class Meta:
-        fields = ['key', 'title']
-
-
-class TypeaheadProgramSearchSerializer(serializers.Serializer):
+class TypeaheadBaseSearchSerializer(serializers.Serializer):
     orgs = serializers.SerializerMethodField()
-    uuid = serializers.CharField()
     title = serializers.CharField()
-    type = serializers.CharField()
     marketing_url = serializers.CharField()
 
     def get_orgs(self, result):
         authoring_organizations = [json.loads(org) for org in result.authoring_organization_bodies]
         return [org['key'] for org in authoring_organizations]
 
-    class Meta:
-        fields = ['uuid', 'title', 'type']
+
+class TypeaheadCourseRunSearchSerializer(TypeaheadBaseSearchSerializer):
+    key = serializers.CharField()
+
+
+class TypeaheadProgramSearchSerializer(TypeaheadBaseSearchSerializer):
+    uuid = serializers.CharField()
+    type = serializers.CharField()
 
 
 class TypeaheadSearchSerializer(serializers.Serializer):

course_discovery/apps/api/tests/test_serializers.py

@@ -1125,14 +1125,14 @@ class ProgramSearchSerializerTests(TestCase):
 
 class TypeaheadCourseRunSearchSerializerTests(TestCase):
     def test_data(self):
-        course_run = CourseRunFactory()
+        authoring_organization = OrganizationFactory()
+        course_run = CourseRunFactory(authoring_organizations=[authoring_organization])
         serialized_course = self.serialize_course_run(course_run)
-        course_run_key = CourseKey.from_string(course_run.key)
 
         expected = {
             'key': course_run.key,
             'title': course_run.title,
-            'org': course_run_key.org,
+            'orgs': [org.key for org in course_run.authoring_organizations.all()],
             'marketing_url': course_run.marketing_url,
         }
         self.assertDictEqual(serialized_course.data, expected)
@@ -1150,7 +1150,7 @@ class TypeaheadProgramSearchSerializerTests(TestCase):
             'uuid': str(program.uuid),
             'title': program.title,
             'type': program.type.name,
-            'orgs': list(program.authoring_organizations.all().values_list('key', flat=True)),
+            'orgs': [org.key for org in program.authoring_organizations.all()],
             'marketing_url': program.marketing_url,
         }
 

course_discovery/apps/api/v1/tests/test_views/test_search.py

@@ -465,3 +465,33 @@ class TypeaheadSearchViewTests(TypeaheadSerializationMixin, LoginMixin, Elastics
             'programs': [self.serialize_program(program)]
         }
         self.assertDictEqual(response.data, expected)
+
+    def test_typeahead_org_course_runs_come_up_first(self):
+        """ Test typeahead response to ensure org is taken into account. """
+        MITx = OrganizationFactory(key='MITx')
+        HarvardX = OrganizationFactory(key='HarvardX')
+        mit_run = CourseRunFactory(
+            authoring_organizations=[MITx, HarvardX],
+            title='MIT Testing1'
+        )
+        harvard_run = CourseRunFactory(
+            authoring_organizations=[HarvardX],
+            title='MIT Testing2'
+        )
+        mit_program = ProgramFactory(
+            authoring_organizations=[MITx, HarvardX],
+            title='MIT Testing1'
+        )
+        harvard_program = ProgramFactory(
+            authoring_organizations=[HarvardX],
+            title='MIT Testing2'
+        )
+        response = self.get_typeahead_response('mit')
+        self.assertEqual(response.status_code, 200)
+        expected = {
+            'course_runs': [self.serialize_course_run(mit_run),
+                            self.serialize_course_run(harvard_run)],
+            'programs': [self.serialize_program(mit_program),
+                         self.serialize_program(harvard_program)]
+        }
+        self.assertDictEqual(response.data, expected)

course_discovery/apps/course_metadata/search_indexes.py

@@ -14,6 +14,10 @@ from course_discovery.apps.course_metadata.models import Course, CourseRun, Prog
 # If we altered our boosting functions to have a max score of 10
 # we would probably want to bump this number.
 TITLE_FIELD_BOOST = 25.0
+# We want to boost org the same as title
+# Primarily this is to prevent courses from other institutions
+# coming up first if there is a partial match on title
+ORG_FIELD_BOOST = TITLE_FIELD_BOOST
 
 
 class OrganizationsMixin:
@@ -30,6 +34,9 @@ class OrganizationsMixin:
     def _prepare_organizations(self, organizations):
         return [self.format_organization(organization) for organization in organizations]
 
+    def prepare_authoring_organization_bodies(self, obj):
+        return [self.format_organization_body(organization) for organization in obj.authoring_organizations.all()]
+
     def prepare_authoring_organizations(self, obj):
         return self._prepare_organizations(obj.authoring_organizations.all())
 
@@ -63,7 +70,8 @@ class BaseCourseIndex(OrganizationsMixin, BaseIndex):
     key = indexes.CharField(model_attr='key', stored=True)
     title = indexes.CharField(model_attr='title', boost=TITLE_FIELD_BOOST)
     title_autocomplete = indexes.NgramField(model_attr='title', boost=TITLE_FIELD_BOOST)
-    authoring_organizations_autocomplete = indexes.NgramField()
+    authoring_organizations_autocomplete = indexes.NgramField(boost=ORG_FIELD_BOOST)
+    authoring_organization_bodies = indexes.MultiValueField()
     short_description = indexes.CharField(model_attr='short_description', null=True)
     full_description = indexes.CharField(model_attr='full_description', null=True)
     subjects = indexes.MultiValueField(faceted=True)
@@ -197,7 +205,7 @@ class ProgramIndex(BaseIndex, indexes.Indexable, OrganizationsMixin):
     marketing_url = indexes.CharField(null=True)
     organizations = indexes.MultiValueField(faceted=True)
     authoring_organizations = indexes.MultiValueField(faceted=True)
-    authoring_organizations_autocomplete = indexes.NgramField()
+    authoring_organizations_autocomplete = indexes.NgramField(boost=ORG_FIELD_BOOST)
     authoring_organization_uuids = indexes.MultiValueField()
     subject_uuids = indexes.MultiValueField()
     staff_uuids = indexes.MultiValueField()
@@ -216,9 +224,6 @@ class ProgramIndex(BaseIndex, indexes.Indexable, OrganizationsMixin):
     def prepare_organizations(self, obj):
         return self.prepare_authoring_organizations(obj) + self.prepare_credit_backing_organizations(obj)
 
-    def prepare_authoring_organization_bodies(self, obj):
-        return [self.format_organization_body(organization) for organization in obj.authoring_organizations.all()]
-
     def prepare_subject_uuids(self, obj):
         return [str(subject.uuid) for course in obj.courses.all() for subject in course.subjects.all()]