Merge pull request #115 from CredoReference/api-to-get-tags-distribution

New API method to return tags distribution info

AUTHORS

@@ -6,3 +6,4 @@ Ed Zarecor <ed@edx.org>
 Gabe Mulley <gabe@edx.org>
 Jason Bau <jbau@stanford.edu>
 John Jarvis <jarv@edx.org>
+Dmitry Viskov <dmitry.viskov@webenterprise.ru>

analytics_data_api/management/commands/generate_fake_course_data.py

@@ -225,6 +225,34 @@ class Command(BaseCommand):
                     course_id=course_id, start_date=start_date, end_date=end_date, metric=metric,
                     range_type='high', low_value=high_floor, high_value=max_value)
 
+    def generate_tags_distribution_data(self, course_id):
+        logger.info("Deleting existed tags distribution data...")
+        models.ProblemsAndTags.objects.all().delete()
+
+        module_id_tpl = 'i4x://test/problem/%d'
+        difficulty_tag = ['Easy', 'Medium', 'Hard']
+        learning_outcome_tag = ['Learned nothing', 'Learned a few things', 'Learned everything']
+        problems_num = 50
+        chance_difficulty = 5
+
+        logger.info("Generating new tags distribution data...")
+        for i in xrange(problems_num):
+            module_id = module_id_tpl % i
+            total_submissions = random.randint(0, 100)
+            correct_submissions = random.randint(0, total_submissions)
+
+            models.ProblemsAndTags.objects.create(
+                course_id=course_id, module_id=module_id,
+                tag_name='learning_outcome', tag_value=random.choice(learning_outcome_tag),
+                total_submissions=total_submissions, correct_submissions=correct_submissions
+            )
+            if random.randint(0, chance_difficulty) != chance_difficulty:
+                models.ProblemsAndTags.objects.create(
+                    course_id=course_id, module_id=module_id,
+                    tag_name='difficulty', tag_value=random.choice(difficulty_tag),
+                    total_submissions=total_submissions, correct_submissions=correct_submissions
+                )
+
     def handle(self, *args, **options):
         course_id = options['course_id']
         username = options['username']
@@ -245,3 +273,4 @@ class Command(BaseCommand):
         self.generate_video_timeline_data(video_id)
         self.generate_learner_engagement_data(course_id, username, start_date, end_date)
         self.generate_learner_engagement_range_data(course_id, start_date, end_date)
+        self.generate_tags_distribution_data(course_id)

analytics_data_api/v0/models.py

@@ -131,6 +131,21 @@ class ProblemResponseAnswerDistribution(BaseProblemResponseAnswerDistribution):
     count = models.IntegerField()
 
 
+class ProblemsAndTags(models.Model):
+    """ Model for the tags_distribution table """
+
+    class Meta(object):
+        db_table = 'tags_distribution'
+
+    course_id = models.CharField(db_index=True, max_length=255)
+    module_id = models.CharField(db_index=True, max_length=255)
+    tag_name = models.CharField(max_length=255)
+    tag_value = models.CharField(max_length=255)
+    total_submissions = models.IntegerField(default=0)
+    correct_submissions = models.IntegerField(default=0)
+    created = models.DateTimeField(auto_now_add=True)
+
+
 class ProblemFirstLastResponseAnswerDistribution(BaseProblemResponseAnswerDistribution):
     """ Updated model for answer_distribution table with counts of first and last attempts at problems. """
 

analytics_data_api/v0/serializers.py

@@ -57,6 +57,18 @@ class ProblemSerializer(serializers.Serializer):
     created = serializers.DateTimeField(format=settings.DATETIME_FORMAT)
 
 
+class ProblemsAndTagsSerializer(serializers.Serializer):
+    """
+    Serializer for problems and tags.
+    """
+
+    module_id = serializers.CharField(required=True)
+    total_submissions = serializers.IntegerField(default=0)
+    correct_submissions = serializers.IntegerField(default=0)
+    tags = serializers.CharField()
+    created = serializers.DateTimeField(format=settings.DATETIME_FORMAT)
+
+
 class ProblemResponseAnswerDistributionSerializer(ModelSerializerWithCreatedField):
     """
     Representation of the Answer Distribution table, without id.

analytics_data_api/v0/tests/views/test_courses.py

@@ -650,6 +650,82 @@ class CourseProblemsListViewTests(DemoCourseMixin, TestCaseWithAuthentication):
         self.assertEquals(response.status_code, 404)
 
 
+class CourseProblemsAndTagsListViewTests(DemoCourseMixin, TestCaseWithAuthentication):
+    def _get_data(self, course_id=None):
+        """
+        Retrieve data for the specified course.
+        """
+
+        course_id = course_id or self.course_id
+        url = '/api/v0/courses/{}/problems_and_tags/'.format(course_id)
+        return self.authenticated_get(url)
+
+    def test_get(self):
+        """
+        The view should return data when data exists for the course.
+        """
+
+        # This data should never be returned by the tests below because the course_id doesn't match.
+        G(models.ProblemsAndTags)
+
+        # Create multiple objects here to test the grouping. Add a model with a different module_id to break up the
+        # natural order and ensure the view properly sorts the objects before grouping.
+        module_id = 'i4x://test/problem/1'
+        alt_module_id = 'i4x://test/problem/2'
+
+        tags = {
+            'difficulty': ['Easy', 'Medium', 'Hard'],
+            'learning_outcome': ['Learned nothing', 'Learned a few things', 'Learned everything']
+        }
+
+        created = datetime.datetime.utcnow()
+        alt_created = created + datetime.timedelta(seconds=2)
+
+        G(models.ProblemsAndTags, course_id=self.course_id, module_id=module_id,
+          tag_name='difficulty', tag_value=tags['difficulty'][0],
+          total_submissions=11, correct_submissions=4, created=created)
+        G(models.ProblemsAndTags, course_id=self.course_id, module_id=module_id,
+          tag_name='learning_outcome', tag_value=tags['learning_outcome'][1],
+          total_submissions=11, correct_submissions=4, created=alt_created)
+        G(models.ProblemsAndTags, course_id=self.course_id, module_id=alt_module_id,
+          tag_name='learning_outcome', tag_value=tags['learning_outcome'][2],
+          total_submissions=4, correct_submissions=0, created=created)
+
+        expected = [
+            {
+                'module_id': module_id,
+                'total_submissions': 11,
+                'correct_submissions': 4,
+                'tags': {
+                    'difficulty': 'Easy',
+                    'learning_outcome': 'Learned a few things',
+                },
+                'created': alt_created.strftime(settings.DATETIME_FORMAT)
+            },
+            {
+                'module_id': alt_module_id,
+                'total_submissions': 4,
+                'correct_submissions': 0,
+                'tags': {
+                    'learning_outcome': 'Learned everything',
+                },
+                'created': created.strftime(settings.DATETIME_FORMAT)
+            }
+        ]
+
+        response = self._get_data(self.course_id)
+        self.assertEquals(response.status_code, 200)
+        self.assertListEqual(sorted(response.data), sorted(expected))
+
+    def test_get_404(self):
+        """
+        The view should return 404 if no data exists for the course.
+        """
+
+        response = self._get_data('foo/bar/course')
+        self.assertEquals(response.status_code, 404)
+
+
 class CourseVideosListViewTests(DemoCourseMixin, TestCaseWithAuthentication):
     def _get_data(self, course_id=None):
         """

analytics_data_api/v0/urls/courses.py

@@ -13,6 +13,7 @@ COURSE_URLS = [
     ('enrollment/gender', views.CourseEnrollmentByGenderView, 'enrollment_by_gender'),
     ('enrollment/location', views.CourseEnrollmentByLocationView, 'enrollment_by_location'),
     ('problems', views.ProblemsListView, 'problems'),
+    ('problems_and_tags', views.ProblemsAndTagsListView, 'problems_and_tags'),
     ('videos', views.VideosListView, 'videos')
 ]
 

analytics_data_api/v0/views/courses.py

@@ -628,9 +628,9 @@ class ProblemsListView(BaseCourseView):
         Returns a collection of submission counts and part IDs for each problem. Each collection contains:
 
             * module_id: The ID of the problem.
-            * total_submissions: Total number of submissions
+            * total_submissions: Total number of submissions.
             * correct_submissions: Total number of *correct* submissions.
-            * part_ids: List of problem part IDs
+            * part_ids: List of problem part IDs.
     """
     serializer_class = serializers.ProblemSerializer
     allow_empty = False
@@ -689,6 +689,53 @@ GROUP BY module_id;
         return rows
 
 
+# pylint: disable=abstract-method
+class ProblemsAndTagsListView(BaseCourseView):
+    """
+    Get the problems with the connected tags.
+
+    **Example request**
+
+        GET /api/v0/courses/{course_id}/problems_and_tags/
+
+    **Response Values**
+
+        Returns a collection of submission counts and tags for each problem. Each collection contains:
+
+            * module_id: The ID of the problem.
+            * total_submissions: Total number of submissions.
+            * correct_submissions: Total number of *correct* submissions.
+            * tags: Dictionary that contains pairs "tag key: tag value".
+    """
+    serializer_class = serializers.ProblemsAndTagsSerializer
+    allow_empty = False
+    model = models.ProblemsAndTags
+
+    def get_queryset(self):
+        queryset = self.model.objects.filter(course_id=self.course_id)
+        items = queryset.all()
+
+        result = {}
+
+        for v in items:
+            if v.module_id in result:
+                result[v.module_id]['tags'][v.tag_name] = v.tag_value
+                if result[v.module_id]['created'] < v.created:
+                    result[v.module_id]['created'] = v.created
+            else:
+                result[v.module_id] = {
+                    'module_id': v.module_id,
+                    'total_submissions': v.total_submissions,
+                    'correct_submissions': v.correct_submissions,
+                    'tags': {
+                        v.tag_name: v.tag_value
+                    },
+                    'created': v.created
+                }
+
+        return result.values()
+
+
 class VideosListView(BaseCourseView):
     """
     Get data for the videos in a course.