edx-enterprise-data 9.2.2__tar.gz → 9.4.0__tar.gz

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/CHANGELOG.rst

@@ -16,6 +16,15 @@ Unreleased
 
 =========================
 
+[9.4.0] - 2024-09-30
+---------------------
+  * chore: upgrade python requirements
+  * pin astriod and edx-lint packages
+
+[9.3.0] - 2024-09-30
+---------------------
+  * refactor: Further improvement in SQL queries for leaderboard API endpoint.
+
 [9.2.2] - 2024-09-27
 ---------------------
   * fix: remove the cache logging on EnterpriseLearnerEnrollmentViewSet.

{edx_enterprise_data-9.2.2/edx_enterprise_data.egg-info → edx_enterprise_data-9.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: edx-enterprise-data
-Version: 9.2.2
+Version: 9.4.0
 Summary: Enterprise Reporting
 Home-page: https://github.com/openedx/edx-enterprise-data
 Author: edX

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0/edx_enterprise_data.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: edx-enterprise-data
-Version: 9.2.2
+Version: 9.4.0
 Summary: Enterprise Reporting
 Home-page: https://github.com/openedx/edx-enterprise-data
 Author: edX

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/enterprise_data/__init__.py

@@ -2,4 +2,4 @@
 Enterprise data api application. This Django app exposes API endpoints used by enterprises.
 """
 
-__version__ = "9.2.2"
+__version__ = "9.4.0"

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/enterprise_data/admin_analytics/database/queries/fact_engagement_admin_dash.py

@@ -111,59 +111,62 @@ class FactEngagementAdminDashQueries:
         """
 
     @staticmethod
-    def get_all_leaderboard_data_query():
+    def get_engagement_data_for_leaderboard_query():
         """
-        Get the query to fetch the leaderboard data.
+        Get the query to fetch the engagement data for leaderboard.
 
-        Query should fetch the leaderboard data for the enterprise customer to show in the data table.
+        Query should fetch the engagement data for like learning time, session length of
+        the enterprise learners to show in the leaderboard.
 
         Returns:
-            (str): Query to fetch the leaderboard data.
+            (str): Query to fetch the engagement data for leaderboard.
         """
         return """
-            WITH Engagement AS (
-                SELECT
-                    email,
-                    ROUND(SUM(learning_time_seconds) / 3600, 1) as learning_time_hours,
-                    SUM(is_engaged) as session_count,
-                    CASE
-                        WHEN SUM(is_engaged) = 0 THEN 0.0
-                        ELSE ROUND(SUM(learning_time_seconds) / 3600 / SUM(is_engaged), 1)
-                    END AS average_session_length
-                FROM fact_enrollment_engagement_day_admin_dash
-                WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
-                    (activity_date BETWEEN %(start_date)s AND %(end_date)s) AND
-                    is_engaged = 1
-                GROUP BY email
-            ),
-            Completions AS (
-                SELECT email, count(course_key) as course_completion_count
-                FROM fact_enrollment_admin_dash
-                WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
-                    (passed_date BETWEEN %(start_date)s AND %(end_date)s) AND
-                    has_passed = 1
-                GROUP BY email
-            )
             SELECT
-                Engagement.email,
-                Engagement.learning_time_hours,
-                Engagement.session_count,
-                Engagement.average_session_length,
-                Completions.course_completion_count
-            FROM Engagement
-            LEFT JOIN Completions
-            ON Engagement.email = Completions.email
-            ORDER BY
-                Engagement.learning_time_hours DESC,
-                Engagement.session_count DESC,
-                Completions.course_completion_count DESC
+                email,
+                ROUND(SUM(learning_time_seconds) / 3600, 1) as learning_time_hours,
+                SUM(is_engaged) as session_count,
+                CASE
+                    WHEN SUM(is_engaged) = 0 THEN 0.0
+                    ELSE ROUND(SUM(learning_time_seconds) / 3600 / SUM(is_engaged), 1)
+                END AS average_session_length
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                (activity_date BETWEEN %(start_date)s AND %(end_date)s) AND
+                is_engaged = 1
+            GROUP BY email
+            ORDER BY learning_time_hours DESC
             LIMIT %(limit)s OFFSET %(offset)s;
         """
 
+    @staticmethod
+    def get_completion_data_for_leaderboard_query(email_list: list):
+        """
+        Get the query to fetch the completions data for leaderboard.
+
+        Query should fetch the completion data for like course completion count of
+        the enterprise learners to show in the leaderboard.
+
+        Arguments:
+            email_list (str): List of emails to filter the completions data.
+
+        Returns:
+            (list<str>): Query to fetch the completions data for leaderboard.
+        """
+        return f"""
+            SELECT email, count(course_key) as course_completion_count
+            FROM fact_enrollment_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                (passed_date BETWEEN %(start_date)s AND %(end_date)s) AND
+                has_passed = 1 AND
+                email IN {str(tuple(email_list))}
+            GROUP BY email;
+        """
+
     @staticmethod
     def get_leaderboard_data_count_query():
         """
-        Get the query to fetch the leaderboard row count.
+        Get the query to fetch the leaderboard row count and null email counter.
 
         Query should fetch the count of rows for the leaderboard data for the enterprise customer.
 
@@ -171,26 +174,13 @@ class FactEngagementAdminDashQueries:
             (str): Query to fetch the leaderboard row count.
         """
         return """
-            WITH Engagement AS (
-                SELECT
-                    email
-                FROM fact_enrollment_engagement_day_admin_dash
-                WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
-                    (activity_date BETWEEN %(start_date)s AND %(end_date)s) AND
-                    is_engaged = 1
-                GROUP BY email
-            ),
-            Completions AS (
-                SELECT email, count(course_key) as course_completion_count
-                FROM fact_enrollment_admin_dash
-                WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
-                    (passed_date BETWEEN %(start_date)s AND %(end_date)s) AND
-                    has_passed = 1
-                GROUP BY email
-            )
             SELECT
-                count(*)
-            FROM Engagement
-            LEFT JOIN Completions
-            ON Engagement.email = Completions.email
+                COUNT(*) OVER () AS record_count,
+                sum(case when email is null then 1 else 0 end) null_count
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                (activity_date BETWEEN %(start_date)s AND %(end_date)s) AND
+                is_engaged = 1
+            GROUP BY email
+            LIMIT 1;
         """

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/enterprise_data/admin_analytics/database/tables/fact_engagement_admin_dash.py

@@ -153,11 +153,14 @@ class FactEngagementAdminDashTable(BaseTable):
             as_dict=True,
         )
 
-    def get_all_leaderboard_data(
+    def _get_engagement_data_for_leaderboard(
             self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, limit: int, offset: int
     ):
         """
-        Get the leaderboard data for the given enterprise customer.
+        Get the engagement data for leaderboard.
+
+        The engagement data would include fields like learning time, session length of
+        the enterprise learners to show in the leaderboard.
 
         Arguments:
             enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
@@ -167,10 +170,10 @@ class FactEngagementAdminDashTable(BaseTable):
             offset (int): The number of records to skip.
 
         Returns:
-            list[dict]: The leaderboard data.
+            list[dict]: The engagement data for leaderboard.
         """
         return run_query(
-            query=self.queries.get_all_leaderboard_data_query(),
+            query=self.queries.get_engagement_data_for_leaderboard_query(),
             params={
                 'enterprise_customer_uuid': enterprise_customer_uuid,
                 'start_date': start_date,
@@ -181,6 +184,70 @@ class FactEngagementAdminDashTable(BaseTable):
             as_dict=True,
         )
 
+    def _get_completion_data_for_leaderboard_query(
+            self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, email_list: list
+    ):
+        """
+        Get the completion data for leaderboard.
+
+        The completion data would include fields like course completion count of enterprise learners of the
+        given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+            email_list (list<str>): List of emails of the enterprise learners.
+
+        Returns:
+            list[dict]: The engagement data for leaderboard.
+        """
+        return run_query(
+            query=self.queries.get_completion_data_for_leaderboard_query(email_list),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )
+
+    def get_all_leaderboard_data(
+            self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, limit: int, offset: int
+    ):
+        """
+        Get the leaderboard data for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+            limit (int): The maximum number of records to return.
+            offset (int): The number of records to skip.
+
+        Returns:
+            list[dict]: The leaderboard data.
+        """
+        engagement_data = self._get_engagement_data_for_leaderboard(
+            enterprise_customer_uuid=enterprise_customer_uuid,
+            start_date=start_date,
+            end_date=end_date,
+            limit=limit,
+            offset=offset,
+        )
+        engagement_data_dict = {engagement['email']: engagement for engagement in engagement_data}
+        completion_data = self._get_completion_data_for_leaderboard_query(
+            enterprise_customer_uuid=enterprise_customer_uuid,
+            start_date=start_date,
+            end_date=end_date,
+            email_list=list(engagement_data_dict.keys()),
+        )
+        for completion in completion_data:
+            email = completion['email']
+            engagement_data_dict[email]['course_completion_count'] = completion['course_completion_count']
+
+        return list(engagement_data_dict.values())
+
     def get_leaderboard_data_count(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
         """
         Get the total number of leaderboard records for the given enterprise customer.

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/requirements/base.txt

@@ -10,27 +10,27 @@ asgiref==3.8.1
     # via django
 asn1crypto==1.5.1
     # via snowflake-connector-python
-awscli==1.33.40
+awscli==1.34.29
     # via -r requirements/reporting.in
 bcrypt==4.2.0
     # via paramiko
-billiard==4.2.0
+billiard==4.2.1
     # via celery
-boto3==1.34.158
+boto3==1.35.29
     # via -r requirements/reporting.in
-botocore==1.34.158
+botocore==1.35.29
     # via
     #   awscli
     #   boto3
     #   s3transfer
 celery==5.3.6
     # via -r requirements/reporting.in
-certifi==2024.7.4
+certifi==2024.8.30
     # via
     #   py2neo
     #   requests
     #   snowflake-connector-python
-cffi==1.17.0
+cffi==1.17.1
     # via
     #   cryptography
     #   pynacl
@@ -54,7 +54,7 @@ click-repl==0.3.0
     # via celery
 colorama==0.4.6
     # via awscli
-cryptography==42.0.8
+cryptography==43.0.1
     # via
     #   -r requirements/reporting.in
     #   django-fernet-fields-v2
@@ -63,7 +63,7 @@ cryptography==42.0.8
     #   pyjwt
     #   pyopenssl
     #   snowflake-connector-python
-django==4.2.15
+django==4.2.16
     # via
     #   -c requirements/common_constraints.txt
     #   -r requirements/base.in
@@ -85,7 +85,7 @@ django-fernet-fields-v2==0.9
     # via -r requirements/base.in
 django-filter==24.3
     # via -r requirements/base.in
-django-model-utils==4.5.1
+django-model-utils==5.0.0
     # via
     #   -r requirements/base.in
     #   edx-rbac
@@ -111,25 +111,25 @@ edx-django-utils==5.15.0
     #   -r requirements/base.in
     #   edx-drf-extensions
     #   edx-rest-api-client
-edx-drf-extensions==10.3.0
+edx-drf-extensions==10.4.0
     # via
     #   -r requirements/base.in
     #   edx-rbac
-edx-opaque-keys==2.10.0
+edx-opaque-keys==2.11.0
     # via
     #   -r requirements/base.in
     #   edx-drf-extensions
-edx-rbac==1.9.0
+edx-rbac==1.10.0
     # via -r requirements/base.in
-edx-rest-api-client==5.7.1
+edx-rest-api-client==6.0.0
     # via -r requirements/base.in
-factory-boy==3.3.0
+factory-boy==3.3.1
     # via -r requirements/base.in
-faker==26.3.0
+faker==30.0.0
     # via factory-boy
-filelock==3.15.4
+filelock==3.16.1
     # via snowflake-connector-python
-idna==3.7
+idna==3.10
     # via
     #   requests
     #   snowflake-connector-python
@@ -139,13 +139,13 @@ jmespath==1.0.1
     # via
     #   boto3
     #   botocore
-kombu==5.4.0
+kombu==5.4.2
     # via celery
 monotonic==1.6
     # via py2neo
 mysql-connector-python==9.0.0
     # via -r requirements/base.in
-newrelic==9.13.0
+newrelic==10.0.0
     # via edx-django-utils
 numpy==1.24.4
     # via
@@ -162,21 +162,21 @@ pandas==2.0.3
     #   -r requirements/base.in
 pansi==2020.7.3
     # via py2neo
-paramiko==3.4.1
+paramiko==3.5.0
     # via -r requirements/reporting.in
-pbr==6.0.0
+pbr==6.1.0
     # via stevedore
 pgpy==0.6.0
     # via -r requirements/reporting.in
-platformdirs==4.2.2
+platformdirs==4.3.6
     # via snowflake-connector-python
-prompt-toolkit==3.0.47
+prompt-toolkit==3.0.48
     # via click-repl
 psutil==6.0.0
     # via edx-django-utils
 py2neo @ https://github.com/overhangio/py2neo/releases/download/2021.2.3/py2neo-2021.2.3.tar.gz
     # via -r requirements/reporting.in
-pyasn1==0.6.0
+pyasn1==0.6.1
     # via
     #   pgpy
     #   rsa
@@ -192,7 +192,7 @@ pyjwt[crypto]==2.9.0
     #   snowflake-connector-python
 pyminizip==0.2.6
     # via -r requirements/reporting.in
-pymongo==4.8.0
+pymongo==4.9.1
     # via edx-opaque-keys
 pynacl==1.5.0
     # via
@@ -207,7 +207,7 @@ python-dateutil==2.9.0.post0
     #   faker
     #   pandas
     #   vertica-python
-pytz==2024.1
+pytz==2024.2
     # via
     #   interchange
     #   pandas
@@ -219,11 +219,10 @@ requests==2.32.3
     #   -r requirements/base.in
     #   edx-drf-extensions
     #   edx-rest-api-client
-    #   slumber
     #   snowflake-connector-python
 rsa==4.7.2
     # via awscli
-rules==3.4
+rules==3.5
     # via -r requirements/base.in
 s3transfer==0.10.2
     # via
@@ -239,31 +238,30 @@ six==1.16.0
     #   py2neo
     #   python-dateutil
     #   vertica-python
-slumber==0.7.1
-    # via edx-rest-api-client
-snowflake-connector-python==3.12.0
+snowflake-connector-python==3.12.2
     # via -r requirements/reporting.in
 sortedcontainers==2.4.0
     # via snowflake-connector-python
 sqlparse==0.5.1
     # via django
-stevedore==5.2.0
+stevedore==5.3.0
     # via
     #   edx-django-utils
     #   edx-opaque-keys
-tomlkit==0.13.0
+tomlkit==0.13.2
     # via snowflake-connector-python
 typing-extensions==4.12.2
     # via
     #   edx-opaque-keys
     #   snowflake-connector-python
-tzdata==2024.1
+tzdata==2024.2
     # via
     #   celery
+    #   kombu
     #   pandas
 unicodecsv==0.14.1
     # via -r requirements/reporting.in
-urllib3==1.26.19
+urllib3==1.26.20
     # via
     #   -c requirements/constraints.txt
     #   botocore

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/requirements/ci.txt

@@ -4,7 +4,7 @@
 #
 #    make upgrade
 #
-cachetools==5.4.0
+cachetools==5.5.0
     # via tox
 chardet==5.2.0
     # via tox
@@ -14,7 +14,7 @@ coverage==7.6.1
     # via -r requirements/ci.in
 distlib==0.3.8
     # via virtualenv
-filelock==3.15.4
+filelock==3.16.1
     # via
     #   tox
     #   virtualenv
@@ -22,15 +22,15 @@ packaging==24.1
     # via
     #   pyproject-api
     #   tox
-platformdirs==4.2.2
+platformdirs==4.3.6
     # via
     #   tox
     #   virtualenv
 pluggy==1.5.0
     # via tox
-pyproject-api==1.7.1
+pyproject-api==1.8.0
     # via tox
-tox==4.17.1
+tox==4.20.0
     # via -r requirements/ci.in
-virtualenv==20.26.3
+virtualenv==20.26.6
     # via tox

{edx_enterprise_data-9.2.2 → edx_enterprise_data-9.4.0}/requirements/constraints.txt

@@ -20,4 +20,10 @@ backports.zoneinfo; python_version<"3.9"
 urllib3<2.0.0
 
 numpy<=1.24.4
-pandas<=2.0.3
+pandas<=2.0.3
+
+# this is the last supported version for python 3.8.x
+astroid<=3.2.4
+
+# upgrade to 5.4.0 causes issues with `importlib`
+edx-lint<=5.3.7