feat: update request to xpert backend (#74)

CHANGELOG.rst

@@ -14,6 +14,10 @@ Change Log
 Unreleased
 **********
 
+4.2.0 - 2024-02-28
+******************
+* Modify call to Xpert backend to prevent use of course index.
+
 4.1.0 - 2024-02-26
 ******************
 * Use course cache to inject course title and course skill names into prompt template.

learning_assistant/__init__.py

@@ -2,6 +2,6 @@
 Plugin for a learning assistant backend, intended for use within edx-platform.
 """
 
-__version__ = '4.1.0'
+__version__ = '4.2.0'
 
 default_app_config = 'learning_assistant.apps.LearningAssistantConfig'  # pylint: disable=invalid-name

learning_assistant/utils.py

@@ -27,15 +27,7 @@ def get_reduced_message_list(prompt_template, message_list):
     """
     If messages are larger than allotted token amount, return a smaller list of messages.
     """
-    # the total number of system tokens is a sum of estimated tokens that includes the prompt template, the
-    # course title, and the course skills. It is necessary to include estimations for the course title and
-    # course skills, as the chat endpoint the prompt is being passed to is responsible for filling in the values
-    # for both of those variables.
-    total_system_tokens = (
-        _estimated_message_tokens(prompt_template)
-        + _estimated_message_tokens('.' * 40)  # average number of characters per course name is 40
-        + _estimated_message_tokens('.' * 116)  # average number of characters for skill names is 116
-    )
+    total_system_tokens = _estimated_message_tokens(prompt_template)
 
     max_tokens = getattr(settings, 'CHAT_COMPLETION_MAX_TOKENS', 16385)
     response_tokens = getattr(settings, 'CHAT_COMPLETION_RESPONSE_TOKENS', 1000)
@@ -55,28 +47,23 @@ def get_reduced_message_list(prompt_template, message_list):
         # insert message at beginning of list, because we are traversing the message list from most recent to oldest
         new_message_list.insert(0, new_message)
 
-    return new_message_list
+    system_message = {'role': 'system', 'content': prompt_template}
 
+    return [system_message] + new_message_list
 
-def create_request_body(prompt_template, message_list, course_id):
+
+def create_request_body(prompt_template, message_list):
     """
     Form request body to be passed to the chat endpoint.
     """
     response_body = {
-        'context': {
-            'content': prompt_template,
-            'render': {
-                'doc_id': course_id,
-                'fields': ['skillNames', 'title']
-            }
-        },
         'message_list': get_reduced_message_list(prompt_template, message_list)
     }
 
     return response_body
 
 
-def get_chat_response(prompt_template, message_list, course_id):
+def get_chat_response(prompt_template, message_list):
     """
     Pass message list to chat endpoint, as defined by the CHAT_COMPLETION_API setting.
     """
@@ -87,7 +74,7 @@ def get_chat_response(prompt_template, message_list, course_id):
         connect_timeout = getattr(settings, 'CHAT_COMPLETION_API_CONNECT_TIMEOUT', 1)
         read_timeout = getattr(settings, 'CHAT_COMPLETION_API_READ_TIMEOUT', 15)
 
-        body = create_request_body(prompt_template, message_list, course_id)
+        body = create_request_body(prompt_template, message_list)
 
         try:
             response = requests.post(

learning_assistant/views.py

@@ -97,7 +97,7 @@ def post(self, request, course_run_id):
 
         prompt_template = render_prompt_template(request, request.user.id, course_run_id, unit_id, course_id)
 
-        status_code, message = get_chat_response(prompt_template, message_list, course_id)
+        status_code, message = get_chat_response(prompt_template, message_list)
 
         return Response(status=status_code, data=message)
 

tests/test_utils.py

@@ -27,7 +27,7 @@ def setUp(self):
         self.course_id = 'edx+test'
 
     def get_response(self):
-        return get_chat_response(self.prompt_template, self.message_list, self.course_id)
+        return get_chat_response(self.prompt_template, self.message_list)
 
     @override_settings(CHAT_COMPLETION_API=None)
     def test_no_endpoint_setting(self):
@@ -89,14 +89,7 @@ def test_post_request_structure(self, mock_requests):
         headers = {'Content-Type': 'application/json', 'x-api-key': settings.CHAT_COMPLETION_API_KEY}
 
         response_body = {
-            'context': {
-                'content': self.prompt_template,
-                'render': {
-                    'doc_id': self.course_id,
-                    'fields': ['skillNames', 'title']
-                }
-            },
-            'message_list': self.message_list
+            'message_list': [{'role': 'system', 'content': self.prompt_template}] + self.message_list
         }
 
         self.get_response()
@@ -120,21 +113,27 @@ def setUp(self):
             {'role': 'user', 'content': 'Goodbye'},
         ]
 
-    @override_settings(CHAT_COMPLETION_MAX_TOKENS=90)
+    @override_settings(CHAT_COMPLETION_MAX_TOKENS=30)
     @override_settings(CHAT_COMPLETION_RESPONSE_TOKENS=1)
     def test_message_list_reduced(self):
         """
         If the number of tokens in the message list is greater than allowed, assert that messages are removed
         """
         # pass in copy of list, as it is modified as part of the reduction
         reduced_message_list = get_reduced_message_list(self.prompt_template, self.message_list)
-        self.assertEqual(len(reduced_message_list), 1)
-        self.assertEqual(reduced_message_list, self.message_list[-1:])
+        self.assertEqual(len(reduced_message_list), 2)
+        self.assertEqual(
+            reduced_message_list,
+            [{'role': 'system', 'content': self.prompt_template}] + self.message_list[-1:]
+        )
 
     def test_message_list(self):
         reduced_message_list = get_reduced_message_list(self.prompt_template, self.message_list)
-        self.assertEqual(len(reduced_message_list), 2)
-        self.assertEqual(reduced_message_list, self.message_list)
+        self.assertEqual(len(reduced_message_list), 3)
+        self.assertEqual(
+            reduced_message_list,
+            [{'role': 'system', 'content': self.prompt_template}] + self.message_list
+        )
 
 
 @ddt.ddt