Azure-Samples
diff --git a/‎app/backend/approaches/chatapproach.py‎
Lines changed: 19 additions & 10 deletions b/‎app/backend/approaches/chatapproach.py‎
Lines changed: 19 additions & 10 deletions
diff --git a/‎app/backend/approaches/retrievethenread.py‎
Lines changed: 17 additions & 16 deletions b/‎app/backend/approaches/retrievethenread.py‎
Lines changed: 17 additions & 16 deletions
diff --git a/‎app/backend/approaches/retrievethenreadvision.py‎
Lines changed: 16 additions & 15 deletions b/‎app/backend/approaches/retrievethenreadvision.py‎
Lines changed: 16 additions & 15 deletions
diff --git a/‎app/backend/requirements.txt‎
Lines changed: 47 additions & 42 deletions b/‎app/backend/requirements.txt‎
Lines changed: 47 additions & 42 deletions
@@ -75,7 +75,9 @@ def get_search_query(self, chat_completion: ChatCompletion, user_query: str):
                 return query_text
         return user_query
 
-    def extract_followup_questions(self, content: str):
+    def extract_followup_questions(self, content: Optional[str]):
+        if content is None:
+            return content, []
         return content.split("<<")[0], re.findall(r"<<([^>>]+)>>", content)
 
     async def run_without_streaming(
@@ -89,15 +91,17 @@ async def run_without_streaming(
             messages, overrides, auth_claims, should_stream=False
         )
         chat_completion_response: ChatCompletion = await chat_coroutine
-        chat_resp = chat_completion_response.model_dump()  # Convert to dict to make it JSON serializable
-        chat_resp = chat_resp["choices"][0]
-        chat_resp["context"] = extra_info
+        content = chat_completion_response.choices[0].message.content
+        role = chat_completion_response.choices[0].message.role
         if overrides.get("suggest_followup_questions"):
-            content, followup_questions = self.extract_followup_questions(chat_resp["message"]["content"])
-            chat_resp["message"]["content"] = content
-            chat_resp["context"]["followup_questions"] = followup_questions
-        chat_resp["session_state"] = session_state
-        return chat_resp
+            content, followup_questions = self.extract_followup_questions(content)
+            extra_info["followup_questions"] = followup_questions
+        chat_app_response = {
+            "message": {"content": content, "role": role},
+            "context": extra_info,
+            "session_state": session_state,
+        }
+        return chat_app_response
 
     async def run_with_streaming(
         self,
@@ -117,7 +121,12 @@ async def run_with_streaming(
             # "2023-07-01-preview" API version has a bug where first response has empty choices
             event = event_chunk.model_dump()  # Convert pydantic model to dict
             if event["choices"]:
-                completion = {"delta": event["choices"][0]["delta"]}
+                completion = {
+                    "delta": {
+                        "content": event["choices"][0]["delta"].get("content"),
+                        "role": event["choices"][0]["delta"]["role"],
+                    }
+                }
                 # if event contains << and not >>, it is start of follow-up question, truncate
                 content = completion["delta"].get("content")
                 content = content or ""  # content may either not exist in delta, or explicitly be None
 
@@ -124,17 +124,15 @@ async def run(
             max_tokens=self.chatgpt_token_limit - response_token_limit,
         )
 
-        chat_completion = (
-            await self.openai_client.chat.completions.create(
-                # Azure OpenAI takes the deployment name as the model name
-                model=self.chatgpt_deployment if self.chatgpt_deployment else self.chatgpt_model,
-                messages=updated_messages,
-                temperature=overrides.get("temperature", 0.3),
-                max_tokens=response_token_limit,
-                n=1,
-                seed=seed,
-            )
-        ).model_dump()
+        chat_completion = await self.openai_client.chat.completions.create(
+            # Azure OpenAI takes the deployment name as the model name
+            model=self.chatgpt_deployment if self.chatgpt_deployment else self.chatgpt_model,
+            messages=updated_messages,
+            temperature=overrides.get("temperature", 0.3),
+            max_tokens=response_token_limit,
+            n=1,
+            seed=seed,
+        )
 
         data_points = {"text": sources_content}
         extra_info = {
@@ -168,8 +166,11 @@ async def run(
             ],
         }
 
-        completion = {}
-        completion["message"] = chat_completion["choices"][0]["message"]
-        completion["context"] = extra_info
-        completion["session_state"] = session_state
-        return completion
+        return {
+            "message": {
+                "content": chat_completion.choices[0].message.content,
+                "role": chat_completion.choices[0].message.role,
+            },
+            "context": extra_info,
+            "session_state": session_state,
+        }
@@ -142,16 +142,14 @@ async def run(
             new_user_content=user_content,
             max_tokens=self.gpt4v_token_limit - response_token_limit,
         )
-        chat_completion = (
-            await self.openai_client.chat.completions.create(
-                model=self.gpt4v_deployment if self.gpt4v_deployment else self.gpt4v_model,
-                messages=updated_messages,
-                temperature=overrides.get("temperature", 0.3),
-                max_tokens=response_token_limit,
-                n=1,
-                seed=seed,
-            )
-        ).model_dump()
+        chat_completion = await self.openai_client.chat.completions.create(
+            model=self.gpt4v_deployment if self.gpt4v_deployment else self.gpt4v_model,
+            messages=updated_messages,
+            temperature=overrides.get("temperature", 0.3),
+            max_tokens=response_token_limit,
+            n=1,
+            seed=seed,
+        )
 
         data_points = {
             "text": sources_content,
@@ -190,8 +188,11 @@ async def run(
             ],
         }
 
-        completion = {}
-        completion["message"] = chat_completion["choices"][0]["message"]
-        completion["context"] = extra_info
-        completion["session_state"] = session_state
-        return completion
+        return {
+            "message": {
+                "content": chat_completion.choices[0].message.content,
+                "role": chat_completion.choices[0].message.role,
+            },
+            "context": extra_info,
+            "session_state": session_state,
+        }
@@ -6,7 +6,9 @@
 #
 aiofiles==24.1.0
     # via quart
-aiohttp==3.9.5
+aiohappyeyeballs==2.3.5
+    # via aiohttp
+aiohttp==3.10.1
     # via
     #   -r requirements.in
     #   microsoft-kiota-authentication-azure
@@ -20,7 +22,7 @@ anyio==4.4.0
     #   openai
 asgiref==3.8.1
     # via opentelemetry-instrumentation-asgi
-attrs==23.2.0
+attrs==24.2.0
     # via aiohttp
 azure-ai-documentintelligence==1.0.0b3
     # via -r requirements.in
@@ -46,13 +48,13 @@ azure-identity==1.17.1
     # via
     #   -r requirements.in
     #   msgraph-sdk
-azure-monitor-opentelemetry==1.6.0
+azure-monitor-opentelemetry==1.6.1
     # via -r requirements.in
-azure-monitor-opentelemetry-exporter==1.0.0b27
+azure-monitor-opentelemetry-exporter==1.0.0b28
     # via azure-monitor-opentelemetry
 azure-search-documents==11.6.0b1
     # via -r requirements.in
-azure-storage-blob==12.21.0
+azure-storage-blob==12.22.0
     # via
     #   -r requirements.in
     #   azure-storage-file-datalake
@@ -70,7 +72,7 @@ certifi==2024.7.4
     #   httpx
     #   msrest
     #   requests
-cffi==1.16.0
+cffi==1.17.0
     # via cryptography
 charset-normalizer==3.3.2
     # via requests
@@ -87,7 +89,9 @@ cryptography==43.0.0
     #   msal
     #   pyjwt
 deprecated==1.2.14
-    # via opentelemetry-api
+    # via
+    #   opentelemetry-api
+    #   opentelemetry-semantic-conventions
 distro==1.9.0
     # via openai
 fixedint==0.1.6
@@ -127,7 +131,7 @@ idna==3.7
     #   httpx
     #   requests
     #   yarl
-importlib-metadata==7.1.0
+importlib-metadata==8.0.0
     # via
     #   opentelemetry-api
     #   opentelemetry-instrumentation-flask
@@ -146,6 +150,8 @@ jinja2==3.1.4
     # via
     #   flask
     #   quart
+jiter==0.5.0
+    # via openai
 markupsafe==2.1.5
     # via
     #   jinja2
@@ -167,7 +173,7 @@ microsoft-kiota-http==1.3.2
     # via
     #   msgraph-core
     #   msgraph-sdk
-microsoft-kiota-serialization-json==1.2.0
+microsoft-kiota-serialization-json==1.3.0
     # via msgraph-sdk
 microsoft-kiota-serialization-text==1.0.0
     # via msgraph-sdk
@@ -192,13 +198,13 @@ numpy==2.0.1
     # via -r requirements.in
 oauthlib==3.2.2
     # via requests-oauthlib
-openai==1.37.0
+openai==1.40.1
     # via
     #   -r requirements.in
     #   openai-messages-token-helper
-openai-messages-token-helper==0.1.5
+openai-messages-token-helper==0.1.9
     # via -r requirements.in
-opentelemetry-api==1.25.0
+opentelemetry-api==1.26.0
     # via
     #   azure-core-tracing-opentelemetry
     #   azure-monitor-opentelemetry-exporter
@@ -221,7 +227,7 @@ opentelemetry-api==1.25.0
     #   opentelemetry-instrumentation-wsgi
     #   opentelemetry-sdk
     #   opentelemetry-semantic-conventions
-opentelemetry-instrumentation==0.46b0
+opentelemetry-instrumentation==0.47b0
     # via
     #   opentelemetry-instrumentation-aiohttp-client
     #   opentelemetry-instrumentation-asgi
@@ -236,49 +242,49 @@ opentelemetry-instrumentation==0.46b0
     #   opentelemetry-instrumentation-urllib
     #   opentelemetry-instrumentation-urllib3
     #   opentelemetry-instrumentation-wsgi
-opentelemetry-instrumentation-aiohttp-client==0.46b0
+opentelemetry-instrumentation-aiohttp-client==0.47b0
     # via -r requirements.in
-opentelemetry-instrumentation-asgi==0.46b0
+opentelemetry-instrumentation-asgi==0.47b0
     # via
     #   -r requirements.in
     #   opentelemetry-instrumentation-fastapi
-opentelemetry-instrumentation-dbapi==0.46b0
+opentelemetry-instrumentation-dbapi==0.47b0
     # via opentelemetry-instrumentation-psycopg2
-opentelemetry-instrumentation-django==0.46b0
+opentelemetry-instrumentation-django==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-fastapi==0.46b0
+opentelemetry-instrumentation-fastapi==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-flask==0.46b0
+opentelemetry-instrumentation-flask==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-httpx==0.46b0
+opentelemetry-instrumentation-httpx==0.47b0
     # via -r requirements.in
-opentelemetry-instrumentation-openai==0.25.5
+opentelemetry-instrumentation-openai==0.26.5
     # via -r requirements.in
-opentelemetry-instrumentation-psycopg2==0.46b0
+opentelemetry-instrumentation-psycopg2==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-requests==0.46b0
+opentelemetry-instrumentation-requests==0.47b0
     # via
     #   -r requirements.in
     #   azure-monitor-opentelemetry
-opentelemetry-instrumentation-urllib==0.46b0
+opentelemetry-instrumentation-urllib==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-urllib3==0.46b0
+opentelemetry-instrumentation-urllib3==0.47b0
     # via azure-monitor-opentelemetry
-opentelemetry-instrumentation-wsgi==0.46b0
+opentelemetry-instrumentation-wsgi==0.47b0
     # via
     #   opentelemetry-instrumentation-django
     #   opentelemetry-instrumentation-flask
 opentelemetry-resource-detector-azure==0.1.5
     # via azure-monitor-opentelemetry
-opentelemetry-sdk==1.25.0
+opentelemetry-sdk==1.26.0
     # via
     #   azure-monitor-opentelemetry
     #   azure-monitor-opentelemetry-exporter
     #   microsoft-kiota-abstractions
     #   microsoft-kiota-authentication-azure
     #   microsoft-kiota-http
     #   opentelemetry-resource-detector-azure
-opentelemetry-semantic-conventions==0.46b0
+opentelemetry-semantic-conventions==0.47b0
     # via
     #   opentelemetry-instrumentation-aiohttp-client
     #   opentelemetry-instrumentation-asgi
@@ -293,9 +299,9 @@ opentelemetry-semantic-conventions==0.46b0
     #   opentelemetry-instrumentation-urllib3
     #   opentelemetry-instrumentation-wsgi
     #   opentelemetry-sdk
-opentelemetry-semantic-conventions-ai==0.3.4
+opentelemetry-semantic-conventions-ai==0.4.1
     # via opentelemetry-instrumentation-openai
-opentelemetry-util-http==0.46b0
+opentelemetry-util-http==0.47b0
     # via
     #   opentelemetry-instrumentation-aiohttp-client
     #   opentelemetry-instrumentation-asgi
@@ -327,13 +333,13 @@ pydantic==2.8.2
     # via openai
 pydantic-core==2.20.1
     # via pydantic
-pyjwt[crypto]==2.8.0
+pyjwt[crypto]==2.9.0
     # via
+    #   -r requirements.in
     #   msal
-    #   pyjwt
-pymupdf==1.24.7
+pymupdf==1.24.9
     # via -r requirements.in
-pymupdfb==1.24.6
+pymupdfb==1.24.9
     # via pymupdf
 pypdf==4.3.1
     # via -r requirements.in
@@ -348,7 +354,7 @@ quart==0.19.6
     #   quart-cors
 quart-cors==0.7.0
     # via -r requirements.in
-regex==2024.5.15
+regex==2024.7.24
     # via tiktoken
 requests==2.32.3
     # via
@@ -362,7 +368,6 @@ requests-oauthlib==2.0.0
 six==1.16.0
     # via
     #   azure-core
-    #   ecdsa
     #   isodate
     #   python-dateutil
 sniffio==1.3.1
@@ -372,22 +377,22 @@ sniffio==1.3.1
     #   openai
 soupsieve==2.5
     # via beautifulsoup4
-std-uritemplate==1.0.3
+std-uritemplate==1.0.5
     # via microsoft-kiota-abstractions
-tenacity==8.5.0
+tenacity==9.0.0
     # via -r requirements.in
 tiktoken==0.7.0
     # via
     #   -r requirements.in
     #   openai-messages-token-helper
     #   opentelemetry-instrumentation-openai
-time-machine==2.14.2
+time-machine==2.15.0
     # via pendulum
-tqdm==4.66.4
+tqdm==4.66.5
     # via openai
 types-beautifulsoup4==4.12.0.20240511
     # via -r requirements.in
-types-html5lib==1.1.11.20240228
+types-html5lib==1.1.11.20240806
     # via types-beautifulsoup4
 types-pillow==10.2.0.20240520
     # via -r requirements.in
@@ -406,7 +411,7 @@ tzdata==2024.1
     # via pendulum
 urllib3==2.2.2
     # via requests
-uvicorn==0.30.3
+uvicorn==0.30.5
     # via -r requirements.in
 werkzeug==3.0.3
     # via