fix some format issues

code/app.py

@@ -247,7 +247,7 @@ def stream_without_data(response):
 
 def conversation_without_data(request):
     azure_endpoint = f"https://{AZURE_OPENAI_RESOURCE}.openai.azure.com/"
-    if AZURE_AUTH_TYPE == 'rbac':
+    if AZURE_AUTH_TYPE == "rbac":
         openai_client = AzureOpenAI(azure_endpoint=azure_endpoint,  api_version=AZURE_OPENAI_API_VERSION, azure_ad_token_provider=AZURE_TOKEN_PROVIDER)
     else:
         openai_client = AzureOpenAI(azure_endpoint=azure_endpoint,  api_version=AZURE_OPENAI_API_VERSION, api_key=AZURE_OPENAI_KEY)
@@ -323,6 +323,7 @@ def conversation_azure_byod():
 @app.route("/api/conversation/custom", methods=["GET", "POST"])
 def conversation_custom():
     from backend.batch.utilities.helpers.OrchestratorHelper import Orchestrator, OrchestrationSettings
+
     message_orchestrator = Orchestrator()
 
     try:
@@ -337,7 +338,12 @@ def conversation_custom():
         chat_history = []
         for i, k in enumerate(user_assistant_messages):
             if i % 2 == 0:
-                chat_history.append((user_assistant_messages[i]["content"],user_assistant_messages[i+1]["content"]))
+                chat_history.append(
+                    (
+                        user_assistant_messages[i]["content"],
+                        user_assistant_messages[i + 1]["content"],
+                    )
+                        )
         from backend.batch.utilities.helpers.ConfigHelper import ConfigHelper
 
         messages = message_orchestrator.handle_message(

code/backend/batch/utilities/helpers/AzureBlobStorageHelper.py

@@ -1,9 +1,14 @@
 from typing import Optional
 from datetime import datetime, timedelta
-from azure.storage.blob import BlobServiceClient, generate_blob_sas, generate_container_sas, ContentSettings, UserDelegationKey
+from azure.storage.blob import (
+    BlobServiceClient,
+    generate_blob_sas,
+    generate_container_sas,
+    ContentSettings,
+    UserDelegationKey,
+)
 from .EnvHelper import EnvHelper
 from azure.identity import DefaultAzureCredential
-import os
 
 
 class AzureBlobStorageClient:
@@ -16,7 +21,7 @@ def __init__(
         env_helper: EnvHelper = EnvHelper()
 
         self.auth_type = env_helper.AZURE_AUTH_TYPE
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             self.account_name = account_name if account_name else env_helper.AZURE_BLOB_ACCOUNT_NAME
             self.container_name : str = container_name if container_name else env_helper.AZURE_BLOB_CONTAINER_NAME
             credential = DefaultAzureCredential()
@@ -49,9 +54,25 @@ def upload_file(self, bytes_data, file_name, content_type='application/pdf'):
             container=self.container_name, blob=file_name
         )
         # Upload the created file
-        blob_client.upload_blob(bytes_data, overwrite=True, content_settings=ContentSettings(content_type=content_type))
+        blob_client.upload_blob(
+            bytes_data,
+            overwrite=True,
+            content_settings=ContentSettings(content_type=content_type),
+        )
         # Generate a SAS URL to the blob and return it, if auth_type is rbac, account_key is None, if not, user_delegation_key is None.
-        return blob_client.url + '?' + generate_blob_sas(self.account_name, self.container_name, file_name, user_delegation_key=self.user_delegation_key, account_key=self.account_key, permission="r", expiry=datetime.utcnow() + timedelta(hours=3))
+        return (
+            blob_client.url
+            + "?"
+            + generate_blob_sas(
+                self.account_name,
+                self.container_name,
+                file_name,
+                user_delegation_key=self.user_delegation_key,
+                account_key=self.account_key,
+                permission="r",
+                expiry=datetime.utcnow() + timedelta(hours=3),
+            )
+        )
 
     def download_file(self, file_name):
         blob_client = self.blob_service_client.get_blob_client(
@@ -81,7 +102,14 @@ def get_all_files(self):
         )
         blob_list = container_client.list_blobs(include="metadata")
         # sas = generate_blob_sas(account_name, container_name, blob.name,account_key=account_key,  permission="r", expiry=datetime.utcnow() + timedelta(hours=3))
-        sas = generate_container_sas(self.account_name, self.container_name, user_delegation_key=self.user_delegation_key, account_key=self.account_key, permission="r", expiry=datetime.utcnow() + timedelta(hours=3))
+        sas = generate_container_sas(
+            self.account_name,
+            self.container_name,
+            user_delegation_key=self.user_delegation_key,
+            account_key=self.account_key,
+            permission="r",
+            expiry=datetime.utcnow() + timedelta(hours=3),
+        )
         files = []
         converted_files = {}
         for blob in blob_list:
@@ -121,10 +149,12 @@ def get_all_files(self):
         return files
 
     def upsert_blob_metadata(self, file_name, metadata):
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             blob_client = self.blob_service_client.get_blob_client(container=self.container_name, blob=file_name)
         else:
-            blob_client = BlobServiceClient.from_connection_string(self.connect_str).get_blob_client(container=self.container_name, blob=file_name)
+            blob_client = BlobServiceClient.from_connection_string(
+                self.connect_str
+            ).get_blob_client(container=self.container_name, blob=file_name)
         # Read metadata from the blob
         blob_metadata = blob_client.get_blob_properties().metadata
         # Update metadata
@@ -134,7 +164,27 @@ def upsert_blob_metadata(self, file_name, metadata):
 
     def get_container_sas(self):
         # Generate a SAS URL to the container and return it
-        return "?" + generate_container_sas(self.account_name, self.container_name, user_delegation_key=self.user_delegation_key, account_key=self.account_key, permission="r", expiry=datetime.utcnow() + timedelta(hours=3))
+        return "?" + generate_container_sas(
+            account_name=self.account_name,
+            container_name=self.container_name,
+            user_delegation_key=self.user_delegation_key,
+            account_key=self.account_key,
+            permission="r",
+            expiry=datetime.utcnow() + timedelta(hours=1),
+        )
+
     def get_blob_sas(self, file_name):
         # Generate a SAS URL to the blob and return it
-        return f"https://{self.account_name}.blob.core.windows.net/{self.container_name}/{file_name}" + "?" + generate_blob_sas(self.account_name, self.container_name, file_name, user_delegation_key=self.user_delegation_key, account_key=self.account_key, permission="r", expiry=datetime.utcnow() + timedelta(hours=3))
+        return (
+            f"https://{self.account_name}.blob.core.windows.net/{self.container_name}/{file_name}"
+            + "?"
+            + generate_blob_sas(
+                account_name=self.account_name,
+                container_name=self.container_name,
+                blob_name=file_name,
+                user_delegation_key=self.user_delegation_key,
+                account_key=self.account_key,
+                permission="r",
+                expiry=datetime.utcnow() + timedelta(hours=1),
+            )
+        )

code/backend/batch/utilities/helpers/AzureFormRecognizerHelper.py

@@ -4,23 +4,32 @@
 import html
 import traceback
 from .EnvHelper import EnvHelper
-import os
 
 
 class AzureFormRecognizerClient:
     def __init__(self) -> None:
-        env_helper : EnvHelper = EnvHelper()
+        env_helper: EnvHelper = EnvHelper()
 
-        self.AZURE_FORM_RECOGNIZER_ENDPOINT : str = env_helper.AZURE_FORM_RECOGNIZER_ENDPOINT
-        if env_helper.AZURE_AUTH_TYPE == 'rbac':
+        self.AZURE_FORM_RECOGNIZER_ENDPOINT: str = (
+            env_helper.AZURE_FORM_RECOGNIZER_ENDPOINT
+        )
+        if env_helper.AZURE_AUTH_TYPE == "rbac":
             self.document_analysis_client = DocumentAnalysisClient(
-                endpoint=self.AZURE_FORM_RECOGNIZER_ENDPOINT, credential=DefaultAzureCredential(), headers={"x-ms-useragent": "chat-with-your-data-solution-accelerator/1.0.0"}
+                endpoint=self.AZURE_FORM_RECOGNIZER_ENDPOINT, 
+                credential=DefaultAzureCredential(), 
+                headers={
+                    "x-ms-useragent": "chat-with-your-data-solution-accelerator/1.0.0"
+                },
             )
         else:
             self.AZURE_FORM_RECOGNIZER_KEY : str = env_helper.AZURE_FORM_RECOGNIZER_KEY
 
             self.document_analysis_client = DocumentAnalysisClient(
-                endpoint=self.AZURE_FORM_RECOGNIZER_ENDPOINT, credential=AzureKeyCredential(self.AZURE_FORM_RECOGNIZER_KEY), headers={"x-ms-useragent": "chat-with-your-data-solution-accelerator/1.0.0"}
+                endpoint=self.AZURE_FORM_RECOGNIZER_ENDPOINT,
+                credential=AzureKeyCredential(self.AZURE_FORM_RECOGNIZER_KEY),
+                headers={
+                    "x-ms-useragent": "chat-with-your-data-solution-accelerator/1.0.0"
+                },
             )
 
     form_recognizer_role_to_html = {

code/backend/batch/utilities/helpers/EnvHelper.py

@@ -29,7 +29,7 @@ def __init__(self, **kwargs) -> None:
         self.AZURE_SEARCH_FIELDS_TAG = os.getenv('AZURE_SEARCH_FIELDS_TAG', 'tag')
         self.AZURE_SEARCH_FIELDS_METADATA = os.getenv('AZURE_SEARCH_FIELDS_METADATA', 'metadata')
         self.AZURE_SEARCH_CONVERSATIONS_LOG_INDEX = os.getenv('AZURE_SEARCH_CONVERSATIONS_LOG_INDEX', 'conversations')
-        self.AZURE_AUTH_TYPE = os.environ.get("AZURE_AUTH_TYPE", "keys")
+        self.AZURE_AUTH_TYPE = os.getenv('AZURE_AUTH_TYPE', 'keys')
         # Azure OpenAI
         self.AZURE_OPENAI_RESOURCE = os.getenv('AZURE_OPENAI_RESOURCE', '')
         self.AZURE_OPENAI_MODEL = os.getenv('AZURE_OPENAI_MODEL', '')
@@ -42,64 +42,64 @@ def __init__(self, **kwargs) -> None:
         self.AZURE_OPENAI_API_VERSION = os.getenv('AZURE_OPENAI_API_VERSION', '')
         self.AZURE_OPENAI_STREAM = os.getenv('AZURE_OPENAI_STREAM', '')
         self.AZURE_OPENAI_EMBEDDING_MODEL = os.getenv('AZURE_OPENAI_EMBEDDING_MODEL', '')
-        self.AZURE_TOKEN_PROVIDER = get_bearer_token_provider(DefaultAzureCredential(), "https://cognitiveservices.azure.com/.default")
+        self.AZURE_TOKEN_PROVIDER = get_bearer_token_provider(DefaultAzureCredential(), 'https://cognitiveservices.azure.com/.default')
         # Initialize Azure keys based on authentication type and environment settings. 
         # When AZURE_AUTH_TYPE is "rbac", azure keys are None or an empty string.
         # When USE_KEY_VAULT environment variable is set, keys are securely fetched from Azure Key Vault using DefaultAzureCredential.
         # Otherwise, keys are obtained from environment variables.
-        if self.AZURE_AUTH_TYPE == "rbac":
+        if self.AZURE_AUTH_TYPE == 'rbac':
             self.AZURE_SEARCH_KEY = None
-            self.AZURE_OPENAI_KEY = ""
+            self.AZURE_OPENAI_KEY = ''
             self.AZURE_SPEECH_KEY = None
-        elif os.environ.get("USE_KEY_VAULT"):
+        elif os.environ.get('USE_KEY_VAULT'):
             credential = DefaultAzureCredential()
-            secret_client = SecretClient(os.environ.get("AZURE_KEY_VAULT_ENDPOINT"), credential)
-            self.AZURE_SEARCH_KEY = secret_client.get_secret(os.environ.get("AZURE_SEARCH_KEY")).value
-            self.AZURE_OPENAI_KEY = secret_client.get_secret(os.environ.get("AZURE_OPENAI_KEY")).value
-            self.AZURE_SPEECH_KEY = secret_client.get_secret(os.environ.get("AZURE_SPEECH_SERVICE_KEY")).value
+            secret_client = SecretClient(os.environ.get('AZURE_KEY_VAULT_ENDPOINT'), credential)
+            self.AZURE_SEARCH_KEY = secret_client.get_secret(os.environ.get('AZURE_SEARCH_KEY')).value
+            self.AZURE_OPENAI_KEY = secret_client.get_secret(os.environ.get('AZURE_OPENAI_KEY')).value
+            self.AZURE_SPEECH_KEY = secret_client.get_secret(os.environ.get('AZURE_SPEECH_SERVICE_KEY')).value
         else:
-            self.AZURE_SEARCH_KEY = os.environ.get("AZURE_SEARCH_KEY")
-            self.AZURE_OPENAI_KEY = os.environ.get("AZURE_OPENAI_KEY")
-            self.AZURE_SPEECH_KEY = os.environ.get("AZURE_SPEECH_SERVICE_KEY")
+            self.AZURE_SEARCH_KEY = os.environ.get('AZURE_SEARCH_KEY')
+            self.AZURE_OPENAI_KEY = os.environ.get('AZURE_OPENAI_KEY')
+            self.AZURE_SPEECH_KEY = os.environ.get('AZURE_SPEECH_SERVICE_KEY')
          # Set env for OpenAI SDK
         self.OPENAI_API_BASE = f"https://{os.getenv('AZURE_OPENAI_RESOURCE')}.openai.azure.com/"
-        self.OPENAI_API_TYPE = "azure" if self.AZURE_AUTH_TYPE == "keys" else "azure_ad"
+        self.OPENAI_API_TYPE = 'azure' if self.AZURE_AUTH_TYPE == 'keys' else 'azure_ad'
         self.OPENAI_API_KEY = self.AZURE_OPENAI_KEY
         self.OPENAI_API_VERSION = self.AZURE_OPENAI_API_VERSION
-        os.environ["OPENAI_API_TYPE"] = self.OPENAI_API_TYPE
-        os.environ["OPENAI_API_BASE"] = f"https://{os.getenv('AZURE_OPENAI_RESOURCE')}.openai.azure.com/"
-        os.environ["OPENAI_API_KEY"] = self.OPENAI_API_KEY
-        os.environ["OPENAI_API_VERSION"] = self.OPENAI_API_VERSION
+        os.environ['OPENAI_API_TYPE'] = self.OPENAI_API_TYPE
+        os.environ['OPENAI_API_BASE'] = f"https://{os.getenv('AZURE_OPENAI_RESOURCE')}.openai.azure.com/"
+        os.environ['OPENAI_API_KEY'] = self.OPENAI_API_KEY
+        os.environ['OPENAI_API_VERSION'] = self.OPENAI_API_VERSION
         # Azure Functions - Batch processing
-        self.BACKEND_URL = os.getenv("BACKEND_URL", "")
-        self.AzureWebJobsStorage = os.getenv("AzureWebJobsStorage", "")
+        self.BACKEND_URL = os.getenv('BACKEND_URL', '')
+        self.AzureWebJobsStorage = os.getenv('AzureWebJobsStorage', '')
         self.DOCUMENT_PROCESSING_QUEUE_NAME = os.getenv(
-            "DOCUMENT_PROCESSING_QUEUE_NAME", ""
+            'DOCUMENT_PROCESSING_QUEUE_NAME', ''
         )
         # Azure Blob Storage
         self.AZURE_BLOB_ACCOUNT_NAME = os.getenv('AZURE_BLOB_ACCOUNT_NAME', '')
-        self.AZURE_BLOB_ACCOUNT_KEY = self.secret_client.get_secret(os.getenv("AZURE_BLOB_ACCOUNT_KEY", '')).value if os.getenv("USE_KEY_VAULT", '') else os.getenv("AZURE_BLOB_ACCOUNT_KEY", '')
+        self.AZURE_BLOB_ACCOUNT_KEY = self.secret_client.get_secret(os.getenv('AZURE_BLOB_ACCOUNT_KEY', '')).value if os.getenv('USE_KEY_VAULT', '') else os.getenv('AZURE_BLOB_ACCOUNT_KEY', '')
         self.AZURE_BLOB_CONTAINER_NAME = os.getenv('AZURE_BLOB_CONTAINER_NAME', '')
         # Azure Form Recognizer
         self.AZURE_FORM_RECOGNIZER_ENDPOINT = os.getenv('AZURE_FORM_RECOGNIZER_ENDPOINT', '')
-        self.AZURE_FORM_RECOGNIZER_KEY = self.secret_client.get_secret(os.getenv("AZURE_FORM_RECOGNIZER_KEY", '')).value if os.getenv("USE_KEY_VAULT", '') else os.getenv('AZURE_FORM_RECOGNIZER_KEY', '')
+        self.AZURE_FORM_RECOGNIZER_KEY = self.secret_client.get_secret(os.getenv('AZURE_FORM_RECOGNIZER_KEY', '')).value if os.getenv('USE_KEY_VAULT', '') else os.getenv('AZURE_FORM_RECOGNIZER_KEY', '')
         # Azure App Insights
         self.APPINSIGHTS_CONNECTION_STRING = os.getenv(
-            "APPINSIGHTS_CONNECTION_STRING", ""
+            'APPINSIGHTS_CONNECTION_STRING', ''
         )
         # Azure AI Content Safety
         self.AZURE_CONTENT_SAFETY_ENDPOINT = os.getenv(
-            "AZURE_CONTENT_SAFETY_ENDPOINT", ""
+            'AZURE_CONTENT_SAFETY_ENDPOINT', ''
         )
         if (
             "https" not in self.AZURE_CONTENT_SAFETY_ENDPOINT
             and "api.cognitive.microsoft.com" not in self.AZURE_CONTENT_SAFETY_ENDPOINT
         ):
             self.AZURE_CONTENT_SAFETY_ENDPOINT = self.AZURE_FORM_RECOGNIZER_ENDPOINT
-        self.AZURE_CONTENT_SAFETY_KEY = self.secret_client.get_secret(os.getenv("AZURE_CONTENT_SAFETY_KEY", '')).value if os.getenv("USE_KEY_VAULT", '') else os.getenv('AZURE_CONTENT_SAFETY_KEY', '')
+        self.AZURE_CONTENT_SAFETY_KEY = self.secret_client.get_secret(os.getenv('AZURE_CONTENT_SAFETY_KEY', '')).value if os.getenv('USE_KEY_VAULT', '') else os.getenv('AZURE_CONTENT_SAFETY_KEY', '')
         # Orchestration Settings
         self.ORCHESTRATION_STRATEGY = os.getenv(
-            "ORCHESTRATION_STRATEGY", "openai_function"
+            'ORCHESTRATION_STRATEGY', 'openai_function'
         )
 
     @staticmethod

code/backend/batch/utilities/helpers/LLMHelper.py

@@ -11,7 +11,7 @@ def __init__(self):
         self.auth_type = env_helper.AZURE_AUTH_TYPE
         self.token_provider = env_helper.AZURE_TOKEN_PROVIDER
 
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             self.openai_client = AzureOpenAI(azure_endpoint=env_helper.OPENAI_API_BASE, api_version=env_helper.AZURE_OPENAI_API_VERSION, azure_ad_token_provider=self.token_provider)
         else:
             self.openai_client = AzureOpenAI(azure_endpoint=env_helper.OPENAI_API_BASE, api_version=env_helper.AZURE_OPENAI_API_VERSION, api_key=env_helper.OPENAI_API_KEY)
@@ -21,22 +21,22 @@ def __init__(self):
         self.embedding_model = env_helper.AZURE_OPENAI_EMBEDDING_MODEL
 
     def get_llm(self):
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             return AzureChatOpenAI(deployment_name=self.llm_model, temperature=0, max_tokens=self.llm_max_tokens, openai_api_version=self.openai_client._api_version, azure_ad_token_provider=self.token_provider)
         else:
             return AzureChatOpenAI(deployment_name=self.llm_model, temperature=0, max_tokens=self.llm_max_tokens, openai_api_version=self.openai_client._api_version)
 
     # TODO: This needs to have a custom callback to stream back to the UI
     def get_streaming_llm(self):
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             return AzureChatOpenAI(streaming=True, callbacks=[StreamingStdOutCallbackHandler], deployment_name=self.llm_model, temperature=0, 
                                max_tokens=self.llm_max_tokens, openai_api_version=self.openai_client._api_version, azure_ad_token_provider=self.token_provider)
         else:
             return AzureChatOpenAI(streaming=True, callbacks=[StreamingStdOutCallbackHandler], deployment_name=self.llm_model, temperature=0, 
                                max_tokens=self.llm_max_tokens, openai_api_version=self.openai_client._api_version)
 
     def get_embedding_model(self):
-        if self.auth_type == 'rbac':
+        if self.auth_type == "rbac":
             return AzureOpenAIEmbeddings(azure_deployment=self.embedding_model, chunk_size=1, azure_ad_token_provider=self.token_provider)
         else:
             return AzureOpenAIEmbeddings(azure_deployment=self.embedding_model, chunk_size=1)