Multilingual support

app/backend/app.py

@@ -7,9 +7,9 @@
 import urllib.parse
 from datetime import datetime, timedelta
 from fastapi.staticfiles import StaticFiles
-import openai
 from fastapi import FastAPI, HTTPException, Request
 from fastapi.responses import RedirectResponse
+import openai
 from approaches.chatreadretrieveread import ChatReadRetrieveReadApproach
 from approaches.approach import Approaches
 from azure.core.credentials import AzureKeyCredential
@@ -65,7 +65,10 @@
     "COSMOSDB_TAGS_CONTAINER_NAME": "tagscontainer",
     "QUERY_TERM_LANGUAGE": "English",
     "TARGET_EMBEDDINGS_MODEL": "BAAI/bge-small-en-v1.5",
-    "ENRICHMENT_APPSERVICE_NAME": "enrichment"
+    "ENRICHMENT_APPSERVICE_NAME": "enrichment",
+    "TARGET_TRANSLATION_LANGUAGE": "en",
+    "ENRICHMENT_ENDPOINT": None,
+    "ENRICHMENT_KEY": None
 }
 
 for key, value in ENV.items():
@@ -97,10 +100,16 @@
 
 # Setup StatusLog to allow access to CosmosDB for logging
 statusLog = StatusLog(
-    ENV["COSMOSDB_URL"], ENV["COSMOSDB_KEY"], ENV["COSMOSDB_LOG_DATABASE_NAME"], ENV["COSMOSDB_LOG_CONTAINER_NAME"]
+    ENV["COSMOSDB_URL"],
+    ENV["COSMOSDB_KEY"],
+    ENV["COSMOSDB_LOG_DATABASE_NAME"],
+    ENV["COSMOSDB_LOG_CONTAINER_NAME"]
 )
 tagsHelper = TagsHelper(
-    ENV["COSMOSDB_URL"], ENV["COSMOSDB_KEY"], ENV["COSMOSDB_TAGS_DATABASE_NAME"], ENV["COSMOSDB_TAGS_CONTAINER_NAME"]
+    ENV["COSMOSDB_URL"],
+    ENV["COSMOSDB_KEY"],
+    ENV["COSMOSDB_TAGS_DATABASE_NAME"],
+    ENV["COSMOSDB_TAGS_CONTAINER_NAME"]
 )
 
 # Comment these two lines out if using keys, set your API key in the OPENAI_API_KEY environment variable instead
@@ -123,7 +132,7 @@
 model_name = ''
 model_version = ''
 
-if (str_to_bool.get(ENV["IS_GOV_CLOUD_DEPLOYMENT"])):
+if str_to_bool.get(ENV["IS_GOV_CLOUD_DEPLOYMENT"]):
     model_name = ENV["AZURE_OPENAI_CHATGPT_MODEL_NAME"]
     model_version = ENV["AZURE_OPENAI_CHATGPT_MODEL_VERSION"]
     embedding_model_name = ENV["AZURE_OPENAI_EMBEDDINGS_MODEL_NAME"]
@@ -171,7 +180,10 @@
                                     model_version,
                                     str_to_bool.get(ENV["IS_GOV_CLOUD_DEPLOYMENT"]),
                                     ENV["TARGET_EMBEDDINGS_MODEL"],
-                                    ENV["ENRICHMENT_APPSERVICE_NAME"]
+                                    ENV["ENRICHMENT_APPSERVICE_NAME"],
+                                    ENV["TARGET_TRANSLATION_LANGUAGE"],
+                                    ENV["ENRICHMENT_ENDPOINT"],
+                                    ENV["ENRICHMENT_KEY"]
                                 )
 }
 
@@ -186,6 +198,7 @@
 
 @app.get("/", include_in_schema=False, response_class=RedirectResponse)
 async def root():
+    """Redirect to the index.html page"""
     return RedirectResponse(url="/index.html")
 
 
@@ -220,7 +233,7 @@ async def chat(request: Request):
 
     except Exception as ex:
         log.error(f"Error in chat:: {ex}")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
 
 @app.get("/getblobclienturl")
 async def get_blob_client_url():
@@ -268,7 +281,7 @@ async def get_all_upload_status(request: Request):
         results = statusLog.read_files_status_by_timeframe(timeframe, State[state])
     except Exception as ex:
         log.exception("Exception in /getalluploadstatus")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
     return results
 
 @app.post("/logstatus")
@@ -280,7 +293,8 @@ async def logstatus(request: Request):
     - request: Request object containing the HTTP request data.
 
     Returns:
-    - A dictionary with the status code 200 if successful, or an error message with status code 500 if an exception occurs.
+    - A dictionary with the status code 200 if successful, or an error
+        message with status code 500 if an exception occurs.
     """
     try:
         json_body = await request.json()
@@ -295,10 +309,10 @@ async def logstatus(request: Request):
                                   state=state,
                                   fresh_start=True)
         statusLog.save_document(document_path=path)
-        
+
     except Exception as ex:
         log.exception("Exception in /logstatus")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
     raise HTTPException(status_code=200, detail="Success")
 
 # Return AZURE_OPENAI_CHATGPT_DEPLOYMENT
@@ -365,7 +379,7 @@ async def get_citation(request: Request):
         results = json.loads(decoded_text)
     except Exception as ex:
         log.exception("Exception in /getalluploadstatus")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
     return results
 
 # Return APPLICATION_TITLE
@@ -393,7 +407,7 @@ async def get_all_tags():
         results = tagsHelper.get_all_tags()
     except Exception as ex:
         log.exception("Exception in /getalltags")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
     return results
 
 @app.post("/retryFile")
@@ -415,7 +429,7 @@ async def retryFile(request: Request):
 
     except Exception as ex:
         logging.exception("Exception in /retryFile")
-        raise HTTPException(status_code=500, detail=str(ex))
+        raise HTTPException(status_code=500, detail=str(ex)) from ex
     return {"status": 200}
 
 

app/backend/approaches/approach.py

@@ -17,6 +17,10 @@ class Approach:
     An approach is a method for answering a question from a query and a set of
     documents.
     """
+    # Chat roles
+    SYSTEM = "system"
+    USER = "user"
+    ASSISTANT = "assistant"
 
     async def run(self, history: list[dict], overrides: dict) -> any:
         """
@@ -28,14 +32,13 @@ async def run(self, history: list[dict], overrides: dict) -> any:
         """
         raise NotImplementedError
 
-     #Aparmar. Custom method to construct Chat History as opposed to single string of chat History.
     def get_messages_from_history(
         self,
         system_prompt: str,
         model_id: str,
         history: Sequence[dict[str, str]],
         user_conv: str,
-        few_shots = [],
+        few_shots = [dict[str, str]],
         max_tokens: int = 4096,
         ) -> []:
         """