Upload app.py with huggingface_hub

app.py

@@ -7,6 +7,7 @@ import pandas as pd
 from datetime import datetime
 import sqlite3
 import weave
+import wandb
 
 from langchain_community.embeddings.sentence_transformer import SentenceTransformerEmbeddings
 from langchain_community.vectorstores import Chroma
@@ -16,26 +17,32 @@ from langchain_community.utilities.sql_database import SQLDatabase
 from langchain_community.agent_toolkits import create_sql_agent
 from langchain.agents import create_tool_calling_agent, AgentExecutor
 from langchain_core.tools import tool
-from openai import AzureOpenAI # NEW
-from langchain_core.caches import BaseCache # NEW
-from langchain_core.callbacks import Callbacks #NEW
-from langchain.chat_models import AzureChatOpenAI # NEW
-AzureChatOpenAI.model_rebuild(_types_namespace={'BaseCache': BaseCache, 'Callbacks': Callbacks})
+
+
+from langchain_openai import AzureChatOpenAI
 
 from huggingface_hub import CommitScheduler
 from pathlib import Path
 
-
+from langchain_openai import AzureChatOpenAI
+from typing import Union
+class NonStreamingAzureChatOpenAI(AzureChatOpenAI):
+    def stream(self, input, config=None, **kwargs):
+        # Emulate streaming by calling invoke and yielding once
+        result = self.invoke(input, config=config, **kwargs)
+        yield result
 #====================================SETUP=====================================#
 # Fetch secrets from Hugging Face Spaces
+wandb.login(key=st.secrets["WANDB_API_KEY"])
 
+model_name = st.secrets["CHATGPT_MODEL"]
 
 # Extract the OpenAI key and endpoint from the configuration
-# Azure not open ai
-api_key = os.getenv("AZURE_OPENAI_KEY")
-endpoint = os.getenv("AZURE_OPENAI_ENDPOINT")
-api_version = os.getenv("AZURE_OPENAI_APIVERSION") #NEW
-model_name = os.getenv("CHATGPT_MODEL")
+
+api_key = st.secrets["AZURE_OPENAI_KEY"]
+azure_endpoint = st.secrets["AZURE_OPENAI_ENDPOINT"]
+api_version = st.secrets["AZURE_OPENAI_APIVERSION"]
+
 
 # Define the location of the SQLite database
 db_loc = 'ecomm.db'
@@ -49,7 +56,7 @@ database_schema = db.get_table_info()
 
 # Let's initiate w&b weave with a project name - this will automatically save all the llm calls made using openai or gemini
 # Make sure to save your w&b api key in secrets as WANDB_API_KEY
-# weave.init('ecomm_support')  <--------------------------------------------------------- Uncomment to log to WANDB
+weave.init('ecomm_support')
 
 
 #=================================Setup Logging=====================================#
@@ -105,21 +112,17 @@ full_prompt = ChatPromptTemplate.from_messages(
         MessagesPlaceholder("agent_scratchpad")
     ]
 )
-##### NEEDS AzureChatOpenAI
+
 # Initialize the ChatOpenAI model with the extracted configuration
-#llm = ChatOpenAI(
-#    openai_api_base=endpoint,
-#    openai_api_key=api_key,
-#    model="gpt-4o-mini",
-#    streaming=False  # Explicitly disabling streaming
-#)
-llm = AzureChatOpenAI(
+llm = NonStreamingAzureChatOpenAI(
     model_name=model_name,
     api_key=api_key,
-    azure_endpoint=endpoint,
+    azure_endpoint=azure_endpoint,
     api_version=api_version,
-    temperature=0
+    temperature=0,
+    streaming=False
 )
+print("Streaming enabled?", llm.streaming)
 
 # Create the SQL agent using the ChatOpenAI model, database, and prompt template
 sqlite_agent = create_sql_agent(
@@ -191,7 +194,7 @@ Here are some documents and their source that may be relevant to the question me
 {question}
 """
 # Load the persisted DB
-persisted_vectordb_location = 'policy_docs/policy_docs'
+persisted_vectordb_location = 'policy_docs'
 #Create a Colelction Name
 collection_name = 'policy_docs'
 
@@ -209,16 +212,13 @@ retriever = vector_store.as_retriever(
     search_kwargs={'k': 5}
 )
 
-###### NEEDS AzureOpenAI
-#client = OpenAI(
-#    api_key=creds["API_KEY"],
-#    base_url=creds["OPENAI_API_BASE"]
-#)
-client = AzureOpenAI(
-    api_key=api_key,
-    azure_endpoint=endpoint,
-    api_version=api_version
-)
+
+# client = AzureOpenAI(
+#     azure_endpoint=creds["AZURE_OPENAI_ENDPOINT"],
+#     api_key=creds["AZURE_OPENAI_KEY"],
+#     api_version=creds["AZURE_OPENAI_APIVERSION"]
+# )
+
 
 @tool
 def rag(user_input: str) -> str:
@@ -245,12 +245,9 @@ def rag(user_input: str) -> str:
     }
     ]
     try:
-        response = client.chat.completions.create(
-        model="gpt-4o-mini",
-        messages=prompt
-        )
-
-        prediction = response.choices[0].message.content
+        response = llm.invoke(prompt)
+        prediction = response.content
+        # prediction = response.choices[0].message.content
     except Exception as e:
         prediction = f'Sorry, I encountered the following error: \n {e}'
 
@@ -354,16 +351,12 @@ def days_since(delivered_date: str) ->str:
 def build_prompt(df):
 
     system_message = f"""
-
         You are an intelligent e-commerce chatbot designed to assist users with pre-order and post-order queries. Your job is to
-
         Gather necessary information from the user to help them with their query.
         If at any point you cannot determine the next steps - defer to human. you do not have clearance to go beyond the scope the following flow.
         Do not provide sql inputs to the sql tool - you only need to ask in natural language what information you need.
         You are only allowed to provide information relevant to the particular customer and the customer information is provided below. you can provide information of this customer only. Following is the information about the customer from the last 2 weeks:
-
         {df}
-
         If this information is not enough to answer question, identify the customer from data above and fetch necessary information usign the sql_tool or rag tool - do not fetch information of other customers.
         use the details provided in the above file to fetch information from sql tool - like customer id, email and phone. Refrain from asking customers details unless necessary.
         If customer asks about a product, you should act as a sales representative and help them understand the product as much as possible and provide all the necessary information for them. You should also provide them the link to the product which you can get from the source of the information.
@@ -375,7 +368,6 @@ def build_prompt(df):
         1. Ask for their feedback and rating out of 5.
         2. then, Use the `register_feedback` tool to log it.  - you MUST ask customer feedback along with asking customer what else they need help with.
         3. After receving customer feedback exit the chat by responding with 'Bye'.
-
         ---
         ### **Handling Out-of-Scope Queries:**
         If the user's query, at any point is not covered by the workflows above:
@@ -500,29 +492,19 @@ def chatbot_interface():
     if 'conversation_history' not in st.session_state:
         st.session_state.conversation_history = [{"role": "assistant", "content": "welcome! I am Raha, how can I help you on this beautiful day?"}]
 
-
     details = fetch_details(st.session_state.email)
-    # st.write(details)
     prompt = build_prompt(details)
-    tools = [sql_tool,defer_to_human, rag, register_feedback, days_since]
-
-    # NEEDS AzureChatOpenAI
-    #chatbot = ChatOpenAI(
-    #openai_api_base=endpoint,
-    #openai_api_key=api_key,
-    #model="gpt-4o-mini",
-    #streaming=False, # Explicitly disabling streaming
-    #temperature=0
-    #)
-
-    chatbot = AzureChatOpenAI(
+    tools = [sql_tool, defer_to_human, rag, register_feedback, days_since]
+
+    chatbot = NonStreamingAzureChatOpenAI(
         model_name=model_name,
         api_key=api_key,
-        azure_endpoint=endpoint,
+        azure_endpoint=azure_endpoint,
         api_version=api_version,
-        temperature=0
+        temperature=0,
+        streaming=False
     )
-
+    print("Streaming enabled?", chatbot.streaming)
 
     agent = create_tool_calling_agent(chatbot, tools, prompt)
     agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
@@ -534,11 +516,8 @@ def chatbot_interface():
 
     # React to user input
     if user_input := st.chat_input("You: ", key="chat_input"):
-        # Display user message in chat message container
         st.chat_message("user").markdown(user_input)
         with st.spinner("Processing..."):
-
-            # Add user message to conversation history
             st.session_state.conversation_history.append({"role": "user", "content": user_input})
 
             conversation_input = "\n".join(
@@ -546,22 +525,19 @@ def chatbot_interface():
             )
 
             try:
-                # Pass the history to the agent
                 response = agent_executor.invoke({"input": conversation_input})
 
-                # Add the chatbot's response to the history
                 chatbot_response = response['output']
                 st.session_state.conversation_history.append({"role": "assistant", "content": chatbot_response})
-                # Check if the assistant's response contains "exit"
+
                 if "bye" in chatbot_response.lower():
-                    log_history(st.session_state.email,st.session_state.conversation_history)
+                    log_history(st.session_state.email, st.session_state.conversation_history)
 
-                # Display the chatbot's response
                 with st.chat_message("assistant"):
                     st.markdown(chatbot_response)
 
             except Exception as e:
-                st.write("Blocked by content policy \n", e )
+                st.write("Blocked by content policy \n", e)
 
 def main():
     # Check if the user is logged in