Update app.py

app.py

@@ -6,7 +6,7 @@ from dotenv import load_dotenv  # Loading environment variables from a .env file
 import json  # Parsing and handling JSON data
 
 # LangChain imports
-#from langchain_openai import ChatOpenAI #uncomment to use OPEN AI
+from langchain_openai import ChatOpenAI
 from langchain_core.documents import Document  # Document data structures
 from langchain_core.runnables import RunnablePassthrough  # LangChain core library for running pipelines
 from langchain_core.output_parsers import StrOutputParser  # String output parser
@@ -20,8 +20,6 @@ from langchain.retrievers import ContextualCompressionRetriever  # Contextual co
 from langchain_community.vectorstores import Chroma  # Implementations of vector stores like Chroma
 from langchain_community.document_loaders import PyPDFDirectoryLoader, PyPDFLoader  # Document loaders for PDFs
 from langchain_community.cross_encoders import HuggingFaceCrossEncoder  # Cross-encoders from HuggingFace
-from langchain_community.llms import HuggingFaceHub #comment to use OPEN AI
-from langchain_community.embeddings import HuggingFaceEmbeddings # comment to use OPEN AI
 from langchain_experimental.text_splitter import SemanticChunker  # Experimental text splitting methods
 from langchain.text_splitter import (
     CharacterTextSplitter,  # Splitting text by characters
@@ -32,8 +30,8 @@ from langchain.agents import create_tool_calling_agent, AgentExecutor
 from langchain_core.prompts import ChatPromptTemplate
 
 # LangChain OpenAI imports
-#from langchain_openai import AzureOpenAIEmbeddings, AzureChatOpenAI  # OpenAI embeddings and models
-#from langchain.embeddings.openai import OpenAIEmbeddings  # OpenAI embeddings for text vectors
+from langchain_openai import AzureOpenAIEmbeddings, AzureChatOpenAI  # OpenAI embeddings and models
+from langchain.embeddings.openai import OpenAIEmbeddings  # OpenAI embeddings for text vectors
 
 # LlamaParse & LlamaIndex imports
 from llama_parse import LlamaParse  # Document parsing library
@@ -64,42 +62,20 @@ MEM0_api_key = os.environ.get("MEM0_API_KEY")
 
 
 # Initialize the OpenAI Embeddings
-#embedding_model = OpenAIEmbeddings(
-#    openai_api_base=endpoint,
-#    openai_api_key=api_key,
-#    model='text-embedding-ada-002'
-#)
-
-# Initialize Hugging Face Embeddings
-embedding_model = HuggingFaceEmbeddings(
-    model_name="sentence-transformers/all-MiniLM-L6-v2"  # good balance of speed & quality
+embedding_model = OpenAIEmbeddings(
+    openai_api_base=endpoint,
+    openai_api_key=api_key,
+    model='text-embedding-ada-002'
 )
 
-# Initialize the Chat OpenAI model
-#llm = ChatOpenAI(
-#    openai_api_base=endpoint,
-#    openai_api_key=api_key,
-#    model="gpt-4o-mini",
-#    streaming=False
-#)
-
-# Initialize Hugging Face LLM
-from langchain_community.chat_models import ChatHuggingFace
-from transformers import pipeline
-
-pipe = pipeline(
-    "text-generation",
-    model="meta-llama/Meta-Llama-3-8B-Instruct",
-    token=os.environ["HF_TOKEN"],
-    max_new_tokens=512,
-    temperature=0.0
-)
 
-llm = ChatHuggingFace(
-    pipeline=pipe,
-    model_id="meta-llama/Meta-Llama-3-8B-Instruct"
+# Initialize the Chat OpenAI model
+llm = ChatOpenAI(
+    openai_api_base=endpoint,
+    openai_api_key=api_key,
+    model="gpt-4o-mini",
+    streaming=False
 )
-
 # This initializes the Chat OpenAI model with the provided endpoint, API key, deployment name, and a temperature setting of 0 (to control response variability).
 
 # set the LLM and embedding model in the LlamaIndex settings.
@@ -125,10 +101,8 @@ class AgentState(TypedDict):
 def expand_query(state: Dict[str, Any]) -> Dict[str, Any]:
     """
     Expands the user query to improve retrieval of bible and spiritual information.
-
     Args:
         state: Workflow state containing at least 'query' and 'query_feedback'.
-
     Returns:
         Workflow state with an additional 'expanded_query' key.
     """
@@ -136,20 +110,15 @@ def expand_query(state: Dict[str, Any]) -> Dict[str, Any]:
 
     print("---------Expanding Query---------")
     system_message = '''You are an assistant that reformulates vague or short user questions into detailed, domain-specific queries related to bible and spiritual questions.
-
 Examples:
 - Input: "David and Goliath?"
   Expanded: "What is the significance of the story of David and Goliath in the context of faith, courage, and divine intervention?"
-
 - Input: "What does Jesus say about love?"
   Expanded: "What teachings did Jesus offer about love in the New Testament, and how do passages like John 13:34–35 and Matthew 22:37–39 reflect those teachings?"
-
 - Input: "Genesis creation"
   Expanded: "How does the Book of Genesis describe the creation of the world, and what are the main theological interpretations of the seven days of creation?"
-
 - Input: "End times?"
   Expanded: "What does the Bible say about the end times, and how do texts like the Book of Revelation, Daniel, and Matthew 24 contribute to Christian eschatology?"
-
 - Input: "Women in the Bible"
   Expanded: "What roles do women play in the Bible, and how are figures such as Mary, Ruth, Esther, and Deborah portrayed in biblical narratives?"
 '''
@@ -183,10 +152,8 @@ retriever = vector_store.as_retriever(
 def retrieve_context(state):
     """
     Retrieves context from the vector store using the expanded or original query.
-
     Args:
         state (Dict): The current state of the workflow, containing the query and expanded query.
-
     Returns:
         Dict: The updated state with the retrieved context.
     """
@@ -218,17 +185,14 @@ def retrieve_context(state):
 def craft_response(state):
     """
     Generates a response using the retrieved context, focusing on biblical teachings and spiritual guidance.
-
     Args:
         state (Dict): The current state of the workflow, containing the query and retrieved context.
-
     Returns:
         Dict: The updated state with the generated response.
     """
     print("---------craft_response---------")
     system_message = '''You are a helpful AI assistant trained to support users in understanding biblical teachings and spiritual guidance, using context retrieved from the Bible and the book *The Purpose Driven Life* by Rick Warren.
 Your responses must strictly adhere to the retrieved context, which is extracted from biblical texts such as the CSB Bible, theological commentaries, or trusted religious sources.
-
 Do not speculate, interpret creatively, or introduce knowledge not found in the provided context. Focus only on scriptural passages, interpretations, historical backgrounds, or theological themes directly supported by the retrieved content.
 If the context does not contain enough information to answer accurately, clearly state that. Aim for clarity, scriptural accuracy, and relevance to the user's query.
     '''
@@ -254,10 +218,8 @@ If the context does not contain enough information to answer accurately, clearly
 def score_groundedness(state):
     """
     Checks whether the response is grounded in the retrieved context.
-
     Args:
         state (Dict): The current state of the workflow, containing the response and context.
-
     Returns:
         Dict: The updated state with the groundedness score.
     """
@@ -266,12 +228,9 @@ def score_groundedness(state):
     system_message = '''You are evaluating whether an AI-generated response is grounded in the retrieved context
 provided from biblical texts (such as the CSB Bible) and the book *The Purpose Driven Life* by Rick Warren.
 The context includes scripture, commentary, and theological content.
-
 Your task is to assign a groundedness score between 0 and 1, where:
-
 - 1.0 means the response is fully supported by the context,
 - 0.0 means the response is entirely unsupported.
-
 Be strict: if even a part of the response is not traceable to the context, reduce the score. Provide only
 the numeric score.'''
 
@@ -297,22 +256,17 @@ the numeric score.'''
 def check_precision(state: Dict) -> Dict:
     """
     Checks whether the response precisely addresses the user’s query.
-
     Args:
         state (Dict): The current state of the workflow, containing the query and response.
-
     Returns:
         Dict: The updated state with the precision score.
     """
     print("---------check_precision---------")
     system_message = '''You are assessing whether an AI-generated response precisely answers the user's query,
 within the domain of biblical interpretation and spiritual guidance drawn from the Bible and *The Purpose Driven Life*.
-
 Provide a precision score between 0 and 1:
-
 - 1.0: The response fully and directly answers the query with clear relevance.
 - 0.0: The response is vague, unrelated, or fails to address the query.
-
 Only return a numeric score.'''
 
     precision_prompt = ChatPromptTemplate.from_messages([
@@ -337,10 +291,8 @@ Only return a numeric score.'''
 def refine_response(state: Dict) -> Dict:
     """
     Suggests improvements for the generated response.
-
     Args:
         state (Dict): The current state of the workflow, containing the query and response.
-
     Returns:
         Dict: The updated state with response refinement suggestions.
     """
@@ -371,10 +323,8 @@ Focus on biblical coherence, faith-based reasoning, and alignment with the theme
 def refine_query(state: Dict) -> Dict:
     """
     Suggests improvements for the expanded query.
-
     Args:
         state (Dict): The current state of the workflow, containing the query and expanded query.
-
     Returns:
         Dict: The updated state with query refinement suggestions.
     """
@@ -539,11 +489,9 @@ llama_guard_client = Groq(api_key=llama_api_key)
 def filter_input_with_llama_guard(user_input, model="meta-llama/llama-guard-4-12b"):
     """
     Filters user input using Llama Guard to ensure it is safe.
-
     Parameters:
     - user_input: The input provided by the user.
     - model: The Llama Guard model to be used for filtering (default is "meta-llama/llama-guard-4-12bb").
-
     Returns:
     - The filtered and safe input.
     """
@@ -571,38 +519,19 @@ class SpiritualBot:
         self.memory = MemoryClient(api_key=os.environ.get("MEM0_API_KEY")) # Complete the code to define the memory client API key
 
         # Initialize the OpenAI client using the provided credentials
-        #self.client = ChatOpenAI(
-        #    model_name="gpt-4o-mini",  # Specify the model to use (e.g., GPT-4 optimized version)
-        #    api_key=os.environ.get("API_KEY"),  # API key for authentication
-        #    openai_api_base = os.environ.get("OPENAI_API_BASE"),
-        #    temperature=0  # Controls randomness in responses; 0 ensures deterministic results
-        #)
-
-        # Initialize Hugging Face client
-        from langchain_community.chat_models import ChatHuggingFace
-        from transformers import pipeline
-
-        pipe = pipeline(
-        "text-generation",
-        model="meta-llama/Meta-Llama-3-8B-Instruct",
-        token=os.environ["HF_TOKEN"],
-        max_new_tokens=512,
-        temperature=0.0
+        self.client = ChatOpenAI(
+            model_name="gpt-4o-mini",  # Specify the model to use (e.g., GPT-4 optimized version)
+            api_key=os.environ.get("API_KEY"),  # API key for authentication
+            openai_api_base = os.environ.get("OPENAI_API_BASE"),
+            temperature=0  # Controls randomness in responses; 0 ensures deterministic results
         )
 
-        # Assign to self.client
-        self.client = ChatHuggingFace(
-        pipeline=pipe,
-        model_id="meta-llama/Meta-Llama-3-8B-Instruct"
-        )
-        
         # Define tools available to the chatbot, such as web search
         tools = [agentic_rag]
 
         # Define the system prompt to set the behavior of the chatbot
         system_prompt = """You are a compassionate and knowledgeable Spiritual Assistant.
     Your purpose is to help users explore biblical teachings and spiritual insights, drawing only from the Bible and *The Purpose Driven Life* by Rick Warren.
-
 Guidelines for Interaction:
 - Maintain a respectful, thoughtful, and non-judgmental tone at all times.
 - Ground every response in scripture or the provided spiritual context — never speculate or invent theology.
@@ -611,7 +540,6 @@ Guidelines for Interaction:
 - When possible, help the user reflect on how biblical principles can apply to personal growth, purpose, and everyday life.
 - Avoid doctrinal debates or denominational bias — focus on shared themes of purpose, love, faith, and spiritual growth.
 - If you cannot answer based on the given sources, humbly acknowledge the limitation and suggest scripture or topics the user might explore further.
-
 Your goal is to walk alongside users on their spiritual journey, offering encouragement, insight, and biblical grounding.
 """
 
@@ -632,7 +560,6 @@ Your goal is to walk alongside users on their spiritual journey, offering encour
     def store_customer_interaction(self, user_id: str, message: str, response: str, metadata: Dict = None):
         """
         Store customer interaction in memory for future reference.
-
         Args:
             user_id (str): Unique identifier for the customer.
             message (str): Customer's query or message.
@@ -663,11 +590,9 @@ Your goal is to walk alongside users on their spiritual journey, offering encour
     def get_relevant_history(self, user_id: str, query: str) -> List[Dict]:
         """
         Retrieve past interactions relevant to the current query.
-
         Args:
             user_id (str): Unique identifier for the customer.
             query (str): The customer's current query.
-
         Returns:
             List[Dict]: A list of relevant past interactions.
         """
@@ -681,11 +606,9 @@ Your goal is to walk alongside users on their spiritual journey, offering encour
     def handle_customer_query(self, user_id: str, query: str) -> str:
         """
         Process a customer's query and provide a response, taking into account past interactions.
-
         Args:
             user_id (str): Unique identifier for the customer.
             query (str): Customer's query.
-
         Returns:
             str: Chatbot's response.
         """
@@ -707,9 +630,7 @@ Your goal is to walk alongside users on their spiritual journey, offering encour
         prompt = f"""
         Context:
         {context}
-
         Current customer query: {query}
-
         Provide a helpful response that takes into account any relevant past interactions.
         """
 
@@ -758,12 +679,9 @@ No user data is **stored**, **shared**, or used for **model training** or any ot
                 st.session_state.chat_history.append({
                     "role": "assistant",
                     "content": f"""👋 **Welcome ようこそ, {user_id}!**
-
 How can I guide you in your spiritual path today?  
 今日、あなたの精神的な道をどのように導くことができるでしょうか?
-
 ---
-
 🔒 **Privacy Notice:**  
 User questions remains private.  
 All processing occurs **within the current session**.  
@@ -820,4 +738,4 @@ No user data is **stored**, **shared**, or used for **model training** or any ot
                 st.session_state.chat_history.append({"role": "assistant", "content": inappropriate_msg})
 
 if __name__ == "__main__":
-    spritual_assistant_streamlit()
+    spritual_assistant_streamlit()