Upload chat_app.py

chat_app.py

@@ -0,0 +1,154 @@
+import os
+from dotenv import load_dotenv
+load_dotenv()
+
+from langchain_astradb import AstraDBVectorStore
+from langchain_google_genai import GoogleGenerativeAIEmbeddings
+from langchain.prompts import PromptTemplate
+from langchain.chains import ConversationalRetrievalChain
+from langchain_google_genai import ChatGoogleGenerativeAI
+import streamlit as st
+import time
+import textwrap
+
+# tokens and all
+os.environ["GOOGLE_API_KEY"] = "AIzaSyCoi9bUBwY5Imto3aInEnYFFQg4xZvSI30"
+os.environ["ASTRA_DB_API_ENDPOINT"] = "https://85494e7a-90c5-40ad-aa7d-3fa4fa3f6c11-us-east-2.apps.astra.datastax.com"
+os.environ["ASTRA_DB_APPLICATION_TOKEN"] = "AstraCS:ZwUXuZDQfZTxkYawpkfRXYgQ:6904895ba7ec606747a82db43fc2169d1913f6e2695d85fb4d3c2937b0cd4d8e"
+
+# embeddings 
+embeddings = GoogleGenerativeAIEmbeddings(
+    model = "models/embedding-001",
+    task_type = "retrieval_document"
+)
+
+# llm 
+llm = ChatGoogleGenerativeAI(
+    model = "gemini-1.5-pro-latest",
+    temperature = 0.7,
+)
+
+# Get Info about the Database 
+vstore = AstraDBVectorStore(
+    collection_name = "Bhagavad_gita_data",
+    embedding = embeddings,
+    token = os.getenv("ASTRA_DB_APPLICATION_TOKEN"),
+    api_endpoint = os.getenv("ASTRA_DB_API_ENDPOINT"),
+)
+
+# Now Retrieve the Documents from Server
+retriever = vstore.as_retriever(search_kwargs = {"k" : 5})
+
+prompt_template = """
+You are a wise counselor drawing from ancient Indian wisdom to offer psychological guidance. Your role is to provide practical, concise advice for modern challenges. 
+You are going to be used for a psychiatrist assitance who gives advices on the context of bhagvad gita.
+Follow these guidelines:
+
+1. Begin with a brief, relatable insight from timeless teachings.
+
+2. Offer 4 to 6 specific, actionable points of advice.
+
+3. Each point should start on a new line and be clear and concise.
+
+4. Connect each piece of advice to universal principles of success and well-being.
+
+5. Use metaphors or examples from ancient texts without explicitly naming them.
+
+6. Conclude with an encouraging statement that motivates the user to apply the advice.
+
+7. Avoid religious terminology. Use phrases like "ancient wisdom" or "timeless teachings" instead.
+
+8. Ensure your response is practical, universally applicable, and inspirational.
+
+9. Be strict that if some gives some wrong or useless input which is not relevant to physcological issue or dilema then reply them to enter the proper question
+
+10. If possible try to give only Bhagavad Gita Verse related to it at end don't get any other verse from any other book. 
+
+Context: {context}
+Question: {question}
+Human: {human_input}
+Chat History: {chat_history}
+"""
+
+
+PROMPT = PromptTemplate(
+    template = prompt_template,
+    input_variables = ["context", "question", "human_input", "chat_history"]
+)
+
+qa_chain = ConversationalRetrievalChain.from_llm(
+    llm,
+    retriever = retriever,
+    combine_docs_chain_kwargs = {"prompt": PROMPT},
+    return_source_documents = False,
+)
+
+# format the output in good format
+def format_and_wrap_text(text, wrap_length=100):
+    # Split the text into main points
+    main_points = text.split('**')
+
+    formatted_text = ""
+    for i in range(1, len(main_points), 2):
+        # Add the main point title
+        formatted_text += f"{main_points[i]}\n"
+
+        # Split the subpoints by '* '
+        subpoints = main_points[i+1].strip().split('* ')
+        for subpoint in subpoints:
+            if subpoint.strip():
+                # Wrap each subpoint and add a bullet
+                wrapped_subpoint = textwrap.fill(subpoint, wrap_length)
+                formatted_text += f"{wrapped_subpoint}\n"
+
+        formatted_text += "\n"
+
+    print(formatted_text)
+
+# Streamlit App Design 
+st.set_page_config(page_title="Arjun AI")
+
+
+# app 
+st.title("Arjun AI")
+st.write("Get Yourself Help from Krishna's Teaching of Bhagavad Gita")
+
+# Initialize chat history
+if "messages" not in st.session_state:
+    st.session_state.messages = []
+
+# Display chat messages from history on app rerun
+for message in st.session_state.messages:
+    with st.chat_message(message["role"]):
+        st.markdown(message["content"])
+
+# React to user input
+if prompt := st.chat_input("What is your question?"):
+    # Display user message in chat message container
+    st.chat_message("user").markdown(prompt)
+    # Add user message to chat history
+    st.session_state.messages.append({"role": "user", "content": prompt})
+
+    with st.chat_message("assistant"):
+        message_placeholder = st.empty()
+        full_response = ""
+
+        # Get response from QA chain
+        result = qa_chain({
+            "question": prompt,
+            "human_input": prompt,
+            "chat_history": [(msg["role"], msg["content"]) for msg in st.session_state.messages]
+        })
+        full_response = result['answer']
+
+        # Simulate stream of response with milliseconds delay
+        for chunk in full_response.split():
+            full_response = f"{full_response}"
+            time.sleep(0.05)
+            # Add a blinking cursor to simulate typing
+            message_placeholder.markdown(full_response)
+        
+        message_placeholder.markdown(full_response)
+    
+    # Add assistant response to chat history
+    st.session_state.messages.append({"role": "assistant", "content": full_response})