Upload 2 files

generate_answer.py

@@ -0,0 +1,95 @@
+import os
+from glob import glob
+import openai
+from dotenv import load_dotenv
+
+from langchain.embeddings import OpenAIEmbeddings
+from langchain.vectorstores import Chroma
+from langchain.document_loaders import PyPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+
+from langchain_community.chat_models import ChatOpenAI
+from langchain.chains import RetrievalQA
+from langchain.memory import ConversationBufferMemory
+
+load_dotenv()
+api_key = os.getenv("OPENAI_API_KEY")
+openai.api_key = api_key
+
+# Helper function to validate response completeness
+def is_response_complete(response: str) -> bool:
+    return response.strip()[-1] in ".!?"
+
+# Retry mechanism for incomplete responses
+def retry_response(messages):
+    response = openai.ChatCompletion.create(
+        model="gpt-3.5-turbo",
+        messages=messages
+    ).choices[0].message['content']
+    if not is_response_complete(response):
+        response += " This is the end of the response. Please let me know if you need further clarification."
+    return response
+
+def base_model_chatbot(messages):
+    system_message = [
+        {"role": "system", "content": "You are a helpful AI chatbot that provides clear, complete, and coherent responses to User's questions. Ensure your answers are in full sentences and complete the thought or idea."}
+    ]
+    messages = system_message + messages
+    response = openai.ChatCompletion.create(
+        model="gpt-3.5-turbo",
+        messages=messages
+    ).choices[0].message['content']
+    # Validate response completeness
+    if not is_response_complete(response):
+        response = retry_response(messages)
+    return response
+
+class VectorDB:
+    """Class to manage document loading and vector database creation."""
+    
+    def __init__(self, docs_directory: str):
+        self.docs_directory = docs_directory
+
+    def create_vector_db(self):
+        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
+
+        files = glob(os.path.join(self.docs_directory, "*.pdf"))
+
+        loadPDFs = [PyPDFLoader(pdf_file) for pdf_file in files]
+
+        pdf_docs = list()
+        for loader in loadPDFs:
+            pdf_docs.extend(loader.load())
+        chunks = text_splitter.split_documents(pdf_docs)
+            
+        return Chroma.from_documents(chunks, OpenAIEmbeddings()) 
+    
+class ConversationalRetrievalChain:
+    """Class to manage the QA chain setup."""
+
+    def __init__(self, model_name="gpt-3.5-turbo", temperature=0):
+        self.model_name = model_name
+        self.temperature = temperature
+
+    def create_chain(self):
+        model = ChatOpenAI(
+            model_name=self.model_name, 
+            temperature=self.temperature,
+            system_prompt="You are a knowledgeable AI that answers questions based on provided documents. Always give responses in clear, complete sentences."
+        )
+        memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
+        vector_db = VectorDB('docs/')
+        retriever = vector_db.create_vector_db().as_retriever(search_type="similarity", search_kwargs={"k": 2})
+        return RetrievalQA.from_chain_type(
+            llm=model,
+            retriever=retriever,
+            memory=memory,
+        )
+    
+def with_pdf_chatbot(messages):
+    query = messages[-1]['content'].strip()
+    qa_chain = ConversationalRetrievalChain().create_chain()
+    result = qa_chain({"query": query})
+    if not is_response_complete(result['result']):
+        result['result'] += " This is the end of the response. Let me know if you need further clarification."
+    return result['result']

helpers.py

@@ -0,0 +1,42 @@
+import base64
+import streamlit as st
+import os
+import openai
+from dotenv import load_dotenv
+from gtts import gTTS
+
+# Function to accept OpenAI API Key as input from the user
+def get_api_key():
+    api_key = st.text_input("Enter your OpenAI API Key", type="password")
+    if api_key:
+        openai.api_key = api_key
+        return api_key
+    else:
+        return None
+
+def speech_to_text(audio_data):
+    """Transcribes audio data to text using OpenAI's API."""
+    with open(audio_data, "rb") as audio_file:
+        transcript = openai.Audio.transcribe(
+            model="whisper-1",
+            file=audio_file
+        )
+    return transcript["text"]
+
+def text_to_speech(input_text):
+    """Generates a TTS audio file from the input text."""
+    tts = gTTS(text=input_text, lang="en")
+    audio_file_path = "temp_audio_play.mp3"
+    tts.save(audio_file_path)
+    return audio_file_path
+
+def autoplay_audio(file_path: str):
+    with open(file_path, "rb") as f:
+        data = f.read()
+    b64 = base64.b64encode(data).decode("utf-8")
+    md = f"""
+    <audio autoplay>
+    <source src="data:audio/mp3;base64,{b64}" type="audio/mp3">
+    </audio>
+    """
+    st.markdown(md, unsafe_allow_html=True)