RAG app

.gitignore

@@ -0,0 +1,2 @@
+.env
+*.pyc

app.py

@@ -0,0 +1,80 @@
+from dotenv import load_dotenv
+import streamlit as st
+from app_style import css, bot_template, user_template
+from llm_chain import RAG_PDF
+
+logo_image_path = "logo.png"
+
+def handle_userinput(user_question):
+    response = st.session_state.conversation({'question': user_question})
+    st.session_state.chat_history = response['chat_history']
+
+    for i, message in enumerate(st.session_state.chat_history):
+        if i % 2 == 0:
+            st.write(user_template.replace(
+                "{{MSG}}", message.content), unsafe_allow_html=True)
+        else:
+            st.write(bot_template.replace(
+                "{{MSG}}", message.content), unsafe_allow_html=True)
+
+
+def main():
+    # loading environment varibales
+    load_dotenv()
+    # Page Config
+    st.set_page_config(page_title="Ask-your-PDFs",
+                       page_icon=":books:")
+    st.write(css, unsafe_allow_html=True)
+    # Chat history session management
+
+    if "conversation" not in st.session_state:
+        st.session_state.conversation = None
+    if "chat_history" not in st.session_state:
+        st.session_state.chat_history = None
+    
+    # for rendering the background image (Uncomment the next line to update the background img of the application)
+    # render_background_img(background_path)
+    
+    # Chat User input
+    st.header("Chat with PDFs :books:")
+    user_question = st.text_input("Ask a question about your documents:")
+    styl = f"""
+            <style>
+                .stTextInput {{
+                position: fixed;
+                bottom: 3rem;
+                }}
+            </style>
+            """
+    st.markdown(styl, unsafe_allow_html=True)
+
+    # Handling user input
+    if user_question:
+        handle_userinput(user_question)
+
+    with st.sidebar:
+        # Loading the Logo
+        st.image(logo_image_path, use_column_width=True)
+        
+        # Header text for the sidebar
+        st.subheader("Your documents")
+
+        # File Uploader (Allowing multiple files upload)
+        pdf_docs = st.file_uploader(
+            "Upload your PDFs here and click on 'Submit'", accept_multiple_files=True)
+        
+        # When the submit button is clicked
+        if st.button("Submit"):
+            # Processing Bar
+            with st.spinner("Processing"):
+                # Creating an object of RAG pipeline
+                RAG_object = RAG_PDF(pdf_docs)
+
+                # Activating the RAG Pipeline
+                st.session_state.conversation = RAG_object.activate_RAG_pipeline()
+                
+                # Posting an update when the upload and processing of RAG architecture done
+                st.write("Processing Completed.")
+
+if __name__ == '__main__':
+    main()

app_style.py

@@ -0,0 +1,73 @@
+import streamlit as st
+import base64
+
+def render_background_img(background_img_path, image_type = "jpg"):
+    '''
+    A function to unpack an image from root folder and set as bg.
+ 
+    Returns
+    -------
+    The background.
+    '''
+    # set bg name
+    main_bg_ext = image_type
+        
+    st.markdown(
+         f"""
+         <style>
+         .stApp {{
+             background: url(data:image/{main_bg_ext};base64,{base64.b64encode(open(background_img_path, "rb").read()).decode()});
+             background-size: cover
+         }}
+         </style>
+         """,
+         unsafe_allow_html=True
+     )
+
+####### CSS for the CHAT UI
+
+css = '''
+<style>
+.chat-message {
+    padding: 1rem; border-radius: 0.5rem; margin-bottom: 0.8rem; display: flex; max-height: 40%; overflow: auto;
+}
+.chat-message.user {
+    background-color: #2b313e
+}
+.chat-message.bot {
+    background-color: #475063
+}
+.chat-message .avatar {
+  width: 15%;
+}
+.chat-message .avatar img {
+  max-width: 60px;
+  max-height: 60px;
+  border-radius: 50%;
+  object-fit: cover;
+}
+.chat-message .message {
+  width: 85%;
+  padding: 0 1.2rem;
+  color: #fff;
+}
+</style>
+'''
+
+bot_template = '''
+<div class="chat-message bot">
+    <div class="avatar">
+        <img src="https://i.ibb.co/cN0nmSj/Screenshot-2023-05-28-at-02-37-21.png" style="max-height: 60px; max-width: 60px; border-radius: 50%; object-fit: cover;">
+    </div>
+    <div class="message">{{MSG}}</div>
+</div>
+'''
+
+user_template = '''
+<div class="chat-message user">
+    <div class="avatar">
+        <img src="https://i.ibb.co/XJBBhsD/IMG-7040.jpg" style="max-height: 60px; max-width: 60px; border-radius: 50%; object-fit: cover;">
+    </div>    
+    <div class="message">{{MSG}}</div>
+</div>
+'''

llm_chain.py

@@ -0,0 +1,87 @@
+import streamlit as st
+from dotenv import load_dotenv
+from PyPDF2 import PdfReader
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.embeddings import OpenAIEmbeddings, HuggingFaceInstructEmbeddings
+from langchain.vectorstores import FAISS
+from langchain.chat_models import ChatOpenAI
+from langchain.memory import ConversationBufferMemory
+from langchain.chains import ConversationalRetrievalChain
+from app_style import css, bot_template, user_template
+from langchain.llms import HuggingFaceHub
+
+class RAG_PDF:
+    '''
+    Class for implementing RAGs for answer questions from PDFs
+    '''
+    def __init__(self, pdf_docs, model = "open-source"):
+        '''
+        Initializing the constructor
+        '''
+        self.pdf_docs = pdf_docs
+        if model=="open-source":
+            # Open Source model to generate embeddings for the text
+            self.embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
+            # Open Source model to generate response (Current model used is T5-XXL)
+            self.llm = HuggingFaceHub(repo_id="google/flan-t5-xxl", model_kwargs={"temperature":0.5, "max_length":512})
+        elif model=="openai":
+            # OpenAI model to generate embeddings for the text
+            self.embeddings = OpenAIEmbeddings()
+            # OpenAI model to generate response
+            self.llm = ChatOpenAI()
+
+
+    def pdf_extract_text(self):
+        '''
+        Extracting text from the PDFs
+        '''
+        text = ""
+        for pdf in self.pdf_docs:
+            pdf_reader = PdfReader(pdf)
+            for page in pdf_reader.pages:
+                text += page.extract_text()
+        return text
+    
+    def pdf_chunkize(self, text):
+        '''
+        Chunking the text into smaller chunks
+        '''
+        text_splitter = CharacterTextSplitter(
+            separator="\n",
+            chunk_size=1000,
+            chunk_overlap=200, #context aware chunking
+            length_function=len
+        )
+        chunks = text_splitter.split_text(text)
+        return chunks
+    
+    def pdf_vectorstore(self, text_chunks):
+        '''
+        Creating vector store for the text chunks
+        '''
+        vectorstore = FAISS.from_texts(texts=text_chunks, embedding=self.embeddings)
+        return vectorstore
+    
+    def pdf_conversation_chain(self, vectorstore):
+        memory = ConversationBufferMemory(
+            memory_key='chat_history', return_messages=True)
+        conversation_chain = ConversationalRetrievalChain.from_llm(
+            llm=self.llm,
+            retriever=vectorstore.as_retriever(),
+            memory=memory
+        )
+        return conversation_chain
+    
+    def activate_RAG_pipeline(self):
+        # get pdf text
+        raw_text = self.pdf_extract_text()
+
+        # get the text chunks
+        text_chunks = self.pdf_chunkize(raw_text)
+
+        # create vector store
+        vectorstore = self.pdf_vectorstore(text_chunks)
+
+        # create conversation chain
+        conversation_chain = self.pdf_conversation_chain(vectorstore)
+        return conversation_chain

requirements.txt

@@ -0,0 +1,11 @@
+langchain==0.1.1
+PyPDF2==3.0.1
+python-dotenv==1.0.0
+streamlit==1.30.0
+openai==0.27.6
+faiss-cpu==1.7.4
+altair==4
+tiktoken==0.4.0
+huggingface-hub
+InstructorEmbedding==1.0.1
+sentence-transformers==2.2.2