Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,116 +1,128 @@
+from dotenv import load_dotenv
 import streamlit as st
-from pypdf import PdfReader
-from docx import Document
-import chromadb
-from chromadb.utils import embedding_functions
-from huggingface_hub import InferenceClient
-import time
+from PyPDF2 import PdfReader
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.embeddings.huggingface import HuggingFaceEmbeddings
+from langchain.vectorstores import FAISS #facebook AI similarity search
+from langchain.chains.question_answering import load_qa_chain
+from langchain import HuggingFaceHub
+import docx
 import os
+from langchain.chains import ConversationalRetrievalChain
+from langchain.memory import ConversationBufferMemory
+from langchain_core.callbacks import StdOutCallbackHandler
+from streamlit_chat import message
 
-# Initialize ChromaDB (ephemeral for HF Spaces)
-client = chromadb.EphemeralClient()
-sentence_transformer_ef = embedding_functions.SentenceTransformerEmbeddingFunction(
-    model_name="all-MiniLM-L6-v2"
-)
-collection = client.get_or_create_collection(
-    name="documents",
-    embedding_function=sentence_transformer_ef
-)
-
-# Initialize HF Inference Client
-hf_client = InferenceClient(model="google/gemma-2b-it")
-
-def chunk_text(text: str, chunk_size: int = 1000) -> List[str]:
-    chunks = []
-    start = 0
-    while start < len(text):
-        end = min(start + chunk_size, len(text))
-        if end < len(text):
-            while end > start and text[end] not in {'.', '!', '?', '\n'}:
-                end -= 1
-            if end == start:
-                end = start + chunk_size
-        chunks.append(text[start:end].strip())
-        start = end
-    return chunks
 
-def process_document(uploaded_file):
+def main():
+    load_dotenv()
+    st.set_page_config(page_title="Ask your PDF")
+    st.header("Ask Your PDF")
+
+    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
+    if "processComplete" not in st.session_state:
+        st.session_state.processComplete = None
+
+    with st.sidebar:
+        uploaded_files =  st.file_uploader("Upload your file",type=['pdf','docx'],accept_multiple_files=True)
+        process = st.button("Process")
+
+    # pdf = st.file_uploader("Upload your pdf",type="pdf")
+
+    if process:
+        files_text = get_files_text(uploaded_files)
+        # get text chunks
+        text_chunks = get_text_chunks(files_text)
+        # create vetore stores
+        vetorestore = get_vectorstore(text_chunks)
+         # create conversation chain
+        st.session_state.conversation = get_conversation_chain(vetorestore) #for openAI
+        # st.session_state.conversation = get_conversation_chain(vetorestore) #for huggingface
+
+        st.session_state.processComplete = True
+
+    if  st.session_state.processComplete == True:
+        user_question = st.chat_input("Ask Question about your files.")
+        if user_question:
+            handel_userinput(user_question)
+
+def get_files_text(uploaded_files):
     text = ""
-    if uploaded_file.type == "application/pdf":
-        reader = PdfReader(uploaded_file)
-        text = "\n".join([page.extract_text() for page in reader.pages])
-    elif uploaded_file.type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
-        doc = Document(uploaded_file)
-        text = "\n".join([para.text for para in doc.paragraphs])
-    elif uploaded_file.type == "text/plain":
-        text = str(uploaded_file.read(), "utf-8")
-    
-    chunks = chunk_text(text)
-    ids = [f"{uploaded_file.name}-{i}" for i in range(len(chunks))]
-    collection.add(
-        documents=chunks,
-        ids=ids,
-        metadatas=[{"source": uploaded_file.name} for _ in chunks]
-    )
-    return len(chunks)
+    for uploaded_file in uploaded_files:
+        split_tup = os.path.splitext(uploaded_file.name)
+        file_extension = split_tup[1]
+        if file_extension == ".pdf":
+            text += get_pdf_text(uploaded_file)
+        elif file_extension == ".docx":
+            text += get_docx_text(uploaded_file)
+        else:
+            text += get_csv_text(uploaded_file)
+    return text
 
-def retrieve_chunks(query: str, k: int = 3) -> Tuple[List[str], List[str]]:
-    results = collection.query(
-        query_texts=[query],
-        n_results=k
-    )
-    return results['documents'][0], results['metadatas'][0]
-
-def generate_response(query: str, context: str) -> str:
-    prompt = f"""Context: {context}\n\nQuestion: {query}\nAnswer:"""
-    return hf_client.text_generation(
-        prompt,
-        max_new_tokens=512,
-        temperature=0.7
+def get_pdf_text(pdf):
+    pdf_reader = PdfReader(pdf)
+    text = ""
+    for page in pdf_reader.pages:
+        text += page.extract_text()
+    return text
+
+def get_docx_text(file):
+    doc = docx.Document(file)
+    allText = []
+    for docpara in doc.paragraphs:
+        allText.append(docpara.text)
+    text = ' '.join(allText)
+    return text
+
+def get_csv_text(file):
+    return "a"
+
+def get_text_chunks(text):
+    # spilit ito chuncks
+    text_splitter = CharacterTextSplitter(
+        separator="\n",
+        chunk_size=900,
+        chunk_overlap=100,
+        length_function=len
     )
+    chunks = text_splitter.split_text(text)
+    return chunks
+
 
-# Streamlit UI
-st.title("📄 Document Q&A Assistant")
+def get_vectorstore(text_chunks):
+    embeddings = HuggingFaceEmbeddings()
+    knowledge_base = FAISS.from_texts(text_chunks,embeddings)
+    return knowledge_base
 
-with st.sidebar:
-    st.header("Upload Documents")
-    uploaded_files = st.file_uploader(
-        "Choose files",
-        type=["pdf", "docx", "txt"],
-        accept_multiple_files=True
+def get_conversation_chain(vetorestore):
+    handler = StdOutCallbackHandler()
+    llm = HuggingFaceHub(repo_id="google/flan-t5-large", model_kwargs={"temperature":5,"max_length":64})
+    memory = ConversationBufferMemory(memory_key='chat_history', return_messages=True)
+    conversation_chain = ConversationalRetrievalChain.from_llm(
+        llm=llm,
+        retriever=vetorestore.as_retriever(),
+        memory=memory,
+        callbacks=[handler]
     )
-    
-    if uploaded_files:
-        with st.spinner("Processing documents..."):
-            for file in uploaded_files:
-                chunks = process_document(file)
-                st.success(f"Processed {file.name} into {chunks} chunks")
-
-if "messages" not in st.session_state:
-    st.session_state.messages = []
-
-for message in st.session_state.messages:
-    with st.chat_message(message["role"]):
-        st.markdown(message["content"])
-
-if prompt := st.chat_input("Ask about your documents"):
-    st.session_state.messages.append({"role": "user", "content": prompt})
-    
-    with st.chat_message("user"):
-        st.markdown(prompt)
-    
-    with st.chat_message("assistant"):
-        with st.spinner("Searching documents..."):
-            chunks, metadata = retrieve_chunks(prompt)
-            context = "\n\n".join(chunks)
-            
-        with st.spinner("Generating response..."):
-            response = generate_response(prompt, context)
-            sources = list(set([m['source'] for m in metadata]))
-            
-            if sources:
-                response += f"\n\nSources: {', '.join(sources)}"
-            
-            st.markdown(response)
-    
-    st.session_state.messages.append({"role": "assistant", "content": response})
+    return conversation_chain
+
+
+def handel_userinput(user_question):
+    response = st.session_state.conversation({'question':user_question})
+    st.session_state.chat_history = response['chat_history']
+
+    # Layout of input/response containers
+    response_container = st.container()
+
+    with response_container:
+        for i, messages in enumerate(st.session_state.chat_history):
+            if i % 2 == 0:
+                message(messages.content, is_user=True, key=str(i))
+            else:
+                message(messages.content, key=str(i))
+
+if __name__ == '__main__':
+    main()