PDF-Insight-PRO / app.py
jatinmehra's picture
Update .gitignore and adjust Max Tokens slider range in app.py
cf23d65
raw
history blame
6.87 kB
import os
import tempfile
import json
import streamlit as st
from streamlit_chat import message
from preprocessing import Model
from io import BytesIO
import pickle
# Home Page Setup
st.set_page_config(
page_title="PDF Insight Pro",
page_icon="πŸ“„",
layout="centered",
)
# Custom CSS for a more polished look
st.markdown("""
<style>
.main {
background-color: #f5f5f5;
}
.stButton button {
background-color: #4CAF50;
color: white;
border-radius: 8px;
}
.stTextInput input {
border-radius: 8px;
padding: 10px;
}
.stFileUploader input {
border-radius: 8px;
}
.stMarkdown h1 {
color: #4CAF50;
}
</style>
""", unsafe_allow_html=True)
# Custom title and header
st.title("πŸ“„ PDF Insight Pro")
st.subheader("Empower Your Documents with AI-Driven Insights")
def display_messages():
"""
Displays the chat messages in the Streamlit UI.
"""
st.subheader("πŸ—¨οΈ Conversation")
st.markdown("---")
for i, (msg, is_user) in enumerate(st.session_state["messages"]):
message(msg, is_user=is_user, key=str(i))
st.session_state["process_input_spinner"] = st.empty()
def process_user_input():
"""
Processes the user input by generating a response from the assistant.
"""
if st.session_state["user_input"] and len(st.session_state["user_input"].strip()) > 0:
user_input = st.session_state["user_input"].strip()
with st.session_state["process_input_spinner"], st.spinner("Analyzing..."):
agent_response = st.session_state["assistant"].get_response(
user_input,
st.session_state["temperature"],
st.session_state["max_tokens"],
st.session_state["model"]
)
st.session_state["messages"].append((user_input, True))
st.session_state["messages"].append((agent_response, False))
st.session_state["user_input"] = ""
# Save chat history temporarily on local storage
with open("chat_history.pkl", "wb") as f:
pickle.dump(st.session_state["messages"], f)
def process_file():
"""
Processes the uploaded PDF file and appends its content to the context.
"""
for file in st.session_state["file_uploader"]:
with tempfile.NamedTemporaryFile(delete=False) as tf:
tf.write(file.getbuffer())
file_path = tf.name
with st.session_state["process_file_spinner"], st.spinner(f"Processing {file.name}..."):
try:
st.session_state["assistant"].add_to_context(file_path)
except Exception as e:
st.error(f"Failed to process file {file.name}: {str(e)}")
os.remove(file_path)
def download_chat_history():
"""
Allows users to download chat history in HTML or JSON format.
"""
# Convert messages to JSON format
chat_data = [{"role": "user" if is_user else "assistant", "content": msg} for msg, is_user in st.session_state["messages"]]
# Download as JSON
json_data = json.dumps(chat_data, indent=4)
st.download_button(
label="πŸ’Ύ Download Chat History as JSON",
data=json_data,
file_name="chat_history.json",
mime="application/json"
)
# Download as HTML
html_data = "<html><body><h1>Chat History</h1><ul>"
for entry in chat_data:
role = "User" if entry["role"] == "user" else "Assistant"
html_data += f"<li><strong>{role}:</strong> {entry['content']}</li>"
html_data += "</ul></body></html>"
st.download_button(
label="πŸ’Ύ Download Chat History as HTML",
data=html_data,
file_name="chat_history.html",
mime="text/html"
)
def main_page():
"""
Main function to set up the Streamlit UI and handle user interactions.
"""
# Initialize session state variables
if "messages" not in st.session_state:
st.session_state["messages"] = []
if "assistant" not in st.session_state:
st.session_state["assistant"] = Model()
if "user_input" not in st.session_state:
st.session_state["user_input"] = ""
if "temperature" not in st.session_state:
st.session_state["temperature"] = 0.5
if "max_tokens" not in st.session_state:
st.session_state["max_tokens"] = 550
if "model" not in st.session_state:
st.session_state["model"] = "llama-3.1-8b-instant"
# File uploader
st.subheader("πŸ“€ Upload Your PDF Documents")
st.file_uploader(
"Choose PDF files to analyze",
type=["pdf"],
key="file_uploader",
on_change=process_file,
accept_multiple_files=True,
)
st.session_state["process_file_spinner"] = st.empty()
# Document management section
if st.session_state["assistant"].contexts:
st.subheader("πŸ—‚οΈ Manage Uploaded Documents")
for i, context in enumerate(st.session_state["assistant"].contexts):
st.text_area(f"Document {i+1} Context", context[:500] + "..." if len(context) > 500 else context, height=100)
if st.button(f"Remove Document {i+1}"):
st.session_state["assistant"].remove_from_context(i)
# Model settings
with st.expander("βš™οΈ Customize AI Settings", expanded=True):
st.slider("Sampling Temperature", min_value=0.0, max_value=1.0, step=0.1, key="temperature", help="Higher values make output more random.")
st.slider("Max Tokens", min_value=750, max_value=5000, step=50, key="max_tokens", help="Limits the length of the response.")
st.selectbox("Choose AI Model", ["llama-3.1-8b-instant", "llama3-70b-8192", "gemma-7b-it"], key="model")
# Display messages and input box
display_messages()
st.text_input("Type your query and hit Enter", key="user_input", on_change=process_user_input, placeholder="Ask something about your documents...")
# Download chat history section
st.subheader("πŸ’Ύ Download Chat History")
download_chat_history()
# Developer info and bug report
st.subheader("🐞 Bug Report")
st.markdown("""
If you encounter any bugs or issues while using the app, please send a bug report to the developer. You can include a screenshot (optional) to help identify the problem.\n
""")
st.subheader("πŸ’‘ Suggestions")
st.markdown("""
Suggestions to improve the app's UI and user interface are also welcome. Feel free to reach out to the developer with your suggestions.\n
""")
st.subheader("πŸ‘¨β€πŸ’» Developer Info")
st.markdown("""
**Developer**: Jatin Mehra\n
**Email**: [email protected]\n
**Mobile**: 9910364780\n
""")
if __name__ == "__main__":
main_page()