app.py

import streamlit as st
import time
import random
import json
from datetime import datetime
import pytz
import platform
import uuid
import extra_streamlit_components as stx
from io import BytesIO
from PIL import Image
import base64
import cv2
import requests
from moviepy.editor import VideoFileClip
from gradio_client import Client
from openai import OpenAI
import openai
import os
from collections import deque
import numpy as np
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Set page config
st.set_page_config(page_title="Personalized Real-Time Chat", page_icon="💬", layout="wide")

# Initialize cookie manager
cookie_manager = stx.CookieManager()

# File to store chat history and user data
CHAT_FILE = "chat_history.txt"

# Function to save chat history and user data to file
def save_data():
    with open(CHAT_FILE, 'w') as f:
        json.dump({
            'messages': st.session_state.messages,
            'users': st.session_state.users
        }, f)

# Function to load chat history and user data from file
def load_data():
    try:
        with open(CHAT_FILE, 'r') as f:
            data = json.load(f)
            st.session_state.messages = data['messages']
            st.session_state.users = data['users']
    except FileNotFoundError:
        st.session_state.messages = []
        st.session_state.users = []

# Load data at the start
load_data()

# Function to get or create user
def get_or_create_user():
    user_id = cookie_manager.get(cookie='user_id')
    if not user_id:
        user_id = str(uuid.uuid4())
        cookie_manager.set('user_id', user_id)
    
    user = next((u for u in st.session_state.users if u['id'] == user_id), None)
    if not user:
        user = {
            'id': user_id,
            'name': random.choice(['Alice', 'Bob', 'Charlie', 'David', 'Eve', 'Frank', 'Grace', 'Henry']),
            'browser': f"{platform.system()} - {st.session_state.get('browser_info', 'Unknown')}"
        }
        st.session_state.users.append(user)
        save_data()
    
    return user

# Initialize session state
if 'messages' not in st.session_state:
    st.session_state.messages = []
if 'users' not in st.session_state:
    st.session_state.users = []
if 'current_user' not in st.session_state:
    st.session_state.current_user = get_or_create_user()

# Initialize OpenAI client
openai.api_key = os.getenv('OPENAI_API_KEY')
openai.organization = os.getenv('OPENAI_ORG_ID')
client = OpenAI(api_key=openai.api_key, organization=openai.organization)
GPT4O_MODEL = "gpt-4o-2024-05-13"

# Initialize HuggingFace client
hf_client = OpenAI(
    base_url="https://api-inference.huggingface.co/v1",
    api_key=os.environ.get('API_KEY')
)

# Create supported models
model_links = {
    "GPT-4o": GPT4O_MODEL,
    "Meta-Llama-3.1-70B-Instruct": "meta-llama/Meta-Llama-3.1-70B-Instruct",
    "Meta-Llama-3.1-405B-Instruct-FP8": "meta-llama/Meta-Llama-3.1-405B-Instruct-FP8",
    "Meta-Llama-3.1-405B-Instruct": "meta-llama/Meta-Llama-3.1-405B-Instruct",
    "Meta-Llama-3.1-8B-Instruct": "meta-llama/Meta-Llama-3.1-8B-Instruct",
    "Meta-Llama-3-70B-Instruct": "meta-llama/Meta-Llama-3-70B-Instruct",
    "Meta-Llama-3-8B-Instruct": "meta-llama/Meta-Llama-3-8B-Instruct",
    "C4ai-command-r-plus": "CohereForAI/c4ai-command-r-plus",
    "Aya-23-35B": "CohereForAI/aya-23-35B",
    "Zephyr-orpo-141b-A35b-v0.1": "HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1",
    "Mixtral-8x7B-Instruct-v0.1": "mistralai/Mixtral-8x7B-Instruct-v0.1",
    "Codestral-22B-v0.1": "mistralai/Codestral-22B-v0.1",
    "Nous-Hermes-2-Mixtral-8x7B-DPO": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
    "Yi-1.5-34B-Chat": "01-ai/Yi-1.5-34B-Chat",
    "Gemma-2-27b-it": "google/gemma-2-27b-it",
    "Meta-Llama-2-70B-Chat-HF": "meta-llama/Llama-2-70b-chat-hf",
    "Meta-Llama-2-7B-Chat-HF": "meta-llama/Llama-2-7b-chat-hf",
    "Meta-Llama-2-13B-Chat-HF": "meta-llama/Llama-2-13b-chat-hf",
    "Mistral-7B-Instruct-v0.1": "mistralai/Mistral-7B-Instruct-v0.1",
    "Mistral-7B-Instruct-v0.2": "mistralai/Mistral-7B-Instruct-v0.2",
    "Mistral-7B-Instruct-v0.3": "mistralai/Mistral-7B-Instruct-v0.3",
    "Gemma-1.1-7b-it": "google/gemma-1.1-7b-it",
    "Gemma-1.1-2b-it": "google/gemma-1.1-2b-it",
    "Zephyr-7B-Beta": "HuggingFaceH4/zephyr-7b-beta",
    "Zephyr-7B-Alpha": "HuggingFaceH4/zephyr-7b-alpha",
    "Phi-3-mini-128k-instruct": "microsoft/Phi-3-mini-128k-instruct",
    "Phi-3-mini-4k-instruct": "microsoft/Phi-3-mini-4k-instruct",
}

# Function to reset conversation
def reset_conversation():
    st.session_state.conversation = []
    st.session_state.messages = []

# Function to generate filenames
def generate_filename(prompt, file_type):
    central = pytz.timezone('US/Central')
    safe_date_time = datetime.now(central).strftime("%m%d_%H%M")
    replaced_prompt = prompt.replace(" ", "_").replace("\n", "_")
    safe_prompt = "".join(x for x in replaced_prompt if x.isalnum() or x == "_")[:90]
    return f"{safe_date_time}_{safe_prompt}.{file_type}"

# Function to create files
def create_file(filename, prompt, response, user_name, timestamp):
    with open(filename, "w", encoding="utf-8") as f:
        f.write(f"User: {user_name}\nTimestamp: {timestamp}\n\nPrompt:\n{prompt}\n\nResponse:\n{response}")

# Function to extract video frames
def extract_video_frames(video_path, seconds_per_frame=2):
    base64Frames = []
    video = cv2.VideoCapture(video_path)
    total_frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT))
    fps = video.get(cv2.CAP_PROP_FPS)
    frames_to_skip = int(fps * seconds_per_frame)
    curr_frame = 0
    while curr_frame < total_frames - 1:
        video.set(cv2.CAP_PROP_POS_FRAMES, curr_frame)
        success, frame = video.read()
        if not success:
            break
        _, buffer = cv2.imencode(".jpg", frame)
        base64Frames.append(base64.b64encode(buffer).decode("utf-8"))
        curr_frame += frames_to_skip
    video.release()
    return base64Frames, None

# Function to process audio for video
def process_audio_for_video(video_input):
    try:
        transcription = client.audio.transcriptions.create(
            model="whisper-1",
            file=video_input,
        )
        return transcription.text
    except:
        return ''

# Function to process text with selected model
def process_text(user_name, text_input, selected_model, temp_values):
    timestamp = datetime.now(pytz.utc).strftime('%Y-%m-%d %H:%M:%S %Z')
    st.session_state.messages.append({"user": user_name, "message": text_input, "timestamp": timestamp})
    
    with st.chat_message(user_name):
        st.markdown(f"{user_name} ({timestamp}): {text_input}")
    
    with st.chat_message("Assistant"):
        try:
            if selected_model == "GPT-4o":
                completion = client.chat.completions.create(
                    model=GPT4O_MODEL,
                    messages=[
                        {"role": "user", "content": m["message"]}
                        for m in st.session_state.messages
                    ],
                    stream=True,
                    temperature=temp_values
                )
                return_text = st.write_stream(completion)
            else:
                messages = [
                    {"content": m["message"]}
                    for m in st.session_state.messages
                ]
                stream = hf_client.chat.completions.create(
                    model=model_links[selected_model],
                    messages=messages,
                    temperature=temp_values,
                    stream=True,
                    max_tokens=3000,
                )
                return_text = st.write_stream(stream)
        except openai.APIError as e:
            return_text = f"OpenAI API Error: {str(e)}"
            st.error(return_text)
        except requests.exceptions.RequestException as e:
            return_text = f"Network Error: {str(e)}"
            st.error(return_text)
        except Exception as e:
            return_text = f"Unexpected Error: {str(e)}"
            st.error(return_text)
        
        if not return_text.startswith("Error:"):
            st.markdown(f"Assistant ({timestamp}): {return_text}")
            filename = generate_filename(text_input, "md")
            create_file(filename, text_input, return_text, user_name, timestamp)
            st.session_state.messages.append({"user": "Assistant", "message": return_text, "timestamp": timestamp})
            save_data()
    
    return return_text

# Function to process image (using GPT-4o)
def process_image(user_name, image_input, user_prompt):
    image = Image.open(BytesIO(image_input))
    base64_image = base64.b64encode(image_input).decode("utf-8")
    
    response = client.chat.completions.create(
        model=GPT4O_MODEL,
        messages=[
            {"role": "system", "content": "You are a helpful assistant that responds in Markdown."},
            {"role": "user", "content": [
                {"type": "text", "text": user_prompt},
                {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_image}"}}
            ]}
        ],
        temperature=0.0,
    )
    image_response = response.choices[0].message.content
    
    timestamp = datetime.now(pytz.utc).strftime('%Y-%m-%d %H:%M:%S %Z')
    st.session_state.messages.append({"user": user_name, "message": image_response, "timestamp": timestamp})
    
    with st.chat_message(user_name):
        st.image(image)
        st.markdown(f"{user_name} ({timestamp}): {user_prompt}")
    
    with st.chat_message("Assistant"):
        st.markdown(image_response)
    
    filename_md = generate_filename(user_prompt, "md")
    create_file(filename_md, user_prompt, image_response, user_name, timestamp)
    save_data()
    return image_response

# Function to process audio (using GPT-4o for transcription)
def process_audio(user_name, audio_input, text_input):
    if audio_input:
        transcription = client.audio.transcriptions.create(
            model="whisper-1",
            file=audio_input,
        )
        timestamp = datetime.now(pytz.utc).strftime('%Y-%m-%d %H:%M:%S %Z')
        st.session_state.messages.append({"user": user_name, "message": transcription.text, "timestamp": timestamp})
        with st.chat_message(user_name):
            st.markdown(f"{user_name} ({timestamp}): {transcription.text}")
        with st.chat_message("Assistant"):
            st.markdown(transcription.text)
            filename = generate_filename(transcription.text, "wav")
            create_file(filename, text_input, transcription.text, user_name, timestamp)
            st.session_state.messages.append({"user": "Assistant", "message": transcription.text, "timestamp": timestamp})
            save_data()
        return transcription.text

# Function to process video (using GPT-4o)
def process_video(user_name, video_input, user_prompt):
    if isinstance(video_input, str):
        with open(video_input, "rb") as video_file:
            video_input = video_file.read()
    base64Frames, audio_path = extract_video_frames(video_input)
    transcript = process_audio_for_video(video_input)
    response = client.chat.completions.create(
        model=GPT4O_MODEL,
        messages=[
            {"role": "system", "content": "You are generating a video summary. Create a summary of the provided video and its transcript. Respond in Markdown"},
            {"role": "user", "content": [
                "These are the frames from the video.",
                *map(lambda x: {"type": "image_url", "image_url": {"url": f'data:image/jpg;base64,{x}', "detail": "low"}}, base64Frames),
                {"type": "text", "text": f"The audio transcription is: {transcript}"},
                {"type": "text", "text": user_prompt}
            ]}
        ],
        temperature=0,
    )
    video_response = response.choices[0].message.content
    st.markdown(video_response)
    timestamp = datetime.now(pytz.utc).strftime('%Y-%m-%d %H:%M:%S %Z')
    filename_md = generate_filename(user_prompt, "md")
    create_file(filename_md, user_prompt, video_response, user_name, timestamp)
    st.session_state.messages.append({"user": user_name, "message": video_response, "timestamp": timestamp})
    save_data()
    return video_response

# Main function for each column
def main_column(column_name):
    st.markdown(f"##### {column_name}")
    selected_model = st.selectbox(f"Select Model for {column_name}", list(model_links.keys()), key=f"{column_name}_model")
    temp_values = st.slider(f'Select a temperature value for {column_name}', 0.0, 1.0, (0.5), key=f"{column_name}_temp")
    
    option = st.selectbox(f"Select an option for {column_name}", ("Text", "Image", "Audio", "Video"), key=f"{column_name}_option")
    
    if option == "Text":
        text_input = st.text_input(f"Enter your text for {column_name}:", key=f"{column_name}_text")
        if st.button(f"Process Text for {column_name}"):
            if text_input:
                process_text(st.session_state.current_user['name'], text_input, selected_model, temp_values)
                # Clear the input after processing
                st.session_state[f"{column_name}_text"] = ""
    elif option == "Image":
        text_input = st.text_input(f"Enter text prompt to use with Image context for {column_name}:", key=f"{column_name}_image_text")
        uploaded_files = st.file_uploader(f"Upload images for {column_name}", type=["png", "jpg", "jpeg"], accept_multiple_files=True, key=f"{column_name}_image_upload")
        if st.button(f"Process Image for {column_name}"):
            for image_input in uploaded_files:
                image_bytes = image_input.read()
                process_image(st.session_state.current_user['name'], image_bytes, text_input)
            # Clear the inputs after processing
            st.session_state[f"{column_name}_image_text"] = ""
            st.session_state[f"{column_name}_image_upload"] = None
    elif option == "Audio":
        text_input = st.text_input(f"Enter text prompt to use with Audio context for {column_name}:", key=f"{column_name}_audio_text")
        uploaded_files = st.file_uploader(f"Upload an audio file for {column_name}", type=["mp3", "wav"], accept_multiple_files=True, key=f"{column_name}_audio_upload")
        if st.button(f"Process Audio for {column_name}"):
            for audio_input in uploaded_files:
                process_audio(st.session_state.current_user['name'], audio_input, text_input)
            # Clear the inputs after processing
            st.session_state[f"{column_name}_audio_text"] = ""
            st.session_state[f"{column_name}_audio_upload"] = None
    elif option == "Video":
        video_input = st.file_uploader(f"Upload a video file for {column_name}", type=["mp4"], key=f"{column_name}_video_upload")
        text_input = st.text_input(f"Enter text prompt to use with Video context for {column_name}:", key=f"{column_name}_video_text")
        if st.button(f"Process Video for {column_name}"):
            if video_input and text_input:
                process_video(st.session_state.current_user['name'], video_input, text_input)
            # Clear the inputs after processing
            st.session_state[f"{column_name}_video_text"] = ""
            st.session_state[f"{column_name}_video_upload"] = None

# Main Streamlit app
st.title("Personalized Real-Time Chat")

# Sidebar
with st.sidebar:
    st.title("User Info")
    st.write(f"Current User: {st.session_state.current_user['name']}")
    st.write(f"Browser: {st.session_state.current_user['browser']}")
    
    new_name = st.text_input("Change your name:")
    if st.button("Update Name"):
        if new_name:
            for user in st.session_state.users:
                if user['id'] == st.session_state.current_user['id']:
                    user['name'] = new_name
                    st.session_state.current_user['name'] = new_name
                    save_data()
                    st.success(f"Name updated to {new_name}")
                    break
    
    st.title("Active Users")
    for user in st.session_state.users:
        st.write(f"{user['name']} ({user['browser']})")
    
    if st.button('Reset Chat'):
        reset_conversation()

# Create two columns
col1, col2 = st.columns(2)

# Run main function for each column
with col1:
    main_column("Column 1")

with col2:
    main_column("Column 2")

# Run the Streamlit app
if __name__ == "__main__":
    st.markdown("\n[by Aaron Wacker](https://huggingface.co/spaces/awacke1/).")