backup10.app.py

# 🚀 Main App - TalkingAIResearcher with Chat, Voice, Media, ArXiv, and More
import streamlit as st
import asyncio
import websockets
import uuid
import argparse
import os
import random
import time
import hashlib
from PIL import Image
import glob
import base64
import io
import streamlit.components.v1 as components
import edge_tts
from audio_recorder_streamlit import audio_recorder
import nest_asyncio
import re
import pytz
import shutil
import anthropic
import openai
from PyPDF2 import PdfReader
import threading
import json
import zipfile
from gradio_client import Client
from dotenv import load_dotenv
from streamlit_marquee import streamlit_marquee
from datetime import datetime
from collections import defaultdict, Counter
import pandas as pd

# 🛠️ Patch asyncio for nesting glory
nest_asyncio.apply()

# 🎨 Page Config
st.set_page_config(
    page_title="🚲TalkingAIResearcher🏆",
    page_icon="🚲🏆",
    layout="wide",
    initial_sidebar_state="auto"
)

# 🌟 Static Config
icons = '🤖🧠🔬📝'
START_ROOM = "Sector 🌌"
FUN_USERNAMES = {
    "CosmicJester 🌌": "en-US-AriaNeural",
    "PixelPanda 🐼": "en-US-JennyNeural",
    "QuantumQuack 🦆": "en-GB-SoniaNeural",
    "StellarSquirrel 🐿️": "en-AU-NatashaNeural",
    "GizmoGuru ⚙️": "en-CA-ClaraNeural",
    "NebulaNinja 🌠": "en-US-GuyNeural",
    "ByteBuster 💾": "en-GB-RyanNeural",
    "GalacticGopher 🌍": "en-AU-WilliamNeural",
    "RocketRaccoon 🚀": "en-CA-LiamNeural",
    "EchoElf 🧝": "en-US-AnaNeural",
}
EDGE_TTS_VOICES = list(set(FUN_USERNAMES.values()))  # 🎙️ Voice options
FILE_EMOJIS = {"md": "📝", "mp3": "🎵", "wav": "🔊"}

# 📁 Directories
for d in ["chat_logs", "vote_logs", "audio_logs", "history_logs", "media_files", "audio_cache"]:
    os.makedirs(d, exist_ok=True)

CHAT_FILE = "chat_logs/global_chat.md"
HISTORY_FILE = "history_logs/chat_history.md"
MEDIA_DIR = "media_files"
AUDIO_CACHE_DIR = "audio_cache"

# 🔑 API Keys
load_dotenv()
anthropic_key = os.getenv('ANTHROPIC_API_KEY', st.secrets.get('ANTHROPIC_API_KEY', ""))
openai_api_key = os.getenv('OPENAI_API_KEY', st.secrets.get('OPENAI_API_KEY', ""))
openai_client = openai.OpenAI(api_key=openai_api_key)

# 🕒 Timestamp Helper
def format_timestamp_prefix(username=""):
    central = pytz.timezone('US/Central')
    now = datetime.now(central)
    return f"{now.strftime('%Y%m%d_%H%M%S')}-by-{username}"

# 📈 Performance Timer
class PerformanceTimer:
    def __init__(self, name): self.name, self.start = name, None
    def __enter__(self): 
        self.start = time.time()
        return self
    def __exit__(self, *args): 
        duration = time.time() - self.start
        st.session_state['operation_timings'][self.name] = duration
        st.session_state['performance_metrics'][self.name].append(duration)

# 🎛️ Session State Init
def init_session_state():
    defaults = {
        'server_running': False, 'server_task': None, 'active_connections': {},
        'media_notifications': [], 'last_chat_update': 0, 'displayed_chat_lines': [],
        'message_text': "", 'audio_cache': {}, 'pasted_image_data': None,
        'quote_line': None, 'refresh_rate': 5, 'base64_cache': {},
        'transcript_history': [], 'last_transcript': "", 'image_hashes': set(),
        'tts_voice': "en-US-AriaNeural", 'chat_history': [], 'marquee_settings': {
            "background": "#1E1E1E", "color": "#FFFFFF", "font-size": "14px",
            "animationDuration": "20s", "width": "100%", "lineHeight": "35px"
        }, 'operation_timings': {}, 'performance_metrics': defaultdict(list),
        'enable_audio': True, 'download_link_cache': {}, 'username': None,
        'autosend': True, 'autosearch': True, 'last_message': "", 'last_query': ""
    }
    for k, v in defaults.items():
        if k not in st.session_state: st.session_state[k] = v

# 🖌️ Marquee Helpers
def update_marquee_settings_ui():
    # 🎨 Sidebar marquee controls
    st.sidebar.markdown("### 🎯 Marquee Settings")
    cols = st.sidebar.columns(2)
    with cols[0]:
        st.session_state['marquee_settings']['background'] = st.color_picker("🎨 Background", "#1E1E1E")
        st.session_state['marquee_settings']['color'] = st.color_picker("✍️ Text", "#FFFFFF")
    with cols[1]:
        st.session_state['marquee_settings']['font-size'] = f"{st.slider('📏 Size', 10, 24, 14)}px"
        st.session_state['marquee_settings']['animationDuration'] = f"{st.slider('⏱️ Speed', 1, 20, 20)}s"

def display_marquee(text, settings, key_suffix=""):
    # 🌈 Show marquee with truncation
    truncated = text[:280] + "..." if len(text) > 280 else text
    streamlit_marquee(content=truncated, **settings, key=f"marquee_{key_suffix}")
    st.write("")

# 📝 Text & File Helpers
def clean_text_for_tts(text): return re.sub(r'[#*!\[\]]+', '', ' '.join(text.split()))[:200] or "No text"
def clean_text_for_filename(text): return '_'.join(re.sub(r'[^\w\s-]', '', text.lower()).split())[:200]
def get_high_info_terms(text, top_n=10):
    stop_words = {'the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'of', 'with'}
    words = re.findall(r'\b\w+(?:-\w+)*\b', text.lower())
    bi_grams = [' '.join(pair) for pair in zip(words, words[1:])]
    filtered = [t for t in words + bi_grams if t not in stop_words and len(t.split()) <= 2]
    return [t for t, _ in Counter(filtered).most_common(top_n)]

def generate_filename(prompt, response, file_type="md"):
    # 📁 Smart filename with info terms
    prefix = format_timestamp_prefix()
    terms = get_high_info_terms(prompt + " " + response, 5)
    snippet = clean_text_for_filename(prompt[:40] + " " + response[:40])
    wct, sw = len(prompt.split()), len(response.split())
    dur = round((wct + sw) / 2.5)
    base = '_'.join(list(dict.fromkeys(terms + [snippet])))[:200 - len(prefix) - len(f"_wct{wct}_sw{sw}_dur{dur}.{file_type}")]
    return f"{prefix}{base}_wct{wct}_sw{sw}_dur{dur}.{file_type}"

def create_file(prompt, response, file_type="md"):
    # 📝 Save file with Q&A
    filename = generate_filename(prompt, response, file_type)
    with open(filename, 'w', encoding='utf-8') as f: f.write(prompt + "\n\n" + response)
    return filename

def get_download_link(file, file_type="mp3"):
    # ⬇️ Cached download link
    cache_key = f"dl_{file}"
    if cache_key not in st.session_state['download_link_cache']:
        with open(file, "rb") as f:
            b64 = base64.b64encode(f.read()).decode()
        st.session_state['download_link_cache'][cache_key] = f'<a href="data:audio/mpeg;base64,{b64}" download="{os.path.basename(file)}">{FILE_EMOJIS.get(file_type, "Download")} Download {os.path.basename(file)}</a>'
    return st.session_state['download_link_cache'][cache_key]

# 🎶 Audio Processing
async def async_edge_tts_generate(text, voice, rate=0, pitch=0, file_format="mp3"):
    # 🎵 Async TTS with caching - Fixed KeyError!
    cache_key = f"{text[:100]}_{voice}_{rate}_{pitch}_{file_format}"
    if cache_key in st.session_state['audio_cache']: return st.session_state['audio_cache'][cache_key], 0
    start_time = time.time()
    text = clean_text_for_tts(text)
    if not text: return None, 0
    filename = f"audio_{format_timestamp_prefix()}_{random.randint(1000, 9999)}.{file_format}"
    communicate = edge_tts.Communicate(text, voice, rate=f"{rate:+d}%", pitch=f"{pitch:+d}Hz")
    await communicate.save(filename)
    st.session_state['audio_cache'][cache_key] = filename
    return filename, time.time() - start_time  # No reliance on operation_timings

def play_and_download_audio(file_path):
    # 🔊 Play + download
    if file_path and os.path.exists(file_path):
        st.audio(file_path)
        st.markdown(get_download_link(file_path), unsafe_allow_html=True)

async def save_chat_entry(username, message, is_markdown=False):
    # 💬 Save chat with multicast broadcast
    central = pytz.timezone('US/Central')
    timestamp = datetime.now(central).strftime("%Y-%m-%d %H:%M:%S")
    entry = f"[{timestamp}] {username}: {message}" if not is_markdown else f"[{timestamp}] {username}:\n```markdown\n{message}\n```"
    with open(CHAT_FILE, 'a') as f: f.write(f"{entry}\n")
    voice = FUN_USERNAMES.get(username, "en-US-AriaNeural")
    audio_file, _ = await async_edge_tts_generate(clean_text_for_tts(message), voice)
    if audio_file:
        with open(HISTORY_FILE, 'a') as f: f.write(f"[{timestamp}] {username}: Audio - {audio_file}\n")
    await broadcast_message(f"{username}|{message}", "chat")
    st.session_state.last_chat_update = time.time()
    st.session_state.chat_history.append(entry)  # Append to history
    return audio_file

async def load_chat():
    # 📜 Load chat history - Numbered like old version
    if not os.path.exists(CHAT_FILE):
        with open(CHAT_FILE, 'a') as f: f.write(f"# {START_ROOM} Chat\n\nWelcome to the cosmic hub! 🎤\n")
    with open(CHAT_FILE, 'r') as f: 
        content = f.read().strip()
    lines = content.split('\n')
    numbered_content = "\n".join(f"{i+1}. {line}" for i, line in enumerate(lines) if line.strip())
    return numbered_content

# 🌐 WebSocket Handling
async def websocket_handler(websocket, path):
    # 🤝 Handle WebSocket clients - Fixed multicast
    client_id = str(uuid.uuid4())
    room_id = "chat"
    if room_id not in st.session_state.active_connections:
        st.session_state.active_connections[room_id] = {}
    st.session_state.active_connections[room_id][client_id] = websocket
    username = st.session_state.get('username', random.choice(list(FUN_USERNAMES.keys())))
    chat_content = await load_chat()
    if not any(f"Client-{client_id}" in line for line in chat_content.split('\n')):
        await save_chat_entry("System 🌟", f"{username} has joined {START_ROOM}!")
    try:
        async for message in websocket:
            if '|' in message:
                username, content = message.split('|', 1)
                await save_chat_entry(username, content)
            else:
                await websocket.send("ERROR|Message format: username|content")
    except websockets.ConnectionClosed:
        await save_chat_entry("System 🌟", f"{username} has left {START_ROOM}!")
    finally:
        if room_id in st.session_state.active_connections and client_id in st.session_state.active_connections[room_id]:
            del st.session_state.active_connections[room_id][client_id]

async def broadcast_message(message, room_id):
    # 📢 Broadcast to all clients - Fixed!
    if room_id in st.session_state.active_connections:
        disconnected = []
        for client_id, ws in st.session_state.active_connections[room_id].items():
            try:
                await ws.send(message)
            except websockets.ConnectionClosed:
                disconnected.append(client_id)
        for client_id in disconnected:
            if client_id in st.session_state.active_connections[room_id]:
                del st.session_state.active_connections[room_id][client_id]

async def run_websocket_server():
    # 🖥️ Start WebSocket server
    if not st.session_state.server_running:
        server = await websockets.serve(websocket_handler, '0.0.0.0', 8765)
        st.session_state.server_running = True
        await server.wait_closed()

# 📚 PDF to Audio
class AudioProcessor:
    def __init__(self):
        self.cache_dir = AUDIO_CACHE_DIR
        os.makedirs(self.cache_dir, exist_ok=True)
        self.metadata = json.load(open(f"{self.cache_dir}/metadata.json")) if os.path.exists(f"{self.cache_dir}/metadata.json") else {}

    def _save_metadata(self):
        with open(f"{self.cache_dir}/metadata.json", 'w') as f: json.dump(self.metadata, f)

    async def create_audio(self, text, voice='en-US-AriaNeural'):
        # 🎶 Generate cached audio
        cache_key = hashlib.md5(f"{text}:{voice}".encode()).hexdigest()
        cache_path = f"{self.cache_dir}/{cache_key}.mp3"
        if cache_key in self.metadata and os.path.exists(cache_path):
            return open(cache_path, 'rb').read()
        text = clean_text_for_tts(text)
        if not text: return None
        communicate = edge_tts.Communicate(text, voice)
        await communicate.save(cache_path)
        self.metadata[cache_key] = {'timestamp': datetime.now().isoformat(), 'text_length': len(text), 'voice': voice}
        self._save_metadata()
        return open(cache_path, 'rb').read()

def process_pdf(pdf_file, max_pages, voice, audio_processor):
    # 📄 Convert PDF to audio
    reader = PdfReader(pdf_file)
    total_pages = min(len(reader.pages), max_pages)
    texts, audios = [], {}
    async def process_page(i, text): audios[i] = await audio_processor.create_audio(text, voice)
    for i in range(total_pages):
        text = reader.pages[i].extract_text()
        texts.append(text)
        threading.Thread(target=lambda: asyncio.run(process_page(i, text))).start()
    return texts, audios, total_pages

# 🔍 ArXiv & AI Lookup
def parse_arxiv_refs(ref_text):
    # 📜 Parse ArXiv refs into dicts
    if not ref_text: return []
    papers = []
    current = {}
    for line in ref_text.split('\n'):
        if line.count('|') == 2:
            if current: papers.append(current)
            date, title, *_ = line.strip('* ').split('|')
            url = re.search(r'(https://arxiv.org/\S+)', line).group(1) if re.search(r'(https://arxiv.org/\S+)', line) else f"paper_{len(papers)}"
            current = {'date': date, 'title': title, 'url': url, 'authors': '', 'summary': '', 'full_audio': None, 'download_base64': ''}
        elif current:
            if not current['authors']: current['authors'] = line.strip('* ')
            else: current['summary'] += ' ' + line.strip() if current['summary'] else line.strip()
    if current: papers.append(current)
    return papers[:20]

def generate_5min_feature_markdown(paper):
    # ✨ 5-min research paper feature
    title, summary, authors, date, url = paper['title'], paper['summary'], paper['authors'], paper['date'], paper['url']
    pdf_url = url.replace("abs", "pdf") + (".pdf" if not url.endswith(".pdf") else "")
    wct, sw = len(title.split()), len(summary.split())
    terms = get_high_info_terms(summary, 15)
    rouge = round((len(terms) / max(sw, 1)) * 100, 2)
    mermaid = "```mermaid\nflowchart TD\n" + "\n".join(f'    T{i+1}["{t}"] --> T{i+2}["{terms[i+1]}"]' for i in range(len(terms)-1)) + "\n```"
    return f"""
## 📄 {title}
**Authors:** {authors} | **Date:** {date} | **Words:** Title: {wct}, Summary: {sw}  
**Links:** [Abstract]({url}) | [PDF]({pdf_url})  
**Terms:** {', '.join(terms)} | **ROUGE:** {rouge}%  
### 🎤 TTF Read Aloud
- **Title:** {title} | **Terms:** {', '.join(terms)} | **ROUGE:** {rouge}%  
#### Concepts Graph
{mermaid}
---
"""

def create_detailed_paper_md(papers): return "# Detailed Summary\n" + "\n".join(generate_5min_feature_markdown(p) for p in papers)

async def create_paper_audio_files(papers, query):
    # 🎧 Generate paper audio
    for p in papers:
        audio_text = clean_text_for_tts(f"{p['title']} by {p['authors']}. {p['summary']}")
        p['full_audio'], _ = await async_edge_tts_generate(audio_text, st.session_state['tts_voice'])
        if p['full_audio']: p['download_base64'] = get_download_link(p['full_audio'])

async def perform_ai_lookup(q, useArxiv=True, useArxivAudio=False):
    # 🔮 AI-powered research
    client = anthropic.Anthropic(api_key=anthropic_key)
    response = client.messages.create(model="claude-3-sonnet-20240229", max_tokens=1000, messages=[{"role": "user", "content": q}])
    result = response.content[0].text
    st.markdown("### Claude's Reply 🧠\n" + result)
    md_file = create_file(q, result)
    audio_file, _ = await async_edge_tts_generate(result, st.session_state['tts_voice'])
    play_and_download_audio(audio_file)

    if useArxiv:
        q += result
        gradio_client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
        refs = gradio_client.predict(q, 10, "Semantic Search", "mistralai/Mixtral-8x7B-Instruct-v0.1", api_name="/update_with_rag_md")[0]
        result = f"🔎 {q}\n\n{refs}"
        md_file, audio_file = create_file(q, result), (await async_edge_tts_generate(result, st.session_state['tts_voice']))[0]
        play_and_download_audio(audio_file)
        papers = parse_arxiv_refs(refs)
        if papers and useArxivAudio: await create_paper_audio_files(papers, q)
        return result, papers
    return result, []

# 📦 Zip Files
def create_zip_of_files(md_files, mp3_files, query):
    # 📦 Zip it up
    all_files = md_files + mp3_files
    if not all_files: return None
    terms = get_high_info_terms(" ".join([open(f, 'r', encoding='utf-8').read() if f.endswith('.md') else os.path.splitext(os.path.basename(f))[0].replace('_', ' ') for f in all_files] + [query]), 5)
    zip_name = f"{format_timestamp_prefix()}_{'-'.join(terms)[:20]}.zip"
    with zipfile.ZipFile(zip_name, 'w') as z: [z.write(f) for f in all_files]
    return zip_name

# 🎮 Main Interface
async def async_interface():
    init_session_state()
    if not st.session_state.username:
        available = [n for n in FUN_USERNAMES if not any(f"{n} has joined" in l for l in (await load_chat()).split('\n'))]
        st.session_state.username = random.choice(available or list(FUN_USERNAMES.keys()))
        st.session_state.tts_voice = FUN_USERNAMES[st.session_state.username]
        await save_chat_entry("System 🌟", f"{st.session_state.username} has joined {START_ROOM}!")

    st.title(f"🤖🧠MMO Chat & Research for {st.session_state.username}📝🔬")
    update_marquee_settings_ui()
    display_marquee(f"🚀 Welcome to {START_ROOM} | 🤖 {st.session_state.username}", st.session_state['marquee_settings'], "welcome")

    if not st.session_state.server_task:
        st.session_state.server_task = asyncio.create_task(run_websocket_server())

    tab_main = st.radio("Action:", ["🎤 Chat & Voice", "📸 Media", "🔍 ArXiv", "📚 PDF to Audio"], horizontal=True)
    useArxiv, useArxivAudio = st.checkbox("Search ArXiv", True), st.checkbox("ArXiv Audio", False)
    st.session_state.autosend = st.checkbox("Autosend Chat", value=True)
    st.session_state.autosearch = st.checkbox("Autosearch ArXiv", value=True)

    # 🎤 Chat & Voice
    if tab_main == "🎤 Chat & Voice":
        st.subheader(f"{START_ROOM} Chat 💬")
        chat_content = await load_chat()
        chat_container = st.container()
        with chat_container:
            st.markdown(chat_content)  # Display numbered chat history
        
        message = st.text_input(f"Message as {st.session_state.username}", key="message_input")
        if message and message != st.session_state.last_message:
            st.session_state.last_message = message
            if st.session_state.autosend or st.button("Send 🚀"):
                await save_chat_entry(st.session_state.username, message, True)
                st.rerun()

        st.subheader("🎤 Speech-to-Chat")
        speech_component = components.declare_component("speech_component", path="mycomponent")
        transcript_data = speech_component(default_value=st.session_state.get('last_transcript', ''))
        if transcript_data and 'value' in transcript_data:
            transcript = transcript_data['value'].strip()
            st.write(f"🎙️ You said: {transcript}")
            if transcript and transcript != st.session_state.last_transcript:
                st.session_state.last_transcript = transcript
                if st.session_state.autosend:
                    await save_chat_entry(st.session_state.username, transcript, True)
                    st.rerun()
                elif st.button("Send to Chat"):
                    await save_chat_entry(st.session_state.username, transcript, True)
                    st.rerun()

    # 📸 Media
    elif tab_main == "📸 Media":
        st.header("📸 Media Gallery")
        tabs = st.tabs(["🎵 Audio", "🖼 Images", "🎥 Video"])
        with tabs[0]:
            for a in glob.glob(f"{MEDIA_DIR}/*.mp3"):
                with st.expander(os.path.basename(a)): play_and_download_audio(a)
        with tabs[1]:
            imgs = glob.glob(f"{MEDIA_DIR}/*.png") + glob.glob(f"{MEDIA_DIR}/*.jpg")
            if imgs:
                cols = st.columns(3)
                for i, f in enumerate(imgs): cols[i % 3].image(f, use_container_width=True)
        with tabs[2]:
            for v in glob.glob(f"{MEDIA_DIR}/*.mp4"):
                with st.expander(os.path.basename(v)): st.video(v)

        uploaded_file = st.file_uploader("Upload Media", type=['png', 'jpg', 'mp4', 'mp3'])
        if uploaded_file:
            filename = f"{format_timestamp_prefix(st.session_state.username)}-{hashlib.md5(uploaded_file.getbuffer()).hexdigest()[:8]}.{uploaded_file.name.split('.')[-1]}"
            with open(f"{MEDIA_DIR}/{filename}", 'wb') as f: f.write(uploaded_file.getbuffer())
            await save_chat_entry(st.session_state.username, f"Uploaded: {filename}")
            st.rerun()

    # 🔍 ArXiv
    elif tab_main == "🔍 ArXiv":
        q = st.text_input("🔍 Query:", key="arxiv_query")
        if q and q != st.session_state.last_query:
            st.session_state.last_query = q
            if st.session_state.autosearch or st.button("🔍 Run"):
                result, papers = await perform_ai_lookup(q, useArxiv, useArxivAudio)
                for i, p in enumerate(papers, 1):
                    with st.expander(f"{i}. 📄 {p['title']}"):
                        st.markdown(f"**{p['date']} | {p['title']}** — [Link]({p['url']})")
                        st.markdown(generate_5min_feature_markdown(p))
                        if p.get('full_audio'): play_and_download_audio(p['full_audio'])

    # 📚 PDF to Audio
    elif tab_main == "📚 PDF to Audio":
        audio_processor = AudioProcessor()
        pdf_file = st.file_uploader("Choose PDF", "pdf")
        max_pages = st.slider('Pages', 1, 100, 10)
        if pdf_file:
            with st.spinner('Processing...'):
                texts, audios, total = process_pdf(pdf_file, max_pages, st.session_state['tts_voice'], audio_processor)
                for i, text in enumerate(texts):
                    with st.expander(f"Page {i+1}"):
                        st.markdown(text)
                        while i not in audios: time.sleep(0.1)
                        if audios[i]:
                            st.audio(audios[i], format='audio/mp3')
                            st.markdown(get_download_link(io.BytesIO(audios[i]), "mp3"), unsafe_allow_html=True)

    # 🗂️ Sidebar
    st.sidebar.subheader("Voice Settings")
    new_username = st.sidebar.selectbox("Change Name/Voice", list(FUN_USERNAMES.keys()), index=list(FUN_USERNAMES.keys()).index(st.session_state.username))
    if new_username != st.session_state.username:
        await save_chat_entry("System 🌟", f"{st.session_state.username} changed to {new_username}")
        st.session_state.username, st.session_state.tts_voice = new_username, FUN_USERNAMES[new_username]
        st.rerun()

    md_files, mp3_files = glob.glob("*.md"), glob.glob("*.mp3")
    st.sidebar.markdown("### 📂 File History")
    for f in sorted(md_files + mp3_files, key=os.path.getmtime, reverse=True)[:10]:
        st.sidebar.write(f"{FILE_EMOJIS.get(f.split('.')[-1], '📄')} {os.path.basename(f)}")
    if st.sidebar.button("⬇️ Zip All"):
        zip_name = create_zip_of_files(md_files, mp3_files, "latest_query")
        if zip_name: st.sidebar.markdown(get_download_link(zip_name, "zip"), unsafe_allow_html=True)

def main():
    # 🎉 Kick it off
    asyncio.run(async_interface())

if __name__ == "__main__":
    main()