Update app.py

app.py

@@ -1,34 +1,48 @@
 import streamlit as st
 import anthropic, openai, base64, cv2, glob, json, math, os, pytz, random, re, requests, time, zipfile
-from datetime import datetime
+from datetime import datetime 
 from audio_recorder_streamlit import audio_recorder
 from collections import defaultdict
 from dotenv import load_dotenv
 from gradio_client import Client
 from huggingface_hub import InferenceClient
 from PIL import Image
+from openai import OpenAI
 from streamlit_marquee import streamlit_marquee
 import asyncio
 import edge_tts
 
+# App Config
 st.set_page_config(page_title="🚲TalkingAIResearcher🏆", page_icon="🚲🏆", layout="wide")
+load_dotenv()
 
 EDGE_TTS_VOICES = ["en-US-AriaNeural", "en-US-GuyNeural", "en-US-JennyNeural", "en-GB-SoniaNeural"]
-FILE_EMOJIS = {"md": "📝", "mp3": "🎵", "wav": "🔊", "txt": "📄", "pdf": "📑"}
+FILE_EMOJIS = {"md": "📝", "mp3": "🎵", "wav": "🔊", "txt": "📄", "pdf": "📑", "html": "🌐"}
 
-# Session state initialization
-for key, default in {
+# Initialize session state
+state_vars = {
     'tts_voice': EDGE_TTS_VOICES[0],
     'audio_format': 'mp3',
     'messages': [],
     'chat_history': [],
     'transcript_history': [],
     'viewing_prefix': None,
-    'should_rerun': False
-}.items():
+    'should_rerun': False,
+    'editing_mode': False,
+    'current_file': None,
+    'file_content': None,
+    'old_val': None,
+    'last_query': ''
+}
+
+for key, default in state_vars.items():
     if key not in st.session_state:
         st.session_state[key] = default
 
+# API clients setup
+openai_client = OpenAI(api_key=os.getenv('OPENAI_API_KEY'))
+claude_client = anthropic.Anthropic(api_key=os.getenv('ANTHROPIC_API_KEY'))
+
 @st.cache_resource
 def get_cached_audio_b64(file_path):
     with open(file_path, "rb") as f:
@@ -38,8 +52,66 @@ def beautify_filename(filename):
     name = os.path.splitext(filename)[0]
     return name.replace('_', ' ').replace('.', ' ')
 
+def clean_speech_text(text):
+    text = re.sub(r'\s+', ' ', text.strip())
+    text = text.replace("</s>", "").replace("#", "")
+    text = re.sub(r"\(https?:\/\/[^\)]+\)", "", text)
+    return text
+
+async def edge_tts_generate(text, voice="en-US-AriaNeural", rate=0, pitch=0, file_format="mp3"):
+    text = clean_speech_text(text)
+    if not text: return None
+    communicate = edge_tts.Communicate(text, voice, rate=f"{rate}%", pitch=f"{pitch}Hz")
+    filename = f"{datetime.now().strftime('%y%m_%H%M')}_{voice}.{file_format}"
+    await communicate.save(filename)
+    return filename
+
+def speak_text(text, voice=None, file_format=None):
+    if not text: return None
+    voice = voice or st.session_state['tts_voice']
+    fmt = file_format or st.session_state['audio_format']
+    return asyncio.run(edge_tts_generate(text, voice, file_format=fmt))
+
+def process_audio_file(audio_path):
+    with open(audio_path, "rb") as f:
+        transcript = openai_client.audio.transcriptions.create(model="whisper-1", file=f)
+        text = transcript.text
+        st.session_state.messages.append({"role": "user", "content": text})
+        return text
+
+def process_with_gpt(text):
+    if not text: return
+    st.session_state.messages.append({"role": "user", "content": text})
+    with st.chat_message("user"): st.markdown(text)
+    with st.chat_message("assistant"):
+        response = openai_client.chat.completions.create(
+            model="gpt-4-turbo-preview",
+            messages=st.session_state.messages,
+            stream=False
+        )
+        answer = response.choices[0].message.content
+        st.write(f"GPT-4: {answer}")
+        create_file(text, answer, "md")
+        st.session_state.messages.append({"role": "assistant", "content": answer})
+        return answer
+
+def process_with_claude(text):
+    if not text: return
+    with st.chat_message("user"): st.markdown(text)
+    with st.chat_message("assistant"):
+        response = claude_client.messages.create(
+            model="claude-3-sonnet-20240229",
+            max_tokens=4000,
+            messages=[{"role": "user", "content": text}]
+        )
+        answer = response.content[0].text
+        st.write(f"Claude-3: {answer}")
+        create_file(text, answer, "md")
+        st.session_state.chat_history.append({"user": text, "claude": answer})
+        return answer
+
 def load_files_for_sidebar():
-    """Load and filter files for sidebar by timestamp prefix"""
+    """Load and filter files by timestamp prefix"""
     files = []
     for f in glob.glob("*.*"):
         basename = os.path.basename(f)
@@ -59,42 +131,6 @@ def load_files_for_sidebar():
                  key=lambda x: max(os.path.getmtime(f) for f in x[1]),
                  reverse=True)
 
-def display_marquee_controls():
-    st.sidebar.markdown("### 🎯 Marquee Settings")
-    cols = st.sidebar.columns(2)
-    with cols[0]:
-        bg_color = st.color_picker("🎨 Background", "#1E1E1E")
-        text_color = st.color_picker("✍️ Text", "#FFFFFF")
-    with cols[1]:
-        font_size = st.slider("📏 Size", 10, 24, 14)
-        duration = st.slider("⏱️ Speed", 1, 20, 10)
-    
-    return {
-        "background": bg_color,
-        "color": text_color,
-        "font-size": f"{font_size}px",
-        "animationDuration": f"{duration}s",
-        "width": "100%",
-        "lineHeight": "35px"
-    }
-
-def get_download_link(file_path, file_type="zip"):
-    with open(file_path, "rb") as f:
-        b64 = base64.b64encode(f.read()).decode()
-    ext_map = {'zip': '📦', 'mp3': '🎵', 'wav': '🔊', 'md': '📝'}
-    emoji = ext_map.get(file_type, '')
-    return f'<a href="data:application/{file_type};base64,{b64}" download="{os.path.basename(file_path)}">{emoji} Download {os.path.basename(file_path)}</a>'
-
-def create_zip_of_files(md_files, mp3_files, wav_files, query=''):
-    all_files = md_files + mp3_files + wav_files
-    if not all_files: return None
-    timestamp = datetime.now().strftime("%y%m_%H%M")
-    zip_name = f"{timestamp}_archive.zip"
-    with zipfile.ZipFile(zip_name, 'w') as z:
-        for f in all_files:
-            z.write(f)
-    return zip_name
-
 def display_file_manager_sidebar(groups_sorted):
     st.sidebar.title("📚 File Manager")
     all_files = {'md': [], 'mp3': [], 'wav': []}
@@ -153,79 +189,126 @@ def display_file_manager_sidebar(groups_sorted):
                             height=0
                         )
 
-async def edge_tts_generate(text, voice, file_format="mp3"):
-    text = re.sub(r'\s+', ' ', text).strip()
-    if not text: return None
-    communicate = edge_tts.Communicate(text, voice)
-    filename = f"{datetime.now().strftime('%y%m_%H%M')}_{voice}.{file_format}"
-    await communicate.save(filename)
-    return filename
+def perform_arxiv_search(query):
+    client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
+    papers = client.predict(
+        query, 20, "Semantic Search",
+        "mistralai/Mixtral-8x7B-Instruct-v0.1",
+        api_name="/update_with_rag_md"
+    )[0]
+    
+    summary = client.predict(
+        query,
+        "mistralai/Mixtral-8x7B-Instruct-v0.1",
+        True,
+        api_name="/ask_llm"
+    )
+
+    result = f"### 🔎 {query}\n\n{summary}\n\n{papers}"
+    st.markdown(result)
+    
+    papers_data = parse_arxiv_refs(papers)
+    if papers_data:
+        create_paper_audio(papers_data, query)
+        display_papers(papers_data)
+    
+    create_file(query, result, "md")
+    return result
 
 def parse_arxiv_refs(text):
     papers = []
-    current_paper = None
+    current = None
     
     for line in text.split('\n'):
         if '|' in line:
-            if current_paper:
-                papers.append(current_paper)
+            if current: papers.append(current)
             parts = line.strip('* ').split('|')
-            current_paper = {
+            current = {
                 'date': parts[0].strip(),
                 'title': parts[1].strip(),
                 'authors': '',
                 'summary': '',
                 'id': re.search(r'(\d{4}\.\d{5})', line).group(1) if re.search(r'(\d{4}\.\d{5})', line) else ''
             }
-        elif current_paper:
-            if not current_paper['authors']:
-                current_paper['authors'] = line.strip('* ')
+        elif current:
+            if not current['authors']:
+                current['authors'] = line.strip('* ')
             else:
-                current_paper['summary'] += ' ' + line.strip()
+                current['summary'] += ' ' + line.strip()
     
-    if current_paper:
-        papers.append(current_paper)
-    return papers
+    if current: papers.append(current)
+    return papers[:20]
 
-def perform_ai_lookup(query):
-    client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
-    response = client.predict(
-        query, 20, "Semantic Search", 
-        "mistralai/Mixtral-8x7B-Instruct-v0.1",
-        api_name="/update_with_rag_md"
-    )
-    
-    papers = parse_arxiv_refs(response[0])
-    marquee_settings = display_marquee_controls()
-    
+def create_paper_audio(papers, query):
+    combined = []
     for paper in papers:
-        content = f"📄 {paper['title']} | 👤 {paper['authors']} | 📝 {paper['summary']}"
-        streamlit_marquee(
-            content=content,
-            **marquee_settings,
-            key=f"paper_{paper['id'] or random.randint(1000,9999)}"
-        )
-        st.write("")
-    
-    return papers
+        try:
+            text = f"{paper['title']} by {paper['authors']}. {paper['summary']}"
+            file_format = st.session_state['audio_format']
+            audio_file = speak_text(text, file_format=file_format)
+            paper['audio'] = audio_file
+            st.write(f"### {FILE_EMOJIS.get(file_format, '')} {os.path.basename(audio_file)}")
+            st.audio(audio_file)
+            combined.append(paper['title'])
+        except Exception as e:
+            st.warning(f"Error generating audio for {paper['title']}: {str(e)}")
+
+    if combined:
+        summary = f"Found papers about: {'; '.join(combined)}. Query was: {query}"
+        summary_audio = speak_text(summary)
+        if summary_audio:
+            st.write("### 📢 Summary")
+            st.audio(summary_audio)
 
 def main():
-    marquee_settings = display_marquee_controls()
+    st.sidebar.markdown("### 🚲BikeAI🏆 Research Assistant")
     
-    streamlit_marquee(
-        content="🚀 Welcome to TalkingAIResearcher | 🤖 Your Research Assistant",
-        **marquee_settings,
-        key="welcome"
-    )
+    # Voice settings
+    st.sidebar.markdown("### 🎤 Voice Config")
+    voice = st.sidebar.selectbox("Voice:", EDGE_TTS_VOICES, 
+                               index=EDGE_TTS_VOICES.index(st.session_state['tts_voice']))
+    fmt = st.sidebar.radio("Format:", ["MP3", "WAV"], index=0)
     
-    tab = st.radio("Action:", ["🎤 Voice", "🔍 ArXiv", "📝 Editor"], horizontal=True)
+    if voice != st.session_state['tts_voice']:
+        st.session_state['tts_voice'] = voice
+        st.rerun()
+    if fmt.lower() != st.session_state['audio_format']:
+        st.session_state['audio_format'] = fmt.lower()
+        st.rerun()
+
+    mode = st.radio("Mode:", ["🎤 Voice", "🔍 ArXiv", "📝 Editor"], horizontal=True)
     
-    if tab == "🔍 ArXiv":
+    if mode == "🔍 ArXiv":
         query = st.text_input("🔍 Search:")
         if query:
-            papers = perform_ai_lookup(query)
-            st.write(f"Found {len(papers)} papers")
+            perform_arxiv_search(query)
+    
+    elif mode == "🎤 Voice":
+        text = st.text_area("Message:", height=100).strip()
+        if st.button("Send"):
+            process_with_gpt(text)
+        
+        st.subheader("History")
+        tab1, tab2 = st.tabs(["Claude", "GPT-4"])
+        with tab1:
+            for msg in st.session_state.chat_history:
+                st.write("You:", msg["user"])
+                st.write("Claude:", msg["claude"])
+        with tab2:
+            for msg in st.session_state.messages:
+                with st.chat_message(msg["role"]):
+                    st.markdown(msg["content"])
     
+    elif mode == "📝 Editor":
+        if st.session_state.current_file:
+            st.subheader(f"Editing: {st.session_state.current_file}")
+            new_content = st.text_area("Content:", st.session_state.file_content, height=300)
+            if st.button("Save"):
+                with open(st.session_state.current_file, 'w') as f:
+                    f.write(new_content)
+                st.success("Saved!")
+                st.session_state.should_rerun = True
+
     groups = load_files_for_sidebar()
     display_file_manager_sidebar(groups)
     
@@ -233,22 +316,20 @@ def main():
         st.session_state.should_rerun = False
         st.rerun()
 
-# Condensed sidebar markdown
 sidebar_md = """# 📚 Research Papers
-
-## 🧠 AGI Levels
+## AGI Levels
 L0 ❌ No AI
-L1 🌱 ChatGPT/Bard [2303.08774v1](https://arxiv.org/abs/2303.08774) [PDF](https://arxiv.org/pdf/2303.08774.pdf)
-L2 💪 Watson [2201.11903v1](https://arxiv.org/abs/2201.11903) [PDF](https://arxiv.org/pdf/2201.11903.pdf)
-L3 🎯 DALL·E [2204.06125v1](https://arxiv.org/abs/2204.06125) [PDF](https://arxiv.org/pdf/2204.06125.pdf)
-L4 🏆 AlphaGo [1712.01815v1](https://arxiv.org/abs/1712.01815) [PDF](https://arxiv.org/pdf/1712.01815.pdf) 
-L5 🚀 AlphaFold [2203.15556v1](https://arxiv.org/abs/2203.15556) [PDF](https://arxiv.org/pdf/2203.15556.pdf)
+L1 🌱 ChatGPT [2303.08774](https://arxiv.org/abs/2303.08774) | [PDF](https://arxiv.org/pdf/2303.08774.pdf)
+L2 💪 Watson [2201.11903](https://arxiv.org/abs/2201.11903) | [PDF](https://arxiv.org/pdf/2201.11903.pdf)
+L3 🎯 DALL·E [2204.06125](https://arxiv.org/abs/2204.06125) | [PDF](https://arxiv.org/pdf/2204.06125.pdf)
+L4 🏆 AlphaGo [1712.01815](https://arxiv.org/abs/1712.01815) | [PDF](https://arxiv.org/pdf/1712.01815.pdf)
+L5 🚀 AlphaFold [L5 🚀 AlphaFold [2203.15556](https://arxiv.org/abs/2203.15556) | [PDF](https://arxiv.org/pdf/2203.15556.pdf)
 
 ## 🧬 AlphaFold2 
-[2203.15556v1](https://arxiv.org/abs/2203.15556) [PDF](https://arxiv.org/pdf/2203.15556.pdf)
-1. 🧬 Input Seq → 2. 🔍 DB Search → 3. 🧩 MSA
-4. 📑 Templates → 5. 🔄 Evoformer → 6. 🧱 Structure
-7. 🎯 3D Predict → 8. ♻️ Recycle x3"""
+[2203.15556](https://arxiv.org/abs/2203.15556) | [PDF](https://arxiv.org/pdf/2203.15556.pdf)
+1. 🧬 Input → 2. 🔍 Search → 3. 🧩 MSA
+4. 📑 Templates → 5. 🔄 Evoformer → 6. 🧱 Structure 
+7. 🎯 3D Predict → 8. ♻️ Recycle"""
 
 st.sidebar.markdown(sidebar_md)