import streamlit as st import anthropic, openai, base64, cv2, glob, json, math, os, pytz, random, re, requests, textract, time, zipfile import plotly.graph_objects as go import streamlit.components.v1 as components from datetime import datetime from audio_recorder_streamlit import audio_recorder from bs4 import BeautifulSoup from collections import defaultdict, deque from dotenv import load_dotenv from gradio_client import Client from huggingface_hub import InferenceClient from io import BytesIO from PIL import Image from PyPDF2 import PdfReader from urllib.parse import quote from xml.etree import ElementTree as ET from openai import OpenAI import extra_streamlit_components as stx from streamlit.runtime.scriptrunner import get_script_run_ctx import asyncio import edge_tts # π§ Config & Setup st.set_page_config( page_title="π²BikeAIπ Claude/GPT Research", page_icon="π²π", layout="wide", initial_sidebar_state="auto", menu_items={ 'Get Help': 'https://huggingface.co/awacke1', 'Report a bug': 'https://huggingface.co/spaces/awacke1', 'About': "π²BikeAIπ Claude/GPT Research AI" } ) load_dotenv() openai_api_key = os.getenv('OPENAI_API_KEY', "") anthropic_key = os.getenv('ANTHROPIC_API_KEY_3', "") if 'OPENAI_API_KEY' in st.secrets: openai_api_key = st.secrets['OPENAI_API_KEY'] if 'ANTHROPIC_API_KEY' in st.secrets: anthropic_key = st.secrets["ANTHROPIC_API_KEY"] openai.api_key = openai_api_key claude_client = anthropic.Anthropic(api_key=anthropic_key) openai_client = OpenAI(api_key=openai.api_key, organization=os.getenv('OPENAI_ORG_ID')) HF_KEY = os.getenv('HF_KEY') API_URL = os.getenv('API_URL') if 'transcript_history' not in st.session_state: st.session_state['transcript_history'] = [] if 'chat_history' not in st.session_state: st.session_state['chat_history'] = [] if 'openai_model' not in st.session_state: st.session_state['openai_model'] = "gpt-4o-2024-05-13" if 'messages' not in st.session_state: st.session_state['messages'] = [] if 'last_voice_input' not in st.session_state: st.session_state['last_voice_input'] = "" if 'editing_file' not in st.session_state: st.session_state['editing_file'] = None if 'edit_new_name' not in st.session_state: st.session_state['edit_new_name'] = "" if 'edit_new_content' not in st.session_state: st.session_state['edit_new_content'] = "" if 'viewing_prefix' not in st.session_state: st.session_state['viewing_prefix'] = None if 'should_rerun' not in st.session_state: st.session_state['should_rerun'] = False # π¨ Minimal Custom CSS st.markdown(""" """, unsafe_allow_html=True) FILE_EMOJIS = { "md": "π", "mp3": "π΅", } def clean_for_speech(text: str) -> str: text = text.replace("\n", " ") text = text.replace("", " ") text = text.replace("#", "") # Remove links like (https://...) text = re.sub(r"\(https?:\/\/[^\)]+\)", "", text) text = re.sub(r"\s+", " ", text).strip() return text def generate_filename(content, file_type="md"): # Prefix: YYMM_HHmm_ -> total 10 chars including underscore # Actually: %y%m_%H%M gives 9 chars, add trailing underscore for total 10 chars. # Example: 23 09 _12 45 _ => '2309_1245_' prefix = datetime.now().strftime("%y%m_%H%M") + "_" # Extract some words from content words = re.findall(r"\w+", content) # Take first 3 words for filename segment name_text = '_'.join(words[:3]) if words else 'file' filename = f"{prefix}{name_text}.{file_type}" return filename def create_file(prompt, response, file_type="md"): # Decide which content to base the filename on (prefer response) base_content = response.strip() if response.strip() else prompt.strip() filename = generate_filename(base_content, file_type) with open(filename, 'w', encoding='utf-8') as f: f.write(prompt + "\n\n" + response) return filename def get_download_link(file): with open(file, "rb") as f: b64 = base64.b64encode(f.read()).decode() return f'π Download {os.path.basename(file)}' @st.cache_resource def speech_synthesis_html(result): html_code = f"""
""" components.html(html_code, height=0) async def edge_tts_generate_audio(text, voice="en-US-AriaNeural", rate=0, pitch=0): text = clean_for_speech(text) if not text.strip(): return None rate_str = f"{rate:+d}%" pitch_str = f"{pitch:+d}Hz" communicate = edge_tts.Communicate(text, voice, rate=rate_str, pitch=pitch_str) out_fn = generate_filename(text,"mp3") await communicate.save(out_fn) return out_fn def speak_with_edge_tts(text, voice="en-US-AriaNeural", rate=0, pitch=0): return asyncio.run(edge_tts_generate_audio(text, voice, rate, pitch)) def play_and_download_audio(file_path): if file_path and os.path.exists(file_path): st.audio(file_path) dl_link = f'Download {os.path.basename(file_path)}' st.markdown(dl_link, unsafe_allow_html=True) def process_image(image_path, user_prompt): with open(image_path, "rb") as imgf: image_data = imgf.read() b64img = base64.b64encode(image_data).decode("utf-8") resp = openai_client.chat.completions.create( model=st.session_state["openai_model"], messages=[ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": [ {"type": "text", "text": user_prompt}, {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{b64img}"}} ]} ], temperature=0.0, ) return resp.choices[0].message.content def process_audio(audio_path): with open(audio_path, "rb") as f: transcription = openai_client.audio.transcriptions.create(model="whisper-1", file=f) st.session_state.messages.append({"role": "user", "content": transcription.text}) return transcription.text def process_video(video_path, seconds_per_frame=1): vid = cv2.VideoCapture(video_path) total = int(vid.get(cv2.CAP_PROP_FRAME_COUNT)) fps = vid.get(cv2.CAP_PROP_FPS) skip = int(fps*seconds_per_frame) frames_b64 = [] for i in range(0, total, skip): vid.set(cv2.CAP_PROP_POS_FRAMES, i) ret, frame = vid.read() if not ret: break _, buf = cv2.imencode(".jpg", frame) frames_b64.append(base64.b64encode(buf).decode("utf-8")) vid.release() return frames_b64 def process_video_with_gpt(video_path, prompt): frames = process_video(video_path) resp = openai_client.chat.completions.create( model=st.session_state["openai_model"], messages=[ {"role":"system","content":"Analyze video frames."}, {"role":"user","content":[ {"type":"text","text":prompt}, *[{"type":"image_url","image_url":{"url":f"data:image/jpeg;base64,{fr}"}} for fr in frames] ]} ] ) return resp.choices[0].message.content def search_arxiv(query): st.write("π Searching ArXiv...") client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern") r1 = client.predict(prompt=query, llm_model_picked="mistralai/Mixtral-8x7B-Instruct-v0.1", stream_outputs=True, api_name="/ask_llm") st.markdown("### Mistral-8x7B-Instruct-v0.1 Result") st.markdown(r1) r2 = client.predict(prompt=query, llm_model_picked="mistralai/Mistral-7B-Instruct-v0.2", stream_outputs=True, api_name="/ask_llm") st.markdown("### Mistral-7B-Instruct-v0.2 Result") st.markdown(r2) return f"{r1}\n\n{r2}" def perform_ai_lookup(q, vocal_summary=True, extended_refs=False, titles_summary=True): start = time.time() client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern") r = client.predict(q,20,"Semantic Search","mistralai/Mixtral-8x7B-Instruct-v0.1",api_name="/update_with_rag_md") refs = r[0] r2 = client.predict(q,"mistralai/Mixtral-8x7B-Instruct-v0.1",True,api_name="/ask_llm") result = f"### π {q}\n\n{r2}\n\n{refs}" st.markdown(result) # Clean for speech before TTS if vocal_summary: main_text = clean_for_speech(r2) audio_file_main = speak_with_edge_tts(main_text) st.write("### ποΈ Vocal Summary (Short Answer)") play_and_download_audio(audio_file_main) if extended_refs: summaries_text = "Here are the summaries from the references: " + refs.replace('"','') summaries_text = clean_for_speech(summaries_text) audio_file_refs = speak_with_edge_tts(summaries_text) st.write("### π Extended References & Summaries") play_and_download_audio(audio_file_refs) if titles_summary: titles = [] for line in refs.split('\n'): m = re.search(r"\[([^\]]+)\]", line) if m: titles.append(m.group(1)) if titles: titles_text = "Here are the titles of the papers: " + ", ".join(titles) titles_text = clean_for_speech(titles_text) audio_file_titles = speak_with_edge_tts(titles_text) st.write("### π Paper Titles") play_and_download_audio(audio_file_titles) elapsed = time.time()-start st.write(f"**Total Elapsed:** {elapsed:.2f} s") # Create MD file from q and result create_file(q, result, "md") return result 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"): c = openai_client.chat.completions.create( model=st.session_state["openai_model"], messages=st.session_state.messages, stream=False ) ans = c.choices[0].message.content st.write("GPT-4o: " + ans) create_file(text, ans, "md") st.session_state.messages.append({"role":"assistant","content":ans}) return ans def process_with_claude(text): if not text: return with st.chat_message("user"): st.markdown(text) with st.chat_message("assistant"): r = claude_client.messages.create( model="claude-3-sonnet-20240229", max_tokens=1000, messages=[{"role":"user","content":text}] ) ans = r.content[0].text st.write("Claude: " + ans) create_file(text, ans, "md") st.session_state.chat_history.append({"user":text,"claude":ans}) return ans def create_zip_of_files(md_files, mp3_files): # Exclude README.md md_files = [f for f in md_files if os.path.basename(f).lower() != 'readme.md'] all_files = md_files + mp3_files if not all_files: return None # Build a descriptive name stems = [os.path.splitext(os.path.basename(f))[0] for f in all_files] joined = "_".join(stems) if len(joined) > 50: joined = joined[:50] + "_etc" zip_name = f"{joined}.zip" with zipfile.ZipFile(zip_name,'w') as z: for f in all_files: z.write(f) return zip_name def load_files_for_sidebar(): # Gather files md_files = glob.glob("*.md") mp3_files = glob.glob("*.mp3") # Exclude README.md md_files = [f for f in md_files if os.path.basename(f).lower() != 'readme.md'] all_files = md_files + mp3_files # Group by first 10 chars of filename groups = defaultdict(list) for f in all_files: fname = os.path.basename(f) prefix = fname[:10] # first 10 chars as group prefix groups[prefix].append(f) # Sort files in each group by mod time descending for prefix in groups: groups[prefix].sort(key=lambda x: os.path.getmtime(x), reverse=True) # Sort prefixes by newest file time sorted_prefixes = sorted(groups.keys(), key=lambda pre: max(os.path.getmtime(x) for x in groups[pre]), reverse=True) return groups, sorted_prefixes def extract_keywords_from_md(files): # Combine all MD content text = "" for f in files: if f.endswith(".md"): c = open(f,'r',encoding='utf-8').read() text += " " + c # Extract first 5 unique words words = re.findall(r"\w+", text.lower()) unique_words = [] for w in words: if w not in unique_words: unique_words.append(w) if len(unique_words) == 5: break return unique_words def display_file_manager_sidebar(groups, sorted_prefixes): st.sidebar.title("π΅ Audio & Document Manager") # Collect all md and mp3 files for zip operations all_md = [] all_mp3 = [] for prefix in groups: for f in groups[prefix]: if f.endswith(".md"): all_md.append(f) elif f.endswith(".mp3"): all_mp3.append(f) top_bar = st.sidebar.columns(3) with top_bar[0]: if st.button("π Del All MD"): for f in all_md: os.remove(f) st.session_state.should_rerun = True with top_bar[1]: if st.button("π Del All MP3"): for f in all_mp3: os.remove(f) st.session_state.should_rerun = True with top_bar[2]: if st.button("β¬οΈ Zip All"): z = create_zip_of_files(all_md, all_mp3) if z: st.sidebar.markdown(get_download_link(z),unsafe_allow_html=True) for prefix in sorted_prefixes: files = groups[prefix] # Extract 5-word keywords from MD in this group kw = extract_keywords_from_md(files) keywords_str = " ".join(kw) if kw else "No Keywords" with st.sidebar.expander(f"{prefix} Files ({len(files)}) - Keywords: {keywords_str}", expanded=True): # Delete group / View group c1,c2 = st.columns(2) with c1: if st.button("πView Group", key="view_group_"+prefix): st.session_state.viewing_prefix = prefix # No rerun needed, just state update with c2: if st.button("πDel Group", key="del_group_"+prefix): for f in files: os.remove(f) st.session_state.should_rerun = True for f in files: fname = os.path.basename(f) ctime = datetime.fromtimestamp(os.path.getmtime(f)).strftime("%Y-%m-%d %H:%M:%S") ext = os.path.splitext(fname)[1].lower().strip('.') st.write(f"**{fname}** - {ctime}") # Individual file actions are less necessary if we have group actions # But we can still provide them if desired. # The user requested grouping primarily, but we can keep minimal file actions if needed. # In instructions now, main focus is group view/delete. # We'll omit individual file view/edit here since we have group view. # If needed, re-add them similarly as before. # For now, rely on "View Group" to see all files. def main(): st.sidebar.markdown("### π²BikeAIπ Multi-Agent Research AI") tab_main = st.radio("Action:",["π€ Voice Input","πΈ Media Gallery","π Search ArXiv","π File Editor"],horizontal=True) model_choice = st.sidebar.radio("AI Model:", ["Arxiv","GPT-4o","Claude-3","GPT+Claude+Arxiv"], index=0) mycomponent = components.declare_component("mycomponent", path="mycomponent") val = mycomponent(my_input_value="Hello") if val: user_input = val.strip() if user_input: if model_choice == "GPT-4o": process_with_gpt(user_input) elif model_choice == "Claude-3": process_with_claude(user_input) elif model_choice == "Arxiv": st.subheader("Arxiv Only Results:") perform_ai_lookup(user_input, vocal_summary=True, extended_refs=False, titles_summary=True) else: col1,col2,col3=st.columns(3) with col1: st.subheader("GPT-4o Omni:") try: process_with_gpt(user_input) except: st.write('GPT 4o error') with col2: st.subheader("Claude-3 Sonnet:") try: process_with_claude(user_input) except: st.write('Claude error') with col3: st.subheader("Arxiv + Mistral:") try: perform_ai_lookup(user_input, vocal_summary=True, extended_refs=False, titles_summary=True) except: st.write("Arxiv error") if tab_main == "π Search ArXiv": st.subheader("π Search ArXiv") q=st.text_input("Research query:") st.markdown("### ποΈ Audio Generation Options") vocal_summary = st.checkbox("ποΈ Vocal Summary (Short Answer)", value=True) extended_refs = st.checkbox("π Extended References & Summaries (Long)", value=False) titles_summary = st.checkbox("π Paper Titles Only", value=True) if q and st.button("Run ArXiv Query"): perform_ai_lookup(q, vocal_summary=vocal_summary, extended_refs=extended_refs, titles_summary=titles_summary) elif tab_main == "π€ Voice Input": st.subheader("π€ Voice Recognition") user_text = st.text_area("Message:", height=100) user_text = user_text.strip() if st.button("Send π¨"): if user_text: if model_choice == "GPT-4o": process_with_gpt(user_text) elif model_choice == "Claude-3": process_with_claude(user_text) elif model_choice == "Arxiv": st.subheader("Arxiv Only Results:") perform_ai_lookup(user_text, vocal_summary=True, extended_refs=False, titles_summary=True) else: col1,col2,col3=st.columns(3) with col1: st.subheader("GPT-4o Omni:") process_with_gpt(user_text) with col2: st.subheader("Claude-3 Sonnet:") process_with_claude(user_text) with col3: st.subheader("Arxiv & Mistral:") res = perform_ai_lookup(user_text, vocal_summary=True, extended_refs=False, titles_summary=True) st.markdown(res) st.subheader("π Chat History") t1,t2=st.tabs(["Claude History","GPT-4o History"]) with t1: for c in st.session_state.chat_history: st.write("**You:**", c["user"]) st.write("**Claude:**", c["claude"]) with t2: for m in st.session_state.messages: with st.chat_message(m["role"]): st.markdown(m["content"]) elif tab_main == "πΈ Media Gallery": st.header("π¬ Media Gallery - Images and Videos") tabs = st.tabs(["πΌοΈ Images", "π₯ Video"]) with tabs[0]: imgs = glob.glob("*.png")+glob.glob("*.jpg") if imgs: c = st.slider("Cols",1,5,3) cols = st.columns(c) for i,f in enumerate(imgs): with cols[i%c]: st.image(Image.open(f),use_container_width=True) if st.button(f"π Analyze {os.path.basename(f)}", key=f"analyze_{f}"): a = process_image(f,"Describe this image.") st.markdown(a) else: st.write("No images found.") with tabs[1]: vids = glob.glob("*.mp4") if vids: for v in vids: with st.expander(f"π₯ {os.path.basename(v)}"): st.markdown(get_media_html(v,"video"),unsafe_allow_html=True) if st.button(f"Analyze {os.path.basename(v)}", key=f"analyze_{v}"): a = process_video_with_gpt(v,"Describe video.") st.markdown(a) else: st.write("No videos found.") elif tab_main == "π File Editor": if getattr(st.session_state,'current_file',None): st.subheader(f"Editing: {st.session_state.current_file}") new_text = st.text_area("Content:", st.session_state.file_content, height=300) if st.button("Save"): with open(st.session_state.current_file,'w',encoding='utf-8') as f: f.write(new_text) st.success("Updated!") st.session_state.should_rerun = True else: st.write("Select a file from the sidebar to edit.") # After main content, load and show file groups in sidebar groups, sorted_prefixes = load_files_for_sidebar() display_file_manager_sidebar(groups, sorted_prefixes) # If viewing a prefix group, show all files in main area if st.session_state.viewing_prefix and st.session_state.viewing_prefix in groups: st.write("---") st.write(f"**Viewing Group:** {st.session_state.viewing_prefix}") # Show all files in this prefix group in order (mp3 and md) # Sort by mod time descending (already sorted) for f in groups[st.session_state.viewing_prefix]: fname = os.path.basename(f) ext = os.path.splitext(fname)[1].lower().strip('.') st.write(f"### {fname}") if ext == "md": content = open(f,'r',encoding='utf-8').read() st.markdown(content) elif ext == "mp3": st.audio(f) else: # just show a download link st.markdown(get_download_link(f), unsafe_allow_html=True) if st.button("Close Group View"): st.session_state.viewing_prefix = None if st.session_state.should_rerun: st.session_state.should_rerun = False st.rerun() if __name__=="__main__": main()