Spaces:
Sleeping
Sleeping
# ------------------ Import ------------------ | |
import streamlit as st | |
import os | |
import time | |
import re | |
import requests | |
from PIL import Image | |
from io import BytesIO | |
from urllib.parse import quote | |
from openai import OpenAI | |
# ------------------ Authentication ------------------ | |
VALID_USERS = { | |
"[email protected]": "Pass.123", | |
"[email protected]": "Pass.123", | |
"[email protected]": "Pass.123", | |
"[email protected]": "Pass.123", | |
} | |
def login(): | |
st.title("π Login Required") | |
email = st.text_input("Email") | |
password = st.text_input("Password", type="password") | |
if st.button("Login"): | |
if VALID_USERS.get(email) == password: | |
st.session_state.authenticated = True | |
st.rerun() | |
else: | |
st.error("β Incorrect email or password.") | |
if "authenticated" not in st.session_state: | |
st.session_state.authenticated = False | |
if not st.session_state.authenticated: | |
login() | |
st.stop() | |
# ------------------ Configuration ------------------ | |
st.set_page_config(page_title="AI Pathology Assistant", layout="wide", initial_sidebar_state="collapsed") | |
st.title("𧬠AI Pathology Assistant") | |
st.caption("AI-powered exploration of pathology, anatomy, and histology documents via OCR + GPT") | |
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY") | |
if not OPENAI_API_KEY: | |
st.error("β Missing OPENAI_API_KEY environment variable.") | |
st.stop() | |
client = OpenAI(api_key=OPENAI_API_KEY) | |
ASSISTANT_ID = "asst_jXDSjCG8LI4HEaFEcjFVq8KB" | |
# ------------------ State Setup ------------------ | |
if "messages" not in st.session_state: | |
st.session_state.messages = [] | |
if "thread_id" not in st.session_state: | |
st.session_state.thread_id = None | |
if "image_urls" not in st.session_state: | |
st.session_state.image_urls = [] | |
if "pending_prompt" not in st.session_state: | |
st.session_state.pending_prompt = None | |
# ------------------ Sidebar ------------------ | |
with st.sidebar: | |
st.header("π§ͺ Pathology Tools") | |
if st.button("π§Ή Clear Chat"): | |
st.session_state.messages = [] | |
st.session_state.thread_id = None | |
st.session_state.image_urls = [] | |
st.session_state.pending_prompt = None | |
st.rerun() | |
show_image = st.toggle("πΈ Show Images", value=True) | |
keyword = st.text_input("Keyword Search", placeholder="e.g. mitosis, carcinoma") | |
if st.button("π Search") and keyword: | |
st.session_state.pending_prompt = f"Find clauses or references related to: {keyword}" | |
section = st.text_input("Section Lookup", placeholder="e.g. Connective Tissue") | |
if section: | |
st.session_state.pending_prompt = f"Summarize or list key points from section: {section}" | |
actions = [ | |
"Select an action...", | |
"List histological features of inflammation", | |
"Summarize features of carcinoma", | |
"List muscle types and features", | |
"Extract diagnostic markers", | |
"Summarize embryology stages" | |
] | |
action = st.selectbox("Common Pathology Queries", actions) | |
if action != actions[0]: | |
st.session_state.pending_prompt = action | |
# ------------------ Chat UI ------------------ | |
chat_col, image_col = st.columns([2, 1]) | |
with chat_col: | |
st.markdown("### π¬ Ask a Pathology-Specific Question") | |
user_input = st.chat_input("Example: What are features of squamous cell carcinoma?") | |
if user_input: | |
st.session_state.messages.append({"role": "user", "content": user_input}) | |
elif st.session_state.pending_prompt: | |
st.session_state.messages.append({"role": "user", "content": st.session_state.pending_prompt}) | |
st.session_state.pending_prompt = None | |
if st.session_state.messages and st.session_state.messages[-1]["role"] == "user": | |
try: | |
if st.session_state.thread_id is None: | |
thread = client.beta.threads.create() | |
st.session_state.thread_id = thread.id | |
client.beta.threads.messages.create( | |
thread_id=st.session_state.thread_id, | |
role="user", | |
content=st.session_state.messages[-1]["content"] | |
) | |
run = client.beta.threads.runs.create( | |
thread_id=st.session_state.thread_id, | |
assistant_id=ASSISTANT_ID | |
) | |
with st.spinner("π¬ Analyzing..."): | |
while True: | |
status = client.beta.threads.runs.retrieve(thread_id=st.session_state.thread_id, run_id=run.id) | |
if status.status in ("completed", "failed", "cancelled"): | |
break | |
time.sleep(1) | |
if status.status == "completed": | |
messages = client.beta.threads.messages.list(thread_id=st.session_state.thread_id) | |
for m in reversed(messages.data): | |
if m.role == "assistant": | |
reply = m.content[0].text.value | |
st.session_state.messages.append({"role": "assistant", "content": reply}) | |
matches = re.findall( | |
r'https://raw\.githubusercontent\.com/AndrewLORTech/witspathologai/main/[^\s]+\.png', | |
reply | |
) | |
st.session_state.image_urls = matches | |
break | |
else: | |
st.error("β Assistant failed to respond.") | |
st.rerun() | |
except Exception as e: | |
st.error(f"β Error: {e}") | |
for msg in st.session_state.messages: | |
with st.chat_message(msg["role"]): | |
st.markdown(msg["content"], unsafe_allow_html=True) | |
# ------------------ Scrollable Image Carousel ------------------ | |
with image_col: | |
if show_image and st.session_state.image_urls: | |
st.markdown("### πΌοΈ Image(s)") | |
st.markdown(""" | |
<style> | |
.carousel-wrapper { | |
display: flex; | |
overflow-x: auto; | |
scroll-snap-type: x mandatory; | |
gap: 1rem; | |
padding: 0.5rem 0; | |
} | |
.carousel-wrapper::-webkit-scrollbar { | |
height: 10px; | |
} | |
.carousel-wrapper::-webkit-scrollbar-thumb { | |
background: #888; | |
border-radius: 5px; | |
} | |
.carousel-wrapper img { | |
scroll-snap-align: start; | |
height: 500px; | |
border-radius: 8px; | |
} | |
</style> | |
""", unsafe_allow_html=True) | |
html = '<div class="carousel-wrapper">' | |
for raw_url in st.session_state.image_urls: | |
try: | |
if "githubusercontent.com" not in raw_url: | |
continue | |
_, raw_path = raw_url.split("githubusercontent.com/", 1) | |
segments = raw_path.strip().split("/") | |
encoded_segments = [quote(seg) for seg in segments] | |
encoded_url = "https://raw.githubusercontent.com/" + "/".join(encoded_segments) | |
html += f'<img src="{encoded_url}" alt="slide preview">' | |
except Exception as e: | |
st.error(f"β Failed to load image: {e}") | |
html += "</div>" | |
st.markdown(html, unsafe_allow_html=True) | |