app.py

#!/usr/bin/env python3
"""
Anton Microscopy Analysis - Clean UI
"""

import streamlit as st
import os
import sys
from pathlib import Path
import numpy as np
from PIL import Image
import random
import logging
import traceback

# Setup logging
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)

# Setup page
st.set_page_config(
    page_title="Anton Microscopy Analysis", 
    page_icon="🔬", 
    layout="centered"
)

# Clean, modern CSS
st.markdown("""
<style>
    /* Hide Streamlit branding */
    #MainMenu {visibility: hidden;}
    footer {visibility: hidden;}
    header {visibility: hidden;}
    
    /* Clean typography */
    .main-header {
        font-size: 2.5rem;
        font-weight: 300;
        color: #1f2937;
        text-align: center;
        margin-bottom: 3rem;
        letter-spacing: -0.02em;
    }
    
    .section-header {
        font-size: 1.1rem;
        font-weight: 500;
        color: #374151;
        margin-bottom: 1rem;
    }
    
    /* Clean upload area */
    .upload-section {
        background: #f8fafc;
        border: 2px dashed #e2e8f0;
        border-radius: 12px;
        padding: 2rem;
        text-align: center;
        margin-bottom: 2rem;
    }
    
    /* Primary button styling */
    .stButton > button {
        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
        color: white;
        border: none;
        border-radius: 8px;
        padding: 0.75rem 2rem;
        font-weight: 500;
        font-size: 1rem;
        width: 100%;
        transition: all 0.2s ease;
        box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
    }
    
    .stButton > button:hover {
        transform: translateY(-1px);
        box-shadow: 0 8px 15px -3px rgba(0, 0, 0, 0.1);
    }
    
    /* Sample selection */
    .sample-selector {
        background: white;
        border-radius: 8px;
        padding: 1rem;
        box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
        margin-bottom: 1rem;
    }
    
    /* Results styling */
    .results-container {
        background: white;
        border-radius: 12px;
        padding: 1.5rem;
        box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
        margin-top: 2rem;
    }
    
    /* Image display */
    .image-container {
        border-radius: 8px;
        overflow: hidden;
        box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
    }
    
    /* Clean spacing */
    .block-container {
        padding-top: 2rem;
        max-width: 1000px;
    }
</style>
""", unsafe_allow_html=True)

# Add Anton to path
sys.path.append(str(Path(__file__).parent))

# Import Anton components
try:
    from anton.core.pipeline import AnalysisPipeline
    from anton.utils.image_io import ImageLoader
    # CMPO mapping removed for simplicity
    anton_available = True
except ImportError as e:
    anton_available = False
    import_error = str(e)

# Main header
st.markdown('<h1 class="main-header">🔬 Anton Microscopy Analysis</h1>', unsafe_allow_html=True)

if not anton_available:
    st.error(f"System Error: {import_error}")
    st.stop()

# Image input section
st.markdown('<div class="section-header">Select Image</div>', unsafe_allow_html=True)

# Check for sample images
sample_images_path = Path("sample_images")
sample_images = sorted(list(sample_images_path.glob("*.BMP"))) if sample_images_path.exists() else []

# Initialize session state
if 'selected_sample_image' not in st.session_state:
    st.session_state.selected_sample_image = sample_images[0].name if sample_images else None

# Image source selection
col1, col2 = st.columns(2)

with col1:
    if sample_images:
        if st.button("🎲 Random Sample", use_container_width=True):
            st.session_state.selected_sample_image = random.choice(sample_images).name
            st.rerun()

with col2:
    use_sample = st.toggle("Use Sample Images", value=bool(sample_images))

# Image selection
current_image = None
image_to_analyze = None

if use_sample and sample_images:
    # Sample image selection
    selected_image = st.selectbox(
        "Choose sample image:",
        [img.name for img in sample_images],
        index=[img.name for img in sample_images].index(st.session_state.selected_sample_image) if st.session_state.selected_sample_image in [img.name for img in sample_images] else 0,
        label_visibility="collapsed"
    )
    
    if selected_image != st.session_state.selected_sample_image:
        st.session_state.selected_sample_image = selected_image
    
    # Load sample image
    sample_image_path = sample_images_path / selected_image
    if sample_image_path.exists():
        try:
            loader = ImageLoader()
            current_image = loader.load(str(sample_image_path))
            image_to_analyze = str(sample_image_path)
        except Exception as e:
            st.error(f"Error loading image: {e}")
else:
    # File upload
    uploaded_file = st.file_uploader(
        "Upload microscopy image",
        type=['png', 'jpg', 'jpeg', 'tiff', 'bmp'],
        label_visibility="collapsed"
    )
    
    if uploaded_file:
        try:
            import tempfile
            with tempfile.NamedTemporaryFile(delete=False, suffix=f"_{uploaded_file.name}") as tmp_file:
                tmp_file.write(uploaded_file.getbuffer())
                temp_path = tmp_file.name
            
            try:
                loader = ImageLoader()
                current_image = loader.load(temp_path)
            except:
                pil_img = Image.open(temp_path)
                current_image = np.array(pil_img)
                if len(current_image.shape) == 3 and current_image.shape[2] == 3:
                    current_image = np.mean(current_image, axis=2).astype(np.uint8)
            
            image_to_analyze = temp_path
        except Exception as e:
            st.error(f"Error loading image: {e}")

# Display image
if current_image is not None:
    st.markdown('<div class="image-container">', unsafe_allow_html=True)
    st.image(current_image, use_container_width=True)
    st.markdown('</div>', unsafe_allow_html=True)
    
    # Analysis button
    st.markdown("---")
    analyze_btn = st.button("🚀 Analyze Image", type="primary", use_container_width=True)
    
    # Run analysis
    if analyze_btn:
        logger.info("Analysis button pressed")
        st.info("Starting analysis...")
        
        # Check API keys
        vlm_provider = "mock"
        if os.getenv('GOOGLE_API_KEY'):
            vlm_provider = "gemini"
            logger.info("Using Gemini VLM provider")
        elif os.getenv('ANTHROPIC_API_KEY'):
            vlm_provider = "claude"
            logger.info("Using Claude VLM provider")
        else:
            logger.info("Using mock VLM provider (no API keys found)")
        
        # Configure pipeline
        if use_sample and sample_images:
            biological_context = {
                "experiment_type": "protein_translocation",
                "cell_line": "U2OS_osteosarcoma", 
                "protein": "FKHR-GFP",
                "readout": "nuclear_vs_cytoplasmic_localization"
            }
            logger.info("Using sample image context: protein translocation")
        else:
            biological_context = {
                "experiment_type": "general_microscopy",
                "readout": "cellular_morphology_and_phenotypes"
            }
            logger.info("Using general microscopy context")
        
        config = {
            "vlm_provider": vlm_provider,
            "biological_context": biological_context
        }
        logger.info(f"Pipeline config: {config}")
        
        # Run analysis with progressive display
        logger.info(f"Creating pipeline with image: {image_to_analyze}")
        pipeline = AnalysisPipeline(config)
        logger.info("Pipeline created, starting progressive analysis...")
        
        # Create containers for each stage
        st.markdown('<div class="results-container">', unsafe_allow_html=True)
        st.markdown("### Analysis Results")
        
        stage_containers = {}
        stage_statuses = {}
        
        stages = [
            ("Global Analysis", "stage_1_global"),
            ("Object Detection", "stage_2_objects"), 
            ("Feature Analysis", "stage_3_features"),
            ("Population Analysis", "stage_4_population")
        ]
        
        # Initialize containers
        for stage_name, stage_key in stages:
            stage_containers[stage_key] = st.empty()
            stage_statuses[stage_key] = st.empty()
        
        def update_stage_display(stage_key, status, data):
            """Update the display for a specific stage."""
            stage_name = next(name for name, key in stages if key == stage_key)
            
            if status == "running":
                stage_statuses[stage_key].info(f"⏳ {stage_name}: {data}")
                stage_containers[stage_key].expander(f"📋 {stage_name}", expanded=False).write("Processing...")
            elif status == "completed":
                stage_statuses[stage_key].success(f"✅ {stage_name}: Completed")
                
                # Extract content from stage data  
                if isinstance(data, dict):
                    logger.info(f"Stage {stage_key} data keys: {list(data.keys())}")
                    
                    # Handle different stage result formats
                    content = data.get('description')
                    if not content:
                        content = data.get('segmentation_guidance')
                    if not content:
                        content = data.get('population_summary')
                    if not content:
                        # Stage 3 feature_descriptions is a list
                        feature_descs = data.get('feature_descriptions', [])
                        logger.info(f"Stage 3 feature_descriptions: {feature_descs}")
                        if feature_descs and isinstance(feature_descs, list):
                            content = '\n'.join(str(desc) for desc in feature_descs)
                        else:
                            content = 'Analysis completed'
                    
                    logger.info(f"Stage {stage_key} final content: {content[:100]}...")
                    
                    if content and content.startswith('```'):
                        lines = content.split('\n')
                        content = '\n'.join([line for line in lines if not line.strip().startswith('```')])
                    
                    # Display in expandable container
                    with stage_containers[stage_key].expander(f"📋 {stage_name}", expanded=True):
                        st.write(content[:1000] + "..." if len(content) > 1000 else content)
                else:
                    stage_containers[stage_key].expander(f"📋 {stage_name}", expanded=True).write(str(data))
        
        try:
            # Create callback to handle stage updates
            def stage_callback(stage, status, data):
                # Map stage names: stage_1 -> stage_1_global
                stage_mapping = {
                    "stage_1": "stage_1_global",
                    "stage_2": "stage_2_objects",
                    "stage_3": "stage_3_features",
                    "stage_4": "stage_4_population"
                }
                mapped_stage = stage_mapping.get(stage, stage)
                update_stage_display(mapped_stage, status, data)
            
            # Run progressive analysis
            results = pipeline.run_pipeline_progressive(
                image_to_analyze, 
                callback=stage_callback
            )
            logger.info(f"Progressive analysis completed. Results keys: {list(results.keys()) if results else 'None'}")
            
            # Clear status messages after completion
            for stage_key in stage_statuses:
                stage_statuses[stage_key].empty()
            
            logger.info("Progressive analysis results displayed successfully")
            
        except Exception as e:
            # Clear any running status messages on error
            for stage_key in stage_statuses:
                stage_statuses[stage_key].empty()
            error_msg = f"Analysis failed: {str(e)}"
            logger.error(error_msg)
            logger.error(f"Full traceback: {traceback.format_exc()}")
            st.error(error_msg)
            st.code(traceback.format_exc())
        
        st.markdown('</div>', unsafe_allow_html=True)
        
        # Cleanup
        if 'temp_path' in locals() and os.path.exists(temp_path):
            try:
                os.remove(temp_path)
            except:
                pass
else:
    st.markdown('<div class="upload-section">', unsafe_allow_html=True)
    st.markdown("**Upload an image or select a sample to begin analysis**")
    st.markdown('</div>', unsafe_allow_html=True)