app.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Oklahoma Damage Assessment System with Comprehensive Sentiment Analysis
Optimized for Hugging Face Spaces deployment
"""

import streamlit as st
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go
from datetime import datetime, timedelta
import random
import json
from typing import Dict, List, Tuple, Optional
import time
import os

# Hugging Face optimization
os.environ['STREAMLIT_BROWSER_GATHER_USAGE_STATS'] = 'false'

# Page configuration - MUST be first Streamlit command
st.set_page_config(
    page_title="🌪️ Oklahoma Damage Assessment",
    page_icon="⚠️",
    layout="wide",
    initial_sidebar_state="expanded",
    menu_items={
        'Get Help': 'https://github.com/your-repo/oklahoma-damage-assessment',
        'Report a bug': 'https://github.com/your-repo/oklahoma-damage-assessment/issues',
        'About': "# Oklahoma Damage Assessment System\nComprehensive disaster impact analysis with sentiment analysis!"
    }
)

# Enhanced CSS for better Hugging Face compatibility
def load_css():
    st.markdown("""
    <style>
    .main-header {
        background: linear-gradient(135deg, #ff7b7b 0%, #d4526e 100%);
        color: white;
        padding: 2rem;
        border-radius: 15px;
        text-align: center;
        margin-bottom: 2rem;
        box-shadow: 0 8px 32px rgba(31, 38, 135, 0.37);
    }
    
    .sentiment-card {
        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
        color: white;
        padding: 1.5rem;
        border-radius: 15px;
        margin: 1rem 0;
        box-shadow: 0 8px 32px rgba(31, 38, 135, 0.37);
    }
    
    .metric-box {
        background: rgba(255, 255, 255, 0.9);
        border: 1px solid #dee2e6;
        padding: 1rem;
        border-radius: 10px;
        text-align: center;
        margin: 0.5rem 0;
        box-shadow: 0 2px 4px rgba(0,0,0,0.1);
    }
    
    .damage-alert {
        background: linear-gradient(135deg, #ff9a56 0%, #ff6b35 100%);
        border-left: 4px solid #ff4500;
        padding: 1.5rem;
        margin: 1rem 0;
        border-radius: 10px;
        color: white;
        box-shadow: 0 4px 15px rgba(255, 107, 53, 0.4);
    }
    
    .psychological-panel {
        background: linear-gradient(135deg, #a8edea 0%, #fed6e3 100%);
        padding: 2rem;
        border-radius: 15px;
        margin: 1rem 0;
        border: 1px solid rgba(255, 255, 255, 0.3);
    }
    
    .stMetric {
        background: white;
        padding: 1rem;
        border-radius: 10px;
        border: 1px solid #e0e0e0;
    }
    
    /* Fix for Hugging Face iframe */
    .main .block-container {
        padding-top: 1rem;
        padding-bottom: 1rem;
    }
    
    /* Responsive design */
    @media (max-width: 768px) {
        .main-header h1 { font-size: 1.5rem; }
        .sentiment-card { padding: 1rem; }
    }
    </style>
    """, unsafe_allow_html=True)

# Data Management with Closures (Simplified for HF)
@st.cache_data
def get_location_data():
    """Get Oklahoma location data with closure pattern"""
    counties = [
        'Oklahoma', 'Tulsa', 'Cleveland', 'Canadian', 'Comanche', 'Garfield',
        'Rogers', 'Washington', 'Pottawatomie', 'Creek', 'Kay', 'Carter',
        'Payne', 'McClain', 'Grady', 'Stephens', 'Pontotoc', 'Lincoln'
    ]
    
    cities = [
        'Oklahoma City', 'Tulsa', 'Norman', 'Lawton', 'Edmond', 'Moore',
        'Midwest City', 'Enid', 'Stillwater', 'Muskogee', 'Bartlesville',
        'Owasso', 'Shawnee', 'Yukon', 'Ardmore', 'Ponca City'
    ]
    
    return {
        'counties': counties,
        'cities': cities
    }

@st.cache_data
def get_sentiment_lexicon():
    """Get emotional lexicon for sentiment analysis"""
    return {
        'devastating': {'sentiment': -0.95, 'emotion': 'despair', 'intensity': 'extreme'},
        'catastrophic': {'sentiment': -0.90, 'emotion': 'terror', 'intensity': 'extreme'},
        'destroyed': {'sentiment': -0.85, 'emotion': 'grief', 'intensity': 'high'},
        'damaged': {'sentiment': -0.65, 'emotion': 'distress', 'intensity': 'moderate'},
        'threatened': {'sentiment': -0.60, 'emotion': 'anxiety', 'intensity': 'moderate'},
        'displaced': {'sentiment': -0.75, 'emotion': 'fear', 'intensity': 'high'},
        'injured': {'sentiment': -0.80, 'emotion': 'pain', 'intensity': 'high'},
        'affected': {'sentiment': -0.45, 'emotion': 'concern', 'intensity': 'moderate'},
        'recovering': {'sentiment': 0.50, 'emotion': 'hope', 'intensity': 'moderate'},
        'resilient': {'sentiment': 0.70, 'emotion': 'strength', 'intensity': 'high'},
        'helping': {'sentiment': 0.60, 'emotion': 'compassion', 'intensity': 'moderate'},
        'support': {'sentiment': 0.55, 'emotion': 'comfort', 'intensity': 'moderate'}
    }

def analyze_sentiment(text):
    """Simplified sentiment analysis"""
    lexicon = get_sentiment_lexicon()
    words = text.lower().split()
    
    sentiments = []
    emotions = {}
    
    for word in words:
        clean_word = ''.join(c for c in word if c.isalnum())
        if clean_word in lexicon:
            data = lexicon[clean_word]
            sentiments.append(data['sentiment'])
            emotions[data['emotion']] = emotions.get(data['emotion'], 0) + 1
    
    avg_sentiment = np.mean(sentiments) if sentiments else 0.0
    dominant_emotion = max(emotions.items(), key=lambda x: x[1])[0] if emotions else 'neutral'
    
    # Classification
    if avg_sentiment <= -0.6:
        classification = 'highly_negative'
    elif avg_sentiment <= -0.3:
        classification = 'negative'
    elif avg_sentiment <= 0.3:
        classification = 'neutral'
    else:
        classification = 'positive'
    
    return {
        'sentiment_score': avg_sentiment,
        'classification': classification,
        'dominant_emotion': dominant_emotion,
        'emotion_distribution': emotions
    }

def generate_damage_data(location):
    """Generate comprehensive damage data"""
    damage_types = [
        {'id': 'drought', 'name': 'Drought', 'color': '#ff8c00', 'icon': '🌵'},
        {'id': 'flood', 'name': 'Flood', 'color': '#1e90ff', 'icon': '🌊'},
        {'id': 'tornado', 'name': 'Tornado', 'color': '#dc143c', 'icon': '🌪️'},
        {'id': 'wildfire', 'name': 'Wildfire', 'color': '#ff4500', 'icon': '🔥'},
        {'id': 'hail', 'name': 'Hail Storm', 'color': '#87ceeb', 'icon': '🌧️'},
        {'id': 'wind', 'name': 'High Winds', 'color': '#a9a9a9', 'icon': '💨'}
    ]
    
    severity_levels = [
        {'id': 1, 'name': 'Minimal', 'color': '#fff3cd'},
        {'id': 2, 'name': 'Minor', 'color': '#ffeeba'},
        {'id': 3, 'name': 'Moderate', 'color': '#ffdf7e'},
        {'id': 4, 'name': 'Severe', 'color': '#ffcc00'},
        {'id': 5, 'name': 'Extreme', 'color': '#ff8c00'}
    ]
    
    damage_type = random.choice(damage_types)
    severity = random.choice(severity_levels)
    
    impact_templates = {
        'drought': "Agricultural operations in {location} are experiencing {severity_adj} stress, causing {emotion} in farming communities",
        'flood': "Flash flooding in {location} has {severity_adj} damaged infrastructure, creating {emotion} among residents",
        'tornado': "Tornado touchdown in {location} has {severity_adj} damaged structures, leaving survivors with {emotion}",
        'wildfire': "Wildfire near {location} has {severity_adj} threatened areas, forcing evacuations and causing {emotion}",
        'hail': "Severe hail storm in {location} has {severity_adj} damaged property, creating {emotion} among homeowners",
        'wind': "High winds in {location} have {severity_adj} damaged infrastructure, causing {emotion} in the community"
    }
    
    severity_adjectives = {
        1: 'slightly', 2: 'moderately', 3: 'significantly', 4: 'severely', 5: 'catastrophically'
    }
    
    emotions = {
        1: 'mild concern', 2: 'growing anxiety', 3: 'widespread distress', 
        4: 'severe trauma', 5: 'overwhelming devastation'
    }
    
    impact_text = impact_templates[damage_type['id']].format(
        location=location,
        severity_adj=severity_adjectives[severity['id']],
        emotion=emotions[severity['id']]
    )
    
    base_multiplier = severity['id']
    population_factor = 1.5 if 'City' in location else 1.0
    
    assessments = {
        'people': {
            'injured': int(random.randint(0, 20) * base_multiplier),
            'displaced': int(random.randint(0, 100) * base_multiplier),
            'affected': int(random.randint(100, 1000) * base_multiplier * population_factor),
            'psychologicalCasualties': int(random.randint(50, 300) * base_multiplier),
            'communityMorale': max(20, 100 - (base_multiplier * random.randint(15, 25)))
        },
        'property': {
            'homes_damaged': int(random.randint(0, 100) * base_multiplier),
            'businesses_affected': int(random.randint(0, 25) * base_multiplier),
            'estimated_cost': int(random.randint(100000, 5000000) * base_multiplier)
        },
        'infrastructure': {
            'roads_damaged_miles': round(random.uniform(1, 30) * base_multiplier, 1),
            'power_outages': int(random.randint(500, 10000) * base_multiplier),
            'water_systems_affected': int(random.randint(0, 10) * base_multiplier)
        },
        'agriculture': {
            'crop_loss_acres': int(random.randint(500, 10000) * base_multiplier),
            'livestock_loss': int(random.randint(0, 200) * base_multiplier),
            'estimated_cost': int(random.randint(100000, 2000000) * base_multiplier)
        }
    }
    
    return {
        'location': location,
        'damageType': damage_type,
        'severity': severity,
        'impact': impact_text,
        'assessments': assessments,
        'timestamp': datetime.now().isoformat()
    }

def calculate_psychological_impact(assessments):
    """Calculate psychological impact score"""
    people_impact = min(1.0, (
        assessments['people']['injured'] * 0.1 +
        assessments['people']['displaced'] * 0.05 +
        assessments['people']['affected'] * 0.0005
    ))
    
    property_impact = min(1.0, assessments['property']['homes_damaged'] * 0.01)
    economic_impact = min(1.0, assessments['property']['estimated_cost'] / 10000000)
    
    total_impact = (people_impact * 0.5 + property_impact * 0.3 + economic_impact * 0.2)
    
    if total_impact < 0.25:
        level = 'minimal'
    elif total_impact < 0.5:
        level = 'moderate'
    elif total_impact < 0.75:
        level = 'severe'
    else:
        level = 'critical'
    
    return {
        'total_score': total_impact,
        'level': level,
        'components': {
            'people_impact': people_impact,
            'property_impact': property_impact,
            'economic_impact': economic_impact
        }
    }

def assess_trauma_risk(assessments):
    """Assess trauma risk indicators"""
    indicators = {
        'direct_victimization': assessments['people']['injured'] > 0,
        'home_destruction': assessments['property']['homes_damaged'] > 0,
        'economic_devastation': assessments['property']['estimated_cost'] > 1000000,
        'social_disruption': assessments['people']['displaced'] > 50,
        'infrastructure_collapse': assessments['infrastructure']['power_outages'] > 5000,
        'community_breakdown': assessments['people']['communityMorale'] < 50
    }
    
    active_indicators = sum(indicators.values())
    risk_score = active_indicators / len(indicators)
    
    if risk_score < 0.3:
        risk_level = 'low'
    elif risk_score < 0.6:
        risk_level = 'moderate'
    else:
        risk_level = 'high'
    
    return {
        'indicators': indicators,
        'active_count': active_indicators,
        'risk_score': risk_score,
        'risk_level': risk_level
    }

def calculate_overall_mental_health(sentiment_data, psych_impact, trauma_risk, community_morale):
    """Calculate overall mental health score"""
    sentiment_score = (sentiment_data['sentiment_score'] + 1) / 2  # Normalize to 0-1
    psychological_score = 1 - psych_impact['total_score']
    trauma_score = 1 - trauma_risk['risk_score']
    morale_score = community_morale / 100
    
    overall_score = (
        sentiment_score * 0.25 +
        psychological_score * 0.35 +
        trauma_score * 0.25 +
        morale_score * 0.15
    )
    
    if overall_score > 0.75:
        level = 'excellent'
    elif overall_score > 0.6:
        level = 'good'
    elif overall_score > 0.4:
        level = 'concerning'
    else:
        level = 'critical'
    
    return {
        'overall_score': overall_score,
        'level': level,
        'components': {
            'sentiment': sentiment_score,
            'psychological': psychological_score,
            'trauma': trauma_score,
            'morale': morale_score
        }
    }

def main():
    """Main application function"""
    # Load CSS
    load_css()
    
    # Header
    st.markdown("""
    <div class="main-header">
        <h1>⚠️ Oklahoma Damage Assessment System</h1>
        <h2>🧠 Comprehensive Sentiment & Psychological Analysis</h2>
        <p>Advanced disaster impact analysis with emotional and mental health assessment</p>
    </div>
    """, unsafe_allow_html=True)
    
    # Sidebar controls
    st.sidebar.title("🎛️ Assessment Controls")
    st.sidebar.markdown("---")
    
    location_data = get_location_data()
    
    # Location selection
    location_type = st.sidebar.radio("📍 Location Type", ["County", "City"])
    
    if location_type == "County":
        selected_location = st.sidebar.selectbox("🏛️ Select County", location_data['counties'])
    else:
        selected_location = st.sidebar.selectbox("🏙️ Select City", location_data['cities'])
    
    # Analysis options
    st.sidebar.markdown("---")
    st.sidebar.subheader("🔧 Analysis Options")
    show_detailed = st.sidebar.checkbox("Show Detailed Analysis", value=True)
    show_trauma = st.sidebar.checkbox("Show Trauma Assessment", value=True)
    show_timeline = st.sidebar.checkbox("Show Recovery Timeline", value=True)
    
    # Generate new assessment button
    if st.sidebar.button("🔄 Generate New Assessment", type="primary"):
        st.rerun()
    
    # Generate data
    if 'damage_data' not in st.session_state or st.sidebar.button("🔄 Refresh Data"):
        st.session_state.damage_data = generate_damage_data(selected_location)
    
    damage_data = st.session_state.damage_data
    
    # Update location if changed
    if damage_data['location'] != selected_location:
        damage_data = generate_damage_data(selected_location)
        st.session_state.damage_data = damage_data
    
    # Perform analysis
    sentiment_data = analyze_sentiment(damage_data['impact'])
    psych_impact = calculate_psychological_impact(damage_data['assessments'])
    trauma_risk = assess_trauma_risk(damage_data['assessments'])
    overall_mental_health = calculate_overall_mental_health(
        sentiment_data, psych_impact, trauma_risk, 
        damage_data['assessments']['people']['communityMorale']
    )
    
    # Main damage alert
    damage_type = damage_data['damageType']
    severity = damage_data['severity']
    
    st.markdown(f"""
    <div class="damage-alert">
        <h3>{damage_type['icon']} {damage_data['location']} Impact Assessment</h3>
        <h4>{damage_type['name']} - {severity['name']} Severity</h4>
        <p><strong>Impact:</strong> {damage_data['impact']}</p>
    </div>
    """, unsafe_allow_html=True)
    
    # Mental health overview
    overall_score = overall_mental_health['overall_score']
    level = overall_mental_health['level']
    
    color_map = {
        'excellent': '#28a745',
        'good': '#17a2b8',
        'concerning': '#ffc107',
        'critical': '#dc3545'
    }
    
    color = color_map.get(level, '#6c757d')
    
    st.markdown(f"""
    <div class="sentiment-card">
        <h4>🎯 Overall Community Mental Health Assessment</h4>
        <div style="display: flex; align-items: center; justify-content: space-between;">
            <div>
                <h2 style="color: white; margin: 0;">{overall_score:.1%}</h2>
                <p style="margin: 0; font-size: 1.2em;">Mental Health Score</p>
            </div>
            <div style="text-align: right;">
                <h3 style="color: white; margin: 0;">{level.upper()}</h3>
                <p style="margin: 0;">Status Level</p>
            </div>
        </div>
    </div>
    """, unsafe_allow_html=True)
    
    # Key metrics
    st.subheader("📊 Key Impact Metrics")
    
    col1, col2, col3, col4 = st.columns(4)
    
    assessments = damage_data['assessments']
    
    with col1:
        st.metric("🚑 Injured", assessments['people']['injured'])
        st.metric("🧠 Psychological", assessments['people']['psychologicalCasualties'])
    
    with col2:
        st.metric("🏠 Homes Damaged", assessments['property']['homes_damaged'])
        st.metric("🚗 Property Cost", f"${assessments['property']['estimated_cost']:,}")
    
    with col3:
        st.metric("👥 Affected", assessments['people']['affected'])
        st.metric("🏠 Displaced", assessments['people']['displaced'])
    
    with col4:
        st.metric("💪 Community Morale", f"{assessments['people']['communityMorale']}%")
        st.metric("⚡ Power Outages", f"{assessments['infrastructure']['power_outages']:,}")
    
    # Detailed analysis sections
    if show_detailed:
        st.subheader("📈 Detailed Sentiment Analysis")
        
        col1, col2 = st.columns(2)
        
        with col1:
            st.markdown("**Text Sentiment Analysis:**")
            classification = sentiment_data['classification']
            
            class_colors = {
                'highly_negative': '#dc3545',
                'negative': '#fd7e14',
                'neutral': '#6c757d',
                'positive': '#28a745'
            }
            
            st.markdown(f"""
            <div style="background: {class_colors.get(classification, '#6c757d')}20; 
                        border-left: 4px solid {class_colors.get(classification, '#6c757d')}; 
                        padding: 1rem; border-radius: 5px;">
                <strong>Classification:</strong> {classification.replace('_', ' ').title()}<br>
                <strong>Sentiment Score:</strong> {sentiment_data['sentiment_score']:.3f}<br>
                <strong>Dominant Emotion:</strong> {sentiment_data['dominant_emotion'].title()}
            </div>
            """, unsafe_allow_html=True)
        
        with col2:
            st.markdown("**Psychological Impact Analysis:**")
            
            level_colors = {
                'minimal': '#28a745',
                'moderate': '#ffc107',
                'severe': '#fd7e14',
                'critical': '#dc3545'
            }
            
            level_color = level_colors.get(psych_impact['level'], '#6c757d')
            
            st.markdown(f"""
            <div style="background: {level_color}20; 
                        border-left: 4px solid {level_color}; 
                        padding: 1rem; border-radius: 5px;">
                <strong>Impact Level:</strong> {psych_impact['level'].title()}<br>
                <strong>Total Score:</strong> {psych_impact['total_score']:.3f}<br>
                <strong>Primary Factors:</strong> People Impact, Property Loss, Economic Stress
            </div>
            """, unsafe_allow_html=True)
    
    # Trauma assessment
    if show_trauma:
        st.subheader("⚠️ Trauma Risk Assessment")
        
        col1, col2, col3 = st.columns(3)
        
        with col1:
            st.metric("⚠️ Risk Level", trauma_risk['risk_level'].upper())
        
        with col2:
            st.metric("🚨 Active Indicators", f"{trauma_risk['active_count']}/6")
        
        with col3:
            st.metric("📊 Risk Score", f"{trauma_risk['risk_score']:.1%}")
        
        # Trauma indicators
        st.markdown("**Trauma Risk Indicators:**")
        
        indicators_data = []
        for indicator, active in trauma_risk['indicators'].items():
            indicators_data.append({
                'Indicator': indicator.replace('_', ' ').title(),
                'Status': 'Active' if active else 'Inactive',
                'Risk_Level': 1 if active else 0
            })
        
        indicators_df = pd.DataFrame(indicators_data)
        
        if not indicators_df.empty:
            fig = px.bar(
                indicators_df,
                x='Indicator',
                y='Risk_Level',
                color='Status',
                title="Trauma Risk Indicators",
                color_discrete_map={'Active': '#dc3545', 'Inactive': '#28a745'}
            )
            fig.update_layout(height=400, xaxis_tickangle=-45)
            st.plotly_chart(fig, use_container_width=True)
    
    # Recovery timeline
    if show_timeline:
        st.subheader("📅 Recovery Timeline")
        
        timeline_phases = [
            ("Immediate Response", "0-72 hours", "Acute stress management"),
            ("Short-term Recovery", "3 days - 4 weeks", "PTSD prevention"),
            ("Medium-term Adjustment", "1-6 months", "Adaptation support"),
            ("Long-term Recovery", "6+ months", "Post-traumatic growth")
        ]
        
        for i, (phase, duration, description) in enumerate(timeline_phases, 1):
            with st.expander(f"Phase {i}: {phase} ({duration})"):
                st.write(f"**Focus:** {description}")
                
                if i == 1:
                    st.write(f"• Expected acute stress in {trauma_risk['risk_score']:.1%} of population")
                    st.write("• Deploy crisis counselors immediately")
                elif i == 2:
                    st.write(f"• PTSD screening for {trauma_risk['active_count'] * 100} individuals")
                    st.write("• Establish support groups")
                elif i == 3:
                    st.write("• Economic counseling for affected households")
                    st.write("• Long-term housing solutions")
                else:
                    st.write(f"• Post-traumatic growth potential: {overall_score:.1%}")
                    st.write("• Community resilience building")
    
    # Mental health resources
    st.subheader("🏥 Mental Health Resource Recommendations")
    
    total_affected = assessments['people']['affected']
    psychological_casualties = assessments['people']['psychologicalCasualties']
    
    col1, col2, col3 = st.columns(3)
    
    with col1:
        st.markdown("**🚨 Immediate Response**")
        counselors_needed = max(5, psychological_casualties // 50)
        st.write(f"• Deploy {counselors_needed} crisis counselors")
        st.write("• Activate 24/7 mental health hotline")
        st.write("• Set up mobile crisis units")
    
    with col2:
        st.markdown("**⏰ Short-term Support**")
        support_groups = max(3, total_affected // 200)
        st.write(f"• Establish {support_groups} support groups")
        st.write("• Trauma screening programs")
        st.write("• Children's mental health services")
    
    with col3:
        st.markdown("**🌱 Long-term Recovery**")
        st.write("• Community resilience workshops")
        st.write("• PTSD treatment programs")
        st.write("• Economic counseling services")
    
    # Data export
    st.sidebar.markdown("---")
    st.sidebar.subheader("📊 Export Data")
    
    if st.sidebar.button("📥 Export Assessment"):
        export_data = {
            'damage_assessment': damage_data,
            'sentiment_analysis': sentiment_data,
            'psychological_impact': psych_impact,
            'trauma_assessment': trauma_risk,
            'mental_health_score': overall_mental_health,
            'generated_at': datetime.now().isoformat()
        }
        
        st.sidebar.download_button(
            label="Download JSON Report",
            data=json.dumps(export_data, indent=2, default=str),
            file_name=f"oklahoma_assessment_{selected_location.replace(' ', '_')}.json",
            mime="application/json"
        )
    
    # Footer
    st.markdown("---")
    st.markdown("""
    <div style="text-align: center; padding: 2rem; background: #f8f9fa; border-radius: 10px;">
        <h4>⚠️ Oklahoma Damage Assessment System</h4>
        <p>🧠 Comprehensive Sentiment & Psychological Analysis Platform</p>
        <p><small>Last Updated: {}</small></p>
    </div>
    """.format(datetime.now().strftime('%Y-%m-%d %H:%M:%S')), unsafe_allow_html=True)

if __name__ == "__main__":
    main()