app.py

import streamlit as st
import requests
import os
import json
import pandas as pd
import plotly.graph_objects as go
import plotly.express as px
import time
import random
from datetime import datetime, timedelta

# Custom CSS for styling
st.markdown("""
    <style>
        .stApp {
            background: #f5f5f5;
        }
        .header {
            font-size: 36px;
            font-weight: bold;
            color: #4CAF50;
            text-align: center;
            margin-bottom: 20px;
        }
        .subheader {
            font-size: 24px;
            font-weight: bold;
            color: #4CAF50;
            text-align: center;
            margin-bottom: 20px;
        }
        .section {
            background: white;
            padding: 20px;
            border-radius: 10px;
            box-shadow: 0px 4px 12px rgba(0, 0, 0, 0.1);
            margin-bottom: 20px;
        }
        .footer {
            text-align: center;
            font-size: 14px;
            color: #777;
            margin-top: 20px;
        }
    </style>
""", unsafe_allow_html=True)

# Function to call the Together AI model
def call_ai_model(all_message):
    url = "https://api.together.xyz/v1/chat/completions"
    payload = {
        "model": "NousResearch/Nous-Hermes-2-Yi-34B",
        "temperature": 1.05,
        "top_p": 0.9,
        "top_k": 50,
        "repetition_penalty": 1,
        "n": 1,
        "messages": [{"role": "user", "content": all_message}],
        "stream_tokens": True,
    }

    TOGETHER_API_KEY = os.getenv('TOGETHER_API_KEY')
    if TOGETHER_API_KEY is None:
        raise ValueError("TOGETHER_API_KEY environment variable not set.")
    
    headers = {
        "accept": "application/json",
        "content-type": "application/json",
        "Authorization": f"Bearer {TOGETHER_API_KEY}",
    }

    response = requests.post(url, json=payload, headers=headers, stream=True)
    response.raise_for_status()

    return response

# Function to generate simulated trust scores
def generate_trust_scores(technologies, issues):
    trust_scores = {}
    for tech in technologies:
        trust_scores[tech] = {}
        for issue in issues:
            trust_scores[tech][issue] = random.uniform(0, 1)
    return trust_scores

# Function to generate simulated kinship impact scores
def generate_kinship_impact(technologies):
    impact_scores = {}
    kinship_aspects = ["Family Communication", "Intergenerational Relationships", "Cultural Traditions"]
    for tech in technologies:
        impact_scores[tech] = {}
        for aspect in kinship_aspects:
            impact_scores[tech][aspect] = random.uniform(-1, 1)
    return impact_scores

# Function to generate simulated gender impact scores
def generate_gender_impact(technologies, genders):
    impact_scores = {}
    for tech in technologies:
        impact_scores[tech] = {}
        for gender in genders:
            impact_scores[tech][gender] = random.uniform(-1, 1)
    return impact_scores

# Function to generate simulated geographic usage data
def generate_geographic_data(technologies, regions):
    geo_data = {}
    for region in regions:
        geo_data[region] = {}
        for tech in technologies:
            geo_data[region][tech] = random.uniform(0, 100)
    return geo_data

# Function to generate simulated long-term economic impact
def generate_economic_impact(technologies):
    impact_data = {}
    indicators = ["GDP Growth", "Employment Rate", "Digital Literacy"]
    for tech in technologies:
        impact_data[tech] = {}
        for indicator in indicators:
            impact_data[tech][indicator] = [random.uniform(-2, 5) for _ in range(5)]  # 5-year projection
    return impact_data

# Streamlit app layout
st.markdown('<div class="header">Digital Technologies, Kinship, and Gender in Kenya</div>', unsafe_allow_html=True)
st.markdown('<div class="subheader">Analyze and visualize the impact of digital technologies on kinship and gender dynamics in Kenya.</div>', unsafe_allow_html=True)

# Input section
with st.container():
    st.markdown('<div class="section">', unsafe_allow_html=True)
    st.subheader("Digital Technology Impacts")
    digital_technologies = st.multiselect("Select digital technologies:", ["Big Data Analytics", "Biometric Authentication", "Blockchain", "E-commerce", "Social Media Platforms"])
    issues = st.multiselect("Select issues of concern:", ["Trust", "Mistrust", "Data Privacy", "Fraud", "Social Classification"])

    st.subheader("Kenya-Specific Inputs")
    regions = st.multiselect("Select regions in Kenya:", ["Nairobi", "Mombasa", "Kisumu", "Nakuru", "Eldoret"])
    gender_focus = st.multiselect("Select gender focus:", ["Male", "Female", "Non-binary"])
    st.markdown('</div>', unsafe_allow_html=True)

# Button to generate analysis
if st.button("Generate Analysis"):
    all_message = (
        f"Analyze the impact of digital technologies on kinship and gender dynamics in Kenya. "
        f"Digital technologies: {', '.join(digital_technologies)}. "
        f"Issues of concern: {', '.join(issues)}. "
        f"Regions: {', '.join(regions)}. Gender focus: {', '.join(gender_focus)}. "
        f"Provide a detailed analysis of how these technologies impact family ties, trust, and gender roles. "
        f"Include specific impacts for each digital technology and issue. "
        f"Organize the information in tables with the following columns: Digital Technology, Impact on Kinship, Impact on Gender Dynamics, Trust Issues. "
        f"Be as accurate and specific to Kenya as possible in your analysis. Make the response short and precise."
    )

    try:
        stages = [
            "Analyzing digital technologies...",
            "Checking location data...",
            "Running simulations...",
            "Processing data...",
            "Assessing impacts...",
            "Calculating predictions...",
            "Compiling results...",
            "Finalizing analysis...",
            "Preparing output..."
        ]
        
        with st.spinner("Analyzing digital technologies and generating analysis..."):
            for stage in stages:
                st.spinner(stage)
                time.sleep(2)
            
            response = call_ai_model(all_message)
    
            analysis_text = ""
            for line in response.iter_lines():
                if line:
                    line_content = line.decode('utf-8')
                    if line_content.startswith("data: "):
                        line_content = line_content[6:]
                    try:
                        json_data = json.loads(line_content)
                        if "choices" in json_data:
                            delta = json_data["choices"][0]["delta"]
                            if "content" in delta:
                                analysis_text += delta["content"]
                    except json.JSONDecodeError:
                        continue
    
            st.success("Analysis completed!")
            
            # Display analysis
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Digital Technologies Impact Analysis in Kenya")
            st.markdown(analysis_text.strip())
            st.markdown('</div>', unsafe_allow_html=True)

            # Generate simulated data
            trust_scores = generate_trust_scores(digital_technologies, issues)
            kinship_impact = generate_kinship_impact(digital_technologies)
            gender_impact = generate_gender_impact(digital_technologies, gender_focus)
            geo_data = generate_geographic_data(digital_technologies, regions)
            economic_impact = generate_economic_impact(digital_technologies)

            # Trust and Fraud Metrics Visualization
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Trust and Fraud Metrics")
            fig_trust = go.Figure()
            for tech in digital_technologies:
                fig_trust.add_trace(go.Bar(
                    x=list(trust_scores[tech].keys()),
                    y=list(trust_scores[tech].values()),
                    name=tech
                ))
            fig_trust.update_layout(barmode='group', title="Trust Scores by Technology and Issue")
            st.plotly_chart(fig_trust)
            st.markdown('</div>', unsafe_allow_html=True)

            # Kinship Structure Analysis
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Impact on Kinship Structures")
            fig_kinship = go.Figure()
            for tech in digital_technologies:
                fig_kinship.add_trace(go.Scatterpolar(
                    r=list(kinship_impact[tech].values()),
                    theta=list(kinship_impact[tech].keys()),
                    fill='toself',
                    name=tech
                ))
            fig_kinship.update_layout(polar=dict(radialaxis=dict(visible=True, range=[-1, 1])), showlegend=True)
            st.plotly_chart(fig_kinship)
            st.markdown('</div>', unsafe_allow_html=True)

            # Gender Impact Visualization
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Gender Impact Analysis")
            fig_gender = go.Figure()
            for tech in digital_technologies:
                fig_gender.add_trace(go.Bar(
                    x=list(gender_impact[tech].keys()),
                    y=list(gender_impact[tech].values()),
                    name=tech
                ))
            fig_gender.update_layout(barmode='group', title="Gender Impact by Technology")
            st.plotly_chart(fig_gender)
            st.markdown('</div>', unsafe_allow_html=True)

            # Geographic Component
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Geographic Distribution of Technology Usage")
            fig_geo = px.choropleth(
                locations=regions,
                locationmode="country names",
                color=[sum(geo_data[region].values()) for region in regions],
                hover_name=regions,
                color_continuous_scale=px.colors.sequential.Viridis,
                title="Technology Usage Intensity Across Kenya"
            )
            fig_geo.update_geos(fitbounds="locations", visible=False)
            st.plotly_chart(fig_geo)
            st.markdown('</div>', unsafe_allow_html=True)

            # Long-term Economic Impact
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Projected Long-term Economic Impact")
            fig_economic = go.Figure()
            years = [datetime.now().year + i for i in range(5)]
            for tech in digital_technologies:
                for indicator in economic_impact[tech]:
                    fig_economic.add_trace(go.Scatter(
                        x=years,
                        y=economic_impact[tech][indicator],
                        mode='lines+markers',
                        name=f"{tech} - {indicator}"
                    ))
            fig_economic.update_layout(title="5-Year Economic Impact Projection", xaxis_title="Year", yaxis_title="Impact (%)")
            st.plotly_chart(fig_economic)
            st.markdown('</div>', unsafe_allow_html=True)

            # Ethical Considerations
            st.markdown('<div class="section">', unsafe_allow_html=True)
            st.subheader("Ethical Considerations")
            ethical_concerns = [
                "Data Privacy: Ensuring user data is protected and used responsibly.",
                "Digital Divide: Addressing inequality in access to digital technologies.",
                "Cultural Preservation: Balancing technological advancement with traditional values.",
                "Algorithmic Bias: Mitigating biases in AI and machine learning systems.",
                "Cybersecurity: Protecting users from fraud and cyber attacks."
            ]
            for concern in ethical_concerns:
                st.write(f"• {concern}")
            st.markdown('</div>', unsafe_allow_html=True)

    except ValueError as ve:
        st.error(f"Configuration error: {ve}")
    except requests.exceptions.RequestException as re:
        st.error(f"Request error: {re}")
    except Exception as e:
        st.error(f"An unexpected error occurred: {e}")

# Footer
st.markdown('<div class="footer">Powered by Together AI and Streamlit</div>', unsafe_allow_html=True)