import plotly.express as px
import plotly.graph_objects as go
import pandas as pd

# Theme settings
BG_COLOR = "#0f1117"
FONT_COLOR = "white"
COLOR_HIGH = "#e63946"
COLOR_MED = "#f4a261"
COLOR_LOW = "#2a9d8f"

COLOR_MAP = {
    "High": COLOR_HIGH,
    "Medium": COLOR_MED,
    "Low": COLOR_LOW
}

def apply_theme(fig):
    fig.update_layout(
        paper_bgcolor=BG_COLOR,
        plot_bgcolor=BG_COLOR,
        font_color=FONT_COLOR,
        margin=dict(l=40, r=40, t=40, b=40)
    )
    return fig

def risk_distribution_chart(df):
    counts = df['risk_level'].value_counts().reset_index()
    counts.columns = ['risk_level', 'count']
    fig = px.pie(counts, values='count', names='risk_level', color='risk_level',
                 color_discrete_map=COLOR_MAP, hole=0.4)
    fig.update_traces(textposition='inside', textinfo='percent+label')
    return apply_theme(fig)

def flagged_transactions_timeline(df):
    flagged_df = df[df['is_flagged'] == 1].copy()
    if flagged_df.empty:
        return px.line(title="No Flagged Transactions")
        
    flagged_df['date'] = flagged_df['timestamp'].dt.date
    daily_counts = flagged_df.groupby('date').size().reset_index(name='count')
    fig = px.line(daily_counts, x='date', y='count', markers=True)
    fig.update_traces(line_color=COLOR_HIGH)
    return apply_theme(fig)

def amount_vs_risk_scatter(df):
    # size needs to be positive
    df['size'] = df['transaction_velocity'].clip(lower=1)
    # create a rule string for hover
    df['rule_str'] = df['rule_flags'].apply(lambda x: ", ".join(x) if isinstance(x, list) and x else "None")
    
    fig = px.scatter(df, x='amount', y='risk_score', color='risk_level',
                     size='size', hover_data=['transaction_id', 'rule_str'],
                     color_discrete_map=COLOR_MAP, log_x=True)
    return apply_theme(fig)

def transaction_type_breakdown(df):
    grouped = df.groupby(['transaction_type', 'is_flagged']).size().reset_index(name='count')
    grouped['Status'] = grouped['is_flagged'].map({1: 'Flagged', 0: 'Clean'})
    fig = px.bar(grouped, x='transaction_type', y='count', color='Status', barmode='group',
                 color_discrete_map={'Flagged': COLOR_HIGH, 'Clean': COLOR_LOW})
    return apply_theme(fig)

def top_flagged_customers_chart(df):
    flagged = df[df['is_flagged'] == 1]
    if flagged.empty:
        return px.bar(title="No Flagged Customers")
        
    cust_stats = flagged.groupby('customer_id').agg(
        flagged_count=('transaction_id', 'count'),
        avg_risk=('risk_score', 'mean')
    ).reset_index().sort_values('flagged_count', ascending=False).head(10)
    
    fig = px.bar(cust_stats, x='flagged_count', y='customer_id', orientation='h',
                 color='avg_risk', color_continuous_scale='Reds')
    fig.update_layout(yaxis={'categoryorder':'total ascending'})
    return apply_theme(fig)

def kyc_tier_distribution(profile_df):
    counts = profile_df['kyc_tier'].value_counts().reset_index()
    counts.columns = ['kyc_tier', 'count']
    fig = px.pie(counts, values='count', names='kyc_tier', color='kyc_tier',
                 color_discrete_map=COLOR_MAP, hole=0.5)
    return apply_theme(fig)

def rule_trigger_frequency(df):
    all_rules = []
    for rules in df['rule_flags']:
        if isinstance(rules, list):
            all_rules.extend(rules)
            
    if not all_rules:
        return px.bar(title="No Rules Triggered")
        
    counts = pd.Series(all_rules).value_counts().reset_index()
    counts.columns = ['Rule', 'Count']
    fig = px.bar(counts, x='Count', y='Rule', orientation='h', color_discrete_sequence=[COLOR_MED])
    fig.update_layout(yaxis={'categoryorder':'total ascending'})
    return apply_theme(fig)