hiffi-artist-intelligence/app.py

"""
app.py
HIFFI Artist Intelligence - Multi-Platform Executive Dashboard
Clean dark-themed design with multi-platform evidence tiers, Spotify & Instagram telemetry,
breakout velocity index, and deep-dive scouting dossier.
"""

import io
from pathlib import Path
import pandas as pd
import plotly.express as px
import streamlit as st

st.set_page_config(
    page_title="HIFFI | Multi-Platform Intelligence",
    page_icon="⚡",
    layout="wide",
    initial_sidebar_state="expanded",
)

DATA_PATH = Path("data/hiffi_multiplatform_500.csv")
FALLBACK_DATA_PATH = Path("enriched_artists.csv")

# Global CSS Polish
st.markdown(
    """
    <style>
    .block-container {
        padding-top: 1.8rem;
        padding-bottom: 2.5rem;
        padding-left: 2.5rem;
        padding-right: 2.5rem;
        max-width: 100% !important;
    }

    /* KPI Cards */
    .kpi-box {
        background: #111827;
        border: 1px solid #1f2937;
        border-radius: 12px;
        padding: 16px 20px;
        margin-bottom: 12px;
    }
    .kpi-title {
        font-size: 0.75rem;
        font-weight: 700;
        color: #9ca3af;
        text-transform: uppercase;
        letter-spacing: 0.05em;
    }
    .kpi-value {
        font-size: 1.9rem;
        font-weight: 800;
        color: #f9fafb;
        margin: 4px 0;
    }
    .kpi-subtitle {
        font-size: 0.75rem;
        color: #6b7280;
    }

    /* Podium Cards */
    .podium-tile {
        background: #111827;
        border-radius: 12px;
        padding: 18px 20px;
        border: 1px solid #1f2937;
        height: 100%;
    }
    .podium-gold { border-top: 4px solid #f59e0b; }
    .podium-silver { border-top: 4px solid #94a3b8; }
    .podium-bronze { border-top: 4px solid #f97316; }

    /* Health & Tier Badges */
    .badge {
        display: inline-block;
        padding: 3px 8px;
        border-radius: 6px;
        font-size: 0.70rem;
        font-weight: 700;
        text-transform: uppercase;
    }
    .badge-tier3 { background: rgba(16, 185, 129, 0.2); color: #10b981; }
    .badge-tier2 { background: rgba(56, 189, 248, 0.2); color: #38bdf8; }
    .badge-tier1 { background: rgba(245, 158, 11, 0.2); color: #f59e0b; }
    .badge-dormant { background: rgba(239, 68, 68, 0.2); color: #ef4444; }

    .custom-divider {
        height: 1px;
        background: #1f2937;
        margin: 24px 0;
    }
    </style>
    """,
    unsafe_allow_html=True,
)


@st.cache_data(ttl=60)
def load_data():
    target = DATA_PATH if DATA_PATH.exists() else FALLBACK_DATA_PATH
    if not target.exists():
        return pd.DataFrame()

    df = pd.read_csv(target)

    # Normalize primary score & rank columns if running from legacy source
    if "hiffi_xp_published_rank" not in df.columns:
        df["hiffi_xp_published_rank"] = df.get("hiffi_xp_rank", df.get("hiffi_rank", df.index + 1))
    if "hiffi_xp_score" not in df.columns:
        df["hiffi_xp_score"] = df.get("production_hiffi_score", df.get("hiffi_score", 0.0))
    if "platform_evidence_tier" not in df.columns:
        has_yt = df.get("production_hiffi_score", df.get("hiffi_score")).notna().astype(int)
        has_sp = df.get("spotify_monthly_listeners", pd.Series(index=df.index)).notna().astype(int)
        has_ig = df.get("instagram_followers", pd.Series(index=df.index)).notna().astype(int)
        df["platform_evidence_tier"] = has_yt + has_sp + has_ig

    numeric_cols = [
        "hiffi_xp_published_rank",
        "hiffi_xp_rank",
        "hiffi_xp_score",
        "production_hiffi_rank",
        "production_hiffi_score",
        "platform_evidence_tier",
        "breakout_velocity_index",
        "subscriber_count",
        "daily_view_velocity",
        "momentum_30d",
        "spotify_monthly_listeners",
        "spotify_daily_change",
        "instagram_followers",
        "confidence_score",
        "breakout_rank"
    ]
    for col in numeric_cols:
        if col in df.columns:
            df[col] = pd.to_numeric(df[col], errors="coerce")

    if "subgenre" not in df.columns:
        def infer_subgenre(r):
            subs = r.get("subscriber_count", 0) or 0
            bvi = r.get("breakout_velocity_index", 0) or 0
            if bvi >= 70:
                return "Breakout / Emerging"
            elif subs >= 15000000:
                return "Legends"
            else:
                return "Trap & Street"
        df["subgenre"] = df.apply(infer_subgenre, axis=1)

    return df


def to_excel_bytes(dataframe: pd.DataFrame) -> bytes:
    output = io.BytesIO()
    with pd.ExcelWriter(output, engine="openpyxl") as writer:
        dataframe.to_excel(writer, index=False, sheet_name="HIFFI_Cohort")
    return output.getvalue()


df = load_data()

# ----------------- Sidebar Controls -----------------
with st.sidebar:
    st.markdown("### ⚡ HIFFI Multi-Platform")
    st.caption("YouTube • Spotify • Instagram Analytics")
    st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)

    if df.empty:
        st.error("No dataset found. Run 'python run_pipeline.py' to generate master data.")
        st.stop()

    view_mode = st.radio(
        "Leaderboard Perspective:",
        ["🏆 Published Multi-Platform Rank", "🚀 Breakout Velocity (BVI)", "🔍 Full Cohort Audit Grid"],
        index=0,
    )

    st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)
    st.markdown("##### 🛡️ Cohort Filters")

    all_verticals = sorted(df["subgenre"].dropna().unique().tolist())
    selected_verticals = st.multiselect("Vertical", options=all_verticals, default=all_verticals)

    tier_options = sorted(df["platform_evidence_tier"].dropna().unique().astype(int).tolist(), reverse=True)
    selected_tiers = st.multiselect("Evidence Tier", options=tier_options, default=tier_options)

    min_xp = st.slider("Min HIFFI-XP Score", 0.0, 100.0, 0.0, 5.0)

filtered_df = df[
    (df["subgenre"].isin(selected_verticals))
    & (df["platform_evidence_tier"].isin(selected_tiers))
    & (df["hiffi_xp_score"].fillna(0.0) >= min_xp)
].copy()

with st.sidebar:
    st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)
    st.markdown("##### 💾 Export Filtered Cohort")
    col_csv, col_xlsx = st.columns(2)

    csv_payload = filtered_df.to_csv(index=False).encode("utf-8")
    col_csv.download_button(
        "📥 CSV",
        data=csv_payload,
        file_name="hiffi_multiplatform_filtered.csv",
        mime="text/csv",
        width="stretch",
    )

    excel_payload = to_excel_bytes(filtered_df)
    col_xlsx.download_button(
        "📊 Excel",
        data=excel_payload,
        file_name="hiffi_multiplatform_filtered.xlsx",
        mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
        width="stretch",
    )

    st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)
    st.caption("Weights: YouTube (50%) • Spotify (30%) • Instagram (20%)")

# ----------------- Main Header -----------------
head_col1, head_col2 = st.columns([3, 1])
with head_col1:
    st.markdown("<h2 style='margin-bottom: 2px;'>HIFFI Multi-Platform Intelligence Hub</h2>", unsafe_allow_html=True)
    st.markdown("<p style='color: #9ca3af; font-size: 0.92rem; margin-top: 0;'>Multi-platform composite rankings with zero-imputation policy and dynamic weight redistribution.</p>", unsafe_allow_html=True)
with head_col2:
    st.markdown(
        f"<div style='text-align: right; color: #6b7280; font-size: 0.8rem; padding-top: 10px;'>Filtered Cohort: <b>{len(filtered_df)} / {len(df)}</b> artists</div>",
        unsafe_allow_html=True,
    )

# ----------------- KPI Cards Row -----------------
k1, k2, k3, k4 = st.columns(4)

with k1:
    st.markdown(
        f"""
        <div class="kpi-box">
            <div class="kpi-title">Tracked Artists</div>
            <div class="kpi-value">{len(df)}</div>
            <div class="kpi-subtitle">Monitored Cohort</div>
        </div>
        """,
        unsafe_allow_html=True,
    )

with k2:
    t3_count = len(df[df["platform_evidence_tier"] == 3])
    st.markdown(
        f"""
        <div class="kpi-box">
            <div class="kpi-title" style="color: #10b981;">Tier 3 (Complete)</div>
            <div class="kpi-value" style="color: #10b981;">{t3_count}</div>
            <div class="kpi-subtitle">YouTube + Spotify + Instagram</div>
        </div>
        """,
        unsafe_allow_html=True,
    )

with k3:
    sp_count = df["spotify_monthly_listeners"].notna().sum()
    st.markdown(
        f"""
        <div class="kpi-box">
            <div class="kpi-title" style="color: #38bdf8;">Spotify Telemetry</div>
            <div class="kpi-value" style="color: #38bdf8;">{sp_count}</div>
            <div class="kpi-subtitle">Verified monthly listeners</div>
        </div>
        """,
        unsafe_allow_html=True,
    )

with k4:
    ig_count = df["instagram_followers"].notna().sum()
    st.markdown(
        f"""
        <div class="kpi-box">
            <div class="kpi-title" style="color: #f59e0b;">Instagram Audience</div>
            <div class="kpi-value" style="color: #f59e0b;">{ig_count}</div>
            <div class="kpi-subtitle">Verified public followings</div>
        </div>
        """,
        unsafe_allow_html=True,
    )

st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)

# ----------------- Podium -----------------
top3 = filtered_df.sort_values(by="hiffi_xp_published_rank", ascending=True).head(3)

if len(top3) >= 3:
    st.markdown("#### 👑 Market Leaders (Published Rankings)")
    p1, p2, p3 = st.columns(3)

    podium_data = [
        (p1, top3.iloc[0], "🥇 PUBLISHED #1", "podium-gold", "#f59e0b"),
        (p2, top3.iloc[1], "🥈 PUBLISHED #2", "podium-silver", "#94a3b8"),
        (p3, top3.iloc[2], "🥉 PUBLISHED #3", "podium-bronze", "#f97316"),
    ]

    for col, row_data, rank_label, css_class, accent_color in podium_data:
        tier_val = int(row_data.get("platform_evidence_tier", 1))
        badge_cls = f"badge-tier{tier_val}"
        sp_str = f"{int(row_data['spotify_monthly_listeners']):,}" if pd.notna(row_data.get("spotify_monthly_listeners")) else "NaN"
        ig_str = f"{int(row_data['instagram_followers']):,}" if pd.notna(row_data.get("instagram_followers")) else "NaN"

        with col:
            st.markdown(
                f"""
                <div class="podium-tile {css_class}">
                    <div style="display: flex; justify-content: space-between; align-items: center; margin-bottom: 6px;">
                        <span style="font-weight: 800; color: {accent_color}; font-size: 0.85rem;">{rank_label}</span>
                        <span class="badge {badge_cls}">Tier {tier_val} Evidence</span>
                    </div>
                    <div style="font-size: 1.35rem; font-weight: 800; color: #ffffff; margin-bottom: 2px;">{row_data['name']}</div>
                    <div style="font-size: 0.8rem; color: #6b7280; margin-bottom: 12px;">{row_data.get('channel_title', '')}</div>
                    <div style="display: flex; justify-content: space-between; background: #0b0f17; padding: 10px 14px; border-radius: 8px;">
                        <div>
                            <div style="font-size: 0.68rem; color: #9ca3af; text-transform: uppercase;">HIFFI-XP</div>
                            <div style="font-size: 1.25rem; font-weight: 800; color: {accent_color};">{row_data['hiffi_xp_score']:.2f}</div>
                        </div>
                        <div>
                            <div style="font-size: 0.68rem; color: #9ca3af; text-transform: uppercase;">Spotify Monthly</div>
                            <div style="font-size: 1.05rem; font-weight: 700; color: #38bdf8;">{sp_str}</div>
                        </div>
                        <div>
                            <div style="font-size: 0.68rem; color: #9ca3af; text-transform: uppercase;">IG Followers</div>
                            <div style="font-size: 1.05rem; font-weight: 700; color: #e5e7eb;">{ig_str}</div>
                        </div>
                    </div>
                </div>
                """,
                unsafe_allow_html=True,
            )

    st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)

# ----------------- Leaderboard Grid -----------------
st.markdown(f"#### 📋 {view_mode}")

if view_mode == "🏆 Published Multi-Platform Rank":
    table_df = filtered_df.sort_values(by="hiffi_xp_published_rank", ascending=True).copy()
    cols = [
        "hiffi_xp_published_rank",
        "hiffi_xp_rank",
        "name",
        "platform_evidence_tier",
        "hiffi_xp_score",
        "spotify_monthly_listeners",
        "instagram_followers",
        "effective_weights",
        "anomaly_flags"
    ]
elif view_mode == "🚀 Breakout Velocity (BVI)":
    table_df = (
        filtered_df[filtered_df["breakout_velocity_index"].notna()]
        .sort_values(by="breakout_velocity_index", ascending=False)
        .copy()
    )
    cols = [
        "name",
        "breakout_velocity_index",
        "hiffi_xp_published_rank",
        "daily_view_velocity",
        "momentum_30d",
        "subscriber_count",
        "subgenre"
    ]
else:
    table_df = filtered_df.sort_values(by="hiffi_xp_published_rank", ascending=True).copy()
    cols = [
        "hiffi_xp_published_rank",
        "name",
        "production_hiffi_rank",
        "hiffi_xp_rank",
        "rank_delta",
        "platform_evidence_tier",
        "hiffi_xp_score",
        "spotify_monthly_listeners",
        "instagram_followers",
        "evidence_exclusions",
        "anomaly_flags"
    ]

render_cols = [c for c in cols if c in table_df.columns]

st.dataframe(
    table_df[render_cols].reset_index(drop=True),
    width="stretch",
    height=320,
    column_config={
        "hiffi_xp_published_rank": st.column_config.NumberColumn("Pub Rank", format="%d", width="small"),
        "hiffi_xp_rank": st.column_config.NumberColumn("Raw XP Rank", format="%d", width="small"),
        "production_hiffi_rank": st.column_config.NumberColumn("YT Rank", format="%d", width="small"),
        "platform_evidence_tier": st.column_config.NumberColumn("Tier", format="%d", width="small"),
        "name": st.column_config.TextColumn("Artist", width="medium"),
        "hiffi_xp_score": st.column_config.ProgressColumn("HIFFI-XP Score", format="%.2f", min_value=0, max_value=100, width="medium"),
        "breakout_velocity_index": st.column_config.ProgressColumn("BVI", format="%.1f", min_value=0, max_value=100, width="small"),
        "spotify_monthly_listeners": st.column_config.NumberColumn("Spotify Monthly", format="%,d", width="medium"),
        "instagram_followers": st.column_config.NumberColumn("IG Followers", format="%,d", width="medium"),
        "effective_weights": st.column_config.TextColumn("Dynamic Redistribution", width="large"),
        "anomaly_flags": st.column_config.TextColumn("Anomaly Status", width="medium"),
        "evidence_exclusions": st.column_config.TextColumn("Exclusions", width="large"),
    },
)

st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)

# ----------------- Visual Insights Charts -----------------
st.markdown("#### 📊 Multi-Platform Analytics")

c_left, c_right = st.columns(2)

with c_left:
    chart_df = filtered_df.dropna(subset=["spotify_monthly_listeners", "instagram_followers", "hiffi_xp_score"]).copy()
    if not chart_df.empty:
        fig_scatter = px.scatter(
            chart_df,
            x="spotify_monthly_listeners",
            y="instagram_followers",
            size="hiffi_xp_score",
            color="platform_evidence_tier",
            hover_name="name",
            log_x=True,
            log_y=True,
            title="Spotify Monthly Listeners vs. Instagram Followers",
            labels={
                "spotify_monthly_listeners": "Spotify Monthly Listeners (Log Scale)",
                "instagram_followers": "Instagram Followers (Log Scale)",
                "platform_evidence_tier": "Tier",
            },
            template="plotly_dark",
        )
        fig_scatter.update_layout(
            paper_bgcolor="#111827",
            plot_bgcolor="#111827",
            font=dict(color="#9ca3af"),
            margin=dict(l=20, r=20, t=40, b=20),
            height=320,
        )
        st.plotly_chart(fig_scatter, width="stretch")
    else:
        st.info("Insufficient complete data points to render multi-platform scatter plot.")

with c_right:
    tier_counts = filtered_df["platform_evidence_tier"].value_counts().reset_index()
    tier_counts.columns = ["Tier", "Count"]
    fig_pie = px.pie(
        tier_counts,
        values="Count",
        names="Tier",
        title="Evidence Tier Cohort Distribution",
        template="plotly_dark",
        color="Tier",
        color_discrete_map={3: "#10b981", 2: "#38bdf8", 1: "#f59e0b", 0: "#ef4444"}
    )
    fig_pie.update_layout(
        paper_bgcolor="#111827",
        plot_bgcolor="#111827",
        font=dict(color="#9ca3af"),
        margin=dict(l=20, r=20, t=40, b=20),
        height=320,
    )
    st.plotly_chart(fig_pie, width="stretch")

# ----------------- Deep Dive Scouting Dossier -----------------
st.markdown("<div class='custom-divider'></div>", unsafe_allow_html=True)
st.markdown("#### 🔎 Artist Dossier")

selected_name = st.selectbox("Inspect Artist Details:", df["name"].unique(), label_visibility="collapsed")
row = df[df["name"] == selected_name].iloc[0]

d1, d2, d3, d4 = st.columns(4)

with d1:
    h_val = row.get("hiffi_xp_score")
    st.metric("HIFFI-XP Score", f"{h_val:.2f}" if pd.notna(h_val) else "N/A")
with d2:
    pub_rank = row.get("hiffi_xp_published_rank")
    st.metric("Published Rank", f"#{int(pub_rank)}" if pd.notna(pub_rank) else "Unranked")
with d3:
    sp_l = row.get("spotify_monthly_listeners")
    st.metric("Spotify Listeners", f"{int(sp_l):,}" if pd.notna(sp_l) else "Unindexed (NaN)")
with d4:
    ig_f = row.get("instagram_followers")
    st.metric("Instagram Followers", f"{int(ig_f):,}" if pd.notna(ig_f) else "Unindexed (NaN)")

st.markdown(
    f"""
    <div style="background: #111827; padding: 14px 18px; border-radius: 8px; border: 1px solid #1f2937; margin-top: 10px; font-size: 0.85rem;">
        <div style="display: flex; gap: 20px; flex-wrap: wrap;">
            <div><b>Artist:</b> {row['name']}</div>
            <div><b>Evidence Tier:</b> Tier {int(row.get('platform_evidence_tier', 1))}</div>
            <div><b>Instagram Status:</b> {row.get('instagram_verification_status', 'N/A')}</div>
            <div><b>Spotify Source:</b> {row.get('spotify_data_source', 'N/A')}</div>
            <div><b>Effective Weights:</b> {row.get('effective_weights', 'N/A')}</div>
            <div><b>Anomaly Flags:</b> {row.get('anomaly_flags', 'CLEAN')}</div>
        </div>
    </div>
    """,
    unsafe_allow_html=True,
)