app.py

# app.py
import streamlit as st
import pandas as pd
import numpy as np
import plotly.express as px
import os
import warnings
warnings.filterwarnings('ignore')

# ---------- Page & Styles ----------
st.set_page_config(page_title="SAP Sales KPI Dashboard", page_icon="📊", layout="wide")
st.markdown("""
<style>
  /* hide default sidebar entirely */
  [data-testid="stSidebar"] { display: none; }
  .main-header { font-size: 2.2rem; font-weight: 800; color: #1f4e79; text-align: left; margin-bottom: .25rem; }
  .subtle { color:#6b7280; margin-bottom:1.25rem; }
  .filter-card { background:#f8f9fa; padding: .9rem 1rem; border-radius:12px; border:1px solid #edf2f7; }
  .kpi-card { background: #ffffff; padding: 1.25rem; border-radius: 14px; border:1px solid #e5e7eb; box-shadow: 0 1px 2px rgba(0,0,0,.03); }
  .kpi-value { font-size: 2.1rem; font-weight: 800; color: #1f4e79; line-height:1; }
  .kpi-label { font-size: .95rem; color: #6b7280; }
  .block-container { padding-top: 1.2rem; }
</style>
""", unsafe_allow_html=True)

# ---------- Kaggle load ----------
@st.cache_data(ttl=3600)
def load_kaggle_sap_data():
    try:
        if 'KAGGLE_USERNAME' not in st.secrets or 'KAGGLE_KEY' not in st.secrets:
            return "Kaggle credentials not found in Streamlit secrets."
        os.environ['KAGGLE_USERNAME'] = st.secrets['KAGGLE_USERNAME']
        os.environ['KAGGLE_KEY'] = st.secrets['KAGGLE_KEY']

        import kaggle
        dataset_name = "mustafakeser4/sap-dataset-bigquery-dataset"
        download_path = "./kaggle_data"

        if not os.path.exists(os.path.join(download_path, 'vbak.csv')):
            with st.spinner("Downloading dataset from Kaggle..."):
                kaggle.api.authenticate()
                kaggle.api.dataset_download_files(dataset_name, path=download_path, unzip=True)

        needed = {'vbak': 'vbak.csv', 'vbap': 'vbap.csv', 'kna1': 'kna1.csv', 'makt': 'makt.csv'}
        tables = {}
        for k, fn in needed.items():
            fp = os.path.join(download_path, fn)
            if not os.path.exists(fp):
                return f"Expected file missing: {fn}"
            tables[k] = pd.read_csv(fp, low_memory=False)
        return tables
    except Exception as e:
        return f"Error during Kaggle data loading: {e}"

# ---------- Processing (robust revenue + safe merges) ----------
@st.cache_data
def create_sales_analytics(_tables):
    try:
        vbak = _tables['vbak'].copy()
        vbap = _tables['vbap'].copy()
        kna1 = _tables['kna1'].copy()
        makt = _tables['makt'].copy()

        # normalize column names
        for df in [vbak, vbap, kna1, makt]:
            df.columns = [c.upper().strip() for c in df.columns]

        # keep SAP keys as strings (avoid leading-zero loss)
        def as_str(df, cols):
            for c in cols:
                if c in df.columns:
                    df[c] = df[c].astype(str).str.strip()

        as_str(vbak, ['VBELN','KUNNR','VKORG','VTWEG'])
        as_str(vbap, ['VBELN','MATNR'])
        as_str(kna1, ['KUNNR'])
        as_str(makt, ['MATNR'])

        # choose numeric helper
        def pick_numeric(df, cols):
            for c in cols:
                if c in df.columns:
                    s = pd.to_numeric(df[c], errors='coerce')
                    if s.notna().sum() > 0 and s.abs().sum() > 0:
                        return s
            return pd.Series(0.0, index=df.index)

        # Build item-level REVENUE
        # primary: NETWR at item level (VBAP)
        netwr_item = pick_numeric(vbap, ['NETWR'])

        # fallback: price * qty using common SAP columns
        price = pick_numeric(vbap, ['NETPR', 'KBETR', 'NETPR_I'])
        qty   = pick_numeric(vbap, ['KWMENG', 'KTMNG', 'MENGE'])
        fallback_rev = (price.fillna(0) * qty.fillna(0)).fillna(0)

        vbap['REVENUE'] = np.where(netwr_item > 0, netwr_item, fallback_rev).astype(float)

        # header fields (include currency if present)
        keep_vbak = ['VBELN','KUNNR','VKORG','VTWEG','ERDAT'] + (['WAERK'] if 'WAERK' in vbak.columns else [])
        vbak_small = vbak[keep_vbak].drop_duplicates('VBELN')
        vbap_small = vbap[['VBELN','MATNR','REVENUE']]
        kna1_small = kna1[['KUNNR','NAME1','LAND1']] if {'KUNNR','NAME1','LAND1'}.issubset(kna1.columns) else pd.DataFrame(columns=['KUNNR','NAME1','LAND1'])

        # product text in English
        makt_en = makt[makt['SPRAS'].eq('E')] if 'SPRAS' in makt.columns else makt
        makt_small = makt_en[['MATNR','MAKTX']].drop_duplicates('MATNR') if {'MATNR','MAKTX'}.issubset(makt_en.columns) else pd.DataFrame(columns=['MATNR','MAKTX'])

        # final sales table
        sales = (vbap_small
                 .merge(vbak_small, on='VBELN', how='inner')
                 .merge(kna1_small, on='KUNNR', how='left')
                 .merge(makt_small, on='MATNR', how='left'))

        if 'ERDAT' in sales.columns:
            sales['ERDAT'] = pd.to_datetime(sales['ERDAT'], errors='coerce')

        # ensure columns exist
        for col in ['REVENUE','LAND1','VTWEG','NAME1','MAKTX','VBELN','KUNNR','VKORG']:
            if col not in sales.columns:
                sales[col] = np.nan if col != 'REVENUE' else 0.0
        if 'WAERK' not in sales.columns:
            sales['WAERK'] = 'N/A'

        # drop obvious junk rows
        sales = sales.replace([np.inf, -np.inf], np.nan).dropna(subset=['REVENUE'])
        return sales
    except Exception as e:
        return f"Error processing sales data: {e}"

# ---------- App ----------
st.markdown('<div class="main-header">🎯 SAP Sales KPI Dashboard</div><div class="subtle">Real SAP ERP sample data (Kaggle)</div>', unsafe_allow_html=True)

tables = load_kaggle_sap_data()
if isinstance(tables, str):
    st.error(tables)
    st.stop()

sales_df = create_sales_analytics(tables)
if isinstance(sales_df, str):
    st.error(sales_df)
    st.stop()

# ---------- Filter Bar (no sidebar) ----------
with st.container():
    st.markdown('<div class="filter-card">', unsafe_allow_html=True)
    c1, c2, c3, c4 = st.columns([1.2, 1.2, 3, 0.9])

    # currency filter
    currencies = [c for c in sales_df['WAERK'].dropna().unique().tolist() if c != 'N/A']
    default_cur = sales_df['WAERK'].mode().iat[0] if len(sales_df) and sales_df['WAERK'].notna().any() else 'N/A'
    with c1:
        currency = st.selectbox("Currency", options=(['All'] + sorted(currencies)) if currencies else ['All'], index=0 if currencies else 0)

    # Top N
    with c2:
        top_n_countries = st.slider("Top N Countries", 5, 20, 10)

    # Region multiselect inside expander to keep tidy
    with c3:
        with st.expander("Region (Country) – click to choose", expanded=False):
            all_countries = sorted(sales_df['LAND1'].dropna().unique().tolist())
            # buttons to select/clear
            b1, b2 = st.columns([1,1])
            if 'selected_countries' not in st.session_state:
                st.session_state.selected_countries = all_countries
            with b1:
                if st.button("Select All"):
                    st.session_state.selected_countries = all_countries
            with b2:
                if st.button("Clear"):
                    st.session_state.selected_countries = []
            selected_region = st.multiselect("Countries", options=all_countries, default=st.session_state.selected_countries, key="countries_ms")

    with c4:
        if st.button("🔄 Clear Cache"):
            st.cache_data.clear()
            st.rerun()
    st.markdown('</div>', unsafe_allow_html=True)

# apply filters
filtered_df = sales_df.copy()
if currency and currency != 'All':
    filtered_df = filtered_df[filtered_df['WAERK'] == currency]
if 'countries_ms' in st.session_state:
    filtered_df = filtered_df[filtered_df['LAND1'].isin(st.session_state.countries_ms)]

st.success(f"✅ Loaded and processed {len(filtered_df):,} sales line-items after filters.")

# ---------- KPIs ----------
st.subheader("Sales KPIs")
k1,k2,k3,k4 = st.columns(4)
with k1: st.markdown(f'<div class="kpi-card"><div class="kpi-value">${float(filtered_df["REVENUE"].sum()):,.0f}</div><div class="kpi-label">Total Revenue</div></div>', unsafe_allow_html=True)
with k2: st.markdown(f'<div class="kpi-card"><div class="kpi-value">{int(filtered_df["KUNNR"].nunique())}</div><div class="kpi-label">Active Customers</div></div>', unsafe_allow_html=True)
with k3:
    aov = float(filtered_df.loc[filtered_df['REVENUE']>0,'REVENUE'].mean() or 0.0)
    st.markdown(f'<div class="kpi-card"><div class="kpi-value">${aov:,.0f}</div><div class="kpi-label">Avg Order Value (item)</div></div>', unsafe_allow_html=True)
with k4: st.markdown(f'<div class="kpi-card"><div class="kpi-value">{int(filtered_df["VBELN"].nunique())}</div><div class="kpi-label">Sales Orders</div></div>', unsafe_allow_html=True)

# ---------- Tabs ----------
tab1, tab2, tab3, tab4 = st.tabs(["👥 Top Customers", "🌍 Regional Analysis", "📈 Distribution Channels", "🛍️ Top Products"])

with tab1:
    st.subheader("Top 10 Customers by Revenue")
    customer_summary = (filtered_df.dropna(subset=['NAME1'])
                        .groupby('NAME1', as_index=False)['REVENUE'].sum()
                        .nlargest(10, 'REVENUE'))
    if customer_summary.empty:
        st.info("No customer data to display.")
    else:
        fig = px.bar(customer_summary, x='REVENUE', y='NAME1', orientation='h',
                     labels={'REVENUE':'Revenue','NAME1':'Customer'}, color='REVENUE')
        st.plotly_chart(fig.update_layout(yaxis={'categoryorder':'total ascending'}), use_container_width=True)

with tab2:
    st.subheader("Revenue by Country")
    regional_summary = (filtered_df.dropna(subset=['LAND1'])
                        .groupby('LAND1', as_index=False)['REVENUE'].sum()
                        .nlargest(top_n_countries, 'REVENUE'))
    if regional_summary.empty:
        st.info("No country data to display.")
    else:
        fig = px.pie(regional_summary, values='REVENUE', names='LAND1',
                     title=f"Top {top_n_countries} Countries by Revenue")
        st.plotly_chart(fig, use_container_width=True)

with tab3:
    st.subheader("Revenue by Distribution Channel")
    channel_summary = (filtered_df.dropna(subset=['VTWEG'])
                       .groupby('VTWEG', as_index=False)['REVENUE'].sum())
    channel_summary['VTWEG'] = channel_summary['VTWEG'].astype(str)
    if channel_summary.empty:
        st.info("No distribution channel data to display.")
    else:
        fig = px.bar(channel_summary, x='VTWEG', y='REVENUE',
                     title="Total Revenue by Distribution Channel",
                     labels={'REVENUE':'Total Revenue','VTWEG':'Distribution Channel'},
                     color='REVENUE')
        st.plotly_chart(fig, use_container_width=True)

with tab4:
    st.subheader("Top 10 Products by Revenue")
    product_summary = (filtered_df.dropna(subset=['MAKTX'])
                       .groupby('MAKTX', as_index=False)['REVENUE'].sum()
                       .nlargest(10, 'REVENUE'))
    if product_summary.empty:
        st.info("No product data to display.")
    else:
        fig = px.bar(product_summary, x='REVENUE', y='MAKTX', orientation='h',
                     labels={'REVENUE':'Revenue','MAKTX':'Product'}, color='REVENUE')
        st.plotly_chart(fig.update_layout(yaxis={'categoryorder':'total ascending'}), use_container_width=True)

st.markdown("---")
st.markdown("<p style='text-align:center;'>Built with Streamlit • Real SAP ERP sample data (Kaggle)</p>", unsafe_allow_html=True)