import streamlit as st
import numpy as np
import pandas as pd
import plotly.graph_objects as go
import pickle
from keras.models import load_model

# ============================================================
# SETTINGS & MODEL LOAD
# ============================================================
st.set_page_config(page_title="Workforce AI Optimizer", layout="wide")

# Sidebar genişliğini sabitleyen CSS
st.markdown(
    """
    <style>
    [data-testid="stSidebar"][aria-expanded="true"]{
        min-width: 360px;
        max-width: 360px;
    }
    </style>
    """,
    unsafe_allow_html=True,
)

@st.cache_resource
def load_assets():
    model = load_model("lstm_traffic_model.keras")
    sc = pickle.load(open("scaler.pkl", "rb"))
    return model, sc

try:
    model, sc = load_assets()
except Exception as e:
    st.error(f"Error: Model files not found! -> {e}")
    st.stop()

# ============================================================
# SIDEBAR / 🛠️ SETUP & GUIDE
# ============================================================
st.sidebar.title("🛠️ Setup & Guide / Rehber")

st.sidebar.markdown("""
**Data Format / Veri Formatı:**
The CSV should contain historical call data. / CSV geçmiş çağrı verilerini içermelidir.
""")

# SLOT MANTIĞI - 6 Örnekli ve Noktalı Versiyon
st.sidebar.info("""
**Slot Logic / Slot Mantığı:**
- **Slot 0:** 08:00-09:00 (100 cals)
- **Slot 1:** 09:00-10:00 (150 cals)
- **Slot 2:** 10:00-11:00 (120 cals)
- **Slot 3:** 11:00-12:00 (180 cals)
- **Slot 4:** 12:00-13:00 (200 cals)
- **Slot 5:** 13:00-14:00 (160 cals)
- **...**
- **Slot 81:** Midnight (5 cals)
""")

st.sidebar.subheader("Sample CSV / Örnek Yapı")
example_df = pd.DataFrame({"calls": [105, 140, 88, 120, 200, 160]})
st.sidebar.dataframe(example_df, use_container_width=True)

st.sidebar.warning("⚠️ **Column Name:** 'calls' or 'Incoming Calls'")

st.sidebar.markdown("---")
st.sidebar.subheader("💰 Cost Settings / Maliyet")
wage = st.sidebar.number_input("Hourly Wage / Saatlik Ücret ($)", value=20)
capacity = st.sidebar.number_input("Calls per Staff / Kapasite", value=15)

# ============================================================
# MAIN UI
# ============================================================
st.title("📞 Workforce Optimization AI / İş Gücü Optimizasyonu")
st.write("Ensuring the right number of people at the right time.")
st.markdown("---")

file = st.file_uploader("Upload CSV / CSV Yükle", type=["csv"])

if file is not None:
    df = pd.read_csv(file)
    target_col = "calls" if "calls" in df.columns else ("Incoming Calls" if "Incoming Calls" in df.columns else None)
    
    if target_col is None:
        st.error("❌ Column not found!")
        st.stop()
    
    raw_data = df[[target_col]].values

    # PREDICTION
    scaled_data = sc.transform(raw_data)
    pred_scaled = model.predict(scaled_data)
    predictions = sc.inverse_transform(pred_scaled)
    needed_staff = np.ceil(predictions / capacity).flatten().astype(int)

    # 🎯 DECISION LOGIC
    st.header("🎯 Decision Logic / Karar Mantığı")
    logic_col1, logic_col2 = st.columns(2)

    with logic_col1:
        st.error("### 🔥 High Intensity (Yüksek Yoğunluk)")
        st.write("**Advice:** INCREASE STAFF to protect quality.")
        st.write("**Öneri:** Kalite için PERSONEL ARTIRIN.")

    with logic_col2:
        st.success("### 💰 Saving Area (Tasarruf Alanı)")
        st.write("**Advice:** REDUCE STAFF to maximize profit.")
        st.write("**Öneri:** Kâr için PERSONELİ AZALTIN.")

    # 📈 GRAPH
    st.markdown("---")
    st.subheader("📈 Capacity Analysis / Kapasite Analizi")
    fig = go.Figure()
    fig.add_trace(go.Scatter(y=raw_data.flatten(), name="Past", line=dict(color="gray")))
    fig.add_trace(go.Scatter(y=predictions.flatten(), name="AI Forecast", line=dict(color="#1C83E1", width=3)))
    fig.add_trace(go.Bar(y=needed_staff * capacity, name="Capacity", opacity=0.2, marker_color="green"))
    fig.update_layout(hovermode="x unified", template="plotly_white", height=400)
    st.plotly_chart(fig, use_container_width=True)

    # 🧠 RECOMMENDATIONS
    st.header("🧠 AI Strategic Recommendations")
    mean_val = np.mean(needed_staff)
    peak_indices = np.where(needed_staff > mean_val * 1.25)[0].tolist()
    low_indices = np.where(needed_staff < mean_val * 0.75)[0].tolist()

    c1, c2 = st.columns(2)
    with c1:
        st.error(f"### 🚨 High Intensity")
        if peak_indices:
            st.write(f"**At:** {', '.join([f'Slot {i}' for i in peak_indices[:5]])}...")
            st.write("Increase staff. / Personel artırın.")
        else:
            st.write("No major peaks.")

    with c2:
        st.success(f"### 📉 Saving Area")
        if low_indices:
            st.write(f"**At:** {', '.join([f'Slot {i}' for i in low_indices[:5]])}...")
            st.write("Reduce staff. / Personeli azaltın.")
        else:
            st.write("No saving opportunity.")

    # 📊 TABLE
    with st.expander("📊 Detailed Schedule Table"):
        res_df = pd.DataFrame({
            "Time Slot": [f"Slot {i}" for i in range(len(predictions))],
            "Predicted Demand": predictions.flatten().astype(int),
            "Suggested Staff": needed_staff
        })
        st.dataframe(res_df, use_container_width=True)

else:
    st.info("👋 Please upload your CSV file to begin. / Başlamak için CSV yükleyin.")