import pandas as pd
import requests
import streamlit as st

# URL of the Flask backend deployed in the Hugging Face Docker space
API_URL = "https://shravanda-extraalearnleadbackend.hf.space"

st.set_page_config(page_title="ExtraaLearn Lead Conversion Predictor", page_icon="🎓", layout="centered")
st.title("🎓 ExtraaLearn Lead Conversion Predictor")
st.write("Predict whether a lead is likely to convert into a paid customer, so the sales team "
         "can prioritise follow-ups.")

tab_single, tab_batch = st.tabs(["Single lead", "Batch scoring (CSV)"])

with tab_single:
    with st.form("lead_form"):
        c1, c2 = st.columns(2)
        with c1:
            age = st.number_input("Age", min_value=18, max_value=100, value=45)
            current_occupation = st.selectbox("Current occupation", ["Professional", "Unemployed", "Student"])
            first_interaction = st.selectbox("First interaction", ["Website", "Mobile App"])
            profile_completed = st.selectbox("Profile completed", ["High", "Medium", "Low"],
                                             help="Low: 0-50%, Medium: 50-75%, High: 75-100%")
            last_activity = st.selectbox("Last activity", ["Email Activity", "Phone Activity", "Website Activity"])
        with c2:
            website_visits = st.number_input("Website visits", min_value=0, max_value=100, value=3)
            time_spent_on_website = st.number_input("Time spent on website (seconds)", min_value=0,
                                                    max_value=10000, value=600)
            page_views_per_visit = st.number_input("Page views per visit", min_value=0.0, max_value=50.0,
                                                   value=2.8, step=0.1)
        st.markdown("**How did the lead hear about ExtraaLearn?**")
        f1, f2, f3, f4, f5 = st.columns(5)
        print_media_type1 = f1.checkbox("Newspaper")
        print_media_type2 = f2.checkbox("Magazine")
        digital_media = f3.checkbox("Digital ads")
        educational_channels = f4.checkbox("Edu channels")
        referral = f5.checkbox("Referral")
        submitted = st.form_submit_button("Predict", type="primary")

    if submitted:
        yes_no = lambda flag: "Yes" if flag else "No"
        lead = {
            "age": int(age), "current_occupation": current_occupation, "first_interaction": first_interaction,
            "profile_completed": profile_completed, "website_visits": int(website_visits),
            "time_spent_on_website": int(time_spent_on_website),
            "page_views_per_visit": float(page_views_per_visit), "last_activity": last_activity,
            "print_media_type1": yes_no(print_media_type1), "print_media_type2": yes_no(print_media_type2),
            "digital_media": yes_no(digital_media), "educational_channels": yes_no(educational_channels),
            "referral": yes_no(referral),
        }
        try:
            response = requests.post(f"{API_URL}/v1/lead", json=lead, timeout=60)
            response.raise_for_status()
            result = response.json()
            prob = result["conversion_probability"]
            if result["prediction"] == 1:
                st.success(f"✅ **Likely to convert**: conversion probability {prob:.0%}. "
                           "Prioritise this lead for a call from a counsellor.")
            else:
                st.warning(f"⚠️ **Unlikely to convert**: conversion probability {prob:.0%}. "
                           "Keep nurturing through low-cost channels (email, content).")
            st.progress(prob)
        except requests.RequestException as e:
            st.error(f"Could not reach the prediction API: {e}")

with tab_batch:
    st.write("Upload a CSV with the same columns as the ExtraaLearn leads data (an `ID` column is optional).")
    uploaded = st.file_uploader("Leads CSV", type="csv")
    if uploaded is not None and st.button("Score leads", type="primary"):
        try:
            response = requests.post(f"{API_URL}/v1/leadbatch",
                                     files={"file": (uploaded.name, uploaded.getvalue(), "text/csv")},
                                     timeout=120)
            response.raise_for_status()
            scored = pd.DataFrame(response.json()).sort_values("conversion_probability", ascending=False)
            st.metric("Leads likely to convert", f"{scored['prediction'].sum()} of {len(scored)}")
            st.dataframe(scored, use_container_width=True, hide_index=True)
            st.download_button("Download scored leads", scored.to_csv(index=False),
                               file_name="scored_leads.csv", mime="text/csv")
        except requests.RequestException as e:
            st.error(f"Could not reach the prediction API: {e}")