import gradio as gr
import pandas as pd
from datetime import datetime, timedelta
import os

CSV_LOG = "baby_log.csv"

# Load or initialize log
if os.path.exists(CSV_LOG):
    log_df = pd.read_csv(CSV_LOG)
    log_df["datetime"] = pd.to_datetime(log_df["datetime"])
else:
    log_df = pd.DataFrame(columns=["datetime", "event"])

def save_logs():
    log_df.to_csv(CSV_LOG, index=False)

def parse_time(hour, minute, ampm):
    hour = int(hour)
    minute = int(minute)
    if ampm == "PM" and hour < 12:
        hour += 12
    if ampm == "AM" and hour == 12:
        hour = 0
    now = datetime.now()
    return now.replace(hour=hour, minute=minute, second=0, microsecond=0)

def get_next_feeding_time(last_time):
    hour = last_time.hour
    if 21 <= hour or hour < 10:
        return last_time + timedelta(hours=4)
    else:
        return last_time + timedelta(hours=2)

def summarize_today(df):
    today = datetime.now().date()
    df_today = df[df["datetime"].dt.date == today]
    return f"🍼 Feedings today: {len(df_today)}"

def avg_feed_gap(df):
    df = df.sort_values("datetime")
    df_today = df[df["datetime"].dt.date == datetime.now().date()]
    times = df_today["datetime"].tolist()
    if len(times) < 2:
        return "⏱️ Avg Gap: N/A"
    gaps = [(t2 - t1).total_seconds() for t1, t2 in zip(times[:-1], times[1:])]
    avg_gap = sum(gaps) / len(gaps)
    avg_td = timedelta(seconds=avg_gap)
    return f"⏱️ Avg Gap: {str(avg_td).split('.')[0]}"

def log_feeding(hour_str, minute_str, ampm):
    global log_df
    try:
        hour = int(hour_str)
        minute = int(minute_str)
    except:
        return "Invalid time input", pd.DataFrame(), "", ""
    feed_time = parse_time(hour, minute, ampm)
    new_entry = pd.DataFrame([{"datetime": feed_time, "event": "Feed"}])
    log_df = pd.concat([log_df, new_entry], ignore_index=True)
    save_logs()
    summary = summarize_today(log_df)
    gap = avg_feed_gap(log_df)
    next_feed_time = get_next_feeding_time(feed_time)
    return next_feed_time.strftime("%I:%M %p"), log_df.tail(50).reset_index(drop=True), summary, gap

def check_next_feeding():
    if log_df.empty:
        return "No previous feeding logged."
    last_time = log_df["datetime"].max()
    next_time = get_next_feeding_time(last_time)
    return f"Next feeding should be around: {next_time.strftime('%I:%M %p')}"

def visualize_log():
    return log_df.tail(50).reset_index(drop=True), summarize_today(log_df), avg_feed_gap(log_df)

with gr.Blocks() as app:
    gr.Markdown("## 👶 Smart Baby Feed Tracker")

    with gr.Row():
        hour_input = gr.Textbox(label="Hour (1-12)", value="10")
        minute_input = gr.Textbox(label="Minute (0-59)", value="00")
        ampm_dropdown = gr.Dropdown(["AM", "PM"], label="AM/PM", value="AM")
        log_btn = gr.Button("✔️ Log Feeding")

    next_feed_time = gr.Textbox(label="⏭️ Next Feeding Time")
    feed_table = gr.Dataframe(label="Recent Feed Log")
    today_summary = gr.Textbox(label="Today's Feed Count")
    avg_gap = gr.Textbox(label="Average Feed Gap")
    log_btn.click(log_feeding, inputs=[hour_input, minute_input, ampm_dropdown],
                  outputs=[next_feed_time, feed_table, today_summary, avg_gap])

    gr.Markdown("### 📋 View Log & Stats Anytime")
    view_btn = gr.Button("📊 Visualize Log")
    view_table = gr.Dataframe(label="Recent Feed Log")
    view_summary = gr.Textbox(label="Today's Feed Count")
    view_gap = gr.Textbox(label="Average Feed Gap")
    view_btn.click(visualize_log, inputs=[], outputs=[view_table, view_summary, view_gap])

    gr.Markdown("### 🔄 Check From Last Feeding")
    check_btn = gr.Button("⏭️ Predict Next Feeding Time")
    next_check = gr.Textbox(label="Predicted Time From Log")
    check_btn.click(check_next_feeding, inputs=[], outputs=next_check)

app.launch()