import streamlit as st
import pandas as pd
from datetime import datetime
import os
import matplotlib.pyplot as plt
import seaborn as sns
from huggingface_hub import HfApi, upload_file, list_repo_files, hf_hub_download

# Configuration for Hugging Face Repository
REPO_ID = "MarcosRodrigo/Breakfast-Poll"
HISTORY_DIR = "history"
TEMP_FILE = "current_selections.csv"

# Hugging Face API (requires a token with write access)
hf_token = st.secrets["HF_TOKEN"]
api = HfApi()

# Initialize all required session state variables
if "users" not in st.session_state:
    st.session_state.users = []
if "current_selections" not in st.session_state:
    st.session_state.current_selections = []
if "step" not in st.session_state:
    st.session_state.step = 1
if "history" not in st.session_state:
    st.session_state.history = []

# Load temporary selections from the shared file
def load_current_selections():
    if os.path.exists(TEMP_FILE):
        return pd.read_csv(TEMP_FILE)
    else:
        return pd.DataFrame(columns=["Name", "Drinks", "Food"])

# Save current user selections to the shared CSV file without overwriting previous data
def save_current_selection_to_file(current_selections):
    current_selections["Drinks"] = current_selections["Drinks"].apply(lambda x: ", ".join(x) if isinstance(x, list) else x)
    current_selections["Food"] = current_selections["Food"].apply(lambda x: ", ".join(x) if isinstance(x, list) else x)

    if os.path.exists(TEMP_FILE):
        existing_selections = pd.read_csv(TEMP_FILE)
        combined_selections = pd.concat([existing_selections, current_selections]).drop_duplicates()
    else:
        combined_selections = current_selections

    combined_selections.to_csv(TEMP_FILE, index=False)

# Upload the shared file to Hugging Face repository for persistence
def upload_temp_file_to_repo():
    if os.path.exists(TEMP_FILE):
        upload_file(
            path_or_fileobj=TEMP_FILE,
            path_in_repo=TEMP_FILE,
            repo_id=REPO_ID,
            token=hf_token,
            repo_type="space"
        )

# Delete a file from the repository (e.g., `current_selections.csv`)
def delete_file_from_repo(filename):
    api.delete_file(
        path_in_repo=filename,
        repo_id=REPO_ID,
        token=hf_token,
        repo_type="space"
    )

# Download the shared file from the repository to ensure persistence and real-time updates
def download_temp_file_from_repo():
    try:
        hf_hub_download(repo_id=REPO_ID, filename=TEMP_FILE, repo_type="space", token=hf_token, local_dir=".")
    except Exception:
        pd.DataFrame(columns=["Name", "Drinks", "Food"]).to_csv(TEMP_FILE, index=False)

# Load history from the repository
def load_history():
    history = []
    files_in_repo = list_repo_files(REPO_ID, token=hf_token, repo_type="space")
    history_files = [f for f in files_in_repo if f.startswith(f"{HISTORY_DIR}/") and f.endswith(".txt")]

    for file in history_files:
        local_filepath = hf_hub_download(repo_id=REPO_ID, filename=file, token=hf_token, repo_type="space")
        summary_df = pd.read_csv(local_filepath)
        date = file.split("/")[-1].split(".txt")[0]
        history.append({"Date": date, "Summary": summary_df})
    return history

# Save the current summary to a text file in the history directory
def save_summary_to_history():
    timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
    history_filename = f"{HISTORY_DIR}/{timestamp}.txt"

    if not os.path.exists(HISTORY_DIR):
        os.makedirs(HISTORY_DIR)

    if os.path.exists(TEMP_FILE):
        summary_df = pd.read_csv(TEMP_FILE)
        summary_df.to_csv(history_filename, index=False)

        upload_file(path_or_fileobj=history_filename, path_in_repo=history_filename, repo_id=REPO_ID, token=hf_token, repo_type="space")
    return timestamp

# Load persistent history and temporary selections on app start
if "history" not in st.session_state:
    download_temp_file_from_repo()
    st.session_state.history = load_history()
    st.session_state.current_selections = load_current_selections().to_dict(orient="records")

# Sidebar for navigating through different views
menu = st.sidebar.selectbox("Select View", ["Poll", "Current", "History", "Graph"])

# Function to reset the current selections after submission
def reset_selections():
    st.session_state.users = []
    st.session_state.current_selections = []

# Poll view with four consecutive steps
if menu == "Poll":
    st.title("Breakfast Poll Application")

    # Step 1: User's Name
    st.header("Step 1: Enter your name")
    name = st.text_input("Name:")
    if st.button("Next", key="step1_next") and name:
        st.session_state.users.append(name)
        st.session_state.step = 2  # Set the next step to be visible

    # Show Step 2 only if Step 1 is completed
    if st.session_state.step >= 2:
        st.header("Step 2: Select your drink(s)")
        drinks_options = [
            "Café con leche", "Colacao", "Descafeinado con leche", "Cortado", 
            "Aguasusia", "Aguasusia susia", "Café descafeinado con leche desnatada", 
            "Italiano", "Café con soja", "Té", "Manzanilla", "Nada"
        ]
        selected_drinks = st.multiselect("Choose your drinks:", drinks_options)

        if st.button("Next", key="step2_next") and selected_drinks:
            st.session_state.current_selections.append({"Name": st.session_state.users[-1], "Drinks": selected_drinks})
            st.session_state.step = 3  # Set the next step to be visible

    # Show Step 3 only if Step 2 is completed
    if st.session_state.step >= 3:
        st.header("Step 3: Select your food(s)")
        food_options = [
            "Barrita con aceite", "Barrita con tomate", "Palmera de chocolate", 
            "Palmera de chocolate blanco", "Yogurt", "Pincho de tortilla", "Nada"
        ]
        selected_food = st.multiselect("Choose your food:", food_options)

        if st.button("Save Selections", key="save_selections") and selected_food:
            st.session_state.current_selections[-1]["Food"] = selected_food
            df = pd.DataFrame(st.session_state.current_selections)
            save_current_selection_to_file(df)
            upload_temp_file_to_repo()
            st.success(f"Selections saved for {st.session_state.users[-1]}!")
            st.session_state.step = 1  # Reset to step 1 for the next user

# "Current" view to display the current summary of all users' selections and submit to history
elif menu == "Current":
    st.title("Current Selections of All Users")
    if st.button("Reload Selections"):
        download_temp_file_from_repo()
    current_df = load_current_selections()
    st.table(current_df)

    if st.button("Submit Summary to History"):
        timestamp = save_summary_to_history()
        st.success(f"Summary saved to history at {timestamp}")
        st.session_state.history = load_history()
        
        # Clear local and remote current selections
        if os.path.exists(TEMP_FILE):
            os.remove(TEMP_FILE)
            delete_file_from_repo(TEMP_FILE)  # Delete the file from the remote repo

            # Create an empty CSV to replace the deleted one
            pd.DataFrame(columns=["Name", "Drinks", "Food"]).to_csv(TEMP_FILE, index=False)
            upload_temp_file_to_repo()

        # st.experimental_set_query_params(step="reset")

# History view to check past summaries
elif menu == "History":
    st.title("Breakfast Poll History")

    # Reload history if it's not already loaded
    if not st.session_state.history:
        st.session_state.history = load_history()
        
    if st.session_state.history:
        # Display history in reverse chronological order
        for record in reversed(st.session_state.history):
            st.subheader(f"Date: {record['Date']}")
            st.table(record["Summary"])
    else:
        st.write("No history records found.")

# "Current" view to display the current summary of all users' selections and submit to history
elif menu == "Current":
    st.title("Current Selections of All Users")
    if st.button("Reload Selections"):
        download_temp_file_from_repo()
    current_df = load_current_selections()
    st.table(current_df)

    if st.button("Submit Summary to History"):
        timestamp = save_summary_to_history()
        st.success(f"Summary saved to history at {timestamp}")
        st.session_state.history = load_history()
        
        # Clear local and remote current selections
        if os.path.exists(TEMP_FILE):
            os.remove(TEMP_FILE)
            delete_file_from_repo(TEMP_FILE)  # Delete the file from the remote repo

            # Create an empty CSV to replace the deleted one
            pd.DataFrame(columns=["Name", "Drinks", "Food"]).to_csv(TEMP_FILE, index=False)
            upload_temp_file_to_repo()

        # st.experimental_set_query_params(step="reset")

# History view to check past summaries
elif menu == "History":
    st.title("Breakfast Poll History")

    # Reload history if it's not already loaded
    if not st.session_state.history:
        st.session_state.history = load_history()
        
    if st.session_state.history:
        # Display history in reverse chronological order
        for record in reversed(st.session_state.history):
            st.subheader(f"Date: {record['Date']}")
            st.table(record["Summary"])
    else:
        st.write("No history records found.")

# Graph view to display a line chart of item selections over time
elif menu == "Graph":
    st.title("Breakfast Poll History - Graph View")

    # Prepare data for plotting
    if st.session_state.history:
        history_data = []
        for record in st.session_state.history:
            date = record['Date']
            for index, row in record['Summary'].iterrows():
                for drink in row['Drinks'].split(', '):
                    history_data.append({'Date': date, 'Item': drink, 'Type': 'Drink'})
                for food in row['Food'].split(', '):
                    history_data.append({'Date': date, 'Item': food, 'Type': 'Food'})

        # Create a DataFrame from history data
        history_df = pd.DataFrame(history_data)
        
        # Count occurrences of each item per date
        item_counts = history_df.groupby(['Date', 'Item']).size().reset_index(name='Count')

        # Create a line plot for each item over time
        plt.figure(figsize=(12, 6))
        sns.lineplot(data=item_counts, x='Date', y='Count', hue='Item', marker='o')

        # Customize the plot
        plt.xticks(rotation=45)
        plt.title('Item Selections Over Time')
        plt.xlabel('Date')
        plt.ylabel('Number of Selections')
        plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.15), ncol=3)
        
        # Display the plot
        st.pyplot(plt.gcf())
    else:
        st.write("No historical data available to plot.")