app_V1.py

import html
import json
import os
from io import StringIO
import streamlit as st
import pandas as pd
from bs4 import BeautifulSoup
from snowflake.snowpark import Session

from Messaging_system.Permes import Permes
from Messaging_system.context_validator import Validator
from dotenv import load_dotenv
load_dotenv()

# -----------------------------------------------------------------------
# Load CSV file
@st.cache_data
def load_data(file_path):
    return pd.read_csv(file_path)

# -----------------------------------------------------------------------

def load_config_(file_path):
    """
    Loads configuration JSON files from the local space. (mostly for loading the Snowflake connection parameters)
    :param file_path: local path to the JSON file
    :return: JSON file
    """
    with open(file_path, 'r') as file:
        return json.load(file)


# -----------------------------------------------------------------------


# Set page configuration and apply custom CSS for black and gold theme
st.set_page_config(page_title="Personalized Message Generator", page_icon=":mailbox_with_mail:", layout="wide")

st.markdown(
    """
    <style>
    body {
        background-color: #000000;
        color: #FFD700;
    }
    .stButton > button {
        background-color: #FFD700;
        color: #000000;
    }
    h1, h2, h3, h4, h5, h6 {
        color: #FFD700;
    }
    .section {
        margin-bottom: 30px;
    }
    .input-label {
        font-size: 18px;
        font-weight: bold;
        margin-top: 10px;
    }
    </style>
    """,
    unsafe_allow_html=True
)


# -----------------------------------------------------------------------

def filter_validated_users(users):
    """
    Filters the input DataFrame by removing rows where the 'valid' column has the value 'False'.

    Parameters:
    users (DataFrame): A pandas DataFrame with a 'valid' column containing strings 'True' or 'False'.

    Returns:
    DataFrame: A filtered DataFrame containing only rows where 'valid' is 'True'.
    """
    # Convert the 'valid' column to boolean for easier filtering
    users['valid'] = users['valid'].map({'True': True, 'False': False})

    # Filter the DataFrame to include only rows where 'valid' is True
    filtered_users = users[users['valid']]

    # Optional: Reset the index of the filtered DataFrame
    filtered_users = filtered_users.reset_index(drop=True)

    return filtered_users

# -----------------------------------------------------------------------

# --------------------------------------------------------------



# --------------------------------------------------------------
def clean_html_tags(users_df):
    """
    accept the data as a Pandas Dataframe and return the preprocessed dataframe.
    This function has access to the columns that contain HTML tags and codes, Therefore it will apply cleaning
    procedures to those columns.
    functions to preprocess the data
    :return: updates users_df
    """

    for col in users_df.columns:
        # Apply the cleaning function to each cell in the column
        users_df[col] = users_df[col].apply(clean_text)

    return users_df


# --------------------------------------------------------------

def clean_text(text):
    if isinstance(text, str):
        # Unescape HTML entities
        text = html.unescape(text)
        # Parse HTML and get text
        soup = BeautifulSoup(text, "html.parser")
        return soup.get_text()
    else:
        return text


# ----------------------------------------------------------------------------
# Load OpenAI API key from Streamlit secrets
openai_api_key = os.environ.get('OPENAI_API')
st.session_state["openai_api_key"] = openai_api_key
# ----------------------------------------------------------------------------


# Main function

def initialize_session_state():
    # Initialize session state variables if not already set
    st.session_state["involve_recsys_result"] = False
    st.session_state["involve_last_interaction"] = False
    st.session_state.valid_instructions = ""
    st.session_state.invalid_instructions = ""

    # Initialize session state variables if not already set
    for key in [
        "data", "brand","recsys_contents", "generated", "csv_output", "users_message", "messaging_mode",
        "messaging_type", "target_column", "ugc_column", "identifier_column", "input_validator", "selected_input_features"
        "selected_features", "additional_instructions", "segment_info", "message_style", "sample_example",
        "CTA", "all_features", "number_of_messages", "instructionset", "segment_name", "number_of_samples",
        "selected_source_features", "platform"
    ]:
        if key not in st.session_state:
            st.session_state[key] = None


def upload_csv_file():
    st.header("Upload CSV File")
    uploaded_file = st.file_uploader("Choose a CSV file", type="csv")

    if uploaded_file is not None:
        users = load_data(uploaded_file)
        st.write(f"Data loaded from {uploaded_file.name}")
        st.session_state.data = users

        columns = users.columns.tolist()
        st.subheader("Available Columns in Uploaded CSV")
        st.write(columns)
        return users
    else:
        return None


def select_identifier_column(users):
    st.header("Select Identifier Column")
    columns = users.columns.tolist()
    identifier_column = st.selectbox("Select the identifier column", columns)
    st.session_state.identifier_column = identifier_column
    st.markdown("---")


def select_target_audience():
    st.header("Select Target Audience")
    options = ["drumeo", "pianote", "guitareo", "singeo"]
    brand = st.selectbox("Choose the brand for the users", options)
    st.session_state.brand = brand
    st.markdown("---")


def select_target_messaging_type():
    st.header("Select Target Messaging Type")
    messaging_type = st.selectbox("Choose the target messaging type", ["Push Notification", "In-App Notification"])

    st.session_state.messaging_type = "push" if messaging_type == "Push Notification" else "app"
    st.markdown("---")


def input_personalization_parameters():
    st.header("Personalization Parameters")
    st.session_state.segment_info = st.text_area("Segment Info", "", placeholder="Tell us more about the users...")
    st.session_state.CTA = st.text_area("CTA", "", placeholder="e.g., check out 'Inspired by your activity' that we have crafted just for you!")
    st.session_state.message_style = st.text_area("Message Style", "", placeholder="(optional) e.g., be kind and friendly (it's better to be as specific as possible)")
    st.session_state.sample_example = st.text_area("Sample Example", "", placeholder="(optional) e.g., Hello! We have crafted a perfect set of courses just for you!")
    number_of_samples = st.text_input("Number of samples to generate messages", "20", placeholder="(optional) default is 20")
    st.session_state.number_of_samples = int(number_of_samples) if number_of_samples else 20
    st.markdown("---")


def input_message_sequence_parameters():
    """Collect settings for sequential message generation (new feature)."""
    st.header("Sequential Messaging Parameters")

    # Number of sequential messages
    number_of_messages = st.number_input(
        "Number of sequential messages to generate (per user)",
        min_value=1, max_value=10, value=1, step=1, key="num_seq_msgs"
    )
    st.session_state.number_of_messages = number_of_messages

    # Segment name for storage / tracking
    segment_name = st.text_input(
        "Segment Name", value="", placeholder="e.g., no_recent_activity", key="segment_name_input"
    )
    st.session_state.segment_name = segment_name

    # Instruction set for each message
    st.subheader("Instructions per Message")
    st.caption("Provide additional tone or style instructions for each sequential message. Leave blank to inherit the main instructions.")

    instructionset = {}
    cols = st.columns(number_of_messages)
    for i in range(1, number_of_messages + 1):
        with cols[(i - 1) % number_of_messages]:
            instr = st.text_input(
                f"Message {i} instructions", value="", placeholder="e.g., Be Cheerful & Motivational", key=f"instr_{i}"
            )
            if instr.strip():
                instructionset[i] = instr.strip()

    # Save to session state
    st.session_state.instructionset = instructionset
    st.markdown("---")


def select_features_from_source_info():
    st.header("Select Features from Available Source Information")
    available_features = ["first_name", "biography", "birthday_reminder", "goals", "Minutes_practiced", "Last_completed_content"]
    selected_source_features = st.multiselect("Select features to use from available source information", available_features)
    selected_source_features.append("instrument")
    st.session_state.selected_source_features = selected_source_features
    st.markdown("---")


def select_features_from_input_file(users):
    st.header("Select Features from your Input file")
    columns = users.columns.tolist()
    selected_features = st.multiselect("Select features to use in generated messages from the input file", columns)
    st.session_state.selected_features = selected_features
    st.markdown("---")


def provide_additional_instructions():
    st.header("Additional Instructions")
    additional_instructions = st.text_area("Provide additional instructions on how to use selected features in the generated message", "")
    st.session_state.additional_instructions = additional_instructions
    st.markdown("---")


def parse_user_generated_context(users):
    st.header("Parsing User-Generated Context")
    user_generated_context = st.checkbox("Do we have a user-generated context provided in the input that you wish to filter?")
    st.session_state.user_generated_context = user_generated_context

    if user_generated_context:
        columns = users.columns.tolist()
        ugc_column = st.selectbox("Select the column that contains User-Generated Context", columns)
        st.session_state.ugc_column = ugc_column

        st.subheader("Provide Additional Instructions for Validation (Optional)")
        valid_instructions = st.text_area("Instructions for valid context", placeholder="Provide instructions for what constitutes valid context...")
        invalid_instructions = st.text_area("Instructions for invalid context", placeholder="Provide instructions for what constitutes invalid context...")
        st.session_state.valid_instructions = valid_instructions
        st.session_state.invalid_instructions = invalid_instructions

        input_validator = Validator(api_key=st.session_state.openai_api_key)
        st.session_state.input_validator = input_validator
    st.markdown("---")


def include_content_recommendations():
    st.header("Include Content Recommendations")
    include_recommendation = st.checkbox("Would you like to include content in the message to recommend to the students?")
    st.session_state.include_recommendation = include_recommendation

    if include_recommendation:
        recommendation_source = st.radio("Select recommendation source", ["Input File", "Musora Recommender System"])
        st.session_state.recommendation_source = recommendation_source

        if recommendation_source == "Musora Recommender System":
            st.session_state.involve_recsys_result = True
            st.session_state.messaging_mode = "recsys_result"

            list_of_content_types = ["song", "workout", "quick_tips", "course"]
            selected_content_types = st.multiselect("Select content_types that you would like to recommend", list_of_content_types)
            st.session_state.recsys_contents = selected_content_types
        else:
            st.session_state.involve_recsys_result = False
            st.session_state.messaging_mode = "message"
            columns = st.session_state.data.columns.tolist()
            target_column = st.selectbox("Select the target column for recommendations", columns)
            st.session_state.target_column = target_column
    else:
        st.session_state.messaging_mode = "message"
        st.session_state.target_column = None
    st.markdown("---")


def generate_personalized_messages(users):
    st.header("Generate Personalized Messages")
    if st.button("Generate Personalized Messages"):
        if st.session_state.CTA.strip() == "" or st.session_state.segment_info.strip() == "":
            st.error("CTA and Segment Info are mandatory fields and cannot be left empty.")
        else:
            conn = {
                "user": os.environ.get("snowflake_user"),
                "password": os.environ.get("snowflake_password"),
                "account": os.environ.get("snowflake_account"),
                "role": os.environ.get("snowflake_role"),
                "database": os.environ.get("snowflake_database"),
                "warehouse": os.environ.get("snowflake_warehouse"),
                "schema": os.environ.get("snowflake_schema")
            }

            config_file_path = 'Config_files/message_system_config.json'
            config_file = load_config_(config_file_path)
            session = Session.builder.configs(conn).create()

            if st.session_state.user_generated_context:
                if st.session_state.valid_instructions.strip() or st.session_state.invalid_instructions.strip():
                    st.session_state.input_validator.set_validator_instructions(
                        valid_instructions=st.session_state.valid_instructions,
                        invalid_instructions=st.session_state.invalid_instructions
                    )
                else:
                    st.session_state.input_validator.set_validator_instructions()

                # Create a progress bar
                progress_bar = st.progress(0)
                status_text = st.empty()

                # Define a callback function to update the progress bar
                def progress_callback(progress, total):
                    percent_complete = int(progress / total * 100)
                    progress_bar.progress(percent_complete)
                    status_text.text(f"Validating user_generated_context: {percent_complete}%")

                st.info("Validating user-generated content. This may take a few moments...")
                users = st.session_state.input_validator.validate_dataframe(
                    dataframe=users, target_column=st.session_state.ugc_column, progress_callback=progress_callback)
                users = filter_validated_users(users)
                st.success("User-generated content has been validated and filtered.")

            st.session_state.all_features = st.session_state.selected_source_features + st.session_state.selected_features

            if "Last_completed_content" in st.session_state.selected_source_features:
                st.session_state.involve_last_interaction = True
            else:
                st.session_state.involve_last_interaction = False

            # Create a progress bar
            progress_bar = st.progress(0)
            status_text = st.empty()

            # Define a callback function to update the progress bar
            def progress_callback(progress, total):
                percent_complete = int(progress / total * 100)
                progress_bar.progress(percent_complete)
                status_text.text(f"Processing: {percent_complete}%")

            permes = Permes()
            users_message = permes.create_personalize_messages(
                session=session,
                users=users,
                brand=st.session_state.brand,
                config_file=config_file,
                openai_api_key=os.environ.get('OPENAI_API'),
                CTA=st.session_state.CTA,
                segment_info=st.session_state.segment_info,
                number_of_samples=st.session_state.number_of_samples,
                message_style=st.session_state.message_style,
                sample_example=st.session_state.sample_example,
                selected_input_features=st.session_state.selected_features,
                selected_source_features=st.session_state.selected_source_features,
                additional_instructions=st.session_state.additional_instructions,
                platform=st.session_state.messaging_type,
                involve_last_interaction=st.session_state.involve_last_interaction,
                involve_recsys_result=st.session_state.involve_recsys_result,
                messaging_mode=st.session_state.messaging_mode,
                identifier_column=st.session_state.identifier_column,
                target_column=st.session_state.target_column,
                recsys_contents=st.session_state.recsys_contents,
                progress_callback=progress_callback,
                # NEW PARAMETERS
                number_of_messages=st.session_state.number_of_messages,
                instructionset=st.session_state.instructionset,
                segment_name=st.session_state.segment_name
            )

            # Clear the progress bar and status text after completion
            progress_bar.empty()
            status_text.empty()

            csv_output = users_message.to_csv(encoding='utf-8-sig', index=False)
            st.session_state.csv_output = csv_output
            st.session_state.users_message = users_message
            st.session_state.generated = True
            st.success("Personalized messages have been generated.")
    st.markdown("---")


def download_generated_messages():
    if st.session_state.get('generated', False):
        st.header("Download Generated Messages")

        # Suppose `df` is your final DataFrame
        df = st.session_state.users_message  # or wherever your DataFrame is

        # Write CSV to an in-memory buffer, with utf-8-sig encoding
        csv_buffer = StringIO()
        df.to_csv(csv_buffer, index=False, encoding='utf-8-sig')
        csv_buffer.seek(0)

        # Convert to bytes (this will include the UTF-8 BOM)
        csv_bytes = csv_buffer.getvalue().encode('utf-8-sig')

        # Provide the bytes to download_button
        st.download_button(
            label="Download output messages as a CSV file",
            data=csv_bytes,
            file_name='personalized_messages.csv',
            mime='text/csv'
        )


def view_generated_messages():
    # Only run if messages have been generated
    if not st.session_state.get('generated', False):
        return

    st.title("🔔 Generated Push Notifications Review")
    df = st.session_state.users_message
    identifier = st.session_state.identifier_column.lower()
    features = st.session_state.all_features

    for idx, (_, user_row) in enumerate(df.iterrows(), start=1):
        user_id = user_row.get(identifier, "N/A")
        # Collapsible container per user
        with st.expander(f"{idx}. User ID: {user_id}", expanded=(idx == 1)):
            st.markdown("##### 👤 User Features")
            # 3-column layout for user metadata
            feature_cols = st.columns(3)
            for i, feat in enumerate(features):
                val = user_row.get(feat, "N/A")
                feature_cols[i % 3].write(f"**{feat}**: {val}")

            st.markdown("---")
            st.markdown("##### 📝 Generated Messages")
            raw = user_row.get('message', '[]')

            try:
                parsed = json.loads(raw)
                # If it's the nested form {"messages_sequence": [ … ]}, grab the list inside.
                if isinstance(parsed, dict) and 'messages_sequence' in parsed:
                    messages = parsed['messages_sequence']
                # If somehow it's already a list, leave it alone.
                elif isinstance(parsed, list):
                    messages = parsed
                else:
                    st.warning(
                        "Unexpected JSON structure for messages; expected a list or {'messages_sequence': [...]}")
                    messages = []
            except json.JSONDecodeError:
                st.error("Could not parse message JSON")
                messages = []

            # Display each push notification
            for m_idx, msg in enumerate(messages, start=1):
                c_img, c_text = st.columns([1, 3])
                with c_img:
                    thumb = msg.get('thumbnail_url')
                    if thumb:
                        st.image(thumb, width=80)
                    else:
                        st.write("No image")

                with c_text:
                    header = msg.get('header', '')
                    body   = msg.get('message', '')
                    link   = msg.get('web_url_path', '#')
                    st.markdown(f"**{m_idx}. {header}**")
                    st.markdown(body)
                    st.markdown(f"[Read more →]({link})")

                st.markdown("---")


if __name__ == "__main__":
    st.title("Personalized Message Generator")

    # Initialize session state variables
    initialize_session_state()

    # Upload CSV File
    users = upload_csv_file()

    if users is not None:
        # Proceed with the rest of the application
        select_identifier_column(users)
        select_target_audience()
        select_target_messaging_type()
        input_personalization_parameters()
        input_message_sequence_parameters()
        select_features_from_source_info()
        select_features_from_input_file(users)
        provide_additional_instructions()
        parse_user_generated_context(users)
        include_content_recommendations()
        generate_personalized_messages(users)
        download_generated_messages()
        view_generated_messages()