app.py

import streamlit as st
import random
import pandas as pd
import requests
from io import BytesIO
from PIL import Image
from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM
import re
import time

# --------------------------- Configuration & CSS ---------------------------
MAX_SIZE = (450, 450)

st.set_page_config(page_title="🔮 Divine Fortune Teller", page_icon=":crystal_ball:")

# Updated CSS: added rules to force text color to black for inputs, text areas, and markdown
st.markdown(
    """
    <style>
    .reportview-container {
        background: linear-gradient(135deg, #f6d365, #fda085);
    }
    .card {
        background: rgba(255, 255, 255, 0.95);
        padding: 30px;
        border-radius: 12px;
        box-shadow: 0 10px 30px rgba(0, 0, 0, 0.1);
        max-width: 800px;
        margin: auto;
        text-align: center;
    }
    /* Force all text to be black */
    body, input, textarea, .stMarkdown, label {
        color: black !important;
    }
    </style>
    """,
    unsafe_allow_html=True,
)

# --------------------------- Session State Initialization ---------------------------
if 'submitted' not in st.session_state:
    st.session_state.submitted = False
if 'error_message' not in st.session_state:
    st.session_state.error_message = ""
if 'question' not in st.session_state:
    st.session_state.question = ""
if 'fortune_number' not in st.session_state:
    st.session_state.fortune_number = None
if 'fortune_row' not in st.session_state:
    st.session_state.fortune_row = None
if "button_count_temp" not in st.session_state:
    st.session_state.button_count_temp = 0
if "cfu_explain_text" not in st.session_state:
    st.session_state.cfu_explain_text = ""

# --------------------------- Load Fortune CSV ---------------------------
if "fortune_data" not in st.session_state:
    try:
        st.session_state.fortune_data = pd.read_csv("/home/user/app/resources/detail.csv")
    except Exception as e:
        st.error(f"Error loading CSV: {e}")
        st.session_state.fortune_data = None

# --------------------------- Helper Functions ---------------------------
def load_and_resize_image(path, max_size=MAX_SIZE):
    """
    Loads an image from a local file path and resizes it to fit within a specified maximum size.
    """
    try:
        img = Image.open(path)
        img.thumbnail(max_size, Image.Resampling.LANCZOS)
        return img
    except Exception as e:
        st.error(f"Error loading image: {e}")
        return None

def download_and_resize_image(url, max_size=MAX_SIZE):
    """
    Downloads an image from a given URL, then resizes it to a predefined maximum size.
    """
    try:
        response = requests.get(url)
        response.raise_for_status()
        image_bytes = BytesIO(response.content)
        img = Image.open(image_bytes)
        img.thumbnail(max_size, Image.Resampling.LANCZOS)
        return img
    except Exception as e:
        st.error(f"Error loading image from URL: {e}")
        return None

def display_text_field(label, text, height):
    """
    Creates and displays a custom-styled text field with a title and scrollable content.
    """
    html = f"""
    <h6 style="display: block; margin-top: 10px;">{label}</h6>
    <div style="border: 1px solid #ccc; border-radius: 4px; background-color: #f0f0f0;
                padding: 10px; height: {height}px; overflow-y: auto; color: black; font-size: 16px;">
        <div>{text}</div>
    </div>
    """
    st.markdown(html, unsafe_allow_html=True)

# --------------------------- Model Functions ---------------------------
def load_finetuned_classifier_model(question):
    """
    Uses a fine-tuned text classification model to categorize the user's question into one of several predefined fortune themes.
    """
    label_list = ["Geomancy", "Lost Property", "Personal Well-Being", "Future Prospect", "Traveling"]
    mapping = {f"LABEL_{i}": label for i, label in enumerate(label_list)}
    pipe = pipeline("text-classification", model="tonyhui2234/CustomModel_classifier_model_10")
    prediction = pipe(question)[0]['label']
    predicted_label = mapping.get(prediction, prediction)
    return predicted_label

def generate_answer(question, fortune):
    """
    Generates a detailed explanation by feeding the question and the selected fortune text into a fine-tuned sequence-to-sequence language model.
    """
    start_time = time.perf_counter()
    tokenizer = AutoTokenizer.from_pretrained("tonyhui2234/finetuned_model_text_gen")
    model = AutoModelForSeq2SeqLM.from_pretrained("tonyhui2234/finetuned_model_text_gen", device_map="auto")
    input_text = "Question: " + question + " Fortune: " + fortune
    inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
    outputs = model.generate(
        **inputs,
        max_length=256,
        num_beams=4,
        early_stopping=True,
        repetition_penalty=2.0,
        no_repeat_ngram_size=3
    )
    answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
    run_time = time.perf_counter() - start_time
    print(f"Runtime: {run_time:.4f} seconds")
    return answer

def analysis(row_detail, classifiy, question):
    """
    Extracts a specific portion of the fortune details based on the classification result and then generates an answer using the text generation model.
    """
    pattern = re.compile(re.escape(classifiy) + r":\s*(.*?)(?:\.|$)", re.IGNORECASE)
    match = pattern.search(row_detail)
    if match:
        result = match.group(1)
        return generate_answer(question, result)
    else:
        return "Heaven's secret cannot be revealed."

def check_sentence_is_english_model(question):
    """
    Checks if the provided text is in English using a language detection model.
    """
    pipe_english = pipeline("text-classification", model="eleldar/language-detection")
    return pipe_english(question)[0]['label'] == 'en'

def check_sentence_is_question_model(question):
    """
    Determines whether the input text is formulated as a question using a question vs. statement classifier.
    """
    pipe_question = pipeline("text-classification", model="shahrukhx01/question-vs-statement-classifier")
    return pipe_question(question)[0]['label'] == 'LABEL_1'

# --------------------------- Callback Functions ---------------------------
def random_draw():
    """
    Randomly selects a fortune entry from the loaded CSV based on a randomly generated number and updates the session state with the fortune’s details.
    """
    st.session_state.fortune_number = random.randint(1, 100)
    df = st.session_state.fortune_data
    if df is not None:
        matching_row = df[df['CNumber'] == st.session_state.fortune_number]
        if not matching_row.empty:
            row = matching_row.iloc[0]
            st.session_state.fortune_row = {
                "Header": row.get("Header", "N/A"),
                "Luck": row.get("Luck", "N/A"),
                "Description": row.get("Description", "No description available."),
                "Detail": row.get("Detail", "No detail available."),
                "HeaderLink": row.get("link", None)
            }
        else:
            st.session_state.fortune_row = {
                "Header": "N/A",
                "Luck": "N/A",
                "Description": "No description available.",
                "Detail": "No detail available.",
                "HeaderLink": None
            }
    else:
        st.session_state.error_message = "Fortune data is not available."
    
    st.session_state.submitted = True
    st.session_state.show_explain = False

def submit_callback():
    """
    Validates the initial user input (ensuring it’s non-empty, in English, and a question), prompts the user to reflect, and then triggers a random fortune draw if the criteria are met.
    """
    question = st.session_state.get("question_input", "").strip()
    if not question:
        st.session_state.error_message = "Please enter a valid question."
        st.session_state.submitted = False
        return
    
    if not check_sentence_is_english_model(question):
        st.session_state.error_message = "Please enter in English!"
        st.session_state.button_count_temp = 0
        return

    if not check_sentence_is_question_model(question):
        st.session_state.error_message = "This is not a question. Please enter again!"
        st.session_state.button_count_temp = 0
        return

    if st.session_state.button_count_temp == 0:
        st.session_state.error_message = "Please take a moment to quietly reflect on your question in your mind, then click submit again!"
        st.session_state.button_count_temp = 1
        return

    st.session_state.error_message = ""
    st.session_state.question = question
    st.session_state.button_count_temp = 0
    random_draw()

def resubmit_callback():
    """
    Allows the user to submit a revised question with similar validations, then updates the fortune selection accordingly.
    """
    new_question = st.session_state.get("resubmit_input", "").strip()
    if new_question == "":
        st.session_state.error_message = "Please enter a valid question."
        return
    
    if not check_sentence_is_english_model(new_question):
        st.session_state.error_message = "Please enter in English!"
        st.session_state.button_count_temp = 0
        return

    if not check_sentence_is_question_model(new_question):
        st.session_state.error_message = "This is not a question. Please enter again!"
        st.session_state.button_count_temp = 0
        return

    if st.session_state.button_count_temp == 0:
        st.session_state.error_message = "Please take a moment to quietly reflect on your question in your mind, then click submit again!"
        st.session_state.button_count_temp = 1
        return

    st.session_state.error_message = ""
    if new_question != st.session_state.question:
        st.session_state.question = new_question
    st.session_state.button_count_temp = 0
    random_draw()

def explain_callback():
    """
    Uses the selected fortune details and the classifier to generate and display a customized explanation for the user's question using the text generation model.
    """
    question = st.session_state.question
    if not st.session_state.fortune_row:
        st.error("Fortune data is not available. Please submit your question first.")
        return
    row_detail = st.session_state.fortune_row.get("Detail", "No detail available.")
    
    classify = load_finetuned_classifier_model(question)
    print(f"classify Checking: {classify}\nQuestion: {question}")
    cfu_explain = analysis(row_detail, classify, question)
    st.session_state.cfu_explain_text = cfu_explain
    st.session_state.show_explain = True

# --------------------------- Layout & Display ---------------------------
st.title("🔮 Divine Fortune Teller")

if not st.session_state.submitted:
    st.image("/home/user/app/resources/front.png", use_container_width=True)
    st.text_input("Ask your fortune question...", key="question_input")
    st.button("Submit", on_click=submit_callback)
    
    if st.session_state.error_message:
        st.error(st.session_state.error_message)
else:
    st.text_input("Your Question", value=st.session_state.question, key="resubmit_input")
    st.button("Resubmit", on_click=resubmit_callback)
    if st.session_state.error_message:
        st.error(st.session_state.error_message)
    
    col1, col2 = st.columns([2, 3])
    with col1:
        if st.session_state.fortune_row and st.session_state.fortune_row.get("HeaderLink"):
            img_from_url = download_and_resize_image(st.session_state.fortune_row.get("HeaderLink"))
            if img_from_url:
                st.markdown("<h6> </h6>", unsafe_allow_html=True)
                st.image(img_from_url, use_container_width=False)
            else:
                default_img = load_and_resize_image("/home/user/app/resources/error.png")
                if default_img:
                    st.image(default_img, caption="Default image", use_container_width=False)
        else:
            default_img = load_and_resize_image("/home/user/app/resources/error.png")
            if default_img:
                st.image(default_img, caption="Default image", use_container_width=False)
    
    with col2:
        if st.session_state.fortune_row:
            luck_text = st.session_state.fortune_row.get("Luck", "N/A")
            summary = f"""
            <div style="font-size: 24px; font-weight: bold;">
                Fortune Stick Number: {st.session_state.fortune_number}<br>
                Luck: {luck_text}
            </div>
            """
            st.markdown(summary, unsafe_allow_html=True)
            description_text = st.session_state.fortune_row.get("Description", "No description available.")
            detail_text = st.session_state.fortune_row.get("Detail", "No detail available.")
            # Replace text_area with our custom text field
            display_text_field("Description:", description_text, 180)
            display_text_field("Detail:", detail_text, 180)
    
    st.button("CFU Explain", on_click=explain_callback)
    if st.session_state.show_explain:
        display_text_field("Explanation:", st.session_state.cfu_explain_text, 200)