import gradio as gr
import pandas as pd
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
from sentence_transformers import SentenceTransformer, util

# ------------------------------
# Load Models
# ------------------------------
qg_model_name = "iarfmoose/t5-base-question-generator"
tokenizer_qg = AutoTokenizer.from_pretrained(qg_model_name)
model_qg = AutoModelForSeq2SeqLM.from_pretrained(qg_model_name)
model_plag = SentenceTransformer('all-MiniLM-L6-v2')
asr = pipeline("automatic-speech-recognition", model="openai/whisper-base")

# ------------------------------
# Quiz Generator
# ------------------------------
def generate_mcqs(text, num_questions=3):
    input_text = f"generate questions: {text.strip()}"
    input_ids = tokenizer_qg.encode(input_text, return_tensors="pt", max_length=512, truncation=True)

    outputs = model_qg.generate(
        input_ids=input_ids,
        max_length=256,
        num_return_sequences=num_questions,
        do_sample=True,
        top_k=50,
        top_p=0.95
    )

    questions = [tokenizer_qg.decode(out, skip_special_tokens=True).strip() for out in outputs]
    return "\n".join([f"{i+1}. {q}" for i, q in enumerate(questions)])

# ------------------------------
# Weakness Analyzer
# ------------------------------
def analyze_weakness(csv_file):
    df = pd.read_csv(csv_file.name)
    summary = df.groupby("Topic")["Score"].mean().sort_values()
    return summary.to_string()

# ------------------------------
# Teaching Assistant (Mock)
# ------------------------------
def chatbot_response(message, history):
    return "This is a placeholder response for now. (LLM not integrated)"

# ------------------------------
# Speech Question Solver (NEW)
# ------------------------------
def speech_answer(audio_file_path):
    transcription = asr(audio_file_path)["text"]
    input_text = f"generate questions: {transcription.strip()}"
    input_ids = tokenizer_qg.encode(input_text, return_tensors="pt", max_length=512, truncation=True)
    outputs = model_qg.generate(input_ids, max_length=256, num_return_sequences=1)
    response = tokenizer_qg.decode(outputs[0], skip_special_tokens=True)
    return f"🗣️ Transcript: {transcription.strip()}\n\n💡 Answer: {response.strip()}"

# ------------------------------
# Summarizer
# ------------------------------
from transformers import pipeline

summarizer = pipeline("summarization", model="facebook/bart-large-cnn")

def summarize_text(text):
    result = summarizer(text, max_length=120, min_length=40, do_sample=False)
    return result[0]["summary_text"]

# ------------------------------
# Engagement Predictor (Mock)
# ------------------------------
def predict_engagement(file):
    df = pd.read_csv(file.name)
    avg_time = df["TimeSpent"].mean()
    return "✅ Engaged student" if avg_time >= 10 else "⚠️ Risk of disengagement"

# ------------------------------
# Badge Generator
# ------------------------------
def generate_badge(file):
    df = pd.read_csv(file.name)
    avg_score = df["Score"].mean()
    if avg_score >= 80:
        return "🏅 Gold Badge"
    elif avg_score >= 50:
        return "🥈 Silver Badge"
    else:
        return "🥉 Bronze Badge"

# ------------------------------
# Translator (Mock)
# ------------------------------
def translate_text(text, target_lang):
    return f"(Translated to {target_lang}) - This is a mock translation."

# ------------------------------
# Plagiarism Checker
# ------------------------------
def check_plagiarism(text1, text2):
    emb1 = model_plag.encode(text1, convert_to_tensor=True)
    emb2 = model_plag.encode(text2, convert_to_tensor=True)
    score = util.cos_sim(emb1, emb2).item()
    return f"Similarity Score: {score:.2f} - {'⚠️ Possible Plagiarism' if score > 0.8 else '✅ Looks Original'}"

# ------------------------------
# Gradio Interface
# ------------------------------
with gr.Blocks() as demo:
    gr.Markdown("# 📚 Smart LMS Suite (Offline)")

    with gr.Tab("🧠 Quiz Generator"):
        quiz_text = gr.Textbox(label="📄 Input Content", lines=6, placeholder="Paste a paragraph here...")
        quiz_slider = gr.Slider(1, 10, value=3, label="🧾 Number of Questions")
        quiz_btn = gr.Button("🚀 Generate Quiz")
        quiz_output = gr.Textbox(label="📋 Generated Questions", lines=10)
        quiz_btn.click(fn=generate_mcqs, inputs=[quiz_text, quiz_slider], outputs=quiz_output)

    with gr.Tab("📉 Weakness Analyzer"):
        weak_file = gr.File(label="Upload CSV with Topic & Score columns")
        weak_btn = gr.Button("Analyze")
        weak_out = gr.Textbox(label="Analysis")
        weak_btn.click(fn=analyze_weakness, inputs=weak_file, outputs=weak_out)

    with gr.Tab("🤖 Teaching Assistant"):
        gr.ChatInterface(fn=chatbot_response)

    with gr.Tab("🎤 Speech Q Solver"):
        audio_in = gr.Audio(label="Upload Audio", type="filepath")
        audio_btn = gr.Button("Transcribe + Generate Answer")
        audio_out = gr.Textbox(label="Answer")
        audio_btn.click(fn=speech_answer, inputs=audio_in, outputs=audio_out)

    with gr.Tab("📄 Summarizer"):
        sum_text = gr.Textbox(lines=5, label="Paste Text")
        sum_btn = gr.Button("Summarize")
        sum_out = gr.Textbox(label="Summary")
        sum_btn.click(fn=summarize_text, inputs=sum_text, outputs=sum_out)

    with gr.Tab("📊 Engagement Predictor"):
        eng_file = gr.File(label="Upload CSV with TimeSpent column")
        eng_btn = gr.Button("Predict")
        eng_out = gr.Textbox()
        eng_btn.click(fn=predict_engagement, inputs=eng_file, outputs=eng_out)

    with gr.Tab("🏅 Badge Generator"):
        badge_file = gr.File(label="Upload CSV with Score column")
        badge_btn = gr.Button("Get Badge")
        badge_out = gr.Textbox()
        badge_btn.click(fn=generate_badge, inputs=badge_file, outputs=badge_out)

    with gr.Tab("🌍 Translator"):
        trans_in = gr.Textbox(label="Enter Text")
        trans_lang = gr.Textbox(label="Target Language")
        trans_btn = gr.Button("Translate")
        trans_out = gr.Textbox()
        trans_btn.click(fn=translate_text, inputs=[trans_in, trans_lang], outputs=trans_out)

    with gr.Tab("📋 Plagiarism Checker"):
        text1 = gr.Textbox(label="Text 1", lines=3)
        text2 = gr.Textbox(label="Text 2", lines=3)
        plag_btn = gr.Button("Check Similarity")
        plag_out = gr.Textbox()
        plag_btn.click(fn=check_plagiarism, inputs=[text1, text2], outputs=plag_out)

# ------------------------------
# Launch
# ------------------------------
demo.launch()