import gradio as gr
import requests
import os
import PyPDF2
from io import BytesIO

# Hugging Face API and model
HF_API_TOKEN = os.getenv("HF_API_TOKEN")
HF_API_URL = "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta"
MODEL_CHECK_PROMPT = "Say hello!"

# ✅ Check whether the API key and model are working correctly
def check_api_ready():
    if not HF_API_TOKEN:
        return False, "❌ HF_API_TOKEN not set. Please add in Space secrets."
    headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
    try:
        resp = requests.post(HF_API_URL, headers=headers, json={"inputs": MODEL_CHECK_PROMPT}, timeout=20)
        if resp.status_code == 200:
            data = resp.json()
            if isinstance(data, dict) and "error" in data:
                return False, f"❌ Model Error: {data['error']}"
            if isinstance(data, list) and data[0].get("generated_text"):
                return True, "✅ Model is ready!"
        elif resp.status_code == 401:
            return False, "❌ Unauthorized. Check your API token."
        else:
            return False, f"❌ Unexpected API response: {resp.text}"
    except Exception as e:
        return False, f"❌ API connection failed: {str(e)}"

# ✅ Tooltip animation HTML from Lottie JSON (place Robotics-Students.json in Space root)
def lottie_html():
    return """
    <script src="https://unpkg.com/@lottiefiles/lottie-player@latest/dist/lottie-player.js"></script>
    <lottie-player src='Robotics-Students.json' background='transparent' speed='1' style='width:340px; height:340px; margin-bottom:12px;' loop autoplay></lottie-player>
    """

# ✅ PDF text extractor
def extract_text_from_pdf(pdf_file):
    reader = PyPDF2.PdfReader(BytesIO(pdf_file.read()))
    return "\n".join([page.extract_text() or "" for page in reader.pages]).strip()

# ✅ Generate questions and answers using language model
def ai_generate_questions(resume_text, job_title):
    prompt = (
        f"You are an AI interview coach.\n"
        f"Candidate Resume:\n{resume_text}\n"
        f"Target Role: {job_title}\n"
        f"Generate 10 realistic interview questions based on this person’s resume, "
        f"and for each question provide a coaching tip to help them answer effectively."
    )
    headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
    payload = {"inputs": prompt, "parameters": {"max_new_tokens": 512, "temperature": 0.7}}

    try:
        response = requests.post(HF_API_URL, headers=headers, json=payload, timeout=60)
        data = response.json()
        if "error" in data:
            return f"<div style='color:red;'>❌ API Error: {data['error']}</div>"
        return data[0]["generated_text"].strip()
    except Exception as e:
        return f"<div style='color:red;'>❌ Exception: {str(e)}</div>"

# ✅ Format final output HTML
def render_output(name, job, status, queue, messages, model_message, show_lottie=True):
    lottie_side = lottie_html() if show_lottie else ""
    model_msg = f"<div style='margin-top:12px;font-size:0.9rem;color:#46a546;'>{model_message}</div>" if model_message else ""
    left_panel = f"""
        {lottie_side}
        <div style='background:#f4f6fa;padding:18px;border-radius:16px;
        max-width:340px;box-shadow:0 2px 14px #ccd4e6b3;font-family:sans-serif;'>
            <div style='font-size:1.25rem;font-weight:700;'>🤖 ROBOT RESUME ANALYZER</div>
            <div style='margin-top:10px;'><b>Status:</b> <span style='color:#058c42'>{status}</span></div>
            <div style='margin-top:10px;'><b>Current:</b> {name}</div>
            <div><b>Position:</b> {job}</div>
            <div style='margin-top:10px;'><b>Analysis Progress:</b></div>
            <div style='font-size:2rem;'>👥</div>
            <div><b>QUEUE</b></div>
            <div style='font-size:1.5rem;font-weight:700;'>{queue}</div>
            <div><i>Analyzing Next...</i></div>
            {model_msg}
        </div>
    """

    return f"""
    <div style='display:flex;gap:30px;'>
        {left_panel}
        <div style='max-width:620px;margin-top:15px;'>{messages}</div>
    </div>
    """

# ✅ Main app logic with staged UI updates
def interface(pdf_file, job_title):
    is_ready, msg = check_api_ready()
    if not is_ready:
        err_html = f"<div style='color:red;'><b>ERROR:</b> {msg}</div>"
        return render_output("---", "---", "Unavailable", 0, err_html, msg, show_lottie=False)

    if not pdf_file:
        return render_output("---", "---", "Waiting for PDF", 1, "<b>Please upload a PDF resume.</b>", msg)

    if not job_title.strip():
        name = pdf_file.name.replace('.pdf','')
        return render_output(name, "---", "Waiting for Job Title", 1, "<b>Please enter a target job title.</b>", msg)

    name = pdf_file.name.replace(".pdf", "")
    try:
        pdf_file.seek(0)
        resume_text = extract_text_from_pdf(pdf_file)
        if not resume_text:
            raise Exception("PDF contains no readable text")
    except Exception as e:
        return render_output(name, job_title, "PDF Error", 1, f"<div style='color:red;'>Failed to extract PDF: {e}</div>", msg)

    yield render_output(name, job_title, "Analyzing...", 1, "<i>Generating questions with AI...</i>", msg)

    ai_output = ai_generate_questions(resume_text, job_title)
    questions = [q.strip() for q in ai_output.split("\n") if q.strip()]

    html = "<h3>✅ Interview Questions & Coaching Tips</h3>"
    for i, q in enumerate(questions[:10], 1):
        html += f"<div style='background:#fff;padding:16px;border-radius:10px;margin-bottom:12px;box-shadow:0 2px 6px #ddd;'>"
        html += f"<b>Q{i}:</b> {q}</div>"

    yield render_output(name, job_title, "Done ✅", 0, html, msg)

# ✅ Launch Gradio app
with gr.Interface(
    fn=interface,
    inputs=[
        gr.File(label="Upload PDF Resume", type="binary", file_types=[".pdf"]),
        gr.Textbox(label="Target Job Title", placeholder="e.g. Data Analyst"),
    ],
    outputs=gr.HTML(),
    allow_flagging="never",
    title="🤖 AI Resume Analyzer + Interview Coach (PDF + Lottie)",
    live=False,
) as demo:
    demo.launch()