import gradio as gr
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate
from langchain.llms import HuggingFacePipeline
from transformers import pipeline
from gtts import gTTS
from bs4 import BeautifulSoup
import tempfile
import requests
import subprocess
import concurrent.futures
from PIL import Image, ImageDraw

# CPU-friendly summarization model
summary_pipe = pipeline("text2text-generation", model="google/flan-t5-base", device=-1)
llm = HuggingFacePipeline(pipeline=summary_pipe)

# Prompt for <50-word promotional summary
summary_prompt = PromptTemplate.from_template("""
Provide a crisp, promotional-style summary (under 50 words) of the following:

{text}

Summary:
""")
summary_chain = LLMChain(llm=llm, prompt=summary_prompt)

# Extract main article content
def extract_main_content(url):
    resp = requests.get(url, timeout=10)
    soup = BeautifulSoup(resp.content, "html.parser")
    for tag in soup(["nav","header","footer","aside","script","style","noscript"]): tag.decompose()
    paras = soup.find_all("p")
    content = "\n".join(p.get_text() for p in paras if len(p.get_text())>60)
    return content or None

# Create gradient background image
def create_background(image_path, size=(1280,720)):
    img = Image.new("RGB", size)
    draw = ImageDraw.Draw(img)
    for i in range(size[1]):
        # gradient from dark blue to black
        r = int(10 + (i/size[1])*20)
        g = int(20 + (i/size[1])*30)
        b = int(50 + (i/size[1])*50)
        draw.line([(0, i), (size[0], i)], fill=(r, g, b))
    img.save(image_path)

# Generate AV summary
def url_to_av_summary(url):
    text = extract_main_content(url)
    if not text:
        return "Failed to extract article content.", None
    text = text[:3000]
    summary = summary_chain.run(text=text)
    summary = summary.replace('"','')
    # TTS
    audio_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3").name
    gTTS(text=summary).save(audio_path)
    # Background image
    bg_path = tempfile.NamedTemporaryFile(delete=False, suffix=".png").name
    create_background(bg_path)
    # Video with animated text
    video_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
    cmd = [
        'ffmpeg', '-y',
        '-loop', '1', '-i', bg_path,
        '-i', audio_path,
        '-vf', (
            "drawtext=fontfile=/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf:text='" + summary +
            "':fontcolor=white:fontsize=48:box=1:boxcolor=black@0.5:boxborderw=5:"
            "x=(w-text_w)/2:y=h-(t*(h+text_h)/15)"
        ),
        '-t', '15',
        '-c:v', 'libx264', '-c:a', 'aac', '-pix_fmt', 'yuv420p', '-shortest', video_path
    ]
    subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
    return summary, video_path

# Timeout wrapper
def safe_summary(url, timeout_secs=60):
    with concurrent.futures.ThreadPoolExecutor() as executor:
        future = executor.submit(url_to_av_summary, url)
        try:
            return future.result(timeout=timeout_secs)
        except concurrent.futures.TimeoutError:
            return "⏱️ Processing timed out.", None

iface = gr.Interface(
    fn=safe_summary,
    inputs=gr.Textbox(label="Article URL"),
    outputs=[gr.Textbox(label="Summary"), gr.Video(label="Video Preview")],
    title="🎥 Promo-Style AV Summary with Gradient Background",
    description="Generates a <15s video with animated text over a gradient background. CPU-only, HuggingFace Spaces-ready."
)

if __name__ == '__main__':
    iface.launch()