app.py

import gradio as gr
import subprocess
import os
from moviepy.editor import VideoFileClip
import whisper
import tempfile
import shutil
import yt_dlp

# Load Whisper model
model = whisper.load_model("small")

def process_video(video):
    """
    Handles:
      - Gradio uploads
      - n8n binary uploads (multipart/form-data)
      - YouTube or web URLs
    """
    try:
        temp_dir = tempfile.mkdtemp()

        # Handle video input from various sources
        if isinstance(video, str) and video.startswith(("http://", "https://")):
            # 🎥 If it's a YouTube link or direct URL, download it
            video_path = os.path.join(temp_dir, "downloaded_video.mp4")
            ydl_opts = {
                "format": "bestvideo+bestaudio/best",
                "outtmpl": video_path,
                "quiet": True,
                "merge_output_format": "mp4"
            }
            with yt_dlp.YoutubeDL(ydl_opts) as ydl:
                ydl.download([video])

        elif isinstance(video, dict) and "name" in video:
            # 📦 n8n file input (dict)
            video_path = video["name"]

        elif hasattr(video, "name"):
            # 📁 Uploaded from Gradio UI
            video_path = os.path.join(temp_dir, os.path.basename(video.name))
            shutil.copy(video.name, video_path)

        elif isinstance(video, str) and os.path.exists(video):
            # 📍 Local file path
            video_path = video

        else:
            raise ValueError("Invalid video input type")

        # 🎧 Extract audio
        video_clip = VideoFileClip(video_path)
        audio_path = os.path.join(temp_dir, "audio.wav")
        video_clip.audio.write_audiofile(audio_path, codec="pcm_s16le")

        # 🗣 Transcribe using Whisper
        result = model.transcribe(audio_path, language="en")
        transcript = result["text"]

        # 💬 Save transcript to SRT
        srt_path = os.path.join(temp_dir, "transcript.srt")
        with open(srt_path, "w", encoding="utf-8") as f:
            for i, seg in enumerate(result["segments"]):
                start = seg["start"]
                end = seg["end"]
                text = seg["text"].strip()
                f.write(f"{i+1}\n{format_time(start)} --> {format_time(end)}\n{text}\n\n")

        # 🔥 Burn subtitles into video
        output_path = os.path.join(temp_dir, "output_with_subs.mp4")
        burn_subtitles(video_path, srt_path, output_path)

        return output_path, transcript

    except Exception as e:
        return None, f"❌ Error: {str(e)}"


def format_time(seconds):
    hrs, rem = divmod(seconds, 3600)
    mins, secs = divmod(rem, 60)
    millis = int((secs - int(secs)) * 1000)
    return f"{int(hrs):02}:{int(mins):02}:{int(secs):02},{millis:03}"


def burn_subtitles(video_path, srt_path, output_path):
    """Use FFmpeg to burn subtitles permanently into the video."""
    cmd = [
        "ffmpeg", "-y",
        "-i", video_path,
        "-vf", f"subtitles={srt_path}",
        "-c:a", "copy",
        output_path
    ]
    subprocess.run(cmd, check=True)


# 🎛️ Gradio Interface
iface = gr.Interface(
    fn=process_video,
    inputs=gr.Textbox(label="Enter YouTube URL or Upload Video File"),
    outputs=[
        gr.Video(label="Video with Subtitles"),
        gr.Textbox(label="Transcript", lines=8)
    ],
    title="🎬 Whisper Subtitle Burner",
    description="Upload a video or paste a YouTube link. This app generates English subtitles using Whisper and burns them into the video."
)

iface.launch()