38702-vm/core/video_engine.py

import os
import subprocess
import requests
import json
from pathlib import Path

WORKSPACE_ROOT = Path(__file__).resolve().parent.parent
OUTPUTS_DIR = WORKSPACE_ROOT / "outputs"
FINAL_VIDEO_DIR = OUTPUTS_DIR / "final_video"
TEMP_AUDIO_DIR = OUTPUTS_DIR / "temp_audio"
AUDIO_DIR = OUTPUTS_DIR / "audio"
BACKGROUNDS_DIR = WORKSPACE_ROOT / "backgrounds"

# Droid Sans Fallback often supports Arabic/CJK
FONT_ARABIC = "/usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf"
# Fallback for English text
FONT_SANS = "/usr/share/fonts/truetype/liberation/LiberationSans-Bold.ttf"

def escape_ffmpeg_text(text):
    if not text:
        return ""
    # Use ord() to check for backslash to avoid syntax errors with literals
    result = ""
    for char in text:
        if ord(char) == 92: # Backslash
            result += "\\"
        elif char == "'":
            result += "'\\''"
        elif char == ":":
            result += ":"
        else:
            result += char
    return result

def ensure_bg_video(bg_path):
    """Ensures a background video exists and is not empty. Creates a placeholder if needed."""
    if not bg_path.exists() or bg_path.stat().st_size == 0:
        bg_path.parent.mkdir(parents=True, exist_ok=True)
        # Create a 5-second 720x1280 (vertical) placeholder video
        color = "black"
        if "nature" in bg_path.name:
            color = "#064E3B" # Deep Emerald
        elif "space" in bg_path.name:
            color = "#1E1B4B" # Deep Indigo
            
        subprocess.run([
            'ffmpeg', '-y', '-f', 'lavfi', '-i', f'color=c={color}:s=720x1280:d=5',
            '-pix_fmt', 'yuv420p', str(bg_path)
        ], check=True)

def get_audio_duration(file_path):
    result = subprocess.run([
        'ffprobe', '-v', 'error', '-show_entries', 'format=duration',
        '-of', 'default=noprint_wrappers=1:nokey=1', str(file_path)
    ], capture_output=True, text=True)
    return float(result.stdout.strip())

def generate_video(task_id):
    from core.models import VideoTask
    task = VideoTask.objects.get(id=task_id)
    task.status = 'processing'
    task.save()

    try:
        # Ensure directories exist
        for d in [FINAL_VIDEO_DIR, TEMP_AUDIO_DIR, AUDIO_DIR, BACKGROUNDS_DIR]:
            d.mkdir(parents=True, exist_ok=True)

        # 1. Fetch Verses Text and Audio
        verses_data = []
        audio_files = []
        
        api_url = f"https://api.alquran.cloud/v1/surah/{task.surah_number}/{task.reciter_identifier}"
        resp = requests.get(api_url)
        resp.raise_for_status()
        data = resp.json()['data']
        
        all_ayahs = data['ayahs']
        selected_ayahs = [a for a in all_ayahs if task.verse_start <= a['numberInSurah'] <= task.verse_end]
        
        # Download audio files and collect text
        for i, ayah in enumerate(selected_ayahs):
            audio_url = ayah['audio']
            audio_path = TEMP_AUDIO_DIR / f"task_{task.id}_ayah_{i}.mp3"
            with requests.get(audio_url, stream=True) as r:
                r.raise_for_status()
                with open(audio_path, 'wb') as f:
                    for chunk in r.iter_content(chunk_size=8192):
                        f.write(chunk)
            audio_files.append(str(audio_path))
            verses_data.append({
                'text': ayah['text'],
                'duration': get_audio_duration(audio_path)
            })

        # 2. Combine Audio
        combined_audio = AUDIO_DIR / f"task_{task.id}_full.mp3"
        filter_complex_audio = "".join([f"[{i}:a]" for i in range(len(audio_files))]) + f"concat=n={len(audio_files)}:v=0:a=1[a]"
        cmd = ['ffmpeg', '-y']
        for f in audio_files:
            cmd.extend(['-i', f])
        cmd.extend(['-filter_complex', filter_complex_audio, '-map', '[a]', str(combined_audio)])
        subprocess.run(cmd, check=True)

        total_duration = get_audio_duration(combined_audio)

        # 3. Generate Video with FFmpeg
        bg_video = BACKGROUNDS_DIR / (task.background_video or "nature.mp4")
        ensure_bg_video(bg_video)

        output_video = FINAL_VIDEO_DIR / f"reels_{task.id}.mp4"
        
        # Build drawtext filters
        drawtext_filters = []
        
        # 1. Surah Name at top
        escaped_surah = escape_ffmpeg_text(task.surah_name)
        drawtext_filters.append(
            f"drawtext=text='{escaped_surah}':fontfile={FONT_SANS}:fontcolor={task.text_color}:fontsize=48:x=(w-text_w)/2:y=100:shadowcolor=black:shadowx=2:shadowy=2"
        )
        
        # 2. Verse text in middle (sequential)
        current_time = 0
        for verse in verses_data:
            escaped_text = escape_ffmpeg_text(verse['text'])
            start = current_time
            end = current_time + verse['duration']
            
            font_to_use = FONT_ARABIC if os.path.exists(FONT_ARABIC) else FONT_SANS
            
            drawtext_filters.append(
                f"drawtext=text='{escaped_text}':fontfile={font_to_use}:fontcolor={task.text_color}:fontsize=42:x=(w-text_w)/2:y=(h-text_h)/2:shadowcolor=black:shadowx=2:shadowy=2:enable='between(t,{start},{end})'"
            )
            current_time += verse['duration']

        vf_chain = [
            "scale=720:1280:force_original_aspect_ratio=increase",
            "crop=720:1280",
            *drawtext_filters
        ]
        vf_string = ",".join(vf_chain)

        subprocess.run([
            'ffmpeg', '-y', 
            '-stream_loop', '-1', '-i', str(bg_video), 
            '-i', str(combined_audio),
            '-t', str(total_duration),
            '-map', '0:v:0', '-map', '1:a:0',
            '-pix_fmt', 'yuv420p', 
            '-vf', vf_string,
            str(output_video)
        ], check=True)

        task.status = 'completed'
        task.output_path = f"/outputs/final_video/{output_video.name}"
        task.save()

    except Exception as e:
        task.status = 'failed'
        task.error_message = str(e)
        task.save()
        import traceback
        print(traceback.format_exc())
        raise e