1217d8cef0
Deploy / Build Production Runtime Images (push) Has been skipped
Deploy / Deploy Production (push) Has been skipped
Deploy / Production Browser E2E (push) Has been skipped
Deploy / Deploy Staging (push) Failing after 210h35m44s
CI/CD Pipeline / Frontend Lint (push) Failing after 210h36m11s
CI/CD Pipeline / Validate Code Quality And Tests (push) Failing after 210h36m17s
155 lines
5.6 KiB
Python
155 lines
5.6 KiB
Python
"""Voice extraction tasks - extract voice tracks and background music from videos."""
|
|
|
|
import json
|
|
import logging
|
|
import os
|
|
import subprocess
|
|
import tempfile
|
|
from typing import Optional
|
|
|
|
from celery import Task
|
|
from sqlalchemy.orm import Session
|
|
from worker_app.celery_app import celery_app
|
|
from worker_app.db import SessionLocal
|
|
|
|
from packages.adapters.sqlalchemy_impl.asset_repository import SQLAlchemyAssetRepository
|
|
from packages.shared.storage import get_storage_service
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
class VoiceExtractor:
|
|
"""Extract voice tracks and background music from videos using FFmpeg."""
|
|
|
|
@staticmethod
|
|
def _run_ffmpeg(cmd: list[str]) -> subprocess.CompletedProcess:
|
|
"""Run FFmpeg command and return result."""
|
|
logger.info(f"Running FFmpeg: {chr(39).join(cmd)}")
|
|
result = subprocess.run(cmd, capture_output=True, text=True)
|
|
if result.returncode != 0:
|
|
logger.error(f"FFmpeg error: {result.stderr}")
|
|
raise RuntimeError(f"FFmpeg failed: {result.stderr}")
|
|
return result
|
|
|
|
def extract_voice(
|
|
self,
|
|
input_path: str,
|
|
output_path: str,
|
|
highpass: int = 200,
|
|
bandpass_freq: int = 300,
|
|
bandpass_width: int = 3000,
|
|
noise_reduction: int = 20,
|
|
) -> str:
|
|
"""Extract voice track from video using FFmpeg."""
|
|
os.makedirs(os.path.dirname(output_path), exist_ok=True)
|
|
cmd = [
|
|
"ffmpeg",
|
|
"-y",
|
|
"-i",
|
|
input_path,
|
|
"-af",
|
|
f"highpass=f={highpass},afftdn=bn={noise_reduction},bandpass=f={bandpass_freq}:width_type=h:width={bandpass_width},loudnorm",
|
|
"-vn",
|
|
"-acodec",
|
|
"libmp3lame",
|
|
"-q:a",
|
|
"2",
|
|
output_path,
|
|
]
|
|
self._run_ffmpeg(cmd)
|
|
logger.info(f"Voice extracted to: {output_path}")
|
|
return output_path
|
|
|
|
def extract_background(self, input_path: str, output_path: str, lowpass: int = 200) -> str:
|
|
"""Extract background music from video."""
|
|
os.makedirs(os.path.dirname(output_path), exist_ok=True)
|
|
cmd = [
|
|
"ffmpeg",
|
|
"-y",
|
|
"-i",
|
|
input_path,
|
|
"-af",
|
|
f"lowpass=f={lowpass},loudnorm",
|
|
"-vn",
|
|
"-acodec",
|
|
"libmp3lame",
|
|
"-q:a",
|
|
"2",
|
|
output_path,
|
|
]
|
|
self._run_ffmpeg(cmd)
|
|
logger.info(f"Background extracted to: {output_path}")
|
|
return output_path
|
|
|
|
|
|
@celery_app.task(bind=True, max_retries=3, name="worker.extract_voice")
|
|
def extract_voice_task(self: Task, asset_id: str) -> dict:
|
|
session = SessionLocal()
|
|
temp_dir = tempfile.mkdtemp()
|
|
try:
|
|
asset_repo = SQLAlchemyAssetRepository(session)
|
|
storage_service = get_storage_service()
|
|
extractor = VoiceExtractor()
|
|
asset = asset_repo.get(asset_id)
|
|
if asset is None:
|
|
raise ValueError(f"Asset {asset_id} not found")
|
|
local_video_path = os.path.join(temp_dir, f"{asset_id}.mp4")
|
|
storage_service.download_file(asset.storage_key, local_video_path)
|
|
voice_output_path = os.path.join(temp_dir, f"{asset_id}_voice.mp3")
|
|
extractor.extract_voice(local_video_path, voice_output_path)
|
|
voice_storage_key = f"assets/{asset_id}/voice.mp3"
|
|
storage_service.upload_file(voice_output_path, voice_storage_key)
|
|
voice_url = storage_service.get_url(voice_storage_key)
|
|
if asset.metadata is None:
|
|
asset.metadata = {}
|
|
asset.metadata["voice_url"] = voice_url
|
|
asset_repo.update(asset)
|
|
session.commit()
|
|
logger.info(f"Voice extraction completed for asset {asset_id}: {voice_url}")
|
|
return {"ok": True, "asset_id": asset_id, "voice_url": voice_url}
|
|
except Exception as e:
|
|
logger.error(f"Voice extraction failed for {asset_id}: {str(e)}")
|
|
session.rollback()
|
|
raise self.retry(exc=e, countdown=60)
|
|
finally:
|
|
session.close()
|
|
import shutil
|
|
|
|
shutil.rmtree(temp_dir, ignore_errors=True)
|
|
|
|
|
|
@celery_app.task(bind=True, max_retries=3, name="worker.extract_background")
|
|
def extract_background_task(self: Task, asset_id: str) -> dict:
|
|
session = SessionLocal()
|
|
temp_dir = tempfile.mkdtemp()
|
|
try:
|
|
asset_repo = SQLAlchemyAssetRepository(session)
|
|
storage_service = get_storage_service()
|
|
extractor = VoiceExtractor()
|
|
asset = asset_repo.get(asset_id)
|
|
if asset is None:
|
|
raise ValueError(f"Asset {asset_id} not found")
|
|
local_video_path = os.path.join(temp_dir, f"{asset_id}.mp4")
|
|
storage_service.download_file(asset.storage_key, local_video_path)
|
|
bg_output_path = os.path.join(temp_dir, f"{asset_id}_background.mp3")
|
|
extractor.extract_background(local_video_path, bg_output_path)
|
|
bg_storage_key = f"assets/{asset_id}/background.mp3"
|
|
storage_service.upload_file(bg_output_path, bg_storage_key)
|
|
bg_url = storage_service.get_url(bg_storage_key)
|
|
if asset.metadata is None:
|
|
asset.metadata = {}
|
|
asset.metadata["background_url"] = bg_url
|
|
asset_repo.update(asset)
|
|
session.commit()
|
|
logger.info(f"Background extraction completed for asset {asset_id}: {bg_url}")
|
|
return {"ok": True, "asset_id": asset_id, "background_url": bg_url}
|
|
except Exception as e:
|
|
logger.error(f"Background extraction failed for {asset_id}: {str(e)}")
|
|
session.rollback()
|
|
raise self.retry(exc=e, countdown=60)
|
|
finally:
|
|
session.close()
|
|
import shutil
|
|
|
|
shutil.rmtree(temp_dir, ignore_errors=True)
|