MogensR's picture
Create models/downloader.py
39ebfbd
raw
history blame
18 kB
"""
Model downloader for BackgroundFX Pro.
Handles downloading, caching, and verification of models.
"""
import os
import shutil
import tempfile
import hashlib
import requests
from pathlib import Path
from typing import Optional, Callable, Dict, Any, List
from dataclasses import dataclass
from enum import Enum
import time
import threading
from urllib.parse import urlparse
from concurrent.futures import ThreadPoolExecutor, Future
import logging
from .registry import ModelInfo, ModelStatus, ModelRegistry
logger = logging.getLogger(__name__)
class DownloadStatus(Enum):
"""Download status."""
PENDING = "pending"
DOWNLOADING = "downloading"
VERIFYING = "verifying"
EXTRACTING = "extracting"
COMPLETED = "completed"
FAILED = "failed"
CANCELLED = "cancelled"
@dataclass
class DownloadProgress:
"""Download progress information."""
model_id: str
status: DownloadStatus
current_bytes: int = 0
total_bytes: int = 0
speed_mbps: float = 0.0
eta_seconds: float = 0.0
error: Optional[str] = None
@property
def progress(self) -> float:
"""Get progress percentage."""
if self.total_bytes > 0:
return (self.current_bytes / self.total_bytes) * 100
return 0.0
class ModelDownloader:
"""Handle model downloading with progress tracking and resume support."""
def __init__(self,
registry: ModelRegistry,
max_workers: int = 3,
chunk_size: int = 8192,
timeout: int = 30,
max_retries: int = 3):
"""
Initialize model downloader.
Args:
registry: Model registry instance
max_workers: Maximum concurrent downloads
chunk_size: Download chunk size in bytes
timeout: Request timeout in seconds
max_retries: Maximum retry attempts
"""
self.registry = registry
self.max_workers = max_workers
self.chunk_size = chunk_size
self.timeout = timeout
self.max_retries = max_retries
# Download management
self.downloads: Dict[str, DownloadProgress] = {}
self.executor = ThreadPoolExecutor(max_workers=max_workers)
self.futures: Dict[str, Future] = {}
self._stop_events: Dict[str, threading.Event] = {}
# Cache directory
self.cache_dir = registry.models_dir / ".cache"
self.cache_dir.mkdir(exist_ok=True)
def download_model(self,
model_id: str,
progress_callback: Optional[Callable[[DownloadProgress], None]] = None,
force: bool = False) -> bool:
"""
Download a model.
Args:
model_id: Model ID to download
progress_callback: Optional progress callback
force: Force re-download even if exists
Returns:
True if download successful
"""
# Get model info
model = self.registry.get_model(model_id)
if not model:
logger.error(f"Model not found: {model_id}")
return False
# Check if already downloaded
if not force and model.status == ModelStatus.AVAILABLE:
logger.info(f"Model already available: {model_id}")
return True
# Initialize progress
progress = DownloadProgress(
model_id=model_id,
status=DownloadStatus.PENDING,
total_bytes=model.file_size
)
self.downloads[model_id] = progress
# Create stop event
self._stop_events[model_id] = threading.Event()
# Submit download task
future = self.executor.submit(
self._download_model_task,
model,
progress,
progress_callback,
force
)
self.futures[model_id] = future
# Wait for completion
try:
return future.result()
except Exception as e:
logger.error(f"Download failed for {model_id}: {e}")
return False
def download_models_async(self,
model_ids: List[str],
progress_callback: Optional[Callable[[str, DownloadProgress], None]] = None,
force: bool = False) -> Dict[str, Future]:
"""
Download multiple models asynchronously.
Args:
model_ids: List of model IDs
progress_callback: Optional progress callback with model_id
force: Force re-download
Returns:
Dictionary of futures
"""
futures = {}
for model_id in model_ids:
model = self.registry.get_model(model_id)
if not model:
logger.warning(f"Model not found: {model_id}")
continue
# Skip if already available
if not force and model.status == ModelStatus.AVAILABLE:
continue
# Initialize progress
progress = DownloadProgress(
model_id=model_id,
status=DownloadStatus.PENDING,
total_bytes=model.file_size
)
self.downloads[model_id] = progress
# Create stop event
self._stop_events[model_id] = threading.Event()
# Wrapper for progress callback
def progress_wrapper(p):
if progress_callback:
progress_callback(model_id, p)
# Submit download task
future = self.executor.submit(
self._download_model_task,
model,
progress,
progress_wrapper,
force
)
futures[model_id] = future
self.futures[model_id] = future
return futures
def _download_model_task(self,
model: ModelInfo,
progress: DownloadProgress,
progress_callback: Optional[Callable],
force: bool) -> bool:
"""
Download model task.
Args:
model: Model information
progress: Progress tracker
progress_callback: Progress callback
force: Force re-download
Returns:
True if successful
"""
try:
# Update status
progress.status = DownloadStatus.DOWNLOADING
self._notify_progress(progress, progress_callback)
# Try primary URL first, then mirrors
urls = [model.url] + model.mirror_urls
for url in urls:
if self._stop_events[model.model_id].is_set():
progress.status = DownloadStatus.CANCELLED
self._notify_progress(progress, progress_callback)
return False
try:
# Download file
output_path = self.registry.models_dir / model.filename
success = self._download_file(
url,
output_path,
progress,
progress_callback,
model.model_id
)
if success:
# Verify file
progress.status = DownloadStatus.VERIFYING
self._notify_progress(progress, progress_callback)
if self._verify_download(output_path, model):
# Update registry
model.status = ModelStatus.AVAILABLE
model.local_path = str(output_path)
model.download_date = time.time()
self.registry._save_registry()
progress.status = DownloadStatus.COMPLETED
self._notify_progress(progress, progress_callback)
logger.info(f"Successfully downloaded: {model.model_id}")
return True
else:
# Verification failed
output_path.unlink(missing_ok=True)
logger.warning(f"Verification failed for {model.model_id}")
except Exception as e:
logger.warning(f"Download failed from {url}: {e}")
continue
# All attempts failed
progress.status = DownloadStatus.FAILED
progress.error = "All download attempts failed"
self._notify_progress(progress, progress_callback)
return False
except Exception as e:
progress.status = DownloadStatus.FAILED
progress.error = str(e)
self._notify_progress(progress, progress_callback)
logger.error(f"Download task failed: {e}")
return False
def _download_file(self,
url: str,
output_path: Path,
progress: DownloadProgress,
progress_callback: Optional[Callable],
model_id: str) -> bool:
"""
Download file with resume support.
Args:
url: Download URL
output_path: Output file path
progress: Progress tracker
progress_callback: Progress callback
model_id: Model ID for stop event
Returns:
True if successful
"""
# Check for partial download
temp_path = output_path.with_suffix('.part')
resume_pos = 0
if temp_path.exists():
resume_pos = temp_path.stat().st_size
logger.info(f"Resuming download from {resume_pos} bytes")
# Prepare headers for resume
headers = {}
if resume_pos > 0:
headers['Range'] = f'bytes={resume_pos}-'
# Start download
start_time = time.time()
bytes_downloaded = resume_pos
try:
response = requests.get(
url,
headers=headers,
stream=True,
timeout=self.timeout
)
response.raise_for_status()
# Get total size
if 'content-length' in response.headers:
total_size = int(response.headers['content-length']) + resume_pos
progress.total_bytes = total_size
else:
total_size = None
# Download with progress
mode = 'ab' if resume_pos > 0 else 'wb'
with open(temp_path, mode) as f:
for chunk in response.iter_content(chunk_size=self.chunk_size):
# Check for cancellation
if self._stop_events[model_id].is_set():
logger.info(f"Download cancelled: {model_id}")
return False
if chunk:
f.write(chunk)
bytes_downloaded += len(chunk)
# Update progress
progress.current_bytes = bytes_downloaded
# Calculate speed and ETA
elapsed = time.time() - start_time
if elapsed > 0:
speed_bps = (bytes_downloaded - resume_pos) / elapsed
progress.speed_mbps = (speed_bps * 8) / 1_000_000
if total_size and speed_bps > 0:
remaining = total_size - bytes_downloaded
progress.eta_seconds = remaining / speed_bps
self._notify_progress(progress, progress_callback)
# Move to final location
shutil.move(str(temp_path), str(output_path))
return True
except requests.exceptions.RequestException as e:
logger.error(f"Download error: {e}")
return False
except Exception as e:
logger.error(f"File write error: {e}")
return False
def _verify_download(self, file_path: Path, model: ModelInfo) -> bool:
"""
Verify downloaded file.
Args:
file_path: Downloaded file path
model: Model information
Returns:
True if verification passed
"""
# Check file exists
if not file_path.exists():
return False
# Check file size
actual_size = file_path.stat().st_size
if model.file_size > 0:
size_diff = abs(actual_size - model.file_size)
if size_diff > 1000: # Allow 1KB difference
logger.warning(f"Size mismatch: expected {model.file_size}, got {actual_size}")
return False
# Check SHA256 if available
if model.sha256:
try:
sha256 = self._calculate_sha256(file_path)
if sha256 != model.sha256:
logger.warning(f"SHA256 mismatch for {model.model_id}")
return False
except Exception as e:
logger.error(f"SHA256 calculation failed: {e}")
return False
return True
def _calculate_sha256(self, file_path: Path) -> str:
"""Calculate SHA256 hash of file."""
sha256_hash = hashlib.sha256()
with open(file_path, "rb") as f:
for byte_block in iter(lambda: f.read(self.chunk_size), b""):
sha256_hash.update(byte_block)
return sha256_hash.hexdigest()
def _notify_progress(self, progress: DownloadProgress, callback: Optional[Callable]):
"""Notify progress callback."""
if callback:
try:
callback(progress)
except Exception as e:
logger.error(f"Progress callback error: {e}")
def cancel_download(self, model_id: str) -> bool:
"""
Cancel ongoing download.
Args:
model_id: Model ID to cancel
Returns:
True if cancelled
"""
if model_id in self._stop_events:
self._stop_events[model_id].set()
# Wait for cancellation
if model_id in self.futures:
try:
self.futures[model_id].result(timeout=5)
except:
pass
del self.futures[model_id]
# Update progress
if model_id in self.downloads:
self.downloads[model_id].status = DownloadStatus.CANCELLED
logger.info(f"Download cancelled: {model_id}")
return True
return False
def get_progress(self, model_id: str) -> Optional[DownloadProgress]:
"""Get download progress for model."""
return self.downloads.get(model_id)
def get_all_progress(self) -> Dict[str, DownloadProgress]:
"""Get all download progress."""
return self.downloads.copy()
def cleanup_partial_downloads(self):
"""Clean up partial download files."""
for file in self.registry.models_dir.glob("*.part"):
try:
file.unlink()
logger.info(f"Removed partial download: {file.name}")
except Exception as e:
logger.error(f"Failed to remove {file}: {e}")
def download_required_models(self,
task: str = None,
gpu_available: bool = True) -> bool:
"""
Download all required models for a task.
Args:
task: Optional task filter
gpu_available: GPU availability
Returns:
True if all downloads successful
"""
# Get required models
required = []
if task:
# Get best model for task
from .registry import ModelTask
task_enum = ModelTask(task)
model = self.registry.get_best_model(
task_enum,
require_gpu=gpu_available if gpu_available else False
)
if model:
required.append(model.model_id)
else:
# Get all essential models
essential = ['rmbg-1.4', 'u2netp', 'modnet']
for model_id in essential:
if self.registry.get_model(model_id):
required.append(model_id)
# Download models
if required:
logger.info(f"Downloading required models: {required}")
futures = self.download_models_async(required)
# Wait for completion
success = True
for model_id, future in futures.items():
try:
if not future.result():
success = False
except Exception:
success = False
return success
return True