Files
modelscope/modelscope/msdatasets/utils/oss_utils.py
2026-01-12 11:53:29 +08:00

241 lines
8.7 KiB
Python

# Copyright (c) Alibaba, Inc. and its affiliates.
from __future__ import print_function
import multiprocessing
import os
import threading
import oss2
from datasets.utils.file_utils import hash_url_to_filename
from oss2 import CredentialsProvider
from oss2.credentials import Credentials
from modelscope.hub.api import HubApi
from modelscope.msdatasets.download.download_config import DataDownloadConfig
from modelscope.utils.config_ds import MS_CACHE_HOME
from modelscope.utils.constant import (DEFAULT_DATA_ACCELERATION_ENDPOINT,
MetaDataFields, UploadMode)
from modelscope.utils.logger import get_logger
logger = get_logger()
ACCESS_ID = 'AccessId'
ACCESS_SECRET = 'AccessSecret'
SECURITY_TOKEN = 'SecurityToken'
BUCKET = 'Bucket'
BACK_DIR = 'BackupDir'
DIR = 'Dir'
class CredentialProviderWrapper(CredentialsProvider):
"""
A custom credentials provider for oss2 that fetches temporary credentials
"""
def __init__(self, api: HubApi, dataset_name: str, namespace: str,
revision: str):
"""
Initializes the CredentialProviderWrapper with dataset information.
Args:
dataset_name (str): The name of the dataset.
namespace (str): The namespace of the dataset.
revision (str): The revision of the dataset.
"""
self.api = api
self.dataset_name = dataset_name
self.namespace = namespace
self.revision = revision
self._lock = threading.Lock()
def get_credentials(self):
"""
oss2 SDK will call this method automatically when it finds the token is expired or needs authentication.
"""
with self._lock:
oss_config = self.api.get_dataset_access_config_session(
dataset_name=self.dataset_name,
namespace=self.namespace,
check_cookie=True,
revision=self.revision)
return Credentials(
access_key_id=oss_config[ACCESS_ID],
access_key_secret=oss_config[ACCESS_SECRET],
security_token=oss_config[SECURITY_TOKEN],
)
class OssUtilities:
"""
A utility class for handling Alibaba Cloud OSS operations such as upload and download.
"""
def __init__(self, dataset_name, namespace, revision):
"""
Initializes the OssUtilities with the given OSS configuration and dataset information.
"""
self.dataset_name = dataset_name
self.namespace = namespace
self.revision = revision
self.api = HubApi()
oss_config = self.api.get_dataset_access_config_session(
dataset_name=self.dataset_name,
namespace=self.namespace,
check_cookie=True,
revision=self.revision)
if os.getenv('ENABLE_DATASET_ACCELERATION') == 'True':
self.endpoint = DEFAULT_DATA_ACCELERATION_ENDPOINT
else:
self.endpoint = f"https://{oss_config['Region']}.aliyuncs.com"
self.resumable_store_root_path = os.path.join(MS_CACHE_HOME,
'tmp/resumable_store')
self.num_threads = multiprocessing.cpu_count()
self.part_size = 1 * 1024 * 1024
self.multipart_threshold = 50 * 1024 * 1024
self.max_retries = 3
self.resumable_store_download = oss2.ResumableDownloadStore(
root=self.resumable_store_root_path)
self.resumable_store_upload = oss2.ResumableStore(
root=self.resumable_store_root_path)
credential_provider = CredentialProviderWrapper(
api=self.api,
dataset_name=self.dataset_name,
namespace=self.namespace,
revision=self.revision)
auth = oss2.ProviderAuthV4(credential_provider)
self.bucket_name = oss_config[BUCKET]
self.bucket = oss2.Bucket(
auth=auth,
endpoint=self.endpoint,
bucket_name=self.bucket_name,
region=oss_config['Region'].lstrip('oss-'),
)
self.oss_dir = oss_config[DIR]
self.oss_backup_dir = oss_config[BACK_DIR]
@staticmethod
def _percentage(consumed_bytes, total_bytes):
if total_bytes:
rate = int(100 * (float(consumed_bytes) / float(total_bytes)))
print('\r{0}% '.format(rate), end='', flush=True)
def download(self, oss_file_name: str,
download_config: DataDownloadConfig) -> str:
"""
Downloads a file from OSS to the local cache.
Args:
oss_file_name (str): The name of the file in OSS to download.
download_config (DataDownloadConfig): Configuration for the download process.
Returns:
str: The local path to the downloaded file.
"""
cache_dir = download_config.cache_dir
candidate_key = os.path.join(self.oss_dir, oss_file_name)
candidate_key_backup = os.path.join(self.oss_backup_dir, oss_file_name)
split = download_config.split
big_data = False
if split:
args_dict = download_config.meta_args_map.get(split)
if args_dict:
big_data = args_dict.get(MetaDataFields.ARGS_BIG_DATA)
retry_count = 0
while True:
try:
# big_data is True when the dataset contains large number of objects
if big_data:
file_oss_key = candidate_key
else:
file_oss_key = candidate_key if self.bucket.object_exists(
candidate_key) else candidate_key_backup
filename = hash_url_to_filename(file_oss_key, etag=None)
local_path = os.path.join(cache_dir, filename)
if download_config.force_download or not os.path.exists(
local_path):
oss2.resumable_download(
bucket=self.bucket,
key=file_oss_key,
filename=local_path,
store=self.resumable_store_download,
multiget_threshold=self.multipart_threshold,
part_size=self.part_size,
progress_callback=self._percentage,
num_threads=self.num_threads)
break
except Exception as e:
logger.warning(
f'Error downloading {oss_file_name}: {e}, trying again...')
retry_count += 1
if retry_count >= self.max_retries:
logger.error(
f'Failed to download {oss_file_name} due to exceeded retries.'
)
raise e
return local_path
def upload(self, oss_object_name: str, local_file_path: str,
indicate_individual_progress: bool,
upload_mode: UploadMode) -> str:
"""
Uploads a local file to OSS.
Args:
oss_object_name (str): The name of the object in OSS.
local_file_path (str): The local file path to upload.
indicate_individual_progress (bool): Whether to show individual progress.
upload_mode (UploadMode): The upload mode (e.g., OVERWRITE, APPEND).
Returns:
str: The OSS object key where the file is uploaded.
"""
retry_count = 0
object_key = os.path.join(self.oss_dir, oss_object_name)
if indicate_individual_progress:
progress_callback = self._percentage
else:
progress_callback = None
while True:
try:
exist = self.bucket.object_exists(object_key)
if upload_mode == UploadMode.APPEND and exist:
logger.info(
f'Skip {oss_object_name} in case of {upload_mode.value} mode.'
)
break
oss2.resumable_upload(
bucket=self.bucket,
key=object_key,
filename=local_file_path,
store=self.resumable_store_upload,
multipart_threshold=self.multipart_threshold,
part_size=self.part_size,
progress_callback=progress_callback,
num_threads=self.num_threads)
break
except Exception as e:
logger.warning(
f'Error uploading {oss_object_name}: {e}, trying again...')
retry_count += 1
if retry_count >= self.max_retries:
logger.error(
f'Failed to upload {oss_object_name} due to exceeded retries.'
)
raise e
return object_key