def stream(self):
url = self.get_url()
if url is None:
return None
# pylint: disable=consider-using-with
# if our url is actually a local file reference
# (can happen in local execution mode)
# then we just open a local file:
if self._looks_like_local_file(url):
fileobj = open(url, 'rb')
else:
fileobj = urlopen(url)
if fileobj:
try:
retval = tarfile.open(fileobj=fileobj, mode='r|*')
return retval
except BaseException as exc:
util.check_for_kb_interrupt()
fileobj.close()
msg: str = 'FAILED to stream artifact {0}: {1}'.format(
url, exc)
util.report_fatal(msg, self.logger)
return None
def _download_no_key_artifact(self):
if self.is_mutable:
self.logger.info("Downloading mutable artifact: %s", self.name)
if self.remote_path is None:
msg: str =\
"CANNOT download artifact without remote path: {0}"\
.format(self.name)
util.report_fatal(msg, self.logger)
key = self._generate_key()
local_path = artifacts_tracker.get_blob_cache(key)
local_path =\
self._get_target_local_path(local_path, self.remote_path)
if os.path.exists(local_path):
msg: str = ('Immutable artifact exists at local_path {0},' +
' skipping the download').format(local_path)
self.logger.debug(msg)
self.local_path = local_path
return local_path
if self.storage_handler.type == StorageType.storageDockerHub or \
self.storage_handler.type == StorageType.storageSHub:
msg: str = ('Qualified {0} points to a shub or dockerhub,' +
' skipping the download').format(self.remote_path)
self.logger.debug(msg)
return self.remote_path
self.storage_handler.download_remote_path(self.remote_path, local_path)
self.logger.debug('Downloaded file %s from external source %s',
local_path, self.remote_path)
self.local_path = local_path
#self.key = key
return self.local_path
def _tar_artifact_single_file(local_path: str, tar_filename: str, key, logger):
tf = None
try:
debug_str: str = ("Tarring artifact single file. " +
"tar_filename = {0}, " + "local_path = {1}, " +
"key = {2}").format(tar_filename, local_path, key)
logger.debug(debug_str)
tf = tarfile.open(tar_filename, 'w')
_, last_name = os.path.split(local_path)
tf.add(local_path, "./" + last_name)
except Exception as exc:
msg: str =\
"FAILED to create tarfile: {0} for artifact {1} reason: {2}"\
.format(tar_filename, local_path, exc)
util.report_fatal(msg, logger)
finally:
if tf is not None:
tf.close()
def _has_newer_artifact(self, local_path) -> bool:
self.logger.debug(
'Comparing date of the artifact %s in storage with local %s',
self.key, local_path)
storage_time = self.storage_handler.get_file_timestamp(self.key)
local_time = os.path.getmtime(local_path)
if storage_time is None:
msg: str = \
("Unable to get storage timestamp for {0}, storage is either " + \
"corrupted or has not finished uploading").format(self.key)
util.report_fatal(msg, self.logger)
return False
timestamp_shift = self.storage_handler.get_timestamp_shift()
if local_time > storage_time - timestamp_shift:
self.logger.debug(
"Local path %s is younger than stored %s, skipping the download",
local_path, self.key)
return False
return True
def _tar_artifact_directory(local_path: str, tar_filename: str, key,
ignore_filepath, logger):
tf = None
try:
debug_str: str = ("Tarring artifact directory. " +
"tar_filename = {0}, " + "local_path = {1}, " +
"key = {2}").format(tar_filename, local_path, key)
if ignore_filepath is not None:
debug_str += ", exclude = {0}".format(ignore_filepath)
logger.debug(debug_str)
tf = tarfile.open(tar_filename, 'w')
files_list = os.listdir(local_path)
for file_name in files_list:
tf.add(os.path.join(local_path, file_name), arcname=file_name)
except Exception as exc:
msg: str =\
"FAILED to create tarfile: {0} for artifact {1} reason: {2}"\
.format(tar_filename, local_path, exc)
util.report_fatal(msg, logger)
finally:
if tf is not None:
tf.close()
def upload(self, local_path=None):
if self.storage_handler is None:
msg: str = "No storage handler is set for artifact {0}"\
.format(self.key)
util.report_fatal(msg, self.logger)
if local_path is None:
local_path = self.local_path
if self.in_blobstore:
msg: str = ('Artifact with key {0} exists in blobstore, ' +
'skipping the upload').format(self.key)
self.logger.debug(msg)
return self.key
if os.path.exists(local_path):
tar_filename =\
tar_artifact(local_path, self.key,
self.get_compression(), self.logger)
if self.key is None:
self.key = 'blobstore/' + util.sha256_checksum(tar_filename) \
+ '.tar' + util.compression_to_extension(self.get_compression())
time_stamp = self.storage_handler.get_file_timestamp(self.key)
if time_stamp is not None:
self.logger.debug(
'Artifact with key %s exists in blobstore, skipping the upload',
self.key)
os.remove(tar_filename)
return self.key
self.storage_handler.upload_file(self.key, tar_filename)
os.remove(tar_filename)
return self.key
self.logger.debug(
"Local path %s does not exist. Not uploading anything.",
local_path)
return None
def download(self, local_path=None, only_newer=True):
if self.storage_handler is None:
msg: str = "No storage handler is set for artifact {0}" \
.format(self.key)
util.report_fatal(msg, self.logger)
if self.key is None:
return self._download_no_key_artifact()
if local_path is None:
if self.local_path is not None and \
os.path.exists(self.local_path):
local_path = self.local_path
else:
if self.is_mutable:
local_path = artifacts_tracker.get_artifact_cache(self.key)
else:
local_path = artifacts_tracker.get_blob_cache(self.key)
if os.path.exists(local_path):
msg: str = (
'Immutable artifact exists at local_path {0},' +
' skipping the download').format(local_path)
self.logger.debug(msg)
self.local_path = local_path
return local_path
local_path = re.sub(r'\/\Z', '', local_path)
self.logger.debug(
"Downloading dir %s to local path %s from studio.storage...",
self.key, local_path)
if only_newer and os.path.exists(local_path):
if not self._has_newer_artifact(local_path):
return local_path
# Now download our artifact from studio.storage and untar it:
return self._download_and_untar_artifact(local_path)
def __init__(self, cred_dict):
self.logger = logs.get_logger(self.__class__.__name__)
self.logger.setLevel(storage_setup.get_storage_verbose_level())
self.type = None
self.key = None
self.secret_key = None
self.session_token = None
self.region = None
self.profile = None
if cred_dict is None:
return
if isinstance(cred_dict, str) and cred_dict == 'none':
return
if not isinstance(cred_dict, dict):
msg: str =\
"NOT SUPPORTED credentials format {0}".format(repr(cred_dict))
util.report_fatal(msg, self.logger)
if len(cred_dict) == 0:
# Empty credentials dictionary is like None:
return
if len(cred_dict) == 1 and AWS_TYPE in cred_dict.keys():
aws_creds = cred_dict[AWS_TYPE]
self.type = AWS_TYPE
self.key = aws_creds.get(AWS_KEY, None)
self.secret_key = aws_creds.get(AWS_SECRET_KEY, None)
self.session_token = aws_creds.get(AWS_SESSION_TOKEN, None)
self.region = self._get_named(AWS_REGION, aws_creds)
self.profile = self._get_named(AWS_PROFILE, aws_creds)
if self.key is None or self.secret_key is None:
msg: str = \
"INVALID aws credentials format {0}".format(repr(cred_dict))
util.report_fatal(msg, self.logger)
else:
msg: str =\
"NOT SUPPORTED credentials format {0}".format(repr(cred_dict))
util.report_fatal(msg, self.logger)
def _report_fatal(self, msg: str):
report_fatal(msg, self.logger)
