def submit():
blob_service = BlobService(account_name=ACCOUNT_NAME, account_key=ACCOUNT_KEY)
# Get a SAS signature (read for 24 hours) for the input container save to a string
inputsig = sasUrl(account=ACCOUNT_NAME, key=ACCOUNT_KEY, container=INPUT_CONTAINER, permission='r')
# Get a SAS signature (write for 24 hours) for the output container save to a string
outputsig = sasUrl(account = ACCOUNT_NAME, key = ACCOUNT_KEY, container = OUTPUT_CONTAINER, permission = 'rwl')
# List all the blobs and dump the content to a string
blobs = blob_service.list_blobs(INPUT_CONTAINER)
bloblist = []
for blob in blobs:
bloblist.append(blob.name)
os.environ[SLURMDEMO_INPUTSIG] = inputsig
os.environ[SLURMDEMO_OUTPUTSIG] = outputsig
os.environ[SLURMDEMO_BLOBLIST] = json.dumps(bloblist)
os.environ[SLURMDEMO_INPUTCONTAINER] = INPUT_CONTAINER
os.environ[SLURMDEMO_OUTPUTCONTAINER] = OUTPUT_CONTAINER
os.environ[SLURMDEMO_INPUTACCOUNT] = ACCOUNT_NAME
os.environ[SLURMDEMO_OUTPUTACCOUNT] = ACCOUNT_NAME
# Call sbatch
cli = "sbatch -N 2 -n 2 --array=0-{nb} slurmdemo.sh".format(nb=len(bloblist))
run(cli, showoutput=True)
def submit():
blob_service = BlobService(account_name=ACCOUNT_NAME,
account_key=ACCOUNT_KEY)
# Get a SAS signature (read for 24 hours) for the input container save to a string
inputsig = sasUrl(account=ACCOUNT_NAME,
key=ACCOUNT_KEY,
container=INPUT_CONTAINER,
permission='r')
# Get a SAS signature (write for 24 hours) for the output container save to a string
outputsig = sasUrl(account=ACCOUNT_NAME,
key=ACCOUNT_KEY,
container=OUTPUT_CONTAINER,
permission='rwl')
# List all the blobs and dump the content to a string
blobs = blob_service.list_blobs(INPUT_CONTAINER)
bloblist = []
for blob in blobs:
bloblist.append(blob.name)
os.environ[SLURMDEMO_INPUTSIG] = inputsig
os.environ[SLURMDEMO_OUTPUTSIG] = outputsig
os.environ[SLURMDEMO_BLOBLIST] = json.dumps(bloblist)
os.environ[SLURMDEMO_INPUTCONTAINER] = INPUT_CONTAINER
os.environ[SLURMDEMO_OUTPUTCONTAINER] = OUTPUT_CONTAINER
os.environ[SLURMDEMO_INPUTACCOUNT] = ACCOUNT_NAME
os.environ[SLURMDEMO_OUTPUTACCOUNT] = ACCOUNT_NAME
# Call sbatch
cli = "sbatch --array=0-{nb} slurmdemo.sh".format(nb=len(bloblist))
run(cli, showoutput=True)
def enumerate_objects(container):
blob_service = BlobService(AZURE_ACCOUNT_NAME, AZURE_ACCOUNT_KEY)
blobs = blob_service.list_blobs(container)
items = []
for blob in blobs:
items.append(blob.name)
return items
def upload_azure_blob(account, account_key, container, filename, file, file_type='file/csv'):
block_blob_service = BlobService(account_name=account, account_key=account_key)
# block_blob_service.put_block_blob_from_path(
# container,
# blockblob,
# file,
# x_ms_blob_content_type='file/csv'
# )
block_blob_service.create_blob_from_stream(container, filename, file)
generator = block_blob_service.list_blobs(container)
for blob in generator:
print(blob.name)
def download_data(key):
blob_service = BlobService(account_name='asosdsrecruiting', account_key=key)
blobs = []
marker = None
while True:
batch = blob_service.list_blobs('recruitingdata', marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
file_name = blob.name
print('Downloading: ' + file_name)
blob_service.get_blob_to_path('recruitingdata', file_name, file_name.replace('/', '_') + '.csv')
class _BlobStorageTestCase(_TestCase):
def _get_container_name(self, handler_name):
container = _get_handler_config_value(handler_name, 'container')
if container:
container = container.replace('_', '-').lower()
return container
def setUp(self):
self.service = BlobService(ACCOUNT_NAME, ACCOUNT_KEY)
# ensure that there's no log file in the container before each test
containers = [c.name for c in self.service.list_containers()]
for handler in LOGGING['handlers']:
container = self._get_container_name(handler)
if container in containers:
filename = _get_handler_config_value(handler, 'filename')
basename = os.path.basename(filename)
for blob in self.service.list_blobs(container, prefix=basename):
self.service.delete_blob(container, blob.name)
class _BlobStorageTestCase(_TestCase):
def _get_container_name(self, handler_name):
container = _get_handler_config_value(handler_name, 'container')
if container:
container = container.replace('_', '-').lower()
return container
def setUp(self):
self.service = BlobService(ACCOUNT_NAME, ACCOUNT_KEY)
# ensure that there's no log file in the container before each test
containers = [c.name for c in self.service.list_containers()]
for handler in LOGGING['handlers']:
container = self._get_container_name(handler)
if container in containers:
filename = _get_handler_config_value(handler, 'filename')
basename = os.path.basename(filename)
for blob in self.service.list_blobs(container,
prefix=basename):
self.service.delete_blob(container, blob.name)
class BlobReader(Reader):
def __init__(self, account, key, container):
self.block_blob_service = BlobService(account_name=account,
account_key=key)
self.container = container
def get_data(self, name):
counter = BLOB_RETRIES
while counter:
try:
data = self.block_blob_service.get_blob_to_bytes(
self.container, name)
except AzureException as azure_exc:
counter -= 1
else:
return data
raise RuntimeError("Couldn't read from blob, %s" % (azure_exc.args[0]))
def to_file(self, handle, blobpath):
counter = BLOB_RETRIES
while counter:
try:
self.block_blob_service.get_blob_to_file(
self.container,
blobpath,
handle,
max_connections=2,
progress_callback=None)
except AzureException as azure_exc:
counter -= 1
else:
return
raise RuntimeError("Couldn't download blob, %s" % (azure_exc.args[0]))
def list(self, prefix):
return self.block_blob_service.list_blobs(self.container, prefix)
def module_impl(rm, log, params, check_mode=False):
if not HAS_AZURE:
raise Exception("The Azure python sdk is not installed (try 'pip install azure')")
if not HAS_REQUESTS:
raise Exception("The requests python module is not installed (try 'pip install requests')")
resource_group = params.get('resource_group')
account_name = params.get('account_name')
container_name = params.get('container_name')
mode = params.get('mode')
x_ms_meta_name_values = params.get('x_ms_meta_name_values')
x_ms_blob_public_access = params.get('x_ms_blob_public_access')
x_ms_blob_cache_control = params.get('x_ms_blob_cache_control')
x_ms_blob_content_encoding = params.get('x_ms_blob_content_encoding')
x_ms_blob_content_language = params.get('x_ms_blob_content_language')
x_ms_blob_content_type = params.get('x_ms_blob_content_type')
prefix = params.get('prefix')
marker = params.get('marker')
max_results = params.get('max_results')
blob_name = params.get('blob_name')
file_path = params.get('file_path')
overwrite = params.get('overwrite')
permissions = params.get('permissions')
hours = params.get('hours')
days = params.get('days')
access_token = params.get('access_token')
results = dict(changed=False)
storage_client = rm.storage_client
if not resource_group:
raise Exception("Parameter error: resource_group cannot be None.")
if not account_name:
raise Exception("Parameter error: account_name cannot be None.")
if not container_name:
raise Exception("Parameter error: container_name cannot be None.")
if not NAME_PATTERN.match(container_name):
raise Exception("Parameter error: container_name must consist of lowercase letters, numbers and hyphens. It must begin with " +
"a letter or number. It may not contain two consecutive hyphens.")
# add file path validation
results['account_name'] = account_name
results['resource_group'] = resource_group
results['container_name'] = container_name
# put (upload), get (download), geturl (return download url (Ansible 1.3+), getstr (download object as string (1.3+)), list (list keys (2.0+)), create (bucket), delete (bucket), and delobj (delete object)
try:
log('Getting keys')
keys = {}
response = storage_client.storage_accounts.list_keys(resource_group, account_name)
keys[KeyName.key1] = response.storage_account_keys.key1
keys[KeyName.key2] = response.storage_account_keys.key2
except AzureHttpError as e:
log('Error getting keys for account %s' % account_name)
raise Exception(str(e.message))
try:
log('Create blob service')
bs = BlobService(account_name, keys[KeyName.key1])
except Exception as e:
log('Error creating blob service.')
raise Exception(str(e.args[0]))
if mode == 'create':
container = get_container_facts(bs, container_name)
if container is not None:
# container exists
results['container'] = container
results['msg'] = "Container already exists."
return results
# create the container
if not check_mode:
log('Create container %s' % container_name)
bs.create_container(container_name, x_ms_meta_name_values, x_ms_blob_public_access)
results['container'] = get_container_facts(bs, container_name)
results['msg'] = "Container created successfully."
results['changed'] = True
return results
if mode == 'update':
container = get_container_facts(bs, container_name)
if container is None:
# container does not exist
if not check_mode:
log('Create container %s' % container_name)
bs.create_container(container_name, x_ms_meta_name_values, x_ms_blob_public_access)
results['changed'] = True
results['msg'] = 'Container created successfully.'
return results
# update existing container
results['msg'] = "Container not changed."
if x_ms_meta_name_values:
if not check_mode:
log('Update x_ms_meta_name_values for container %s' % container_name)
bs.set_container_metadata(container_name, x_ms_meta_name_values)
results['changed'] = True
results['msg'] = 'Container meta data updated successfully.'
if x_ms_blob_public_access:
access = x_ms_blob_public_access
if x_ms_blob_public_access == 'private':
access = None
if not check_mode:
log('Set access to %s for container %s' % (access, container_name))
bs.set_container_acl(container_name=container_name, x_ms_blob_public_access=access)
results['changed'] = True
results['msg'] = 'Container ACL updated successfully.'
if permissions:
if hours == 0 and days == 0:
raise Exception("Parameter error: expecting hours > 0 or days > 0")
id = "%s-%s" % (container_name, permissions)
si = get_identifier(id, hours, days, permissions)
identifiers = SignedIdentifiers()
identifiers.signed_identifiers.append(si)
if not check_mode:
log('Set permissions to %s for container %s' % (permissions, container_name))
bs.set_container_acl(container_name=container_name,signed_identifiers=identifiers)
results['changed'] = True
results['msg'] = 'Container ACL updated successfully.'
results['container'] = get_container_facts(bs, container_name)
return results
if mode == 'delete':
container = get_container_facts(bs, container_name)
if container is None:
results['msg'] = "Container %s could not be found." % container_name
return results
if not check_mode:
log('Deleting container %s' % container_name)
bs.delete_container(container_name)
results['changed'] = True
results['msg'] = 'Container deleted successfully.'
return results
if mode == 'delete_blob':
if blob_name is None:
raise Exception("Parameter error: blob_name cannot be None.")
container = container_check(bs, container_name)
blob = get_blob_facts(bs, container_name, blob_name)
if not blob:
results['msg'] = 'Blob %s could not be found in container %s.' % (blob_name, container_name)
return results
if not check_mode:
log('Deleteing %s from container %s.' % (blob_name, container_name))
bs.delete_blob(container_name, blob_name)
results['changed'] = True
results['msg'] = 'Blob successfully deleted.'
return results
if mode == 'put':
if not blob_name:
raise Exception("Parameter error: blob_name cannot be None.")
if not file_path :
raise Exception("Parameter error: file_path cannot be None.")
if not path_check(file_path):
raise Exception("File %s does not exist." % file_path)
container = get_container_facts(bs, container_name)
blob = None
if container is not None:
blob = get_blob_facts(bs, container_name, blob_name)
if container is not None and blob is not None:
# both container and blob already exist
md5_remote = blob['content-md5']
md5_local = get_md5(file_path)
results['container'] = container
results['blob'] = blob
if md5_local == md5_remote:
sum_matches = True
results['msg'] = 'File checksums match. File not uploaded.'
if overwrite == 'always':
if not check_mode:
log('Uploading %s to container %s.' % (file_path, container_name))
put_block_blob(
bs,
container_name,
blob_name,
file_path,
x_ms_meta_name_values,
x_ms_blob_cache_control,
x_ms_blob_content_encoding,
x_ms_blob_content_language,
x_ms_blob_content_type
)
results['blob'] = get_blob_facts(bs, container_name, blob_name)
results['changed'] = True
results['msg'] = 'File successfully uploaded.'
else:
sum_matches = False
if overwrite in ('always', 'different'):
if not check_mode:
log('Uploading %s to container %s.' % (file_path, container_name))
put_block_blob(
bs,
container_name,
blob_name,
file_path,
x_ms_meta_name_values,
x_ms_blob_cache_control,
x_ms_blob_content_encoding,
x_ms_blob_content_language,
x_ms_blob_content_type
)
results['blob'] = get_blob_facts(bs, container_name, blob_name)
results['changed'] = True
results['msg'] = 'File successfully uploaded.'
else:
results['msg'] = "WARNING: Checksums do not match. Use overwrite parameter to force upload."
return results
if container is None:
# container does not exist. create container and upload.
if not check_mode:
log('Creating container %s.' % container_name)
bs.create_container(container_name, x_ms_meta_name_values, x_ms_blob_public_access)
log('Uploading %s to container %s.' % (file_path, container_name))
put_block_blob(
bs,
container_name,
blob_name,
file_path,
x_ms_meta_name_values,
x_ms_blob_cache_control,
x_ms_blob_content_encoding,
x_ms_blob_content_language,
x_ms_blob_content_type
)
results['conainer'] = get_container_facts(bs, container_name)
results['blob'] = get_blob_facts(bs, container_name, blob_name)
results['changed'] = True
results['msg'] = 'Successfully created container and uploaded file.'
return results
if container is not None:
# container exists. just upload.
if not check_mode:
log('Uploading %s to container %s.' % (file_path, container_name))
put_block_blob(
bs,
container_name,
blob_name,
file_path,
x_ms_meta_name_values,
x_ms_blob_cache_control,
x_ms_blob_content_encoding,
x_ms_blob_content_language,
x_ms_blob_content_type
)
results['blob'] = get_blob_facts(bs, container_name, blob_name)
results['changed'] = True
results['msg'] = 'Successfully updloaded file.'
return results
if mode == 'list':
container = container_check(bs, container_name)
response = bs.list_blobs(
container_name,
prefix,
marker,
max_results
)
results['blobs'] = []
for blob in response.blobs:
b = dict(
name = blob.name,
snapshot = blob.snapshot,
last_modified = blob.properties.last_modified,
content_length = blob.properties.content_length,
blob_type = blob.properties.blob_type,
)
results['blobs'].append(b)
return results
if mode == 'get':
if file_path is None:
raise Exception("Parameter error: file_path cannot be None.")
container = container_check(bs, container_name)
blob = blob_check(bs, container_name, blob_name)
path_exists = path_check(file_path)
if not path_exists or overwrite == 'always':
if not check_mode:
bs.get_blob_to_path(container_name, blob_name, file_path)
results['changed'] = True
results['msg'] = "Blob %s successfully downloaded to %s." % (blob_name, file_path)
return results
if path_exists:
md5_remote = blob['content-md5']
md5_local = get_md5(file_path)
if md5_local == md5_remote:
sum_matches = True
if overwrite == 'always':
if not check_mode:
bs.get_blob_to_path(container_name, blob_name, file_path)
results['changed'] = True
results['msg'] = "Blob %s successfully downloaded to %s." % (blob_name, file_path)
else:
results['msg'] = "Local and remote object are identical, ignoring. Use overwrite parameter to force."
else:
sum_matches = False
if overwrite in ('always', 'different'):
if not check_mode:
bs.get_blob_to_path(container_name, blob_name, file_path)
results['changed'] = True
results['msg'] = "Blob %s successfully downloaded to %s." % (blob_name, file_path)
else:
results['msg'] ="WARNING: Checksums do not match. Use overwrite parameter to force download."
if sum_matches is True and overwrite == 'never':
results['msg'] = "Local and remote object are identical, ignoring. Use overwrite parameter to force."
return results
if mode == 'get_url':
if not blob_name:
raise Exception("Parameter error: blob_name cannot be None.")
container = container_check(bs, container_name)
blob = blob_check(bs, container_name, blob_name)
url = bs.make_blob_url(
container_name=container_name,
blob_name=blob_name,
sas_token=access_token)
results['url'] = url
results['msg'] = "Url: %s" % url
return results
if mode == 'get_token':
if hours == 0 and days == 0:
raise Exception("Parameter error: expecting hours > 0 or days > 0")
container = container_check(bs, container_name)
blob = blob_check(bs, container_name, blob_name)
results['blob_name'] = blob_name
sap = get_shared_access_policy(permissions, hours=hours, days=days)
token = bs.generate_shared_access_signature(container_name, blob_name, sap)
results['access_token'] = token
return results
class AzureIOStore(IOStore):
"""
A class that lets you get input from and send output to Azure Storage.
"""
def __init__(self, account_name, container_name, name_prefix=""):
"""
Make a new AzureIOStore that reads from and writes to the given
container in the given account, adding the given prefix to keys. All
paths will be interpreted as keys or key prefixes.
If the name prefix does not end with a trailing slash, and is not empty,
one will be added automatically.
Account keys are retrieved from the AZURE_ACCOUNT_KEY environment
variable or from the ~/.toilAzureCredentials file, as in Toil itself.
"""
# Make sure azure libraries actually loaded
assert (have_azure)
self.account_name = account_name
self.container_name = container_name
self.name_prefix = name_prefix
if self.name_prefix != "" and not self.name_prefix.endswith("/"):
# Make sure it has the trailing slash required.
self.name_prefix += "/"
# Sneak into Toil and use the same keys it uses
self.account_key = toil.jobStores.azureJobStore._fetchAzureAccountKey(
self.account_name)
# This will hold out Azure blob store connection
self.connection = None
def __getstate__(self):
"""
Return the state to use for pickling. We don't want to try and pickle
an open Azure connection.
"""
return (self.account_name, self.account_key, self.container_name,
self.name_prefix)
def __setstate__(self, state):
"""
Set up after unpickling.
"""
self.account_name = state[0]
self.account_key = state[1]
self.container_name = state[2]
self.name_prefix = state[3]
self.connection = None
def __connect(self):
"""
Make sure we have an Azure connection, and set one up if we don't.
"""
if self.connection is None:
RealtimeLogger.debug("Connecting to account {}, using "
"container {} and prefix {}".format(
self.account_name, self.container_name,
self.name_prefix))
# Connect to the blob service where we keep everything
self.connection = BlobService(account_name=self.account_name,
account_key=self.account_key)
@backoff
def read_input_file(self, input_path, local_path):
"""
Get input from Azure.
"""
self.__connect()
RealtimeLogger.debug("Loading {} from AzureIOStore".format(input_path))
# Download the blob. This is known to be synchronous, although it can
# call a callback during the process.
self.connection.get_blob_to_path(self.container_name,
self.name_prefix + input_path,
local_path)
def list_input_directory(self,
input_path,
recursive=False,
with_times=False):
"""
Loop over fake /-delimited directories on Azure. The prefix may or may
not not have a trailing slash; if not, one will be added automatically.
Returns the names of files and fake directories in the given input fake
directory, non-recursively.
If with_times is specified, will yield (name, time) pairs including
modification times as datetime objects. Times on directories are None.
"""
self.__connect()
RealtimeLogger.info(
"Enumerating {} from AzureIOStore".format(input_path))
# Work out what the directory name to list is
fake_directory = self.name_prefix + input_path
if fake_directory != "" and not fake_directory.endswith("/"):
# We have a nonempty prefix, and we need to end it with a slash
fake_directory += "/"
# This will hold the marker that we need to send back to get the next
# page, if there is one. See <http://stackoverflow.com/a/24303682>
marker = None
# This holds the subdirectories we found; we yield each exactly once if
# we aren't recursing.
subdirectories = set()
while True:
# Get the results from Azure. We don't use delimiter since Azure
# doesn't seem to provide the placeholder entries it's supposed to.
result = self.connection.list_blobs(self.container_name,
prefix=fake_directory,
marker=marker)
RealtimeLogger.info("Found {} files".format(len(result)))
for blob in result:
# Yield each result's blob name, but directory names only once
# Drop the common prefix
relative_path = blob.name[len(fake_directory):]
if (not recursive) and "/" in relative_path:
# We found a file in a subdirectory, and we aren't supposed
# to be recursing.
subdirectory, _ = relative_path.split("/", 1)
if subdirectory not in subdirectories:
# It's a new subdirectory. Yield and remember it
subdirectories.add(subdirectory)
if with_times:
yield subdirectory, None
else:
yield subdirectory
else:
# We found an actual file
if with_times:
mtime = blob.properties.last_modified
if isinstance(mtime, datetime.datetime):
# Make sure we're getting proper localized datetimes
# from the new Azure Storage API.
assert (mtime.tzinfo is not None and
mtime.tzinfo.utcoffset(mtime) is not None)
else:
# Convert mtime from a string as in the old API.
mtime = dateutil.parser.parse(mtime).replace(
tzinfo=dateutil.tz.tzutc())
yield relative_path, mtime
else:
yield relative_path
# Save the marker
marker = result.next_marker
if not marker:
break
@backoff
def write_output_file(self, local_path, output_path):
"""
Write output to Azure. Will create the container if necessary.
"""
self.__connect()
RealtimeLogger.debug("Saving {} to AzureIOStore".format(output_path))
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Upload the blob (synchronously)
# TODO: catch no container error here, make the container, and retry
self.connection.put_block_blob_from_path(
self.container_name, self.name_prefix + output_path, local_path)
@backoff
def exists(self, path):
"""
Returns true if the given input or output file exists in Azure already.
"""
self.__connect()
marker = None
while True:
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
return True
# Save the marker
marker = result.next_marker
if not marker:
break
return False
@backoff
def get_mtime(self, path):
"""
Returns the modification time of the given blob if it exists, or None
otherwise.
"""
self.__connect()
marker = None
while True:
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
mtime = blob.properties.last_modified
if isinstance(mtime, datetime.datetime):
# Make sure we're getting proper localized datetimes
# from the new Azure Storage API.
assert (mtime.tzinfo is not None
and mtime.tzinfo.utcoffset(mtime) is not None)
else:
# Convert mtime from a string as in the old API.
mtime = dateutil.parser.parse(mtime).replace(
tzinfo=dateutil.tz.tzutc())
return mtime
# Save the marker
marker = result.next_marker
if not marker:
break
return None
@backoff
def get_size(self, path):
"""
Returns the size in bytes of the given blob if it exists, or None
otherwise.
"""
self.__connect()
marker = None
while True:
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
size = blob.properties.content_length
return size
# Save the marker
marker = result.next_marker
if not marker:
break
return None
class AzureIOStore(IOStore):
"""
A class that lets you get input from and send output to Azure Storage.
"""
def __init__(self, account_name, container_name, name_prefix=""):
"""
Make a new AzureIOStore that reads from and writes to the given
container in the given account, adding the given prefix to keys. All
paths will be interpreted as keys or key prefixes.
If the name prefix does not end with a trailing slash, and is not empty,
one will be added automatically.
Account keys are retrieved from the AZURE_ACCOUNT_KEY environment
variable or from the ~/.toilAzureCredentials file, as in Toil itself.
"""
# Make sure azure libraries actually loaded
assert(have_azure)
self.account_name = account_name
self.container_name = container_name
self.name_prefix = name_prefix
if self.name_prefix != "" and not self.name_prefix.endswith("/"):
# Make sure it has the trailing slash required.
self.name_prefix += "/"
# Sneak into Toil and use the same keys it uses
self.account_key = toil.jobStores.azureJobStore._fetchAzureAccountKey(
self.account_name)
# This will hold out Azure blob store connection
self.connection = None
def __getstate__(self):
"""
Return the state to use for pickling. We don't want to try and pickle
an open Azure connection.
"""
return (self.account_name, self.account_key, self.container_name,
self.name_prefix)
def __setstate__(self, state):
"""
Set up after unpickling.
"""
self.account_name = state[0]
self.account_key = state[1]
self.container_name = state[2]
self.name_prefix = state[3]
self.connection = None
def __connect(self):
"""
Make sure we have an Azure connection, and set one up if we don't.
"""
if self.connection is None:
RealTimeLogger.get().debug("Connecting to account {}, using "
"container {} and prefix {}".format(self.account_name,
self.container_name, self.name_prefix))
# Connect to the blob service where we keep everything
self.connection = BlobService(
account_name=self.account_name, account_key=self.account_key)
@backoff
def read_input_file(self, input_path, local_path):
"""
Get input from Azure.
"""
self.__connect()
RealTimeLogger.get().debug("Loading {} from AzureIOStore".format(
input_path))
# Download the blob. This is known to be synchronous, although it can
# call a callback during the process.
self.connection.get_blob_to_path(self.container_name,
self.name_prefix + input_path, local_path)
def list_input_directory(self, input_path, recursive=False,
with_times=False):
"""
Loop over fake /-delimited directories on Azure. The prefix may or may
not not have a trailing slash; if not, one will be added automatically.
Returns the names of files and fake directories in the given input fake
directory, non-recursively.
If with_times is specified, will yield (name, time) pairs including
modification times as datetime objects. Times on directories are None.
"""
self.__connect()
RealTimeLogger.get().info("Enumerating {} from AzureIOStore".format(
input_path))
# Work out what the directory name to list is
fake_directory = self.name_prefix + input_path
if fake_directory != "" and not fake_directory.endswith("/"):
# We have a nonempty prefix, and we need to end it with a slash
fake_directory += "/"
# This will hold the marker that we need to send back to get the next
# page, if there is one. See <http://stackoverflow.com/a/24303682>
marker = None
# This holds the subdirectories we found; we yield each exactly once if
# we aren't recursing.
subdirectories = set()
while True:
# Get the results from Azure. We don't use delimiter since Azure
# doesn't seem to provide the placeholder entries it's supposed to.
result = self.connection.list_blobs(self.container_name,
marker=marker)
RealTimeLogger.get().info("Found {} files".format(len(result)))
for blob in result:
# Yield each result's blob name, but directory names only once
# Drop the common prefix
relative_path = blob.name
if (not recursive) and "/" in relative_path:
# We found a file in a subdirectory, and we aren't supposed
# to be recursing.
subdirectory, _ = relative_path.split("/", 1)
if subdirectory not in subdirectories:
# It's a new subdirectory. Yield and remember it
subdirectories.add(subdirectory)
if with_times:
yield subdirectory, None
else:
yield subdirectory
else:
# We found an actual file
if with_times:
mtime = dateutil.parser.parse(
blob.properties.last_modified).replace(
tzinfo=dateutil.tz.tzutc())
yield relative_path, mtime
else:
yield relative_path
# Save the marker
marker = result.next_marker
if not marker:
break
@backoff
def write_output_file(self, local_path, output_path):
"""
Write output to Azure. Will create the container if necessary.
"""
self.__connect()
RealTimeLogger.get().debug("Saving {} to AzureIOStore".format(
output_path))
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Upload the blob (synchronously)
# TODO: catch no container error here, make the container, and retry
self.connection.put_block_blob_from_path(self.container_name,
self.name_prefix + output_path, local_path)
@backoff
def exists(self, path):
"""
Returns true if the given input or output file exists in Azure already.
"""
self.__connect()
marker = None
while True:
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path, marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
return True
# Save the marker
marker = result.next_marker
if not marker:
break
return False
@backoff
def get_mtime(self, path):
"""
Returns the modification time of the given blob if it exists, or None
otherwise.
"""
self.__connect()
marker = None
while True:
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path, marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
return dateutil.parser.parse(
blob.properties.last_modified).replace(
tzinfo=dateutil.tz.tzutc())
# Save the marker
marker = result.next_marker
if not marker:
break
return None
blob_service = BlobService(account_name="<account_name>", account_key="<account_key>")
blob_service.create_container("datacontainer")
blob_service.create_container("datacontainer", x_ms_blob_public_access="container")
blob_service.set_container_acl("datacontainer", x_ms_blob_public_access="container")
blob_service.put_block_blob_from_path(
"datacontainer", "datablob", "StorageClientPy.py", x_ms_blob_content_type="text/x-script.phyton"
)
blobs = []
marker = None
while True:
batch = blob_service.list_blobs("datacontainer", marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
print(blob.name)
blob_service.get_blob_to_path("datacontainer", "datablob", "out-StorageClientPy.py")
blob_service.delete_blob("datacontainer", "datablob")
class Azure(object):
'''
A class used to connect to the Azure storage and
upload/download files using blob storage
'''
def __init__(self, params={}):
'''
Constructor for the Azure object
'''
if "user" in params:
self.user = params["user"]
else:
self.user = None
if "key" in params:
self.key = params["key"]
else:
self.key = None
def connect(self, host, port, user, password, secure):
'''
Connect to the Azure service with given user and key
@param user - username to use to connect to
@param key - key to use to connect
'''
kwargs = {}
err = None
if not host is None:
kwargs["host_base"] = "." + host
if not user is None:
kwargs["account_name"] = user
elif not self.user is None:
kwargs["account_name"] = self.user
if not password is None:
kwargs["account_key"] = password
elif not self.key is None:
kwargs["account_key"] = self.key
kwargs["protocol"] = "https" if secure else "http"
try:
self.service = BlobService(**kwargs)
except Exception as e:
err = e.message
self.service = None
if self.service is None:
raise OsakaException("Failed to connect to Azure:" +
("" if err is None else err))
@classmethod
def getSchemes(clazz):
'''
Returns a list of schemes this handler handles
Note: handling the scheme of another handler produces unknown results
@returns list of handled schemes
'''
return ["azure", "azures"]
def close(self):
'''
Close this service
'''
pass
def put(self, path, url):
'''
Put a file up to the cloud
@param path - path to upload
@param url - path in cloud to upload too
'''
if os.path.isdir(path):
return walk(self.put, path, url)
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
self.service.create_container(cont)
self.service.put_block_blob_from_path(cont, blob, path)
return True
def get(self, url, dest):
'''
Get file(s) from the cloud
@param url - url on cloud to pull down (on cloud)
@param dest - dest to download too
'''
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
for b in self.service.list_blobs(cont, prefix=blob):
destination = os.path.join(dest, os.path.relpath(
b.name, blob)) if blob != b.name else dest
if not os.path.exists(os.path.dirname(destination)):
os.mkdir(os.path.dirname(destination))
self.service.get_blob_to_path(cont, b.name, destination)
return True
def rm(self, url):
'''
Remove this url and all children urls
@param url - url to remove
'''
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
for b in self.service.list_blobs(cont, prefix=blob):
self.service.delete_blob(cont, b.name)
return True
blob_service.create_container(CONTAINER, x_ms_blob_public_access='container')
def putblob(fileid, filename):
global ACCOUNT
blob_service.put_block_blob_from_path(
CONTAINER,
fileid,
filename,
x_ms_blob_content_type=guess_type(filename)
)
return 'https://%s.blob.core.windows.net/%s/%s' %(ACCOUNT, CONTAINER, fileid)
putblob('quotes.pkl', 'quotes.pkl')
blobs = []
marker = None
while True:
batch = blob_service.list_blobs(CONTAINER, marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
print(blob.name)
#blob_service.delete_blob(CONTAINER, 'quotes.pkl')
# -*- coding: utf-8 -*-
"""
Created on Fri Oct 09 16:36:56 2015
@author: justin.malinchak
"""
# List blobs in container
from azure.storage.blob import BlobService
blob_service = BlobService(account_name='portalvhdss5m831rhl98hj', account_key='Z1MliCYE7p9Ks9kYQoGeM4V99hODtiJL82BVi/zIm06jLYh7n0tV8YaZHzITKixMwUUmjJ1Vp05XrgHG+gXFlg==')
blobs = []
marker = None
while True:
batch = blob_service.list_blobs('mycontainer', marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
bname = blob.name
print('')
print(bname)
print('')
bpathname = 'C:\\Batches\\$Work\\' + bname
blob_service.get_blob_to_path('mycontainer', bname, bpathname)
print('')
print('blob downloaded ' + bpathname)
print('')
class Command(BaseCommand):
help = "Synchronizes static media to cloud files."
option_list = BaseCommand.option_list + (
optparse.make_option('-w', '--wipe',
action='store_true', dest='wipe', default=False,
help="Wipes out entire contents of container first."),
optparse.make_option('-t', '--test-run',
action='store_true', dest='test_run', default=False,
help="Performs a test run of the sync."),
optparse.make_option('-c', '--container',
dest='container', help="Override STATIC_CONTAINER."),
)
# settings from azurite.settings
ACCOUNT_NAME = AZURITE['ACCOUNT_NAME']
ACCOUNT_KEY = AZURITE['ACCOUNT_KEY']
STATIC_CONTAINER = AZURITE['STATIC_CONTAINER']
# paths
DIRECTORY = os.path.abspath(settings.STATIC_ROOT)
STATIC_URL = settings.STATIC_URL
if not DIRECTORY.endswith('/'):
DIRECTORY = DIRECTORY + '/'
if STATIC_URL.startswith('/'):
STATIC_URL = STATIC_URL[1:]
local_object_names = []
create_count = 0
upload_count = 0
update_count = 0
skip_count = 0
delete_count = 0
service = None
def handle(self, *args, **options):
self.wipe = options.get('wipe')
self.test_run = options.get('test_run')
self.verbosity = int(options.get('verbosity'))
if hasattr(options, 'container'):
self.STATIC_CONTAINER = options.get('container')
self.sync_files()
def sync_files(self):
self.service = BlobService(account_name=self.ACCOUNT_NAME,
account_key=self.ACCOUNT_KEY)
try:
self.service.get_container_properties(self.STATIC_CONTAINER)
except AzureMissingResourceHttpError:
self.service.create_container(self.STATIC_CONTAINER,
x_ms_blob_public_access='blob')
self.service.set_container_acl(self.STATIC_CONTAINER, x_ms_blob_public_access='blob')
# if -w option is provided, wipe out the contents of the container
if self.wipe:
blob_count = len(self.service.list_blobs(self.STATIC_CONTAINER))
if self.test_run:
print "Wipe would delete %d objects." % blob_count
else:
print "Deleting %d objects..." % blob_count
for blob in self.service.list_blobs(self.STATIC_CONTAINER):
self.service.delete_blob(self.STATIC_CONTAINER, blob.name)
# walk through the directory, creating or updating files on the cloud
os.path.walk(self.DIRECTORY, self.upload_files, "foo")
# remove any files on remote that don't exist locally
self.delete_files()
# print out the final tally to the cmd line
self.update_count = self.upload_count - self.create_count
print
if self.test_run:
print "Test run complete with the following results:"
print "Skipped %d. Created %d. Updated %d. Deleted %d." % (
self.skip_count, self.create_count, self.update_count, self.delete_count)
def upload_files(self, arg, dirname, names):
# upload or skip items
for item in names:
file_path = os.path.join(dirname, item)
if os.path.isdir(file_path):
continue # Don't try to upload directories
object_name = self.STATIC_URL + file_path.split(self.DIRECTORY)[1]
self.local_object_names.append(object_name)
try:
properties = self.service.get_blob_properties(self.STATIC_CONTAINER,
object_name)
except AzureMissingResourceHttpError:
properties = {}
self.create_count += 1
cloud_datetime = None
if 'last-modified' in properties:
cloud_datetime = (properties['last-modified'] and
datetime.datetime.strptime(
properties['last-modified'],
"%a, %d %b %Y %H:%M:%S %Z"
) or None)
local_datetime = datetime.datetime.utcfromtimestamp(
os.stat(file_path).st_mtime)
if cloud_datetime and local_datetime < cloud_datetime:
self.skip_count += 1
if self.verbosity > 1:
print "Skipped %s: not modified." % object_name
continue
if not self.test_run:
file_contents = open(file_path, 'r').read()
content_type, encoding = mimetypes.guess_type(file_path)
self.service.put_blob(self.STATIC_CONTAINER, object_name, file_contents,
x_ms_blob_type='BlockBlob', x_ms_blob_content_type=content_type,
content_encoding=encoding)
# sync_headers(cloud_obj)
self.upload_count += 1
if self.verbosity > 1:
print "Uploaded", object_name
def delete_files(self):
# remove any objects in the container that don't exist locally
for blob in self.service.list_blobs(self.STATIC_CONTAINER):
if blob.name not in self.local_object_names:
self.delete_count += 1
if self.verbosity > 1:
print "Deleted %s" % blob.name
if not self.test_run:
self.service.delete_blob(self.STATIC_CONTAINER, blob.name)
class AzureFS(LoggingMixIn, Operations):
"""Azure Blob Storage filesystem"""
blobs = None
containers = dict() # <cname, dict(stat:dict,
#files:None|dict<fname, stat>)
fds = dict() # <fd, (path, bytes, dirty)>
fd = 0
def __init__(self, account, key):
self.blobs = BlobService(account, key)
self.rebuild_container_list()
def convert_to_epoch(self, date):
"""Converts Tue, 31 Jul 2012 07:17:34 GMT format to epoch"""
return int(time.mktime(time.strptime(date, TIME_FORMAT)))
def rebuild_container_list(self):
cmap = dict()
cnames = set()
for c in self.blobs.list_containers():
date = c.properties.last_modified
cstat = dict(st_mode=(S_IFDIR | 0755), st_uid=getuid(), st_size=0,
st_mtime=self.convert_to_epoch(date))
cname = c.name
cmap['/' + cname] = dict(stat=cstat, files=None)
cnames.add(cname)
cmap['/'] = dict(files={},
stat=dict(st_mode=(S_IFDIR | 0755),
st_uid=getuid(), st_size=0,
st_mtime=int(time.time())))
self.containers = cmap # destroys fs tree cache resistant to misses
def _parse_path(self, path): # returns </dir, file(=None)>
if path.count('/') > 1: # file
return str(path[:path.rfind('/')]), str(path[path.rfind('/') + 1:])
else: # dir
pos = path.rfind('/', 1)
if pos == -1:
return path, None
else:
return str(path[:pos]), None
def parse_container(self, path):
base_container = path[1:] # /abc/def/g --> abc
if base_container.find('/') > -1:
base_container = base_container[:base_container.find('/')]
return str(base_container)
def _get_dir(self, path, contents_required=False):
if not self.containers:
self.rebuild_container_list()
if path in self.containers and not (contents_required and \
self.containers[path]['files'] is None):
return self.containers[path]
cname = self.parse_container(path)
if '/' + cname not in self.containers:
raise FuseOSError(ENOENT)
else:
if self.containers['/' + cname]['files'] is None:
# fetch contents of container
log.info("------> CONTENTS NOT FOUND: %s" % cname)
blobs = self.blobs.list_blobs(cname)
dirstat = dict(st_mode=(S_IFDIR | 0755), st_size=0,
st_uid=getuid(), st_mtime=time.time())
if self.containers['/' + cname]['files'] is None:
self.containers['/' + cname]['files'] = dict()
for f in blobs:
blob_name = f.name
blob_date = f.properties.last_modified
blob_size = long(f.properties.content_length)
node = dict(st_mode=(S_IFREG | 0644), st_size=blob_size,
st_mtime=self.convert_to_epoch(blob_date),
st_uid=getuid())
if blob_name.find('/') == -1: # file just under container
self.containers['/' + cname]['files'][blob_name] = node
return self.containers['/' + cname]
return None
def _get_file(self, path):
d, f = self._parse_path(path)
dir = self._get_dir(d, True)
if dir is not None and f in dir['files']:
return dir['files'][f]
def getattr(self, path, fh=None):
d, f = self._parse_path(path)
if f is None:
dir = self._get_dir(d)
return dir['stat']
else:
file = self._get_file(path)
if file:
return file
raise FuseOSError(ENOENT)
# FUSE
def mkdir(self, path, mode):
if path.count('/') <= 1: # create on root
name = path[1:]
if not 3 <= len(name) <= 63:
log.error("Container names can be 3 through 63 chars long.")
raise FuseOSError(ENAMETOOLONG)
if name is not name.lower():
log.error("Container names cannot contain uppercase \
characters.")
raise FuseOSError(EACCES)
if name.count('--') > 0:
log.error('Container names cannot contain consecutive \
dashes (-).')
raise FuseOSError(EAGAIN)
#TODO handle all "-"s must be preceded by letter or numbers
#TODO starts with only letter or number, can contain letter, nr,'-'
resp = self.blobs.create_container(name)
if resp:
self.rebuild_container_list()
log.info("CONTAINER %s CREATED" % name)
else:
raise FuseOSError(EACCES)
log.error("Invalid container name or container already \
exists.")
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def rmdir(self, path):
if path.count('/') == 1:
c_name = path[1:]
resp = self.blobs.delete_container(c_name)
if resp:
if path in self.containers:
del self.containers[path]
else:
raise FuseOSError(EACCES)
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def create(self, path, mode):
node = dict(st_mode=(S_IFREG | mode), st_size=0, st_nlink=1,
st_uid=getuid(), st_mtime=time.time())
d, f = self._parse_path(path)
if not f:
log.error("Cannot create files on root level: /")
raise FuseOSError(ENOSYS)
dir = self._get_dir(d, True)
if not dir:
raise FuseOSError(EIO)
dir['files'][f] = node
return self.open(path, data='') # reusing handler provider
def open(self, path, flags=0, data=None):
if data == None: # download contents
c_name = self.parse_container(path)
f_name = path[path.find('/', 1) + 1:]
try:
data = self.blobs.get_blob(c_name, f_name)
except AzureMissingResourceHttpError:
dir = self._get_dir('/' + c_name, True)
if f_name in dir['files']:
del dir['files'][f_name]
raise FuseOSError(ENOENT)
except AzureException as e:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
self.fd += 1
self.fds[self.fd] = (path, data, False)
return self.fd
def flush(self, path, fh=None):
if not fh:
raise FuseOSError(EIO)
else:
if fh not in self.fds:
raise FuseOSError(EIO)
path = self.fds[fh][0]
data = self.fds[fh][1]
dirty = self.fds[fh][2]
if not dirty:
return 0 # avoid redundant write
d, f = self._parse_path(path)
c_name = self.parse_container(path)
if data is None:
data = ''
try:
if len(data) < 64 * 1024 * 1024: # 64 mb
self.blobs.put_blob(c_name, f, data, 'BlockBlob')
else:
# divide file by blocks and upload
block_size = 8 * 1024 * 1024
num_blocks = int(math.ceil(len(data) * 1.0 / block_size))
rd = str(random.randint(1, 1e8))
block_ids = list()
for i in range(num_blocks):
part = data[i * block_size:min((i + 1) * block_size,
len(data))]
block_id = base64.encodestring('%s_%s' % (rd,
(8 - len(str(i))) * '0' + str(i)))
self.blobs.put_block(c_name, f, part, block_id)
block_ids.append(block_id)
self.blobs.put_block_list(c_name, f, block_ids)
except AzureException:
raise FuseOSError(EAGAIN)
dir = self._get_dir(d, True)
if not dir or f not in dir['files']:
raise FuseOSError(EIO)
# update local data
dir['files'][f]['st_size'] = len(data)
dir['files'][f]['st_mtime'] = time.time()
self.fds[fh] = (path, data, False) # mark as not dirty
return 0
def release(self, path, fh=None):
if fh is not None and fh in self.fds:
del self.fds[fh]
def truncate(self, path, length, fh=None):
return 0 # assume done, no need
def write(self, path, data, offset, fh=None):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
else:
d = self.fds[fh][1]
if d is None:
d = ""
self.fds[fh] = (self.fds[fh][0], d[:offset] + data, True)
return len(data)
def unlink(self, path):
c_name = self.parse_container(path)
d, f = self._parse_path(path)
try:
self.blobs.delete_blob(c_name, f)
_dir = self._get_dir(path, True)
if _dir and f in _dir['files']:
del _dir['files'][f]
return 0
except AzureMissingResourceHttpError:
raise FuseOSError(ENOENT)
except Exception as e:
raise FuseOSError(EAGAIN)
def readdir(self, path, fh):
if path == '/':
return ['.', '..'] + [x[1:] for x in self.containers.keys() \
if x is not '/']
dir = self._get_dir(path, True)
if not dir:
raise FuseOSError(ENOENT)
return ['.', '..'] + dir['files'].keys()
def read(self, path, size, offset, fh):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
f_name = path[path.find('/', 1) + 1:]
c_name = path[1:path.find('/', 1)]
try:
data = self.blobs.get_blob(c_name, f_name)
self.fds[fh] = (self.fds[fh][0], data, False)
return data[offset:offset + size]
except URLError, e:
if e.code == 404:
raise FuseOSError(ENOENT)
elif e.code == 403:
raise FUSEOSError(EPERM)
else:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
data = self.fds[fh][1]
if data is None:
data = ""
return data[offset:offset + size]
source_account = BlobService(
account_name='cngssd',
account_key=
'DVvOtpOVW71er9ztR3mooJk4Zc3ZNovW9YV3qu4Y6bkN0eCfHutpcNVXW6gtpfolRk4CcAlmftz/+SDwm2BQag=='
)
dest_account = BlobService(
account_name='testcng',
account_key=
'piWr6zleZ1sL8aopv5Y4NRYyrVWaW2/QrXcPpsxRec4IxtEoR1IyRmZCkbdyq50Bfu0qidF8SicQahdM+OExvg=='
)
# list blobs
source_blobs = []
source_marker = None
while True:
batch = source_account.list_blobs('vhds', marker=source_marker)
source_blobs.extend(batch)
if not batch.next_marker:
break
source_marker = batch.next_marker
for blob in source_blobs:
print(blob.name)
dest_blobs = []
dest_marker = None
while True:
batch = dest_account.list_blobs('vhds', marker=dest_marker)
dest_blobs.extend(batch)
if not batch.next_marker:
break
dest_marker = batch.next_marker
class BlobSource(DataSource):
def __init__(self):
self.storage_account = getenv('STORAGE_ACCOUNT')
self.blob_service = BlobService(self.storage_account,
getenv('STORAGE_KEY'))
def load(self, sparkContext, container, path):
path = ('/' if path[0] != '/' else '') + path
uri = 'wasb://%s@%s.blob.core.windows.net%s' % (
container, self.storage_account, path)
print 'Loading from %s' % uri
return sparkContext.textFile(uri)
def download(self, container, path):
print 'Downloading blob from %s/%s' % (container, path)
self.blob_service.get_blob_to_path(container, path, path)
print 'Downloaded blob to ' + path
def saveAsJson(self, payload, container, path):
path = path.lstrip('/')
print path
print 'Saving to %s/%s' % (container, path)
json_string = json.dumps(payload, ensure_ascii=False).encode('utf-8')
try:
self.blob_service.put_blob(
container,
path,
json_string,
'BlockBlob',
x_ms_blob_cache_control='max-age=3600',
x_ms_blob_content_type='application/json')
except Exception as e:
print 'Failed to save %s/%s: %s' % (container, path, str(e))
raise
def saveAsText(self, rdd, container, path):
path = path.lstrip('/')
path = '/' + path
print 'Saving rdd to %s%s' % (container, path)
uri = 'wasb://%s@%s.blob.core.windows.net%s' % (
container, self.storage_account, path)
try:
rdd.saveAsTextFile(uri)
except Exception as e:
print 'Failed to save %s%s: %s' % (container, path, str(e))
raise
def deleteAllBut(self, container, exceptFolderName):
print 'deleteAllBut called'
try:
bloblistingresult = self.blob_service.list_blobs(container)
for i in bloblistingresult:
print i.name
if not exceptFolderName in i.name:
try:
print 'deleting'
self.blob_service.delete_blob(container, i.name)
print 'deleted'
except Exception as e:
print 'Failed to delete %s/%s: %s' % (container,
i.name, str(e))
raise
except Exception as e:
print 'Failed to list things in %s: %s' % (container, str(e))
raise
)
blob_service.create_container(CONTAINER, x_ms_blob_public_access='container')
def putblob(fileid, filename):
global ACCOUNT
blob_service.put_block_blob_from_path(
CONTAINER,
fileid,
filename,
x_ms_blob_content_type=guess_type(filename))
return 'https://%s.blob.core.windows.net/%s/%s' % (ACCOUNT, CONTAINER,
fileid)
putblob('quotes.pkl', 'quotes.pkl')
blobs = []
marker = None
while True:
batch = blob_service.list_blobs(CONTAINER, marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
print(blob.name)
#blob_service.delete_blob(CONTAINER, 'quotes.pkl')
class AzureFS(LoggingMixIn, Operations):
"""Azure Blob Storage filesystem"""
blobs = None
containers = dict() # <cname, dict(stat:dict,
#files:None|dict<fname, stat>)
fds = dict() # <fd, (path, bytes, dirty)>
fd = 0
def __init__(self, account, key):
self.blobs = BlobService(account, key)
self.rebuild_container_list()
def convert_to_epoch(self, date):
"""Converts Tue, 31 Jul 2012 07:17:34 GMT format to epoch"""
return int(time.mktime(time.strptime(date, TIME_FORMAT)))
def rebuild_container_list(self):
cmap = dict()
cnames = set()
for c in self.blobs.list_containers():
date = c.properties.last_modified
cstat = dict(st_mode=(S_IFDIR | 0755),
st_uid=getuid(),
st_size=0,
st_mtime=self.convert_to_epoch(date))
cname = c.name
cmap['/' + cname] = dict(stat=cstat, files=None)
cnames.add(cname)
cmap['/'] = dict(files={},
stat=dict(st_mode=(S_IFDIR | 0755),
st_uid=getuid(),
st_size=0,
st_mtime=int(time.time())))
self.containers = cmap # destroys fs tree cache resistant to misses
def _parse_path(self, path): # returns </dir, file(=None)>
if path.count('/') > 1: # file
return str(path[:path.rfind('/')]), str(path[path.rfind('/') + 1:])
else: # dir
pos = path.rfind('/', 1)
if pos == -1:
return path, None
else:
return str(path[:pos]), None
def parse_container(self, path):
base_container = path[1:] # /abc/def/g --> abc
if base_container.find('/') > -1:
base_container = base_container[:base_container.find('/')]
return str(base_container)
def _get_dir(self, path, contents_required=False):
if not self.containers:
self.rebuild_container_list()
if path in self.containers and not (contents_required and \
self.containers[path]['files'] is None):
return self.containers[path]
cname = self.parse_container(path)
if '/' + cname not in self.containers:
raise FuseOSError(ENOENT)
else:
if self.containers['/' + cname]['files'] is None:
# fetch contents of container
log.info("------> CONTENTS NOT FOUND: %s" % cname)
blobs = self.blobs.list_blobs(cname)
dirstat = dict(st_mode=(S_IFDIR | 0755),
st_size=0,
st_uid=getuid(),
st_mtime=time.time())
if self.containers['/' + cname]['files'] is None:
self.containers['/' + cname]['files'] = dict()
for f in blobs:
blob_name = f.name
blob_date = f.properties.last_modified
blob_size = long(f.properties.content_length)
node = dict(st_mode=(S_IFREG | 0644),
st_size=blob_size,
st_mtime=self.convert_to_epoch(blob_date),
st_uid=getuid())
if blob_name.find('/') == -1: # file just under container
self.containers['/' + cname]['files'][blob_name] = node
return self.containers['/' + cname]
return None
def _get_file(self, path):
d, f = self._parse_path(path)
dir = self._get_dir(d, True)
if dir is not None and f in dir['files']:
return dir['files'][f]
def getattr(self, path, fh=None):
d, f = self._parse_path(path)
if f is None:
dir = self._get_dir(d)
return dir['stat']
else:
file = self._get_file(path)
if file:
return file
raise FuseOSError(ENOENT)
# FUSE
def mkdir(self, path, mode):
if path.count('/') <= 1: # create on root
name = path[1:]
if not 3 <= len(name) <= 63:
log.error("Container names can be 3 through 63 chars long.")
raise FuseOSError(ENAMETOOLONG)
if name is not name.lower():
log.error("Container names cannot contain uppercase \
characters.")
raise FuseOSError(EACCES)
if name.count('--') > 0:
log.error('Container names cannot contain consecutive \
dashes (-).')
raise FuseOSError(EAGAIN)
#TODO handle all "-"s must be preceded by letter or numbers
#TODO starts with only letter or number, can contain letter, nr,'-'
resp = self.blobs.create_container(name)
if resp:
self.rebuild_container_list()
log.info("CONTAINER %s CREATED" % name)
else:
raise FuseOSError(EACCES)
log.error("Invalid container name or container already \
exists.")
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def rmdir(self, path):
if path.count('/') == 1:
c_name = path[1:]
resp = self.blobs.delete_container(c_name)
if resp:
if path in self.containers:
del self.containers[path]
else:
raise FuseOSError(EACCES)
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def create(self, path, mode):
node = dict(st_mode=(S_IFREG | mode),
st_size=0,
st_nlink=1,
st_uid=getuid(),
st_mtime=time.time())
d, f = self._parse_path(path)
if not f:
log.error("Cannot create files on root level: /")
raise FuseOSError(ENOSYS)
dir = self._get_dir(d, True)
if not dir:
raise FuseOSError(EIO)
dir['files'][f] = node
return self.open(path, data='') # reusing handler provider
def open(self, path, flags=0, data=None):
if data == None: # download contents
c_name = self.parse_container(path)
f_name = path[path.find('/', 1) + 1:]
try:
data = self.blobs.get_blob(c_name, f_name)
except AzureMissingResourceHttpError:
dir = self._get_dir('/' + c_name, True)
if f_name in dir['files']:
del dir['files'][f_name]
raise FuseOSError(ENOENT)
except AzureException as e:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
self.fd += 1
self.fds[self.fd] = (path, data, False)
return self.fd
def flush(self, path, fh=None):
if not fh:
raise FuseOSError(EIO)
else:
if fh not in self.fds:
raise FuseOSError(EIO)
path = self.fds[fh][0]
data = self.fds[fh][1]
dirty = self.fds[fh][2]
if not dirty:
return 0 # avoid redundant write
d, f = self._parse_path(path)
c_name = self.parse_container(path)
if data is None:
data = ''
try:
if len(data) < 64 * 1024 * 1024: # 64 mb
self.blobs.put_blob(c_name, f, data, 'BlockBlob')
else:
# divide file by blocks and upload
block_size = 8 * 1024 * 1024
num_blocks = int(math.ceil(len(data) * 1.0 / block_size))
rd = str(random.randint(1, 1e8))
block_ids = list()
for i in range(num_blocks):
part = data[i * block_size:min((i + 1) *
block_size, len(data))]
block_id = base64.encodestring(
'%s_%s' % (rd, (8 - len(str(i))) * '0' + str(i)))
self.blobs.put_block(c_name, f, part, block_id)
block_ids.append(block_id)
self.blobs.put_block_list(c_name, f, block_ids)
except AzureException:
raise FuseOSError(EAGAIN)
dir = self._get_dir(d, True)
if not dir or f not in dir['files']:
raise FuseOSError(EIO)
# update local data
dir['files'][f]['st_size'] = len(data)
dir['files'][f]['st_mtime'] = time.time()
self.fds[fh] = (path, data, False) # mark as not dirty
return 0
def release(self, path, fh=None):
if fh is not None and fh in self.fds:
del self.fds[fh]
def truncate(self, path, length, fh=None):
return 0 # assume done, no need
def write(self, path, data, offset, fh=None):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
else:
d = self.fds[fh][1]
if d is None:
d = ""
self.fds[fh] = (self.fds[fh][0], d[:offset] + data, True)
return len(data)
def unlink(self, path):
c_name = self.parse_container(path)
d, f = self._parse_path(path)
try:
self.blobs.delete_blob(c_name, f)
_dir = self._get_dir(path, True)
if _dir and f in _dir['files']:
del _dir['files'][f]
return 0
except AzureMissingResourceHttpError:
raise FuseOSError(ENOENT)
except Exception as e:
raise FuseOSError(EAGAIN)
def readdir(self, path, fh):
if path == '/':
return ['.', '..'] + [x[1:] for x in self.containers.keys() \
if x is not '/']
dir = self._get_dir(path, True)
if not dir:
raise FuseOSError(ENOENT)
return ['.', '..'] + dir['files'].keys()
def read(self, path, size, offset, fh):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
f_name = path[path.find('/', 1) + 1:]
c_name = path[1:path.find('/', 1)]
try:
data = self.blobs.get_blob(c_name, f_name)
self.fds[fh] = (self.fds[fh][0], data, False)
return data[offset:offset + size]
except URLError, e:
if e.code == 404:
raise FuseOSError(ENOENT)
elif e.code == 403:
raise FUSEOSError(EPERM)
else:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
data = self.fds[fh][1]
if data is None:
data = ""
return data[offset:offset + size]
@author: justin.malinchak
"""
# List blobs in container
from azure.storage.blob import BlobService
blob_service = BlobService(
account_name='portalvhdss5m831rhl98hj',
account_key=
'Z1MliCYE7p9Ks9kYQoGeM4V99hODtiJL82BVi/zIm06jLYh7n0tV8YaZHzITKixMwUUmjJ1Vp05XrgHG+gXFlg=='
)
blobs = []
marker = None
while True:
batch = blob_service.list_blobs('mycontainer', marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
bname = blob.name
print('')
print(bname)
print('')
bpathname = 'C:\\Batches\\$Work\\' + bname
blob_service.get_blob_to_path('mycontainer', bname, bpathname)
print('')
print('blob downloaded ' + bpathname)
print('')
class AzureConnector():
def __init__(self, config):
tree = ET.parse('SharedConfig.xml')
self.myMachineName = tree.find('.//Instance').get("id")
self.sms = ServiceManagementService(
subscription_id=config.get("azure", "subscription_id"),
cert_file=config.get("azure", "cert_file")
);
self.bus_service = ServiceBusService(
service_namespace=config.get("azure", "bus_namespace"),
shared_access_key_name=config.get("azure", "bus_shared_access_key_name"),
shared_access_key_value=config.get("azure", "bus_shared_access_key_value"))
self.command_queue = config.get("azure", "commandQueuePath")
for tries in range(1,10):
try:
self.bus_service.create_queue(self.command_queue)
break
except:
print "Esperando"
self.status_topic = config.get("azure", "statusTopicPath")
self.bus_service.create_queue(self.status_topic)
self.storage = BlobService(account_name=config.get("azure", "account_name"),
account_key=config.get("azure", "account_key"))
self.algo_storage_name = config.get("azure", "algorithm_storage_name")
self.storage.create_container(self.algo_storage_name, fail_on_exist=False)
self.proj_storage_name = config.get("azure", "project_storage_name")
self.storage.create_container(self.proj_storage_name, fail_on_exist=False)
def check_new_tasks(self):
for tries in range(1,2):
try:
message = self.bus_service.receive_queue_message(self.command_queue, peek_lock=False, timeout=60)
break
except:
message = None
if message is None or message.body is None:
return None
job_description = json.loads(message.body.replace('/AzureBlobStorage/', ''))
command = CommandMetadata(
command_id = job_description["command_id"],
algorithm_directory = job_description["algorithm_prfx"],
project_prfx = job_description["project_prfx"],
project_input_files = job_description["project_input_files"],
algorithm_executable_name = job_description["algorithm_executable_name"],
algorithm_parameters = job_description["algorithm_parameters"],
sent_timestamp = datetime.datetime.strptime(job_description["sent_timestamp"], "%d/%m/%Y %H:%M:%S"),
machine_size=job_description["machine_size"])
# Retornar dados sobre o comando consumido da fila
return command
# Não há nada na fila
return None
def list_algo_files(self, prfx):
list = self.storage.list_blobs(container_name=self.algo_storage_name, prefix=prfx)
result = []
for blob in list:
result.append(blob.name)
return result
def download_algo_zip(self, algorithm_bin_file, tmp_file):
print "download_algo_zip(algorithm_bin_file="+algorithm_bin_file+", tmp_file="+tmp_file+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.algo_storage_name, algorithm_bin_file, tmp_file,
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_project(self, project_name, blob_name, dir):
print "download_file_to_project(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_project(self, project_name, blob_name, dir):
print "download_file_to_project(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def upload_proj_file(self, project_name, blob_name, dir):
print "upload_proj_file(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
if blob_name[0] == '/':
blob_name = blob_name[1:]
for tries in range(1,5):
try:
self.storage.put_block_blob_from_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)))
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_algo(self, blob_name, dir):
print "download_file_to_algo(blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(container_name=self.algo_storage_name,
blob_name=os.path.join(blob_name),
file_path=os.path.join(dir,blob_name),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def send_status(self, main_status):
for tries in range(1,5):
try:
self.bus_service.send_topic_message(topic_name=self.status_topic,
message=Message(main_status.encode('utf-8')))
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def shutdown_myself(self):
# A máquina virtual irá cometer suicídio.
print("Removendo máquina virtual da nuvem...")
for tries in range(1,5):
try:
self.sms.delete_deployment(
service_name=self.myMachineName,
deployment_name=self.myMachineName, delete_vhd=True)
exit(0)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
