def storage_uri(uri_str, default_scheme='file', debug=False):
"""Instantiate a StorageUri from a URI string.
:type uri_str: string
:param uri_str: URI naming bucket + optional object.
:type default_scheme: string
:param default_scheme: default scheme for scheme-less URIs.
:rtype: :class:`boto.StorageUri` subclass
:return: StorageUri subclass for given URI.
uri_str must be one of the following formats:
gs://bucket/name
s3://bucket/name
gs://bucket
s3://bucket
filename
The last example uses the default scheme ('file', unless overridden)
"""
# Manually parse URI components instead of using urlparse.urlparse because
# what we're calling URIs don't really fit the standard syntax for URIs
# (the latter includes an optional host/net location part).
end_scheme_idx = uri_str.find('://')
if end_scheme_idx == -1:
scheme = default_scheme.lower()
path = uri_str
else:
scheme = uri_str[0:end_scheme_idx].lower()
path = uri_str[end_scheme_idx + 3:]
if scheme not in ['file', 's3', 'gs']:
raise InvalidUriError('Unrecognized scheme "%s"' % scheme)
if scheme == 'file':
# For file URIs we have no bucket name, and use the complete path
# (minus 'file://') as the object name.
return FileStorageUri(path, debug)
else:
path_parts = path.split('/', 1)
bucket_name = path_parts[0]
# Ensure the bucket name is valid, to avoid possibly confusing other
# parts of the code. (For example if we didn't catch bucket names
# containing ':', when a user tried to connect to the server with that
# name they might get a confusing error about non-integer port numbers.)
if (bucket_name and not re.match(
'^[a-z0-9][a-z0-9\._-]{1,253}[a-z0-9]$', bucket_name)):
raise InvalidUriError('Invalid bucket name in URI "%s"' % uri_str)
object_name = ''
if len(path_parts) > 1:
object_name = path_parts[1]
return BucketStorageUri(scheme, bucket_name, object_name, debug)
def set_acl(self, acl_or_str, key_name='', validate=True, headers=None,
version_id=None):
if not self.bucket_name:
raise InvalidUriError('set_acl on bucket-less URI (%s)' %
self.uri)
self.get_bucket(validate, headers).set_acl(acl_or_str, key_name,
headers, version_id)
def add_group_email_grant(self,
permission,
email_address,
recursive=False,
validate=False,
headers=None):
self._check_bucket_uri('add_group_email_grant')
if self.scheme != 'gs':
raise ValueError('add_group_email_grant() not supported for %s '
'URIs.' % self.scheme)
if self.object_name:
if recursive:
raise ValueError(
'add_group_email_grant() on key-ful URI cannot '
'specify recursive=True')
key = self.get_key(validate, headers)
self.check_response(key, 'key', self.uri)
key.add_group_email_grant(permission, email_address, headers)
elif self.bucket_name:
bucket = self.get_bucket(validate, headers)
bucket.add_group_email_grant(permission, email_address, recursive,
headers)
else:
raise InvalidUriError('add_group_email_grant() on bucket-less URI '
'%s' % self.uri)
def connect(self, access_key_id=None, secret_access_key=None, **kwargs):
"""
Opens a connection to appropriate provider, depending on provider
portion of URI. Requires Credentials defined in boto config file (see
boto/pyami/config.py).
@type storage_uri: StorageUri
@param storage_uri: StorageUri specifying a bucket or a bucket+object
@rtype: L{AWSAuthConnection<boto.gs.connection.AWSAuthConnection>}
@return: A connection to storage service provider of the given URI.
"""
if not self.connection:
if self.scheme == 's3':
from boto.s3.connection import S3Connection
self.connection = S3Connection(access_key_id,
secret_access_key, **kwargs)
elif self.scheme == 'gs':
from boto.gs.connection import GSConnection
self.connection = GSConnection(access_key_id,
secret_access_key, **kwargs)
elif self.scheme == 'file':
from boto.file.connection import FileConnection
self.connection = FileConnection(self)
else:
raise InvalidUriError('Unrecognized scheme "%s"' % self.scheme)
self.connection.debug = self.debug
return self.connection
def get_acl(self, validate=True, headers=None):
if not self.bucket_name:
raise InvalidUriError('get_acl on bucket-less URI (%s)' % self.uri)
bucket = self.get_bucket(validate, headers)
# This works for both bucket- and object- level ACLs (former passes
# key_name=None):
return bucket.get_acl(self.object_name, headers)
def enable_logging(self, target_bucket, target_prefix=None, validate=True,
headers=None, version_id=None):
if not self.bucket_name:
raise InvalidUriError(
'disable_logging on bucket-less URI (%s)' % self.uri)
bucket = self.get_bucket(validate, headers)
bucket.enable_logging(target_bucket, target_prefix, headers=headers)
def check_response(self, resp, level, uri):
if resp is None:
raise InvalidUriError('\n'.join(textwrap.wrap(
'Attempt to get %s for "%s" failed. This can happen if '
'the URI refers to a non-existent object or if you meant to '
'operate on a directory (e.g., leaving off -R option on gsutil '
'cp, mv, or ls of a bucket)' % (level, uri), 80)))
def get_key(self, validate=True, headers=None, version_id=None):
if not self.object_name:
raise InvalidUriError('get_key on object-less URI (%s)' % self.uri)
bucket = self.get_bucket(validate, headers)
key = bucket.get_key(self.object_name, headers, version_id)
self.check_response(key, 'key', self.uri)
return key
def set_cors(self, cors, validate=True, headers=None):
"""sets or updates a bucket's CORS XML"""
if not self.bucket_name:
raise InvalidUriError('set_cors on bucket-less URI (%s)' %
self.uri)
bucket = self.get_bucket(validate, headers)
bucket.set_cors(cors.to_xml(), headers)
def exists(self, headers=None):
"""Returns True if the object exists or False if it doesn't"""
if not self.object_name:
raise InvalidUriError('exists on object-less URI (%s)' % self.uri)
bucket = self.get_bucket()
key = bucket.get_key(self.object_name, headers=headers)
return bool(key)
def delete_key(self, validate=True, headers=None, version_id=None,
mfa_token=None):
if not self.object_name:
raise InvalidUriError('delete_key on object-less URI (%s)' %
self.uri)
bucket = self.get_bucket(validate, headers)
return bucket.delete_key(self.object_name, headers, version_id,
mfa_token)
def set_def_acl(self, acl_or_str, key_name='', validate=True, headers=None,
version_id=None):
"""sets or updates a bucket's default object acl"""
if not self.bucket_name:
raise InvalidUriError('set_acl on bucket-less URI (%s)' %
self.uri)
self.get_bucket(validate, headers).set_def_acl(acl_or_str, key_name,
headers)
def check_response(self, resp, level, uri):
if resp is None:
raise InvalidUriError(
'Attempt to get %s for "%s" failed.\nThis '
'can happen if the URI refers to a non-'
'existent object or if you meant to\noperate '
'on a directory (e.g., leaving off -R option '
'on gsutil cp, mv, or ls of a\nbucket)' % (level, uri))
def set_canned_acl(self, acl_str, validate=True, headers=None,
version_id=None):
if not self.object_name:
raise InvalidUriError('set_canned_acl on object-less URI (%s)' %
self.uri)
key = self.get_key(validate, headers)
self.check_response(key, 'key', self.uri)
key.set_canned_acl(acl_str, headers, version_id)
def canned_acls(self):
if self.bucket_name is None:
raise InvalidUriError('canned_acls on bucket-less URI (%s)' %
self.uri)
conn = self.connect()
canned_acls = conn.provider.canned_acls
self.check_response(canned_acls, 'canned_acls', self.uri)
return canned_acls
def get_bucket(self, validate=True, headers=None):
if self.bucket_name is None:
raise InvalidUriError('get_bucket on bucket-less URI (%s)' %
self.uri)
conn = self.connect()
bucket = conn.get_bucket(self.bucket_name, validate, headers)
self.check_response(bucket, 'bucket', self.uri)
return bucket
def acl_class(self):
if self.bucket_name is None:
raise InvalidUriError('acl_class on bucket-less URI (%s)' %
self.uri)
conn = self.connect()
acl_class = conn.provider.acl_class
self.check_response(acl_class, 'acl_class', self.uri)
return acl_class
def set_subresource(self, subresource, value, validate=True, headers=None,
version_id=None):
if not self.bucket_name:
raise InvalidUriError(
'set_subresource on bucket-less URI (%s)' % self.uri)
bucket = self.get_bucket(validate, headers)
bucket.set_subresource(subresource, value, self.object_name, headers,
version_id)
def get_acl(self, validate=True, headers=None, version_id=None):
if not self.bucket_name:
raise InvalidUriError('get_acl on bucket-less URI (%s)' % self.uri)
bucket = self.get_bucket(validate, headers)
# This works for both bucket- and object- level ACLs (former passes
# key_name=None):
acl = bucket.get_acl(self.object_name, headers, version_id)
self.check_response(acl, 'acl', self.uri)
return acl
def connect(self, access_key_id=None, secret_access_key=None, **kwargs):
"""
Opens a connection to appropriate provider, depending on provider
portion of URI. Requires Credentials defined in boto config file (see
boto/pyami/config.py).
@type storage_uri: StorageUri
@param storage_uri: StorageUri specifying a bucket or a bucket+object
@rtype: L{AWSAuthConnection<boto.gs.connection.AWSAuthConnection>}
@return: A connection to storage service provider of the given URI.
"""
connection_args = dict(self.connection_args or ())
if (hasattr(self, 'suppress_consec_slashes') and
'suppress_consec_slashes' not in connection_args):
connection_args['suppress_consec_slashes'] = (
self.suppress_consec_slashes)
connection_args.update(kwargs)
if not self.connection:
if self.scheme in self.provider_pool:
self.connection = self.provider_pool[self.scheme]
elif self.scheme == 's3':
from boto.s3.connection import S3Connection
self.connection = S3Connection(access_key_id,
secret_access_key,
**connection_args)
self.provider_pool[self.scheme] = self.connection
elif self.scheme == 'gs':
from boto.gs.connection import GSConnection
# Use OrdinaryCallingFormat instead of boto-default
# SubdomainCallingFormat because the latter changes the hostname
# that's checked during cert validation for HTTPS connections,
# which will fail cert validation (when cert validation is
# enabled).
#
# The same is not true for S3's HTTPS certificates. In fact,
# we don't want to do this for S3 because S3 requires the
# subdomain to match the location of the bucket. If the proper
# subdomain is not used, the server will return a 301 redirect
# with no Location header.
#
# Note: the following import can't be moved up to the
# start of this file else it causes a config import failure when
# run from the resumable upload/download tests.
from boto.s3.connection import OrdinaryCallingFormat
connection_args['calling_format'] = OrdinaryCallingFormat()
self.connection = GSConnection(access_key_id,
secret_access_key,
**connection_args)
self.provider_pool[self.scheme] = self.connection
elif self.scheme == 'file':
from boto.file.connection import FileConnection
self.connection = FileConnection(self)
else:
raise InvalidUriError('Unrecognized scheme "%s"' %
self.scheme)
self.connection.debug = self.debug
return self.connection
def get_cors(self, validate=True, headers=None):
"""returns a bucket's CORS XML"""
if not self.bucket_name:
raise InvalidUriError('get_cors on bucket-less URI (%s)' %
self.uri)
bucket = self.get_bucket(validate, headers)
cors = bucket.get_cors(headers)
self.check_response(cors, 'cors', self.uri)
return cors
def set_canned_acl(self, acl_str, validate=True, headers=None,
version_id=None):
"""sets or updates a bucket's acl to a predefined (canned) value"""
if not self.object_name:
raise InvalidUriError('set_canned_acl on object-less URI (%s)' %
self.uri)
key = self.get_key(validate, headers)
self.check_response(key, 'key', self.uri)
key.set_canned_acl(acl_str, headers, version_id)
def _set_tracker_uri(self, uri):
"""
Called when we start a new resumable upload or get a new tracker
URI for the upload. Saves URI and resets upload state.
Raises InvalidUriError if URI is syntactically invalid.
"""
parse_result = urlparse.urlparse(uri)
if (parse_result.scheme.lower() not in ['http', 'https']
or not parse_result.netloc or not parse_result.query):
raise InvalidUriError('Invalid tracker URI (%s)' % uri)
qdict = cgi.parse_qs(parse_result.query)
if not qdict or not 'upload_id' in qdict:
raise InvalidUriError('Invalid tracker URI (%s)' % uri)
self.tracker_uri = uri
self.tracker_uri_host = parse_result.netloc
self.tracker_uri_path = '/?%s' % parse_result.query
self.server_has_bytes = 0
def get_contents_as_string(self, validate=True, headers=None, cb=None,
num_cb=10, torrent=False, version_id=None):
if not self.object_name:
raise InvalidUriError('get_contents_as_string on object-less URI '
'(%s)' % self.uri)
key = self.get_key(validate, headers)
self.check_response(key, 'key', self.uri)
return key.get_contents_as_string(headers, cb, num_cb, torrent,
version_id)
def check_response(self, resp, level, uri):
if resp is None:
raise InvalidUriError('Attempt to get %s for "%s" failed.\nThis '
'can happen if the URI refers to a non-'
'existent object or if you meant to\noperate '
'on a directory (e.g., leaving off -R option '
'on gsutil cp, mv, or ls of a\nbucket). If a '
'version-ful object was specified, you may '
'have neglected to\nuse a -v flag.' %
(level, uri))
def get_contents_as_string(self,
validate=True,
headers=None,
cb=None,
num_cb=10,
torrent=False):
if not self.object_name:
raise InvalidUriError('get_contents_as_string on object-less URI '
'(%s)' % self.uri)
key = self.get_key(validate, headers)
return key.get_contents_as_string(headers, cb, num_cb, torrent)
def storage_uri_for_key(key):
"""Returns a StorageUri for the given key.
:type key: :class:`boto.s3.key.Key` or subclass
:param key: URI naming bucket + optional object.
"""
if not isinstance(key, boto.s3.key.Key):
raise InvalidUriError('Requested key (%s) is not a subclass of '
'boto.s3.key.Key' % str(type(key)))
prov_name = key.bucket.connection.provider.get_provider_name()
uri_str = '%s://%s/%s' % (prov_name, key.bucket.name, key.name)
return storage_uri(uri_str)
def add_user_grant(self, permission, user_id, recursive=False,
validate=True, headers=None):
if not self.bucket_name:
raise InvalidUriError('add_user_grant on bucket-less URI (%s)' %
self.uri)
if not self.object_name:
bucket = self.get_bucket(validate, headers)
bucket.add_user_grant(permission, user_id, recursive, headers)
else:
key = self.get_key(validate, headers)
self.check_response(key, 'key', self.uri)
key.add_user_grant(permission, user_id)
def clone_replace_name(self, new_name):
"""Instantiate a BucketStorageUri from the current BucketStorageUri,
but replacing the object_name.
@type new_name: string
@param new_name: new object name
"""
if not self.bucket_name:
raise InvalidUriError(
'clone_replace_name() on bucket-less URI %s' % self.uri)
return BucketStorageUri(self.scheme, self.bucket_name, new_name,
self.debug)
def add_email_grant(self,
permission,
email_address,
recursive=False,
validate=True,
headers=None):
if not self.bucket_name:
raise InvalidUriError('add_email_grant on bucket-less URI (%s)' %
self.uri)
if not self.object_name:
bucket = self.get_bucket(validate, headers)
bucket.add_email_grant(permission, email_address, recursive,
headers)
else:
key = self.get_key(validate, headers)
key.add_email_grant(permission, email_address)