def download_dir_contents(self, source_bucket, source_dir, dest_dir):
"""Recursively download contents of a Google Storage directory to local disk
params:
source_bucket: GS bucket to copy the files from
source_dir: full path (Posix-style) within that bucket; read the files
from this directory
dest_dir: full path (local-OS-style) on local disk of directory to copy
the files into
The copy operates as a "merge with overwrite": any files in source_dir will
be "overlaid" on top of the existing content in dest_dir. Existing files
with the same names will be overwritten.
TODO(epoger): Download multiple files simultaneously to reduce latency.
"""
_makedirs_if_needed(dest_dir)
b = self._connect_to_bucket(bucket=source_bucket)
(dirs, files) = self.list_bucket_contents(
bucket=source_bucket, subdir=source_dir)
for filename in files:
key = Key(b)
key.name = posixpath.join(source_dir, filename)
dest_path = os.path.join(dest_dir, filename)
with open(dest_path, 'w') as f:
try:
key.get_contents_to_file(fp=f)
except BotoServerError, e:
e.body = (repr(e.body) +
' while downloading gs://%s/%s to local_path=%s' % (
b.name, key.name, dest_path))
raise
def download_dir_contents(self, source_bucket, source_dir, dest_dir):
"""Recursively download contents of a Google Storage directory to local disk
params:
source_bucket: GS bucket to copy the files from
source_dir: full path (Posix-style) within that bucket; read the files
from this directory
dest_dir: full path (local-OS-style) on local disk of directory to copy
the files into
The copy operates as a "merge with overwrite": any files in source_dir will
be "overlaid" on top of the existing content in dest_dir. Existing files
with the same names will be overwritten.
TODO(epoger): Download multiple files simultaneously to reduce latency.
"""
_makedirs_if_needed(dest_dir)
b = self._connect_to_bucket(bucket=source_bucket)
(dirs, files) = self.list_bucket_contents(bucket=source_bucket,
subdir=source_dir)
for filename in files:
key = Key(b)
key.name = posixpath.join(source_dir, filename)
dest_path = os.path.join(dest_dir, filename)
with open(dest_path, 'w') as f:
try:
key.get_contents_to_file(fp=f)
except BotoServerError, e:
e.body = (
repr(e.body) +
' while downloading gs://%s/%s to local_path=%s' %
(b.name, key.name, dest_path))
raise
def download_file(self, source_bucket, source_path, dest_path,
create_subdirs_if_needed=False, source_generation=None):
"""Downloads a single file from Google Cloud Storage to local disk.
Args:
source_bucket: GS bucket to download the file from
source_path: full path (Posix-style) within that bucket
dest_path: full path (local-OS-style) on local disk to copy the file to
create_subdirs_if_needed: boolean; whether to create subdirectories as
needed to create dest_path
source_generation: the generation version of the source
"""
b = self._connect_to_bucket(bucket=source_bucket)
key = Key(b)
key.name = source_path
if source_generation:
key.generation = source_generation
if create_subdirs_if_needed:
_makedirs_if_needed(os.path.dirname(dest_path))
with open(dest_path, 'w') as f:
try:
key.get_contents_to_file(fp=f)
except BotoServerError, e:
e.body = (repr(e.body) +
' while downloading gs://%s/%s to local_path=%s' % (
b.name, source_path, dest_path))
raise
def download_file(self,
source_bucket,
source_path,
dest_path,
create_subdirs_if_needed=False,
source_generation=None):
"""Downloads a single file from Google Cloud Storage to local disk.
Args:
source_bucket: GS bucket to download the file from
source_path: full path (Posix-style) within that bucket
dest_path: full path (local-OS-style) on local disk to copy the file to
create_subdirs_if_needed: boolean; whether to create subdirectories as
needed to create dest_path
source_generation: the generation version of the source
"""
b = self._connect_to_bucket(bucket=source_bucket)
key = Key(b)
key.name = source_path
if source_generation:
key.generation = source_generation
if create_subdirs_if_needed:
_makedirs_if_needed(os.path.dirname(dest_path))
with open(dest_path, 'w') as f:
try:
key.get_contents_to_file(fp=f)
except BotoServerError, e:
e.body = (repr(e.body) +
' while downloading gs://%s/%s to local_path=%s' %
(b.name, source_path, dest_path))
raise
def delete_file(self, bucket, path):
"""Delete a single file within a GS bucket.
TODO(epoger): what if bucket or path does not exist? Should probably raise
an exception. Implement, and add a test to exercise this.
Params:
bucket: GS bucket to delete a file from
path: full path (Posix-style) of the file within the bucket to delete
"""
b = self._connect_to_bucket(bucket=bucket)
key = Key(b)
key.name = path
try:
key.delete()
except BotoServerError, e:
e.body = (repr(e.body) +
' while deleting gs://%s/%s' % (b.name, path))
raise
def delete_file(self, bucket, path):
"""Delete a single file within a GS bucket.
TODO(epoger): what if bucket or path does not exist? Should probably raise
an exception. Implement, and add a test to exercise this.
Params:
bucket: GS bucket to delete a file from
path: full path (Posix-style) of the file within the bucket to delete
"""
b = self._connect_to_bucket(bucket=bucket)
key = Key(b)
key.name = path
try:
key.delete()
except BotoServerError, e:
e.body = (repr(e.body) + ' while deleting gs://%s/%s' %
(b.name, path))
raise
def upload_file(self, source_path, dest_bucket, dest_path,
upload_if=UploadIf.ALWAYS,
predefined_acl=None,
fine_grained_acl_list=None):
"""Upload contents of a local file to Google Storage.
params:
source_path: full path (local-OS-style) on local disk to read from
dest_bucket: GS bucket to copy the file to
dest_path: full path (Posix-style) within that bucket
upload_if: one of the UploadIf values, describing in which cases we should
upload the file
predefined_acl: which predefined ACL to apply to the file on Google
Storage; must be one of the PredefinedACL values defined above.
If None, inherits dest_bucket's default object ACL.
fine_grained_acl_list: list of (id_type, id_value, permission) tuples
to apply to the uploaded file (on top of the predefined_acl),
or None if predefined_acl is sufficient
TODO(epoger): Consider adding a do_compress parameter that would compress
the file using gzip before upload, and add a "Content-Encoding:gzip" header
so that HTTP downloads of the file would be unzipped automatically.
See https://developers.google.com/storage/docs/gsutil/addlhelp/
WorkingWithObjectMetadata#content-encoding
"""
b = self._connect_to_bucket(bucket=dest_bucket)
local_md5 = None # filled in lazily
if upload_if == self.UploadIf.IF_NEW:
old_key = b.get_key(key_name=dest_path)
if old_key:
print ('Skipping upload of existing file gs://%s/%s' % (
b.name, dest_path))
return
elif upload_if == self.UploadIf.IF_MODIFIED:
old_key = b.get_key(key_name=dest_path)
if old_key:
if not local_md5:
local_md5 = _get_local_md5(path=source_path)
if ('"%s"' % local_md5) == old_key.etag:
print (
'Skipping upload of unmodified file gs://%s/%s : %s' % (
b.name, dest_path, local_md5))
return
elif upload_if != self.UploadIf.ALWAYS:
raise Exception('unknown value of upload_if: %s' % upload_if)
# Upload the file using a temporary name at first, in case the transfer
# is interrupted partway through.
if not local_md5:
local_md5 = _get_local_md5(path=source_path)
initial_key = Key(b)
initial_key.name = dest_path + '-uploading-' + local_md5
try:
initial_key.set_contents_from_filename(filename=source_path,
policy=predefined_acl)
except BotoServerError, e:
e.body = (repr(e.body) +
' while uploading source_path=%s to gs://%s/%s' % (
source_path, b.name, initial_key.name))
raise
def upload_file(self,
source_path,
dest_bucket,
dest_path,
upload_if=UploadIf.ALWAYS,
predefined_acl=None,
fine_grained_acl_list=None):
"""Upload contents of a local file to Google Storage.
params:
source_path: full path (local-OS-style) on local disk to read from
dest_bucket: GS bucket to copy the file to
dest_path: full path (Posix-style) within that bucket
upload_if: one of the UploadIf values, describing in which cases we should
upload the file
predefined_acl: which predefined ACL to apply to the file on Google
Storage; must be one of the PredefinedACL values defined above.
If None, inherits dest_bucket's default object ACL.
fine_grained_acl_list: list of (id_type, id_value, permission) tuples
to apply to the uploaded file (on top of the predefined_acl),
or None if predefined_acl is sufficient
TODO(epoger): Consider adding a do_compress parameter that would compress
the file using gzip before upload, and add a "Content-Encoding:gzip" header
so that HTTP downloads of the file would be unzipped automatically.
See https://developers.google.com/storage/docs/gsutil/addlhelp/
WorkingWithObjectMetadata#content-encoding
"""
b = self._connect_to_bucket(bucket=dest_bucket)
local_md5 = None # filled in lazily
if upload_if == self.UploadIf.IF_NEW:
old_key = b.get_key(key_name=dest_path)
if old_key:
print('Skipping upload of existing file gs://%s/%s' %
(b.name, dest_path))
return
elif upload_if == self.UploadIf.IF_MODIFIED:
old_key = b.get_key(key_name=dest_path)
if old_key:
if not local_md5:
local_md5 = _get_local_md5(path=source_path)
if ('"%s"' % local_md5) == old_key.etag:
print(
'Skipping upload of unmodified file gs://%s/%s : %s' %
(b.name, dest_path, local_md5))
return
elif upload_if != self.UploadIf.ALWAYS:
raise Exception('unknown value of upload_if: %s' % upload_if)
# Upload the file using a temporary name at first, in case the transfer
# is interrupted partway through.
if not local_md5:
local_md5 = _get_local_md5(path=source_path)
initial_key = Key(b)
initial_key.name = dest_path + '-uploading-' + local_md5
try:
initial_key.set_contents_from_filename(filename=source_path,
policy=predefined_acl)
except BotoServerError, e:
e.body = (repr(e.body) +
' while uploading source_path=%s to gs://%s/%s' %
(source_path, b.name, initial_key.name))
raise
