def RunCommand(self):
"""Command entry point for the mv command."""
# Check each source arg up, refusing to delete a bucket src URL (force users
# to explicitly do that as a separate operation).
for arg_to_check in self.args[0:-1]:
url = StorageUrlFromString(arg_to_check)
if url.IsCloudUrl() and (url.IsBucket() or url.IsProvider()):
raise CommandException(
'You cannot move a source bucket using the mv '
'command. If you meant to move\nall objects in '
'the bucket, you can use a command like:\n'
'\tgsutil mv %s/* %s' % (arg_to_check, self.args[-1]))
# Insert command-line opts in front of args so they'll be picked up by cp
# and rm commands (e.g., for -p option). Use undocumented (internal
# use-only) cp -M option, which causes each original object to be deleted
# after successfully copying to its destination, and also causes naming
# behavior consistent with Unix mv naming behavior (see comments in
# ConstructDstUrl).
unparsed_args = ['-M']
if self.recursion_requested:
unparsed_args.append('-R')
unparsed_args.extend(self.unparsed_args)
self.command_runner.RunNamedCommand('cp', unparsed_args, self.headers,
self.debug,
self.parallel_operations)
return 0
def _EnumerateNotificationsFromArgs(self,
accept_notification_configs=True):
"""Yields bucket/notification tuples from command-line args.
Given a list of strings that are bucket names (gs://foo) or notification
config IDs, yield tuples of bucket names and their associated notifications.
Args:
accept_notification_configs: whether notification configs are valid args.
Yields:
Tuples of the form (bucket_name, Notification)
"""
path_regex = self._GetNotificationPathRegex()
for list_entry in self.args:
match = path_regex.match(list_entry)
if match:
if not accept_notification_configs:
raise CommandException(
'%s %s accepts only bucket names, but you provided %s'
%
(self.command_name, self.subcommand_name, list_entry))
bucket_name = match.group('bucket')
notification_id = match.group('notification')
found = False
for notification in self.gsutil_api.ListNotificationConfigs(
bucket_name, provider='gs'):
if notification.id == notification_id:
yield (bucket_name, notification)
found = True
break
if not found:
raise NotFoundException('Could not find notification %s' %
list_entry)
else:
storage_url = StorageUrlFromString(list_entry)
if not storage_url.IsCloudUrl():
raise CommandException(
'The %s command must be used on cloud buckets or notification '
'config names.' % self.command_name)
if storage_url.scheme != 'gs':
raise CommandException(
'The %s command only works on gs:// buckets.')
path = None
if storage_url.IsProvider():
path = 'gs://*'
elif storage_url.IsBucket():
path = list_entry
if not path:
raise CommandException(
'The %s command cannot be used on cloud objects, only buckets'
% self.command_name)
for blr in self.WildcardIterator(path).IterBuckets(
bucket_fields=['id']):
for notification in self.gsutil_api.ListNotificationConfigs(
blr.storage_url.bucket_name, provider='gs'):
yield (blr.storage_url.bucket_name, notification)
def HaveProviderUrls(args_to_check):
"""Checks whether args_to_check contains any provider URLs (like 'gs://').
Args:
args_to_check: Command-line argument subset to check.
Returns:
True if args_to_check contains any provider URLs.
"""
for url_str in args_to_check:
storage_url = StorageUrlFromString(url_str)
if storage_url.IsCloudUrl() and storage_url.IsProvider():
return True
return False
def RunCommand(self):
"""Command entry point for the ls command."""
got_nomatch_errors = False
got_bucket_nomatch_errors = False
listing_style = ListingStyle.SHORT
get_bucket_info = False
self.recursion_requested = False
self.all_versions = False
self.include_etag = False
self.human_readable = False
if self.sub_opts:
for o, a in self.sub_opts:
if o == '-a':
self.all_versions = True
elif o == '-e':
self.include_etag = True
elif o == '-b':
get_bucket_info = True
elif o == '-h':
self.human_readable = True
elif o == '-l':
listing_style = ListingStyle.LONG
elif o == '-L':
listing_style = ListingStyle.LONG_LONG
elif o == '-p':
self.project_id = a
elif o == '-r' or o == '-R':
self.recursion_requested = True
if not self.args:
# default to listing all gs buckets
self.args = ['gs://']
total_objs = 0
total_bytes = 0
def MaybePrintBucketHeader(blr):
if len(self.args) > 1:
print '%s:' % blr.url_string.encode(UTF8)
print_bucket_header = MaybePrintBucketHeader
for url_str in self.args:
storage_url = StorageUrlFromString(url_str)
if storage_url.IsFileUrl():
raise CommandException('Only cloud URLs are supported for %s' %
self.command_name)
bucket_fields = None
if (listing_style == ListingStyle.SHORT
or listing_style == ListingStyle.LONG):
bucket_fields = ['id']
elif listing_style == ListingStyle.LONG_LONG:
bucket_fields = [
'location', 'storageClass', 'versioning', 'acl',
'defaultObjectAcl', 'website', 'logging', 'cors',
'lifecycle'
]
if storage_url.IsProvider():
# Provider URL: use bucket wildcard to list buckets.
for blr in self.WildcardIterator(
'%s://*' % storage_url.scheme).IterBuckets(
bucket_fields=bucket_fields):
self._PrintBucketInfo(blr, listing_style)
elif storage_url.IsBucket() and get_bucket_info:
# ls -b bucket listing request: List info about bucket(s).
total_buckets = 0
for blr in self.WildcardIterator(url_str).IterBuckets(
bucket_fields=bucket_fields):
if not ContainsWildcard(url_str) and not blr.root_object:
# Iterator does not make an HTTP call for non-wildcarded
# listings with fields=='id'. Ensure the bucket exists by calling
# GetBucket.
self.gsutil_api.GetBucket(blr.storage_url.bucket_name,
fields=['id'],
provider=storage_url.scheme)
self._PrintBucketInfo(blr, listing_style)
total_buckets += 1
if not ContainsWildcard(url_str) and not total_buckets:
got_bucket_nomatch_errors = True
else:
# URL names a bucket, object, or object subdir ->
# list matching object(s) / subdirs.
def _PrintPrefixLong(blr):
print '%-33s%s' % ('', blr.url_string.encode(UTF8))
if listing_style == ListingStyle.SHORT:
# ls helper by default readies us for a short listing.
ls_helper = LsHelper(
self.WildcardIterator,
self.logger,
all_versions=self.all_versions,
print_bucket_header_func=print_bucket_header,
should_recurse=self.recursion_requested)
elif listing_style == ListingStyle.LONG:
bucket_listing_fields = ['name', 'updated', 'size']
if self.all_versions:
bucket_listing_fields.extend(
['generation', 'metageneration'])
if self.include_etag:
bucket_listing_fields.append('etag')
ls_helper = LsHelper(
self.WildcardIterator,
self.logger,
print_object_func=self._PrintLongListing,
print_dir_func=_PrintPrefixLong,
print_bucket_header_func=print_bucket_header,
all_versions=self.all_versions,
should_recurse=self.recursion_requested,
fields=bucket_listing_fields)
elif listing_style == ListingStyle.LONG_LONG:
# List all fields
bucket_listing_fields = None
ls_helper = LsHelper(
self.WildcardIterator,
self.logger,
print_object_func=PrintFullInfoAboutObject,
print_dir_func=_PrintPrefixLong,
print_bucket_header_func=print_bucket_header,
all_versions=self.all_versions,
should_recurse=self.recursion_requested,
fields=bucket_listing_fields)
else:
raise CommandException('Unknown listing style: %s' %
listing_style)
exp_dirs, exp_objs, exp_bytes = ls_helper.ExpandUrlAndPrint(
storage_url)
if storage_url.IsObject() and exp_objs == 0 and exp_dirs == 0:
got_nomatch_errors = True
total_bytes += exp_bytes
total_objs += exp_objs
if total_objs and listing_style != ListingStyle.SHORT:
print('TOTAL: %d objects, %d bytes (%s)' %
(total_objs, total_bytes,
MakeHumanReadable(float(total_bytes))))
if got_nomatch_errors:
raise CommandException('One or more URLs matched no objects.')
if got_bucket_nomatch_errors:
raise NotFoundException(
'One or more bucket URLs matched no buckets.')
return 0
def RunCommand(self):
"""Command entry point for the rm command."""
# self.recursion_requested is initialized in command.py (so it can be
# checked in parent class for all commands).
self.continue_on_error = self.parallel_operations
self.read_args_from_stdin = False
self.all_versions = False
if self.sub_opts:
for o, unused_a in self.sub_opts:
if o == '-a':
self.all_versions = True
elif o == '-f':
self.continue_on_error = True
elif o == '-I':
self.read_args_from_stdin = True
elif o == '-r' or o == '-R':
self.recursion_requested = True
self.all_versions = True
if self.read_args_from_stdin:
if self.args:
raise CommandException(
'No arguments allowed with the -I flag.')
url_strs = StdinIterator()
else:
if not self.args:
raise CommandException(
'The rm command (without -I) expects at '
'least one URL.')
url_strs = self.args
# Tracks number of object deletes that failed.
self.op_failure_count = 0
# Tracks if any buckets were missing.
self.bucket_not_found_count = 0
# Tracks buckets that are slated for recursive deletion.
bucket_urls_to_delete = []
self.bucket_strings_to_delete = []
if self.recursion_requested:
bucket_fields = ['id']
for url_str in url_strs:
url = StorageUrlFromString(url_str)
if url.IsBucket() or url.IsProvider():
for blr in self.WildcardIterator(url_str).IterBuckets(
bucket_fields=bucket_fields):
bucket_urls_to_delete.append(blr.storage_url)
self.bucket_strings_to_delete.append(url_str)
self.preconditions = PreconditionsFromHeaders(self.headers or {})
try:
# Expand wildcards, dirs, buckets, and bucket subdirs in URLs.
name_expansion_iterator = NameExpansionIterator(
self.command_name,
self.debug,
self.logger,
self.gsutil_api,
url_strs,
self.recursion_requested,
project_id=self.project_id,
all_versions=self.all_versions,
continue_on_error=self.continue_on_error
or self.parallel_operations)
seek_ahead_iterator = None
# Cannot seek ahead with stdin args, since we can only iterate them
# once without buffering in memory.
if not self.read_args_from_stdin:
seek_ahead_iterator = SeekAheadNameExpansionIterator(
self.command_name,
self.debug,
self.GetSeekAheadGsutilApi(),
url_strs,
self.recursion_requested,
all_versions=self.all_versions,
project_id=self.project_id)
# Perform remove requests in parallel (-m) mode, if requested, using
# configured number of parallel processes and threads. Otherwise,
# perform requests with sequential function calls in current process.
self.Apply(
_RemoveFuncWrapper,
name_expansion_iterator,
_RemoveExceptionHandler,
fail_on_error=(not self.continue_on_error),
shared_attrs=['op_failure_count', 'bucket_not_found_count'],
seek_ahead_iterator=seek_ahead_iterator)
# Assuming the bucket has versioning enabled, url's that don't map to
# objects should throw an error even with all_versions, since the prior
# round of deletes only sends objects to a history table.
# This assumption that rm -a is only called for versioned buckets should be
# corrected, but the fix is non-trivial.
except CommandException as e:
# Don't raise if there are buckets to delete -- it's valid to say:
# gsutil rm -r gs://some_bucket
# if the bucket is empty.
if _ExceptionMatchesBucketToDelete(self.bucket_strings_to_delete,
e):
DecrementFailureCount()
else:
raise
except ServiceException, e:
if not self.continue_on_error:
raise
def RunCommand(self):
"""Command entry point for the rm command."""
# self.recursion_requested is initialized in command.py (so it can be
# checked in parent class for all commands).
self.continue_on_error = self.parallel_operations
self.read_args_from_stdin = False
self.all_versions = False
if self.sub_opts:
for o, unused_a in self.sub_opts:
if o == '-a':
self.all_versions = True
elif o == '-f':
self.continue_on_error = True
elif o == '-I':
self.read_args_from_stdin = True
elif o == '-r' or o == '-R':
self.recursion_requested = True
self.all_versions = True
if self.read_args_from_stdin:
if self.args:
raise CommandException(
'No arguments allowed with the -I flag.')
url_strs = StdinIterator()
else:
if not self.args:
raise CommandException(
'The rm command (without -I) expects at '
'least one URL.')
url_strs = self.args
# Tracks number of object deletes that failed.
self.op_failure_count = 0
# Tracks if any buckets were missing.
self.bucket_not_found_count = 0
# Tracks buckets that are slated for recursive deletion.
bucket_urls_to_delete = []
self.bucket_strings_to_delete = []
if self.recursion_requested:
bucket_fields = ['id']
for url_str in url_strs:
url = StorageUrlFromString(url_str)
if url.IsBucket() or url.IsProvider():
for blr in self.WildcardIterator(url_str).IterBuckets(
bucket_fields=bucket_fields):
bucket_urls_to_delete.append(blr.storage_url)
self.bucket_strings_to_delete.append(url_str)
self.preconditions = PreconditionsFromHeaders(self.headers or {})
try:
# Expand wildcards, dirs, buckets, and bucket subdirs in URLs.
name_expansion_iterator = NameExpansionIterator(
self.command_name,
self.debug,
self.logger,
self.gsutil_api,
url_strs,
self.recursion_requested,
project_id=self.project_id,
all_versions=self.all_versions,
continue_on_error=self.continue_on_error
or self.parallel_operations)
seek_ahead_iterator = None
# Cannot seek ahead with stdin args, since we can only iterate them
# once without buffering in memory.
if not self.read_args_from_stdin:
seek_ahead_iterator = SeekAheadNameExpansionIterator(
self.command_name,
self.debug,
self.GetSeekAheadGsutilApi(),
url_strs,
self.recursion_requested,
all_versions=self.all_versions,
project_id=self.project_id)
# Perform remove requests in parallel (-m) mode, if requested, using
# configured number of parallel processes and threads. Otherwise,
# perform requests with sequential function calls in current process.
self.Apply(
_RemoveFuncWrapper,
name_expansion_iterator,
_RemoveExceptionHandler,
fail_on_error=(not self.continue_on_error),
shared_attrs=['op_failure_count', 'bucket_not_found_count'],
seek_ahead_iterator=seek_ahead_iterator)
# Assuming the bucket has versioning enabled, url's that don't map to
# objects should throw an error even with all_versions, since the prior
# round of deletes only sends objects to a history table.
# This assumption that rm -a is only called for versioned buckets should be
# corrected, but the fix is non-trivial.
except CommandException as e:
# Don't raise if there are buckets to delete -- it's valid to say:
# gsutil rm -r gs://some_bucket
# if the bucket is empty.
if _ExceptionMatchesBucketToDelete(self.bucket_strings_to_delete,
e):
DecrementFailureCount()
else:
raise
except ServiceException as e:
if not self.continue_on_error:
raise
if self.bucket_not_found_count:
raise CommandException(
'Encountered non-existent bucket during listing')
if self.op_failure_count and not self.continue_on_error:
raise CommandException('Some files could not be removed.')
# If this was a gsutil rm -r command covering any bucket subdirs,
# remove any dir_$folder$ objects (which are created by various web UI
# tools to simulate folders).
if self.recursion_requested:
folder_object_wildcards = []
for url_str in url_strs:
url = StorageUrlFromString(url_str)
if url.IsObject():
folder_object_wildcards.append('%s**_$folder$' % url_str)
if folder_object_wildcards:
self.continue_on_error = True
try:
name_expansion_iterator = NameExpansionIterator(
self.command_name,
self.debug,
self.logger,
self.gsutil_api,
folder_object_wildcards,
self.recursion_requested,
project_id=self.project_id,
all_versions=self.all_versions)
# When we're removing folder objects, always continue on error
self.Apply(_RemoveFuncWrapper,
name_expansion_iterator,
_RemoveFoldersExceptionHandler,
fail_on_error=False)
except CommandException as e:
# Ignore exception from name expansion due to an absent folder file.
if not e.reason.startswith(NO_URLS_MATCHED_GENERIC):
raise
# Now that all data has been deleted, delete any bucket URLs.
for url in bucket_urls_to_delete:
self.logger.info('Removing %s...', url)
@Retry(NotEmptyException, tries=3, timeout_secs=1)
def BucketDeleteWithRetry():
self.gsutil_api.DeleteBucket(url.bucket_name,
provider=url.scheme)
BucketDeleteWithRetry()
if self.op_failure_count:
plural_str = 's' if self.op_failure_count else ''
raise CommandException(
'%d file%s/object%s could not be removed.' %
(self.op_failure_count, plural_str, plural_str))
return 0
def RunCommand(self):
"""Command entry point for the rm command."""
# self.recursion_requested is initialized in command.py (so it can be
# checked in parent class for all commands).
self.continue_on_error = False
self.read_args_from_stdin = False
self.all_versions = False
if self.sub_opts:
for o, unused_a in self.sub_opts:
if o == '-a':
self.all_versions = True
elif o == '-f':
self.continue_on_error = True
elif o == '-I':
self.read_args_from_stdin = True
elif o == '-r' or o == '-R':
self.recursion_requested = True
self.all_versions = True
if self.read_args_from_stdin:
if self.args:
raise CommandException(
'No arguments allowed with the -I flag.')
url_strs = StdinIterator()
else:
if not self.args:
raise CommandException(
'The rm command (without -I) expects at '
'least one URL.')
url_strs = self.args
bucket_urls_to_delete = []
bucket_strings_to_delete = []
if self.recursion_requested:
bucket_fields = ['id']
for url_str in url_strs:
url = StorageUrlFromString(url_str)
if url.IsBucket() or url.IsProvider():
for blr in self.WildcardIterator(url_str).IterBuckets(
bucket_fields=bucket_fields):
bucket_urls_to_delete.append(blr.storage_url)
bucket_strings_to_delete.append(url_str)
self.preconditions = PreconditionsFromHeaders(self.headers or {})
# Used to track if any files failed to be removed.
self.everything_removed_okay = True
try:
# Expand wildcards, dirs, buckets, and bucket subdirs in URLs.
name_expansion_iterator = NameExpansionIterator(
self.command_name,
self.debug,
self.logger,
self.gsutil_api,
url_strs,
self.recursion_requested,
project_id=self.project_id,
all_versions=self.all_versions,
continue_on_error=self.continue_on_error
or self.parallel_operations)
# Perform remove requests in parallel (-m) mode, if requested, using
# configured number of parallel processes and threads. Otherwise,
# perform requests with sequential function calls in current process.
self.Apply(_RemoveFuncWrapper,
name_expansion_iterator,
_RemoveExceptionHandler,
fail_on_error=(not self.continue_on_error))
# Assuming the bucket has versioning enabled, url's that don't map to
# objects should throw an error even with all_versions, since the prior
# round of deletes only sends objects to a history table.
# This assumption that rm -a is only called for versioned buckets should be
# corrected, but the fix is non-trivial.
except CommandException as e:
# Don't raise if there are buckets to delete -- it's valid to say:
# gsutil rm -r gs://some_bucket
# if the bucket is empty.
if not bucket_urls_to_delete and not self.continue_on_error:
raise
# Reset the failure count if we failed due to an empty bucket that we're
# going to delete.
msg = 'No URLs matched: '
if msg in str(e):
parts = str(e).split(msg)
if len(parts) == 2 and parts[1] in bucket_strings_to_delete:
ResetFailureCount()
except ServiceException, e:
if not self.continue_on_error:
raise
def RunCommand(self):
"""Command entry point for the du command."""
self.line_ending = '\n'
self.all_versions = False
self.produce_total = False
self.human_readable = False
self.summary_only = False
self.exclude_patterns = []
if self.sub_opts:
for o, a in self.sub_opts:
if o == '-0':
self.line_ending = '\0'
elif o == '-a':
self.all_versions = True
elif o == '-c':
self.produce_total = True
elif o == '-e':
self.exclude_patterns.append(a)
elif o == '-h':
self.human_readable = True
elif o == '-s':
self.summary_only = True
elif o == '-X':
if a == '-':
f = sys.stdin
else:
f = open(a, 'r')
try:
for line in f:
line = line.strip()
if line:
self.exclude_patterns.append(line)
finally:
f.close()
if not self.args:
# Default to listing all gs buckets.
self.args = ['gs://']
total_bytes = 0
got_nomatch_errors = False
def _PrintObjectLong(blr):
return self._PrintInfoAboutBucketListingRef(blr)
def _PrintNothing(unused_blr=None):
pass
def _PrintDirectory(num_bytes, name):
if not self.summary_only:
self._PrintSummaryLine(num_bytes, name)
for url_arg in self.args:
top_level_storage_url = StorageUrlFromString(url_arg)
if top_level_storage_url.IsFileUrl():
raise CommandException('Only cloud URLs are supported for %s' %
self.command_name)
bucket_listing_fields = ['size']
ls_helper = LsHelper(self.WildcardIterator,
self.logger,
print_object_func=_PrintObjectLong,
print_dir_func=_PrintNothing,
print_dir_header_func=_PrintNothing,
print_dir_summary_func=_PrintDirectory,
print_newline_func=_PrintNothing,
all_versions=self.all_versions,
should_recurse=True,
exclude_patterns=self.exclude_patterns,
fields=bucket_listing_fields)
# ls_helper expands to objects and prefixes, so perform a top-level
# expansion first.
if top_level_storage_url.IsProvider():
# Provider URL: use bucket wildcard to iterate over all buckets.
top_level_iter = self.WildcardIterator(
'%s://*' % top_level_storage_url.scheme).IterBuckets(
bucket_fields=['id'])
elif top_level_storage_url.IsBucket():
top_level_iter = self.WildcardIterator(
'%s://%s' %
(top_level_storage_url.scheme,
top_level_storage_url.bucket_name)).IterBuckets(
bucket_fields=['id'])
else:
top_level_iter = [BucketListingObject(top_level_storage_url)]
for blr in top_level_iter:
storage_url = blr.storage_url
if storage_url.IsBucket() and self.summary_only:
storage_url = StorageUrlFromString(
storage_url.CreatePrefixUrl(wildcard_suffix='**'))
_, exp_objs, exp_bytes = ls_helper.ExpandUrlAndPrint(
storage_url)
if (storage_url.IsObject() and exp_objs == 0
and ContainsWildcard(url_arg)
and not self.exclude_patterns):
got_nomatch_errors = True
total_bytes += exp_bytes
if self.summary_only:
self._PrintSummaryLine(exp_bytes,
blr.url_string.rstrip('/'))
if self.produce_total:
self._PrintSummaryLine(total_bytes, 'total')
if got_nomatch_errors:
raise CommandException('One or more URLs matched no objects.')
return 0
