# TODO:
# --md5sum - display md5 of each file as read from disk
-import apiclient.errors
import argparse
import arvados
import base64
import socket
import sys
import tempfile
+from apiclient import errors as apiclient_errors
import arvados.commands._util as arv_cmd
CAUGHT_SIGNALS = [signal.SIGINT, signal.SIGQUIT, signal.SIGTERM]
+api_client = None
upload_opts = argparse.ArgumentParser(add_help=False)
stream per filesystem directory that contains files.
""")
-upload_opts.add_argument('--project-uuid', metavar='UUID', help="""
-Store the collection in the specified project, instead of your Home
-project.
-""")
-
-upload_opts.add_argument('--name', help="""
-Save the collection with the specified name, rather than the default
-generic name "Saved at {time} by {username}@{host}".
-""")
-
_group = upload_opts.add_mutually_exclusive_group()
_group.add_argument('--as-stream', action='store_true', dest='stream',
it is not a directory. Implies --manifest.
""")
+upload_opts.add_argument('--portable-data-hash', action='store_true',
+ help="""
+Print the portable data hash instead of the Arvados UUID for the collection
+created by the upload.
+""")
+
run_opts = argparse.ArgumentParser(add_help=False)
+
+run_opts.add_argument('--project-uuid', metavar='UUID', help="""
+Store the collection in the specified project, instead of your Home
+project.
+""")
+
+run_opts.add_argument('--name', help="""
+Save the collection with the specified name.
+""")
+
_group = run_opts.add_mutually_exclusive_group()
_group.add_argument('--progress', action='store_true',
help="""
arg_parser = argparse.ArgumentParser(
description='Copy data from the local filesystem to Keep.',
- parents=[upload_opts, run_opts])
+ parents=[upload_opts, run_opts, arv_cmd.retry_opt])
def parse_arguments(arguments):
args = arg_parser.parse_args(arguments)
STATE_PROPS = (arvados.ResumableCollectionWriter.STATE_PROPS +
['bytes_written', '_seen_inputs'])
- def __init__(self, cache=None, reporter=None, bytes_expected=None):
+ def __init__(self, cache=None, reporter=None, bytes_expected=None,
+ api_client=None, num_retries=0):
self.bytes_written = 0
self._seen_inputs = []
self.cache = cache
self.reporter = reporter
self.bytes_expected = bytes_expected
- super(ArvPutCollectionWriter, self).__init__()
+ super(ArvPutCollectionWriter, self).__init__(
+ api_client, num_retries=num_retries)
@classmethod
- def from_cache(cls, cache, reporter=None, bytes_expected=None):
+ def from_cache(cls, cache, reporter=None, bytes_expected=None,
+ num_retries=0):
try:
state = cache.load()
state['_data_buffer'] = [base64.decodestring(state['_data_buffer'])]
- writer = cls.from_state(state, cache, reporter, bytes_expected)
+ writer = cls.from_state(state, cache, reporter, bytes_expected,
+ num_retries=num_retries)
except (TypeError, ValueError,
arvados.errors.StaleWriterStateError) as error:
- return cls(cache, reporter, bytes_expected)
+ return cls(cache, reporter, bytes_expected, num_retries=num_retries)
else:
return writer
def exit_signal_handler(sigcode, frame):
sys.exit(-sigcode)
-def check_project_exists(project_uuid):
- try:
- arvados.api('v1').groups().get(uuid=project_uuid).execute()
- except (apiclient.errors.Error, arvados.errors.NotFoundError) as error:
- raise ValueError("Project {} not found ({})".format(project_uuid,
- error))
+def desired_project_uuid(api_client, project_uuid, num_retries):
+ if not project_uuid:
+ query = api_client.users().current()
+ elif arvados.util.user_uuid_pattern.match(project_uuid):
+ query = api_client.users().get(uuid=project_uuid)
+ elif arvados.util.group_uuid_pattern.match(project_uuid):
+ query = api_client.groups().get(uuid=project_uuid)
else:
- return True
+ raise ValueError("Not a valid project UUID: {}".format(project_uuid))
+ return query.execute(num_retries=num_retries)['uuid']
+
+def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr):
+ global api_client
-def prep_project_link(args, stderr, project_exists=check_project_exists):
- # Given the user's command line arguments, return a dictionary with data
- # to create the desired project link for this Collection, or None.
- # Raises ValueError if the arguments request something impossible.
- making_collection = not (args.raw or args.stream)
- if not making_collection:
- if args.name or args.project_uuid:
- raise ValueError("Requested a Link without creating a Collection")
- return None
- link = {'tail_uuid': args.project_uuid,
- 'link_class': 'name',
- 'name': args.name}
- if not link['tail_uuid']:
- link['tail_uuid'] = arvados.api('v1').users().current().execute()['uuid']
- elif not project_exists(link['tail_uuid']):
- raise ValueError("Project {} not found".format(args.project_uuid))
- if not link['name']:
- link['name'] = "Saved at {} by {}@{}".format(
+ args = parse_arguments(arguments)
+ status = 0
+ if api_client is None:
+ api_client = arvados.api('v1')
+
+ # Determine the name to use
+ if args.name:
+ if args.stream or args.raw:
+ print >>stderr, "Cannot use --name with --stream or --raw"
+ sys.exit(1)
+ collection_name = args.name
+ else:
+ collection_name = "Saved at {} by {}@{}".format(
datetime.datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S UTC"),
pwd.getpwuid(os.getuid()).pw_name,
socket.gethostname())
- stderr.write(
- "arv-put: No --name specified. Saving as \"%s\"\n" % link['name'])
- link['owner_uuid'] = link['tail_uuid']
- return link
-
-def create_project_link(locator, link):
- link['head_uuid'] = locator
- return arvados.api('v1').links().create(body=link).execute()
-def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr):
- status = 0
+ if args.project_uuid and (args.stream or args.raw):
+ print >>stderr, "Cannot use --project-uuid with --stream or --raw"
+ sys.exit(1)
- args = parse_arguments(arguments)
+ # Determine the parent project
try:
- project_link = prep_project_link(args, stderr)
- except ValueError as error:
- print >>stderr, "arv-put: {}.".format(error)
- sys.exit(2)
+ project_uuid = desired_project_uuid(api_client, args.project_uuid,
+ args.retries)
+ except (apiclient_errors.Error, ValueError) as error:
+ print >>stderr, error
+ sys.exit(1)
if args.progress:
reporter = progress_writer(human_progress)
sys.exit(1)
if resume_cache is None:
- writer = ArvPutCollectionWriter(resume_cache, reporter, bytes_expected)
+ writer = ArvPutCollectionWriter(resume_cache, reporter, bytes_expected,
+ num_retries=args.retries)
else:
writer = ArvPutCollectionWriter.from_cache(
- resume_cache, reporter, bytes_expected)
+ resume_cache, reporter, bytes_expected, num_retries=args.retries)
# Install our signal handler for each code in CAUGHT_SIGNALS, and save
# the originals.
elif args.raw:
output = ','.join(writer.data_locators())
else:
- # Register the resulting collection in Arvados.
- collection = arvados.api().collections().create(
- body={
- 'uuid': writer.finish(),
- 'manifest_text': writer.manifest_text(),
- },
- ).execute()
-
- # Print the locator (uuid) of the new collection.
- output = collection['uuid']
- if project_link is not None:
- try:
- create_project_link(output, project_link)
- except apiclient.errors.Error as error:
- print >>stderr, (
- "arv-put: Error adding Collection to project: {}.".format(
- error))
- status = 1
-
+ try:
+ # Register the resulting collection in Arvados.
+ collection = api_client.collections().create(
+ body={
+ 'owner_uuid': project_uuid,
+ 'name': collection_name,
+ 'manifest_text': writer.manifest_text()
+ },
+ ensure_unique_name=True
+ ).execute(num_retries=args.retries)
+
+ print >>stderr, "Collection saved as '%s'" % collection['name']
+
+ if args.portable_data_hash and 'portable_data_hash' in collection and collection['portable_data_hash']:
+ output = collection['portable_data_hash']
+ else:
+ output = collection['uuid']
+
+ except apiclient_errors.Error as error:
+ print >>stderr, (
+ "arv-put: Error creating Collection on project: {}.".format(
+ error))
+ status = 1
+
+ # Print the locator (uuid) of the new collection.
stdout.write(output)
if not output.endswith('\n'):
stdout.write('\n')
projects / arvados.git / blobdiff