#!/usr/bin/env python
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
import argparse
import hashlib
import arvados
import arvados.commands._util as arv_cmd
+import arvados.util as util
from arvados._version import __version__
overwritten. This option causes even devices, sockets, and fifos to be
skipped.
""")
+group.add_argument('--strip-manifest', action='store_true', default=False,
+ help="""
+When getting a collection manifest, strip its access tokens before writing
+it.
+""")
def parse_arguments(arguments, stdout, stderr):
args = parser.parse_args(arguments)
def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr):
global api_client
-
+
+ if stdout is sys.stdout and hasattr(stdout, 'buffer'):
+ # in Python 3, write to stdout as binary
+ stdout = stdout.buffer
+
args = parse_arguments(arguments, stdout, stderr)
if api_client is None:
api_client = arvados.api('v1')
r = re.search(r'^(.*?)(/.*)?$', args.locator)
- collection = r.group(1)
+ col_loc = r.group(1)
get_prefix = r.group(2)
if args.r and not get_prefix:
get_prefix = os.sep
- try:
- reader = arvados.CollectionReader(collection, num_retries=args.retries)
- except Exception as error:
- logger.error("failed to read collection: {}".format(error))
- return 1
# User asked to download the collection's manifest
if not get_prefix:
open_flags |= os.O_EXCL
try:
if args.destination == "-":
- stdout.write(reader.manifest_text())
+ write_block_or_manifest(dest=stdout, src=col_loc,
+ api_client=api_client, args=args)
else:
out_fd = os.open(args.destination, open_flags)
with os.fdopen(out_fd, 'wb') as out_file:
- out_file.write(reader.manifest_text())
+ write_block_or_manifest(dest=out_file,
+ src=col_loc, api_client=api_client,
+ args=args)
except (IOError, OSError) as error:
logger.error("can't write to '{}': {}".format(args.destination, error))
return 1
except (arvados.errors.ApiError, arvados.errors.KeepReadError) as error:
- logger.error("failed to download '{}': {}".format(collection, error))
+ logger.error("failed to download '{}': {}".format(col_loc, error))
return 1
+ except arvados.errors.ArgumentError as error:
+ if 'Argument to CollectionReader' in str(error):
+ logger.error("error reading collection: {}".format(error))
+ return 1
+ else:
+ raise
return 0
+ try:
+ reader = arvados.CollectionReader(col_loc, num_retries=args.retries)
+ except Exception as error:
+ logger.error("failed to read collection: {}".format(error))
+ return 1
+
# Scan the collection. Make an array of (stream, file, local
# destination filename) tuples, and add up total size to extract.
todo = []
if args.hash:
digestor = hashlib.new(args.hash)
try:
- with s.open(f.name, 'r') as file_reader:
+ with s.open(f.name, 'rb') as file_reader:
for data in file_reader.readall():
if outfile:
outfile.write(data)
def files_in_collection(c):
# Sort first by file type, then alphabetically by file path.
- for i in sorted(c.keys(),
+ for i in sorted(list(c.keys()),
key=lambda k: (
isinstance(c[k], arvados.collection.Subcollection),
k.upper())):
elif isinstance(c[i], arvados.collection.Subcollection):
for s, f in files_in_collection(c[i]):
yield (s, f)
+
+def write_block_or_manifest(dest, src, api_client, args):
+ if '+A' in src:
+ # block locator
+ kc = KeepClient(api_client=api_client)
+ dest.write(kc.get(src, num_retries=args.retries))
+ else:
+ # collection UUID or portable data hash
+ reader = arvados.CollectionReader(src, num_retries=args.retries)
+ dest.write(reader.manifest_text(strip=args.strip_manifest).encode())