sdk/python/arvados/commands/ls.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import argparse
   6 import collections
   7 import logging
   8 import re
   9 import sys
  10
  11 import arvados
  12 import arvados.commands._util as arv_cmd
  13
  14 from arvados._version import __version__
  15
  16 FileInfo = collections.namedtuple('FileInfo', ['stream_name', 'name', 'size'])
  17
  18 def parse_args(args):
  19     parser = argparse.ArgumentParser(
  20         description='List contents of a manifest',
  21         parents=[arv_cmd.retry_opt])
  22
  23     parser.add_argument('locator', type=str,
  24                         help="""Collection UUID or locator, optionally with a subdir path.""")
  25     parser.add_argument('-s', action='store_true',
  26                         help="""List file sizes, in KiB.""")
  27     parser.add_argument('--version', action='version',
  28                         version="%s %s" % (sys.argv[0], __version__),
  29                         help='Print version and exit.')
  30
  31     return parser.parse_args(args)
  32
  33 def size_formatter(coll_file):
  34     return "{:>10}".format((coll_file.size + 1023) // 1024)
  35
  36 def name_formatter(coll_file):
  37     return "{}/{}".format(coll_file.stream_name, coll_file.name)
  38
  39 def main(args, stdout, stderr, api_client=None, logger=None):
  40     args = parse_args(args)
  41
  42     if api_client is None:
  43         api_client = arvados.api('v1', num_retries=args.retries)
  44
  45     if logger is None:
  46         logger = logging.getLogger('arvados.arv-ls')
  47
  48     try:
  49         r = re.search(r'^(.*?)(/.*)?$', args.locator)
  50         collection = r.group(1)
  51         get_prefix = r.group(2)
  52
  53         cr = arvados.CollectionReader(collection, api_client=api_client,
  54                                       num_retries=args.retries)
  55         if get_prefix:
  56             if get_prefix[-1] == '/':
  57                 get_prefix = get_prefix[:-1]
  58             stream_name = '.' + get_prefix
  59             reader = cr.find(stream_name)
  60             if not (isinstance(reader, arvados.CollectionReader) or
  61                     isinstance(reader, arvados.collection.Subcollection)):
  62                 logger.error("'{}' is not a subdirectory".format(get_prefix))
  63                 return 1
  64         else:
  65             stream_name = '.'
  66             reader = cr
  67     except (arvados.errors.ApiError,
  68             arvados.errors.ArgumentError,
  69             arvados.errors.NotFoundError) as error:
  70         logger.error("error fetching collection: {}".format(error))
  71         return 1
  72
  73     formatters = []
  74     if args.s:
  75         formatters.append(size_formatter)
  76     formatters.append(name_formatter)
  77
  78     for f in files_in_collection(reader, stream_name):
  79         print(*(info_func(f) for info_func in formatters), file=stdout)
  80
  81     return 0
  82
  83 def files_in_collection(c, stream_name='.'):
  84     # Sort first by file type, then alphabetically by file path.
  85     for i in sorted(c.keys(),
  86                     key=lambda k: (
  87                         isinstance(c[k], arvados.collection.Subcollection),
  88                         k.upper())):
  89         if isinstance(c[i], arvados.arvfile.ArvadosFile):
  90             yield FileInfo(stream_name=stream_name,
  91                            name=i,
  92                            size=c[i].size())
  93         elif isinstance(c[i], arvados.collection.Subcollection):
  94             for f in files_in_collection(c[i], "{}/{}".format(stream_name, i)):
  95                 yield f