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