sdk/python/arvados/commands/ls.py

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