crunch_scripts/decompress-all.py

   1 #!/usr/bin/env python
   2
   3 import arvados
   4 import re
   5 import subprocess
   6
   7 arvados.job_setup.one_task_per_input_file(if_sequence=0, and_end_task=True,
   8                                           input_as_path=True)
   9
  10 task = arvados.current_task()
  11
  12 input_file = arvados.gettaskparam('input')
  13
  14 result = re.match(r"(^[a-f0-9]{32}\+\d+)(\+\S+)*(/.*)(/.*)?$", input_file)
  15
  16 outdir = os.path.join(task.tmpdir, "output")
  17 os.mkdirs(outdir)
  18 os.chdir(outdir)
  19
  20 if result != None:
  21     cr = arvados.CollectionReader(re.group(1))
  22     streamname = '.'
  23     if re.group(3) != None:
  24         streamname += re.group(2)
  25         filename = re.group(3)[1:]
  26     else:
  27         filename = re.group(2)[1:]
  28
  29     os.mkdirs(streamname)
  30     os.chdir(streamname)
  31     streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
  32     filereader = stream.files()[filename]
  33     rc = subprocess.call("dtrx", "-r", "-n", arvados.get_task_param_mount('input'))
  34     if rc == 0:
  35         out.write_directory_tree(outdir, max_manifest_depth=0)
  36         arvados.task_set_output(out.finish())
  37     else:
  38         arvados.task_set_output(streamname + filereader.as_manifest()[1:])
  39 else:
  40     sys.exit(1)