sdk/cwl/arvados_cwl/fsaccess.py

   1 import fnmatch
   2 import os
   3 import errno
   4
   5 import cwltool.stdfsaccess
   6 from cwltool.pathmapper import abspath
   7
   8 import arvados.util
   9 import arvados.collection
  10 import arvados.arvfile
  11
  12 class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
  13     """Implement the cwltool FsAccess interface for Arvados Collections."""
  14
  15     def __init__(self, basedir, api_client=None):
  16         super(CollectionFsAccess, self).__init__(basedir)
  17         self.api_client = api_client
  18         self.collections = {}
  19
  20     def get_collection(self, path):
  21         p = path.split("/")
  22         if p[0].startswith("keep:") and arvados.util.keep_locator_pattern.match(p[0][5:]):
  23             pdh = p[0][5:]
  24             if pdh not in self.collections:
  25                 self.collections[pdh] = arvados.collection.CollectionReader(pdh, api_client=self.api_client)
  26             return (self.collections[pdh], "/".join(p[1:]))
  27         else:
  28             return (None, path)
  29
  30     def _match(self, collection, patternsegments, parent):
  31         if not patternsegments:
  32             return []
  33
  34         if not isinstance(collection, arvados.collection.RichCollectionBase):
  35             return []
  36
  37         ret = []
  38         # iterate over the files and subcollections in 'collection'
  39         for filename in collection:
  40             if patternsegments[0] == '.':
  41                 # Pattern contains something like "./foo" so just shift
  42                 # past the "./"
  43                 ret.extend(self._match(collection, patternsegments[1:], parent))
  44             elif fnmatch.fnmatch(filename, patternsegments[0]):
  45                 cur = os.path.join(parent, filename)
  46                 if len(patternsegments) == 1:
  47                     ret.append(cur)
  48                 else:
  49                     ret.extend(self._match(collection[filename], patternsegments[1:], cur))
  50         return ret
  51
  52     def glob(self, pattern):
  53         collection, rest = self.get_collection(pattern)
  54         if collection and not rest:
  55             return [pattern]
  56         patternsegments = rest.split("/")
  57         return self._match(collection, patternsegments, "keep:" + collection.manifest_locator())
  58
  59     def open(self, fn, mode):
  60         collection, rest = self.get_collection(fn)
  61         if collection:
  62             return collection.open(rest, mode)
  63         else:
  64             return super(CollectionFsAccess, self).open(self._abs(fn), mode)
  65
  66     def exists(self, fn):
  67         collection, rest = self.get_collection(fn)
  68         if collection:
  69             return collection.exists(rest)
  70         else:
  71             return super(CollectionFsAccess, self).exists(fn)
  72
  73     def isfile(self, fn):  # type: (unicode) -> bool
  74         collection, rest = self.get_collection(fn)
  75         if collection:
  76             if rest:
  77                 return isinstance(collection.find(rest), arvados.arvfile.ArvadosFile)
  78             else:
  79                 return False
  80         else:
  81             return super(CollectionFsAccess, self).isfile(fn)
  82
  83     def isdir(self, fn):  # type: (unicode) -> bool
  84         collection, rest = self.get_collection(fn)
  85         if collection:
  86             if rest:
  87                 return isinstance(collection.find(rest), arvados.collection.Collection)
  88             else:
  89                 return True
  90         else:
  91             return super(CollectionFsAccess, self).isdir(fn)
  92
  93     def listdir(self, fn):  # type: (unicode) -> List[unicode]
  94         collection, rest = self.get_collection(fn)
  95         if collection:
  96             if rest:
  97                 dir = collection.find(rest)
  98             else:
  99                 dir = collection
 100             if dir is None:
 101                 raise IOError(errno.ENOENT, "Directory '%s' in '%s' not found" % (rest, collection.portable_data_hash()))
 102             if not isinstance(dir, arvados.collection.Collection):
 103                 raise IOError(errno.ENOENT, "Path '%s' in '%s' is not a Directory" % (rest, collection.portable_data_hash()))
 104             return [abspath(l, fn) for l in dir.keys()]
 105         else:
 106             return super(CollectionFsAccess, self).listdir(fn)
 107
 108     def join(self, path, *paths): # type: (unicode, *unicode) -> unicode
 109         if paths and paths[-1].startswith("keep:") and arvados.util.keep_locator_pattern.match(paths[-1][5:]):
 110             return paths[-1]
 111         return os.path.join(path, *paths)
 112
 113     def realpath(self, path):
 114         if path.startswith("$(task.tmpdir)") or path.startswith("$(task.outdir)"):
 115             return path
 116         collection, rest = self.get_collection(path)
 117         if collection:
 118             return path
 119         else:
 120             return os.path.realpath(path)