sdk/cwl/arvados_cwl/pathmapper.py

   1 import re
   2 import logging
   3 import uuid
   4
   5 import arvados.commands.run
   6 import arvados.collection
   7
   8 from cwltool.pathmapper import PathMapper, MapperEnt, abspath
   9 from cwltool.workflow import WorkflowException
  10
  11 logger = logging.getLogger('arvados.cwl-runner')
  12
  13 class ArvPathMapper(PathMapper):
  14     """Convert container-local paths to and from Keep collection ids."""
  15
  16     pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
  17     pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$')
  18
  19     def __init__(self, arvrunner, referenced_files, input_basedir,
  20                  collection_pattern, file_pattern, name=None, **kwargs):
  21         self.arvrunner = arvrunner
  22         self.input_basedir = input_basedir
  23         self.collection_pattern = collection_pattern
  24         self.file_pattern = file_pattern
  25         self.name = name
  26         super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None)
  27
  28     def visit(self, srcobj, uploadfiles):
  29         src = srcobj["location"]
  30         if srcobj["class"] == "File":
  31             if "#" in src:
  32                 src = src[:src.index("#")]
  33             if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src):
  34                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "File")
  35             if src not in self._pathmap:
  36                 # Local FS ref, may need to be uploaded or may be on keep
  37                 # mount.
  38                 ab = abspath(src, self.input_basedir)
  39                 st = arvados.commands.run.statfile("", ab, fnPattern=self.file_pattern)
  40                 if isinstance(st, arvados.commands.run.UploadFile):
  41                     uploadfiles.add((src, ab, st))
  42                 elif isinstance(st, arvados.commands.run.ArvFile):
  43                     self._pathmap[src] = MapperEnt(ab, st.fn, "File")
  44                 elif src.startswith("_:") and "contents" in srcobj:
  45                     pass
  46                 else:
  47                     raise WorkflowException("Input file path '%s' is invalid" % st)
  48             if "secondaryFiles" in srcobj:
  49                 for l in srcobj["secondaryFiles"]:
  50                     self.visit(l, uploadfiles)
  51         elif srcobj["class"] == "Directory":
  52             if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  53                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "Directory")
  54             else:
  55                 for l in srcobj["listing"]:
  56                     self.visit(l, uploadfiles)
  57
  58     def addentry(self, obj, c, path, subdirs):
  59         if obj["location"] in self._pathmap:
  60             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  61             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  62             for l in obj.get("secondaryFiles", []):
  63                 self.addentry(l, c, path, subdirs)
  64         elif obj["class"] == "Directory":
  65             for l in obj["listing"]:
  66                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  67             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  68         elif obj["location"].startswith("_:") and "contents" in obj:
  69             with c.open(path + "/" + obj["basename"], "w") as f:
  70                 f.write(obj["contents"].encode("utf-8"))
  71         else:
  72             raise WorkflowException("Don't know what to do with '%s'" % obj["location"])
  73
  74     def setup(self, referenced_files, basedir):
  75         # type: (List[Any], unicode) -> None
  76         self._pathmap = self.arvrunner.get_uploaded()
  77         uploadfiles = set()
  78
  79         for srcobj in referenced_files:
  80             self.visit(srcobj, uploadfiles)
  81
  82         if uploadfiles:
  83             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
  84                                              self.arvrunner.api,
  85                                              dry_run=False,
  86                                              num_retries=self.arvrunner.num_retries,
  87                                              fnPattern=self.file_pattern,
  88                                              name=self.name,
  89                                              project=self.arvrunner.project_uuid)
  90
  91         for src, ab, st in uploadfiles:
  92             self._pathmap[src] = MapperEnt("keep:" + st.keepref, st.fn, "File")
  93             self.arvrunner.add_uploaded(src, self._pathmap[src])
  94
  95         for srcobj in referenced_files:
  96             if srcobj["class"] == "Directory":
  97                 if srcobj["location"] not in self._pathmap:
  98                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
  99                                                       num_retries=self.arvrunner.num_retries)
 100                     subdirs = []
 101                     for l in srcobj["listing"]:
 102                         self.addentry(l, c, ".", subdirs)
 103
 104                     check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 105                     if not check["items"]:
 106                         c.save_new(owner_uuid=self.arvrunner.project_uuid)
 107
 108                     ab = self.collection_pattern % c.portable_data_hash()
 109                     self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "Directory")
 110                     for loc, sub in subdirs:
 111                         ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 112                         self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 113             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
 114                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 115
 116                 c = arvados.collection.Collection(api_client=self.arvrunner.api,
 117                                                   num_retries=self.arvrunner.num_retries                                                  )
 118                 subdirs = []
 119                 self.addentry(srcobj, c, ".", subdirs)
 120
 121                 check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 122                 if not check["items"]:
 123                     c.save_new(owner_uuid=self.arvrunner.project_uuid)
 124
 125                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
 126                 self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File")
 127                 if srcobj.get("secondaryFiles"):
 128                     ab = self.collection_pattern % c.portable_data_hash()
 129                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory")
 130                 for loc, sub in subdirs:
 131                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 132                     self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 133
 134         self.keepdir = None
 135
 136     def reversemap(self, target):
 137         if target.startswith("keep:"):
 138             return (target, target)
 139         elif self.keepdir and target.startswith(self.keepdir):
 140             return (target, "keep:" + target[len(self.keepdir)+1:])
 141         else:
 142             return super(ArvPathMapper, self).reversemap(target)
 143
 144 class InitialWorkDirPathMapper(PathMapper):
 145     def setup(self, referenced_files, basedir):
 146         # type: (List[Any], unicode) -> None
 147
 148         # Go through each file and set the target to its own directory along
 149         # with any secondary files.
 150         stagedir = self.stagedir
 151         for fob in referenced_files:
 152             self.visit(fob, stagedir, basedir)
 153
 154         for path, (ab, tgt, type) in self._pathmap.items():
 155             if type in ("File", "Directory") and ab.startswith("keep:"):
 156                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)