sdk/cwl/arvados_cwl/pathmapper.py

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