sdk/cwl/arvados_cwl/pathmapper.py

   1 import re
   2 import logging
   3 import uuid
   4 import os
   5
   6 import arvados.commands.run
   7 import arvados.collection
   8
   9 from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
  10 from cwltool.workflow import WorkflowException
  11
  12 logger = logging.getLogger('arvados.cwl-runner')
  13
  14 class ArvPathMapper(PathMapper):
  15     """Convert container-local paths to and from Keep collection ids."""
  16
  17     pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
  18     pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$')
  19
  20     def __init__(self, arvrunner, referenced_files, input_basedir,
  21                  collection_pattern, file_pattern, name=None, **kwargs):
  22         self.arvrunner = arvrunner
  23         self.input_basedir = input_basedir
  24         self.collection_pattern = collection_pattern
  25         self.file_pattern = file_pattern
  26         self.name = name
  27         super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None)
  28
  29     def visit(self, srcobj, uploadfiles):
  30         src = srcobj["location"]
  31         if srcobj["class"] == "File":
  32             if "#" in src:
  33                 src = src[:src.index("#")]
  34             if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src):
  35                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "File")
  36             if src not in self._pathmap:
  37                 # Local FS ref, may need to be uploaded or may be on keep
  38                 # mount.
  39                 ab = abspath(src, self.input_basedir)
  40                 st = arvados.commands.run.statfile("", ab, fnPattern="keep:%s/%s")
  41                 if isinstance(st, arvados.commands.run.UploadFile):
  42                     uploadfiles.add((src, ab, st))
  43                 elif isinstance(st, arvados.commands.run.ArvFile):
  44                     self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
  45                 elif src.startswith("_:"):
  46                     if "contents" in srcobj:
  47                         pass
  48                     else:
  49                         raise WorkflowException("File literal '%s' is missing contents" % src)
  50                 elif src.startswith("arvwf:"):
  51                     self._pathmap[src] = MapperEnt(src, src, "File")
  52                 else:
  53                     raise WorkflowException("Input file path '%s' is invalid" % st)
  54             if "secondaryFiles" in srcobj:
  55                 for l in srcobj["secondaryFiles"]:
  56                     self.visit(l, uploadfiles)
  57         elif srcobj["class"] == "Directory":
  58             if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  59                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "Directory")
  60             for l in srcobj.get("listing", []):
  61                 self.visit(l, uploadfiles)
  62
  63     def addentry(self, obj, c, path, subdirs):
  64         if obj["location"] in self._pathmap:
  65             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  66             if srcpath == "":
  67                 srcpath = "."
  68             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  69             for l in obj.get("secondaryFiles", []):
  70                 self.addentry(l, c, path, subdirs)
  71         elif obj["class"] == "Directory":
  72             for l in obj["listing"]:
  73                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  74             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  75         elif obj["location"].startswith("_:") and "contents" in obj:
  76             with c.open(path + "/" + obj["basename"], "w") as f:
  77                 f.write(obj["contents"].encode("utf-8"))
  78         else:
  79             raise WorkflowException("Don't know what to do with '%s'" % obj["location"])
  80
  81     def setup(self, referenced_files, basedir):
  82         # type: (List[Any], unicode) -> None
  83         uploadfiles = set()
  84
  85         for k,v in self.arvrunner.get_uploaded().iteritems():
  86             self._pathmap[k] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File")
  87
  88         for srcobj in referenced_files:
  89             self.visit(srcobj, uploadfiles)
  90
  91         if uploadfiles:
  92             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
  93                                              self.arvrunner.api,
  94                                              dry_run=False,
  95                                              num_retries=self.arvrunner.num_retries,
  96                                              fnPattern="keep:%s/%s",
  97                                              name=self.name,
  98                                              project=self.arvrunner.project_uuid)
  99
 100         for src, ab, st in uploadfiles:
 101             self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
 102             self.arvrunner.add_uploaded(src, self._pathmap[src])
 103
 104         for srcobj in referenced_files:
 105             if srcobj["class"] == "Directory":
 106                 if srcobj["location"] not in self._pathmap:
 107                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
 108                                                       keep_client=self.arvrunner.keep_client,
 109                                                       num_retries=self.arvrunner.num_retries)
 110                     subdirs = []
 111                     for l in srcobj["listing"]:
 112                         self.addentry(l, c, ".", subdirs)
 113
 114                     check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 115                     if not check["items"]:
 116                         c.save_new(owner_uuid=self.arvrunner.project_uuid)
 117
 118                     ab = self.collection_pattern % c.portable_data_hash()
 119                     self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "Directory")
 120                     for loc, sub in subdirs:
 121                         ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 122                         self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 123             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
 124                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 125
 126                 c = arvados.collection.Collection(api_client=self.arvrunner.api,
 127                                                   keep_client=self.arvrunner.keep_client,
 128                                                   num_retries=self.arvrunner.num_retries                                                  )
 129                 subdirs = []
 130                 self.addentry(srcobj, c, ".", subdirs)
 131
 132                 check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 133                 if not check["items"]:
 134                     c.save_new(owner_uuid=self.arvrunner.project_uuid)
 135
 136                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
 137                 self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File")
 138                 if srcobj.get("secondaryFiles"):
 139                     ab = self.collection_pattern % c.portable_data_hash()
 140                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory")
 141                 for loc, sub in subdirs:
 142                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 143                     self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 144
 145         self.keepdir = None
 146
 147     def reversemap(self, target):
 148         if target.startswith("keep:"):
 149             return (target, target)
 150         elif self.keepdir and target.startswith(self.keepdir):
 151             return (target, "keep:" + target[len(self.keepdir)+1:])
 152         else:
 153             return super(ArvPathMapper, self).reversemap(target)
 154
 155 class StagingPathMapper(PathMapper):
 156     _follow_dirs = True
 157
 158     def visit(self, obj, stagedir, basedir, copy=False):
 159         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
 160         loc = obj["location"]
 161         tgt = os.path.join(stagedir, obj["basename"])
 162         if obj["class"] == "Directory":
 163             self._pathmap[loc] = MapperEnt(loc, tgt, "Directory")
 164             if loc.startswith("_:") or self._follow_dirs:
 165                 self.visitlisting(obj.get("listing", []), tgt, basedir)
 166         elif obj["class"] == "File":
 167             if loc in self._pathmap:
 168                 return
 169             if "contents" in obj and loc.startswith("_:"):
 170                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
 171             else:
 172                 if copy:
 173                     self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile")
 174                 else:
 175                     self._pathmap[loc] = MapperEnt(loc, tgt, "File")
 176                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 177
 178
 179 class InitialWorkDirPathMapper(StagingPathMapper):
 180     def setup(self, referenced_files, basedir):
 181         # type: (List[Any], unicode) -> None
 182
 183         # Go through each file and set the target to its own directory along
 184         # with any secondary files.
 185         self.visitlisting(referenced_files, self.stagedir, basedir)
 186
 187         for path, (ab, tgt, type) in self._pathmap.items():
 188             if type in ("File", "Directory") and ab.startswith("keep:"):
 189                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)
 190
 191
 192 class FinalOutputPathMapper(StagingPathMapper):
 193     _follow_dirs = False
 194     def setup(self, referenced_files, basedir):
 195         # type: (List[Any], unicode) -> None
 196         self.visitlisting(referenced_files, self.stagedir, basedir)