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 schema_salad.sourceline import SourceLine
  10
  11 from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
  12 from cwltool.workflow import WorkflowException
  13
  14 logger = logging.getLogger('arvados.cwl-runner')
  15
  16 class ArvPathMapper(PathMapper):
  17     """Convert container-local paths to and from Keep collection ids."""
  18
  19     pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
  20     pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$')
  21
  22     def __init__(self, arvrunner, referenced_files, input_basedir,
  23                  collection_pattern, file_pattern, name=None, **kwargs):
  24         self.arvrunner = arvrunner
  25         self.input_basedir = input_basedir
  26         self.collection_pattern = collection_pattern
  27         self.file_pattern = file_pattern
  28         self.name = name
  29         super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None)
  30
  31     def visit(self, srcobj, uploadfiles):
  32         src = srcobj["location"]
  33         if "#" in src:
  34             src = src[:src.index("#")]
  35
  36         if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  37             self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], srcobj["class"], True)
  38
  39         if src not in self._pathmap:
  40             # Local FS ref, may need to be uploaded or may be on keep
  41             # mount.
  42             ab = abspath(src, self.input_basedir)
  43             st = arvados.commands.run.statfile("", ab,
  44                                                fnPattern="keep:%s/%s",
  45                                                dirPattern="keep:%s/%s")
  46             with SourceLine(srcobj, "location", WorkflowException):
  47                 if isinstance(st, arvados.commands.run.UploadFile):
  48                     uploadfiles.add((src, ab, st))
  49                 elif isinstance(st, arvados.commands.run.ArvFile):
  50                     self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File", True)
  51                 elif src.startswith("_:"):
  52                     if "contents" in srcobj:
  53                         pass
  54                     else:
  55                         raise WorkflowException("File literal '%s' is missing contents" % src)
  56                 elif src.startswith("arvwf:"):
  57                     self._pathmap[src] = MapperEnt(src, src, "File", True)
  58                 else:
  59                     raise WorkflowException("Input file path '%s' is invalid" % st)
  60
  61         with SourceLine(srcobj, "secondaryFiles", WorkflowException):
  62             for l in srcobj.get("secondaryFiles", []):
  63                 self.visit(l, uploadfiles)
  64         with SourceLine(srcobj, "listing", WorkflowException):
  65             for l in srcobj.get("listing", []):
  66                 self.visit(l, uploadfiles)
  67
  68     def addentry(self, obj, c, path, subdirs):
  69         if obj["location"] in self._pathmap:
  70             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  71             if srcpath == "":
  72                 srcpath = "."
  73             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  74             for l in obj.get("secondaryFiles", []):
  75                 self.addentry(l, c, path, subdirs)
  76         elif obj["class"] == "Directory":
  77             for l in obj.get("listing", []):
  78                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  79             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  80         elif obj["location"].startswith("_:") and "contents" in obj:
  81             with c.open(path + "/" + obj["basename"], "w") as f:
  82                 f.write(obj["contents"].encode("utf-8"))
  83         else:
  84             raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
  85
  86     def setup(self, referenced_files, basedir):
  87         # type: (List[Any], unicode) -> None
  88         uploadfiles = set()
  89
  90         already_uploaded = self.arvrunner.get_uploaded()
  91         for k in referenced_files:
  92             loc = k["location"]
  93             if loc in already_uploaded:
  94                 v = already_uploaded[loc]
  95                 self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File", True)
  96
  97         for srcobj in referenced_files:
  98             self.visit(srcobj, uploadfiles)
  99
 100         if uploadfiles:
 101             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
 102                                              self.arvrunner.api,
 103                                              dry_run=False,
 104                                              num_retries=self.arvrunner.num_retries,
 105                                              fnPattern="keep:%s/%s",
 106                                              name=self.name,
 107                                              project=self.arvrunner.project_uuid)
 108
 109         for src, ab, st in uploadfiles:
 110             self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.keepref,
 111                                            "Directory" if os.path.isdir(ab) else "File", True)
 112             self.arvrunner.add_uploaded(src, self._pathmap[src])
 113
 114         for srcobj in referenced_files:
 115             subdirs = []
 116             if srcobj["class"] == "Directory":
 117                 if srcobj["location"] not in self._pathmap:
 118                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
 119                                                       keep_client=self.arvrunner.keep_client,
 120                                                       num_retries=self.arvrunner.num_retries)
 121                     for l in srcobj.get("listing", []):
 122                         self.addentry(l, c, ".", subdirs)
 123
 124                     check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 125                     if not check["items"]:
 126                         c.save_new(owner_uuid=self.arvrunner.project_uuid)
 127
 128                     ab = self.collection_pattern % c.portable_data_hash()
 129                     self._pathmap[srcobj["location"]] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
 130             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
 131                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 132
 133                 c = arvados.collection.Collection(api_client=self.arvrunner.api,
 134                                                   keep_client=self.arvrunner.keep_client,
 135                                                   num_retries=self.arvrunner.num_retries                                                  )
 136                 self.addentry(srcobj, c, ".", subdirs)
 137
 138                 check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 139                 if not check["items"]:
 140                     c.save_new(owner_uuid=self.arvrunner.project_uuid)
 141
 142                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
 143                 self._pathmap[srcobj["location"]] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), srcobj["basename"]),
 144                                                               ab, "File", True)
 145                 if srcobj.get("secondaryFiles"):
 146                     ab = self.collection_pattern % c.portable_data_hash()
 147                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
 148
 149             if subdirs:
 150                 for loc, sub in subdirs:
 151                     # subdirs will all start with "./", strip it off
 152                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 153                     self._pathmap[loc] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), sub[2:]),
 154                                                    ab, "Directory", True)
 155
 156         self.keepdir = None
 157
 158     def reversemap(self, target):
 159         if target.startswith("keep:"):
 160             return (target, target)
 161         elif self.keepdir and target.startswith(self.keepdir):
 162             return (target, "keep:" + target[len(self.keepdir)+1:])
 163         else:
 164             return super(ArvPathMapper, self).reversemap(target)
 165
 166 class StagingPathMapper(PathMapper):
 167     _follow_dirs = True
 168
 169     def visit(self, obj, stagedir, basedir, copy=False, staged=False):
 170         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
 171         loc = obj["location"]
 172         tgt = os.path.join(stagedir, obj["basename"])
 173         if obj["class"] == "Directory":
 174             self._pathmap[loc] = MapperEnt(loc, tgt, "Directory", staged)
 175             if loc.startswith("_:") or self._follow_dirs:
 176                 self.visitlisting(obj.get("listing", []), tgt, basedir)
 177         elif obj["class"] == "File":
 178             if loc in self._pathmap:
 179                 return
 180             if "contents" in obj and loc.startswith("_:"):
 181                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile", staged)
 182             else:
 183                 if copy:
 184                     self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile", staged)
 185                 else:
 186                     self._pathmap[loc] = MapperEnt(loc, tgt, "File", staged)
 187                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 188
 189
 190 class VwdPathMapper(StagingPathMapper):
 191     def setup(self, referenced_files, basedir):
 192         # type: (List[Any], unicode) -> None
 193
 194         # Go through each file and set the target to its own directory along
 195         # with any secondary files.
 196         self.visitlisting(referenced_files, self.stagedir, basedir)
 197
 198         for path, (ab, tgt, type, staged) in self._pathmap.items():
 199             if type in ("File", "Directory") and ab.startswith("keep:"):
 200                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type, staged)
 201
 202
 203 class NoFollowPathMapper(StagingPathMapper):
 204     _follow_dirs = False
 205     def setup(self, referenced_files, basedir):
 206         # type: (List[Any], unicode) -> None
 207         self.visitlisting(referenced_files, self.stagedir, basedir)