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