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