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                 else:
  51                     raise WorkflowException("Input file path '%s' is invalid" % st)
  52             if "secondaryFiles" in srcobj:
  53                 for l in srcobj["secondaryFiles"]:
  54                     self.visit(l, uploadfiles)
  55         elif srcobj["class"] == "Directory":
  56             if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  57                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "Directory")
  58             for l in srcobj.get("listing", []):
  59                 self.visit(l, uploadfiles)
  60
  61     def addentry(self, obj, c, path, subdirs):
  62         if obj["location"] in self._pathmap:
  63             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  64             if srcpath == "":
  65                 srcpath = "."
  66             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  67             for l in obj.get("secondaryFiles", []):
  68                 self.addentry(l, c, path, subdirs)
  69         elif obj["class"] == "Directory":
  70             for l in obj["listing"]:
  71                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  72             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  73         elif obj["location"].startswith("_:") and "contents" in obj:
  74             with c.open(path + "/" + obj["basename"], "w") as f:
  75                 f.write(obj["contents"].encode("utf-8"))
  76         else:
  77             raise WorkflowException("Don't know what to do with '%s'" % obj["location"])
  78
  79     def setup(self, referenced_files, basedir):
  80         # type: (List[Any], unicode) -> None
  81         uploadfiles = set()
  82
  83         for k,v in self.arvrunner.get_uploaded().iteritems():
  84             self._pathmap[k] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File")
  85
  86         for srcobj in referenced_files:
  87             self.visit(srcobj, uploadfiles)
  88
  89         if uploadfiles:
  90             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
  91                                              self.arvrunner.api,
  92                                              dry_run=False,
  93                                              num_retries=self.arvrunner.num_retries,
  94                                              fnPattern="keep:%s/%s",
  95                                              name=self.name,
  96                                              project=self.arvrunner.project_uuid)
  97
  98         for src, ab, st in uploadfiles:
  99             self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
 100             self.arvrunner.add_uploaded(src, self._pathmap[src])
 101
 102         for srcobj in referenced_files:
 103             if srcobj["class"] == "Directory":
 104                 if srcobj["location"] not in self._pathmap:
 105                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
 106                                                       keep_client=self.arvrunner.keep_client,
 107                                                       num_retries=self.arvrunner.num_retries)
 108                     subdirs = []
 109                     for l in srcobj["listing"]:
 110                         self.addentry(l, c, ".", subdirs)
 111
 112                     check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 113                     if not check["items"]:
 114                         c.save_new(owner_uuid=self.arvrunner.project_uuid)
 115
 116                     ab = self.collection_pattern % c.portable_data_hash()
 117                     self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "Directory")
 118                     for loc, sub in subdirs:
 119                         ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 120                         self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 121             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
 122                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 123
 124                 c = arvados.collection.Collection(api_client=self.arvrunner.api,
 125                                                   keep_client=self.arvrunner.keep_client,
 126                                                   num_retries=self.arvrunner.num_retries                                                  )
 127                 subdirs = []
 128                 self.addentry(srcobj, c, ".", subdirs)
 129
 130                 check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 131                 if not check["items"]:
 132                     c.save_new(owner_uuid=self.arvrunner.project_uuid)
 133
 134                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
 135                 self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File")
 136                 if srcobj.get("secondaryFiles"):
 137                     ab = self.collection_pattern % c.portable_data_hash()
 138                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory")
 139                 for loc, sub in subdirs:
 140                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 141                     self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 142
 143         self.keepdir = None
 144
 145     def reversemap(self, target):
 146         if target.startswith("keep:"):
 147             return (target, target)
 148         elif self.keepdir and target.startswith(self.keepdir):
 149             return (target, "keep:" + target[len(self.keepdir)+1:])
 150         else:
 151             return super(ArvPathMapper, self).reversemap(target)
 152
 153 class StagingPathMapper(PathMapper):
 154     _follow_dirs = True
 155
 156     def visit(self, obj, stagedir, basedir, copy=False):
 157         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
 158         loc = obj["location"]
 159         tgt = os.path.join(stagedir, obj["basename"])
 160         if obj["class"] == "Directory":
 161             self._pathmap[loc] = MapperEnt(loc, tgt, "Directory")
 162             if loc.startswith("_:") or self._follow_dirs:
 163                 self.visitlisting(obj.get("listing", []), tgt, basedir)
 164         elif obj["class"] == "File":
 165             if loc in self._pathmap:
 166                 return
 167             if "contents" in obj and loc.startswith("_:"):
 168                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
 169             else:
 170                 if copy:
 171                     self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile")
 172                 else:
 173                     self._pathmap[loc] = MapperEnt(loc, tgt, "File")
 174                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 175
 176
 177 class InitialWorkDirPathMapper(StagingPathMapper):
 178     def setup(self, referenced_files, basedir):
 179         # type: (List[Any], unicode) -> None
 180
 181         # Go through each file and set the target to its own directory along
 182         # with any secondary files.
 183         self.visitlisting(referenced_files, self.stagedir, basedir)
 184
 185         for path, (ab, tgt, type) in self._pathmap.items():
 186             if type in ("File", "Directory") and ab.startswith("keep:"):
 187                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)
 188
 189
 190 class FinalOutputPathMapper(StagingPathMapper):
 191     _follow_dirs = False
 192     def setup(self, referenced_files, basedir):
 193         # type: (List[Any], unicode) -> None
 194         self.visitlisting(referenced_files, self.stagedir, basedir)