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