sdk/cwl/arvados_cwl/pathmapper.py

   1 import re
   2 import logging
   3 import uuid
   4 import os
   5 import urllib
   6
   7 import arvados.commands.run
   8 import arvados.collection
   9
  10 from schema_salad.sourceline import SourceLine
  11
  12 from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
  13 from cwltool.workflow import WorkflowException
  14
  15 logger = logging.getLogger('arvados.cwl-runner')
  16
  17 class ArvPathMapper(PathMapper):
  18     """Convert container-local paths to and from Keep collection ids."""
  19
  20     pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
  21     pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.*)?$')
  22
  23     def __init__(self, arvrunner, referenced_files, input_basedir,
  24                  collection_pattern, file_pattern, name=None, **kwargs):
  25         self.arvrunner = arvrunner
  26         self.input_basedir = input_basedir
  27         self.collection_pattern = collection_pattern
  28         self.file_pattern = file_pattern
  29         self.name = name
  30         super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None)
  31
  32     def visit(self, srcobj, uploadfiles):
  33         src = srcobj["location"]
  34         if srcobj["class"] == "File":
  35             if "#" in src:
  36                 src = src[:src.index("#")]
  37             if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src):
  38                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), "File")
  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, fnPattern="keep:%s/%s")
  44                 with SourceLine(srcobj, "location", WorkflowException):
  45                     if isinstance(st, arvados.commands.run.UploadFile):
  46                         uploadfiles.add((src, ab, st))
  47                     elif isinstance(st, arvados.commands.run.ArvFile):
  48                         self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.unquote(st.fn[5:]), "File")
  49                     elif src.startswith("_:"):
  50                         if "contents" in srcobj:
  51                             pass
  52                         else:
  53                             raise WorkflowException("File literal '%s' is missing contents" % src)
  54                     elif src.startswith("arvwf:"):
  55                         self._pathmap[src] = MapperEnt(src, src, "File")
  56                     else:
  57                         raise WorkflowException("Input file path '%s' is invalid" % st)
  58             if "secondaryFiles" in srcobj:
  59                 for l in srcobj["secondaryFiles"]:
  60                     self.visit(l, uploadfiles)
  61         elif srcobj["class"] == "Directory":
  62             if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  63                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), "Directory")
  64             for l in srcobj.get("listing", []):
  65                 self.visit(l, uploadfiles)
  66
  67     def addentry(self, obj, c, path, subdirs):
  68         if obj["location"] in self._pathmap:
  69             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  70             if srcpath == "":
  71                 srcpath = "."
  72             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  73             for l in obj.get("secondaryFiles", []):
  74                 self.addentry(l, c, path, subdirs)
  75         elif obj["class"] == "Directory":
  76             for l in obj["listing"]:
  77                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  78             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  79         elif obj["location"].startswith("_:") and "contents" in obj:
  80             with c.open(path + "/" + obj["basename"], "w") as f:
  81                 f.write(obj["contents"].encode("utf-8"))
  82         else:
  83             raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
  84
  85     def setup(self, referenced_files, basedir):
  86         # type: (List[Any], unicode) -> None
  87         uploadfiles = set()
  88
  89         already_uploaded = self.arvrunner.get_uploaded()
  90         for k in referenced_files:
  91             loc = k["location"]
  92             if loc in already_uploaded:
  93                 v = already_uploaded[loc]
  94                 self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % urllib.unquote(v.resolved[5:]), "File")
  95
  96         for srcobj in referenced_files:
  97             self.visit(srcobj, uploadfiles)
  98
  99         if uploadfiles:
 100             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
 101                                              self.arvrunner.api,
 102                                              dry_run=False,
 103                                              num_retries=self.arvrunner.num_retries,
 104                                              fnPattern="keep:%s/%s",
 105                                              name=self.name,
 106                                              project=self.arvrunner.project_uuid)
 107
 108         for src, ab, st in uploadfiles:
 109             self._pathmap[src] = MapperEnt(urllib.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:], "File")
 110             self.arvrunner.add_uploaded(src, self._pathmap[src])
 111
 112         for srcobj in referenced_files:
 113             if srcobj["class"] == "Directory":
 114                 if srcobj["location"] not in self._pathmap:
 115                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
 116                                                       keep_client=self.arvrunner.keep_client,
 117                                                       num_retries=self.arvrunner.num_retries)
 118                     subdirs = []
 119                     for l in srcobj["listing"]:
 120                         self.addentry(l, 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.collection_pattern % c.portable_data_hash()
 127                     self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "Directory")
 128                     for loc, sub in subdirs:
 129                         ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 130                         self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 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                 subdirs = []
 138                 self.addentry(srcobj, c, ".", subdirs)
 139
 140                 check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 141                 if not check["items"]:
 142                     c.save_new(owner_uuid=self.arvrunner.project_uuid)
 143
 144                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
 145                 self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File")
 146                 if srcobj.get("secondaryFiles"):
 147                     ab = self.collection_pattern % c.portable_data_hash()
 148                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory")
 149                 for loc, sub in subdirs:
 150                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 151                     self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
 152
 153         self.keepdir = None
 154
 155     def reversemap(self, target):
 156         if target.startswith("keep:"):
 157             return (target, target)
 158         elif self.keepdir and target.startswith(self.keepdir):
 159             return (target, "keep:" + target[len(self.keepdir)+1:])
 160         else:
 161             return super(ArvPathMapper, self).reversemap(target)
 162
 163 class StagingPathMapper(PathMapper):
 164     _follow_dirs = True
 165
 166     def visit(self, obj, stagedir, basedir, copy=False):
 167         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
 168         loc = obj["location"]
 169         tgt = os.path.join(stagedir, obj["basename"])
 170         if obj["class"] == "Directory":
 171             self._pathmap[loc] = MapperEnt(loc, tgt, "Directory")
 172             if loc.startswith("_:") or self._follow_dirs:
 173                 self.visitlisting(obj.get("listing", []), tgt, basedir)
 174         elif obj["class"] == "File":
 175             if loc in self._pathmap:
 176                 return
 177             if "contents" in obj and loc.startswith("_:"):
 178                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
 179             else:
 180                 if copy:
 181                     self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile")
 182                 else:
 183                     self._pathmap[loc] = MapperEnt(loc, tgt, "File")
 184                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 185
 186
 187 class VwdPathMapper(StagingPathMapper):
 188     def setup(self, referenced_files, basedir):
 189         # type: (List[Any], unicode) -> None
 190
 191         # Go through each file and set the target to its own directory along
 192         # with any secondary files.
 193         self.visitlisting(referenced_files, self.stagedir, basedir)
 194
 195         for path, (ab, tgt, type) in self._pathmap.items():
 196             if type in ("File", "Directory") and ab.startswith("keep:"):
 197                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)
 198
 199
 200 class NoFollowPathMapper(StagingPathMapper):
 201     _follow_dirs = False
 202     def setup(self, referenced_files, basedir):
 203         # type: (List[Any], unicode) -> None
 204         self.visitlisting(referenced_files, self.stagedir, basedir)