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