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 "#" in src:
  35             src = src[:src.index("#")]
  36
  37         if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
  38             self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), srcobj["class"], True)
  39
  40         if src not in self._pathmap:
  41             if src.startswith("file:"):
  42                 # Local FS ref, may need to be uploaded or may be on keep
  43                 # mount.
  44                 ab = abspath(src, self.input_basedir)
  45                 st = arvados.commands.run.statfile("", ab,
  46                                                    fnPattern="keep:%s/%s",
  47                                                    dirPattern="keep:%s/%s")
  48                 with SourceLine(srcobj, "location", WorkflowException):
  49                     if isinstance(st, arvados.commands.run.UploadFile):
  50                         uploadfiles.add((src, ab, st))
  51                     elif isinstance(st, arvados.commands.run.ArvFile):
  52                         self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.unquote(st.fn[5:]), "File", True)
  53                     else:
  54                         raise WorkflowException("Input file path '%s' is invalid" % st)
  55             elif src.startswith("_:"):
  56                 if srcobj["class"] == "File" and "contents" not in srcobj:
  57                     raise WorkflowException("File literal '%s' is missing `contents`" % src)
  58                 if srcobj["class"] == "Directory" and "listing" not in srcobj:
  59                     raise WorkflowException("Directory literal '%s' is missing `listing`" % src)
  60             else:
  61                 self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True)
  62
  63         with SourceLine(srcobj, "secondaryFiles", WorkflowException):
  64             for l in srcobj.get("secondaryFiles", []):
  65                 self.visit(l, uploadfiles)
  66         with SourceLine(srcobj, "listing", WorkflowException):
  67             for l in srcobj.get("listing", []):
  68                 self.visit(l, uploadfiles)
  69
  70     def addentry(self, obj, c, path, subdirs):
  71         if obj["location"] in self._pathmap:
  72             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
  73             if srcpath == "":
  74                 srcpath = "."
  75             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
  76             for l in obj.get("secondaryFiles", []):
  77                 self.addentry(l, c, path, subdirs)
  78         elif obj["class"] == "Directory":
  79             for l in obj.get("listing", []):
  80                 self.addentry(l, c, path + "/" + obj["basename"], subdirs)
  81             subdirs.append((obj["location"], path + "/" + obj["basename"]))
  82         elif obj["location"].startswith("_:") and "contents" in obj:
  83             with c.open(path + "/" + obj["basename"], "w") as f:
  84                 f.write(obj["contents"].encode("utf-8"))
  85         else:
  86             raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
  87
  88     def setup(self, referenced_files, basedir):
  89         # type: (List[Any], unicode) -> None
  90         uploadfiles = set()
  91
  92         already_uploaded = self.arvrunner.get_uploaded()
  93         for k in referenced_files:
  94             loc = k["location"]
  95             if loc in already_uploaded:
  96                 v = already_uploaded[loc]
  97                 self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % urllib.unquote(v.resolved[5:]), "File", True)
  98
  99         for srcobj in referenced_files:
 100             self.visit(srcobj, uploadfiles)
 101
 102         if uploadfiles:
 103             arvados.commands.run.uploadfiles([u[2] for u in uploadfiles],
 104                                              self.arvrunner.api,
 105                                              dry_run=False,
 106                                              num_retries=self.arvrunner.num_retries,
 107                                              fnPattern="keep:%s/%s",
 108                                              name=self.name,
 109                                              project=self.arvrunner.project_uuid)
 110
 111         for src, ab, st in uploadfiles:
 112             self._pathmap[src] = MapperEnt(urllib.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:],
 113                                            "Directory" if os.path.isdir(ab) else "File", True)
 114             self.arvrunner.add_uploaded(src, self._pathmap[src])
 115
 116         for srcobj in referenced_files:
 117             subdirs = []
 118             if srcobj["class"] == "Directory":
 119                 if srcobj["location"] not in self._pathmap:
 120                     c = arvados.collection.Collection(api_client=self.arvrunner.api,
 121                                                       keep_client=self.arvrunner.keep_client,
 122                                                       num_retries=self.arvrunner.num_retries)
 123                     for l in srcobj.get("listing", []):
 124                         self.addentry(l, c, ".", subdirs)
 125
 126                     check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries)
 127                     if not check["items"]:
 128                         c.save_new(owner_uuid=self.arvrunner.project_uuid)
 129
 130                     ab = self.collection_pattern % c.portable_data_hash()
 131                     self._pathmap[srcobj["location"]] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
 132             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
 133                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 134
 135                 c = arvados.collection.Collection(api_client=self.arvrunner.api,
 136                                                   keep_client=self.arvrunner.keep_client,
 137                                                   num_retries=self.arvrunner.num_retries                                                  )
 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("keep:%s/%s" % (c.portable_data_hash(), srcobj["basename"]),
 146                                                               ab, "File", True)
 147                 if srcobj.get("secondaryFiles"):
 148                     ab = self.collection_pattern % c.portable_data_hash()
 149                     self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
 150
 151             if subdirs:
 152                 for loc, sub in subdirs:
 153                     # subdirs will all start with "./", strip it off
 154                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
 155                     self._pathmap[loc] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), sub[2:]),
 156                                                    ab, "Directory", True)
 157
 158         self.keepdir = None
 159
 160     def reversemap(self, target):
 161         if target.startswith("keep:"):
 162             return (target, target)
 163         elif self.keepdir and target.startswith(self.keepdir):
 164             return (target, "keep:" + target[len(self.keepdir)+1:])
 165         else:
 166             return super(ArvPathMapper, self).reversemap(target)
 167
 168 class StagingPathMapper(PathMapper):
 169     _follow_dirs = True
 170
 171     def visit(self, obj, stagedir, basedir, copy=False, staged=False):
 172         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
 173         loc = obj["location"]
 174         tgt = os.path.join(stagedir, obj["basename"])
 175         if obj["class"] == "Directory":
 176             self._pathmap[loc] = MapperEnt(loc, tgt, "Directory", staged)
 177             if loc.startswith("_:") or self._follow_dirs:
 178                 self.visitlisting(obj.get("listing", []), tgt, basedir)
 179         elif obj["class"] == "File":
 180             if loc in self._pathmap:
 181                 return
 182             if "contents" in obj and loc.startswith("_:"):
 183                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile", staged)
 184             else:
 185                 if copy:
 186                     self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile", staged)
 187                 else:
 188                     self._pathmap[loc] = MapperEnt(loc, tgt, "File", staged)
 189                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 190
 191
 192 class VwdPathMapper(StagingPathMapper):
 193     def setup(self, referenced_files, basedir):
 194         # type: (List[Any], unicode) -> None
 195
 196         # Go through each file and set the target to its own directory along
 197         # with any secondary files.
 198         self.visitlisting(referenced_files, self.stagedir, basedir)
 199
 200         for path, (ab, tgt, type, staged) in self._pathmap.items():
 201             if type in ("File", "Directory") and ab.startswith("keep:"):
 202                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type, staged)
 203
 204
 205 class NoFollowPathMapper(StagingPathMapper):
 206     _follow_dirs = False
 207     def setup(self, referenced_files, basedir):
 208         # type: (List[Any], unicode) -> None
 209         self.visitlisting(referenced_files, self.stagedir, basedir)