import logging
import uuid
import os
+import urllib
import arvados.commands.run
import arvados.collection
"""Convert container-local paths to and from Keep collection ids."""
pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
- pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$')
+ pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.*)?$')
def __init__(self, arvrunner, referenced_files, input_basedir,
collection_pattern, file_pattern, name=None, **kwargs):
src = src[:src.index("#")]
if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
- self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], srcobj["class"], True)
+ self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), srcobj["class"], True)
if src not in self._pathmap:
- # Local FS ref, may need to be uploaded or may be on keep
- # mount.
- ab = abspath(src, self.input_basedir)
- st = arvados.commands.run.statfile("", ab,
- fnPattern="keep:%s/%s",
- dirPattern="keep:%s/%s")
- with SourceLine(srcobj, "location", WorkflowException):
- if isinstance(st, arvados.commands.run.UploadFile):
- uploadfiles.add((src, ab, st))
- elif isinstance(st, arvados.commands.run.ArvFile):
- self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File", True)
- elif src.startswith("_:"):
- if "contents" in srcobj:
- pass
+ if src.startswith("file:"):
+ # Local FS ref, may need to be uploaded or may be on keep
+ # mount.
+ ab = abspath(src, self.input_basedir)
+ st = arvados.commands.run.statfile("", ab,
+ fnPattern="keep:%s/%s",
+ dirPattern="keep:%s/%s")
+ with SourceLine(srcobj, "location", WorkflowException):
+ if isinstance(st, arvados.commands.run.UploadFile):
+ uploadfiles.add((src, ab, st))
+ elif isinstance(st, arvados.commands.run.ArvFile):
+ self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.unquote(st.fn[5:]), "File", True)
else:
- raise WorkflowException("File literal '%s' is missing contents" % src)
- elif src.startswith("arvwf:"):
- self._pathmap[src] = MapperEnt(src, src, "File", True)
- else:
- raise WorkflowException("Input file path '%s' is invalid" % st)
+ raise WorkflowException("Input file path '%s' is invalid" % st)
+ elif src.startswith("_:"):
+ if srcobj["class"] == "File" and "contents" not in srcobj:
+ raise WorkflowException("File literal '%s' is missing `contents`" % src)
+ if srcobj["class"] == "Directory" and "listing" not in srcobj:
+ raise WorkflowException("Directory literal '%s' is missing `listing`" % src)
+ else:
+ self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True)
with SourceLine(srcobj, "secondaryFiles", WorkflowException):
for l in srcobj.get("secondaryFiles", []):
loc = k["location"]
if loc in already_uploaded:
v = already_uploaded[loc]
- self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File", True)
+ self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % urllib.unquote(v.resolved[5:]), "File", True)
for srcobj in referenced_files:
self.visit(srcobj, uploadfiles)
project=self.arvrunner.project_uuid)
for src, ab, st in uploadfiles:
- self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.keepref,
+ self._pathmap[src] = MapperEnt(urllib.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:],
"Directory" if os.path.isdir(ab) else "File", True)
self.arvrunner.add_uploaded(src, self._pathmap[src])