X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/c60bff923fdd7616eeb1bd8f5998331cbd262970..5624fec61db977d386ce03ca333241c74ca251b5:/sdk/cwl/arvados_cwl/pathmapper.py diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py index e17264bcdd..1f6aa577c1 100644 --- a/sdk/cwl/arvados_cwl/pathmapper.py +++ b/sdk/cwl/arvados_cwl/pathmapper.py @@ -6,7 +6,9 @@ import os import arvados.commands.run import arvados.collection -from cwltool.pathmapper import PathMapper, MapperEnt, abspath +from schema_salad.sourceline import SourceLine + +from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs from cwltool.workflow import WorkflowException logger = logging.getLogger('arvados.cwl-runner') @@ -37,27 +39,35 @@ class ArvPathMapper(PathMapper): # 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=self.file_pattern) - if isinstance(st, arvados.commands.run.UploadFile): - uploadfiles.add((src, ab, st)) - elif isinstance(st, arvados.commands.run.ArvFile): - self._pathmap[src] = MapperEnt(ab, st.fn, "File") - elif src.startswith("_:") and "contents" in srcobj: - pass - else: - raise WorkflowException("Input file path '%s' is invalid" % st) + st = arvados.commands.run.statfile("", ab, fnPattern="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") + elif src.startswith("_:"): + if "contents" in srcobj: + pass + else: + raise WorkflowException("File literal '%s' is missing contents" % src) + elif src.startswith("arvwf:"): + self._pathmap[src] = MapperEnt(src, src, "File") + else: + raise WorkflowException("Input file path '%s' is invalid" % st) if "secondaryFiles" in srcobj: for l in srcobj["secondaryFiles"]: self.visit(l, uploadfiles) elif srcobj["class"] == "Directory": if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src): self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "Directory") - for l in srcobj["listing"]: + for l in srcobj.get("listing", []): self.visit(l, uploadfiles) def addentry(self, obj, c, path, subdirs): if obj["location"] in self._pathmap: src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved) + if srcpath == "": + srcpath = "." c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True) for l in obj.get("secondaryFiles", []): self.addentry(l, c, path, subdirs) @@ -69,13 +79,19 @@ class ArvPathMapper(PathMapper): with c.open(path + "/" + obj["basename"], "w") as f: f.write(obj["contents"].encode("utf-8")) else: - raise WorkflowException("Don't know what to do with '%s'" % obj["location"]) + raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"]) def setup(self, referenced_files, basedir): # type: (List[Any], unicode) -> None - self._pathmap = self.arvrunner.get_uploaded() uploadfiles = set() + already_uploaded = self.arvrunner.get_uploaded() + for k in referenced_files: + 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") + for srcobj in referenced_files: self.visit(srcobj, uploadfiles) @@ -84,18 +100,19 @@ class ArvPathMapper(PathMapper): self.arvrunner.api, dry_run=False, num_retries=self.arvrunner.num_retries, - fnPattern=self.file_pattern, + fnPattern="keep:%s/%s", name=self.name, project=self.arvrunner.project_uuid) for src, ab, st in uploadfiles: - self._pathmap[src] = MapperEnt("keep:" + st.keepref, st.fn, "File") + self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File") self.arvrunner.add_uploaded(src, self._pathmap[src]) for srcobj in referenced_files: if srcobj["class"] == "Directory": if srcobj["location"] not in self._pathmap: c = arvados.collection.Collection(api_client=self.arvrunner.api, + keep_client=self.arvrunner.keep_client, num_retries=self.arvrunner.num_retries) subdirs = [] for l in srcobj["listing"]: @@ -114,6 +131,7 @@ class ArvPathMapper(PathMapper): (srcobj["location"].startswith("_:") and "contents" in srcobj)): c = arvados.collection.Collection(api_client=self.arvrunner.api, + keep_client=self.arvrunner.keep_client, num_retries=self.arvrunner.num_retries ) subdirs = [] self.addentry(srcobj, c, ".", subdirs) @@ -141,36 +159,45 @@ class ArvPathMapper(PathMapper): else: return super(ArvPathMapper, self).reversemap(target) -class InitialWorkDirPathMapper(PathMapper): +class StagingPathMapper(PathMapper): + _follow_dirs = True def visit(self, obj, stagedir, basedir, copy=False): # type: (Dict[unicode, Any], unicode, unicode, bool) -> None + loc = obj["location"] + tgt = os.path.join(stagedir, obj["basename"]) if obj["class"] == "Directory": - self._pathmap[obj["location"]] = MapperEnt(obj["location"], stagedir, "Directory") - self.visitlisting(obj.get("listing", []), stagedir, basedir) + self._pathmap[loc] = MapperEnt(loc, tgt, "Directory") + if loc.startswith("_:") or self._follow_dirs: + self.visitlisting(obj.get("listing", []), tgt, basedir) elif obj["class"] == "File": - loc = obj["location"] if loc in self._pathmap: return - tgt = os.path.join(stagedir, obj["basename"]) - if "contents" in obj and obj["location"].startswith("_:"): + if "contents" in obj and loc.startswith("_:"): self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile") else: if copy: - self._pathmap[loc] = MapperEnt(obj["path"], tgt, "WritableFile") + self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile") else: - self._pathmap[loc] = MapperEnt(obj["path"], tgt, "File") + self._pathmap[loc] = MapperEnt(loc, tgt, "File") self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir) + +class VwdPathMapper(StagingPathMapper): def setup(self, referenced_files, basedir): # type: (List[Any], unicode) -> None # Go through each file and set the target to its own directory along # with any secondary files. - stagedir = self.stagedir - for fob in referenced_files: - self.visit(fob, stagedir, basedir) + self.visitlisting(referenced_files, self.stagedir, basedir) for path, (ab, tgt, type) in self._pathmap.items(): if type in ("File", "Directory") and ab.startswith("keep:"): self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type) + + +class NoFollowPathMapper(StagingPathMapper): + _follow_dirs = False + def setup(self, referenced_files, basedir): + # type: (List[Any], unicode) -> None + self.visitlisting(referenced_files, self.stagedir, basedir)