X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/e34ddeba03a6a9b4a1f9913ff5871938e83ee1d2..5624fec61db977d386ce03ca333241c74ca251b5:/sdk/cwl/arvados_cwl/pathmapper.py diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py index 0a11925ca7..1f6aa577c1 100644 --- a/sdk/cwl/arvados_cwl/pathmapper.py +++ b/sdk/cwl/arvados_cwl/pathmapper.py @@ -1,13 +1,23 @@ import re +import logging +import uuid +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') class ArvPathMapper(PathMapper): """Convert container-local paths to and from Keep collection ids.""" - pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+') + pdh_path = 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): @@ -18,43 +28,127 @@ class ArvPathMapper(PathMapper): self.name = name super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None) + def visit(self, srcobj, uploadfiles): + src = srcobj["location"] + if srcobj["class"] == "File": + if "#" in src: + src = src[:src.index("#")] + if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src): + self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "File") + 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") + 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.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) + elif obj["class"] == "Directory": + for l in obj["listing"]: + self.addentry(l, c, path + "/" + obj["basename"], subdirs) + subdirs.append((obj["location"], path + "/" + obj["basename"])) + elif obj["location"].startswith("_:") and "contents" in obj: + with c.open(path + "/" + obj["basename"], "w") as f: + f.write(obj["contents"].encode("utf-8")) + else: + 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: - if srcobj["class"] == "File": - src = srcobj["location"] - if "#" in src: - src = src[:src.index("#")] - if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src): - self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "File") - 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=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") - else: - raise cwltool.workflow.WorkflowException("Input file path '%s' is invalid" % st) + self.visit(srcobj, uploadfiles) if uploadfiles: arvados.commands.run.uploadfiles([u[2] for u in uploadfiles], self.arvrunner.api, dry_run=False, - num_retries=3, - fnPattern=self.file_pattern, + num_retries=self.arvrunner.num_retries, + fnPattern="keep:%s/%s", name=self.name, project=self.arvrunner.project_uuid) for src, ab, st in uploadfiles: - self._pathmap[src] = MapperEnt(ab, 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"]: + self.addentry(l, c, ".", subdirs) + + check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries) + if not check["items"]: + c.save_new(owner_uuid=self.arvrunner.project_uuid) + + ab = self.collection_pattern % c.portable_data_hash() + self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "Directory") + for loc, sub in subdirs: + ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) + self._pathmap[loc] = MapperEnt(ab, ab, "Directory") + elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or + (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) + + check = self.arvrunner.api.collections().list(filters=[["portable_data_hash", "=", c.portable_data_hash()]], limit=1).execute(num_retries=self.arvrunner.num_retries) + if not check["items"]: + c.save_new(owner_uuid=self.arvrunner.project_uuid) + + ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"]) + self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File") + if srcobj.get("secondaryFiles"): + ab = self.collection_pattern % c.portable_data_hash() + self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory") + for loc, sub in subdirs: + ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) + self._pathmap[loc] = MapperEnt(ab, ab, "Directory") + self.keepdir = None def reversemap(self, target): @@ -64,3 +158,46 @@ class ArvPathMapper(PathMapper): return (target, "keep:" + target[len(self.keepdir)+1:]) else: return super(ArvPathMapper, self).reversemap(target) + +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[loc] = MapperEnt(loc, tgt, "Directory") + if loc.startswith("_:") or self._follow_dirs: + self.visitlisting(obj.get("listing", []), tgt, basedir) + elif obj["class"] == "File": + if loc in self._pathmap: + return + if "contents" in obj and loc.startswith("_:"): + self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile") + else: + if copy: + self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile") + else: + 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. + 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)