X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/1d77dec0f7bc1cca34d52288bd39ecd0a79be250..f56ca98bda57d61b6b9b26b5fa1e845f8d567e5b:/sdk/cwl/arvados_cwl/pathmapper.py diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py index 52a65ff580..998890a31c 100644 --- a/sdk/cwl/arvados_cwl/pathmapper.py +++ b/sdk/cwl/arvados_cwl/pathmapper.py @@ -1,3 +1,7 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + import re import logging import uuid @@ -54,6 +58,8 @@ class ArvPathMapper(PathMapper): if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src): self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), srcobj["class"], True) + debug = logger.isEnabledFor(logging.DEBUG) + if src not in self._pathmap: if src.startswith("file:"): # Local FS ref, may need to be uploaded or may be on keep @@ -61,8 +67,9 @@ class ArvPathMapper(PathMapper): 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): + dirPattern="keep:%s/%s", + raiseOSError=True) + with SourceLine(srcobj, "location", WorkflowException, debug): if isinstance(st, arvados.commands.run.UploadFile): uploadfiles.add((src, ab, st)) elif isinstance(st, arvados.commands.run.ArvFile): @@ -77,25 +84,26 @@ class ArvPathMapper(PathMapper): else: self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True) - with SourceLine(srcobj, "secondaryFiles", WorkflowException): + with SourceLine(srcobj, "secondaryFiles", WorkflowException, debug): for l in srcobj.get("secondaryFiles", []): self.visit(l, uploadfiles) - with SourceLine(srcobj, "listing", WorkflowException): + with SourceLine(srcobj, "listing", WorkflowException, debug): for l in srcobj.get("listing", []): self.visit(l, uploadfiles) - def addentry(self, obj, c, path, subdirs): + def addentry(self, obj, c, path, remap): 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) + remap.append((obj["location"], path + "/" + obj["basename"])) for l in obj.get("secondaryFiles", []): - self.addentry(l, c, path, subdirs) + self.addentry(l, c, path, remap) elif obj["class"] == "Directory": for l in obj.get("listing", []): - self.addentry(l, c, path + "/" + obj["basename"], subdirs) - subdirs.append((obj["location"], path + "/" + obj["basename"])) + self.addentry(l, c, path + "/" + obj["basename"], remap) + remap.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")) @@ -147,13 +155,13 @@ class ArvPathMapper(PathMapper): self._pathmap[loc] = MapperEnt(urllib.quote(fn, "/:+@"), self.collection_pattern % fn[5:], cls, True) for srcobj in referenced_files: - subdirs = [] + remap = [] if srcobj["class"] == "Directory" and 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) for l in srcobj.get("listing", []): - self.addentry(l, c, ".", subdirs) + self.addentry(l, c, ".", remap) 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"]: @@ -167,7 +175,7 @@ class ArvPathMapper(PathMapper): c = arvados.collection.Collection(api_client=self.arvrunner.api, keep_client=self.arvrunner.keep_client, num_retries=self.arvrunner.num_retries ) - self.addentry(srcobj, c, ".", subdirs) + self.addentry(srcobj, c, ".", remap) 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"]: @@ -180,32 +188,53 @@ class ArvPathMapper(PathMapper): ab = self.collection_pattern % c.portable_data_hash() self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True) - if subdirs: - for loc, sub in subdirs: - # subdirs will all start with "./", strip it off - ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) + if remap: + for loc, sub in remap: + # subdirs start with "./", strip it off + if sub.startswith("./"): + ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) + else: + ab = self.file_pattern % (c.portable_data_hash(), sub) self._pathmap[loc] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), sub[2:]), ab, "Directory", True) self.keepdir = None def reversemap(self, target): - if target.startswith("keep:"): + p = super(ArvPathMapper, self).reversemap(target) + if p: + return p + elif target.startswith("keep:"): return (target, target) elif self.keepdir and target.startswith(self.keepdir): - return (target, "keep:" + target[len(self.keepdir)+1:]) + kp = "keep:" + target[len(self.keepdir)+1:] + return (kp, kp) else: - return super(ArvPathMapper, self).reversemap(target) + return None class StagingPathMapper(PathMapper): _follow_dirs = True + def __init__(self, referenced_files, basedir, stagedir, separateDirs=True): + self.targets = set() + super(StagingPathMapper, self).__init__(referenced_files, basedir, stagedir, separateDirs) + def visit(self, obj, stagedir, basedir, copy=False, staged=False): # type: (Dict[unicode, Any], unicode, unicode, bool) -> None loc = obj["location"] tgt = os.path.join(stagedir, obj["basename"]) + basetgt, baseext = os.path.splitext(tgt) + n = 1 + if tgt in self.targets and (self.reversemap(tgt)[0] != loc): + while tgt in self.targets: + n += 1 + tgt = "%s_%i%s" % (basetgt, n, baseext) + self.targets.add(tgt) if obj["class"] == "Directory": - self._pathmap[loc] = MapperEnt(loc, tgt, "Directory", staged) + if obj.get("writable"): + self._pathmap[loc] = MapperEnt(loc, tgt, "WritableDirectory", staged) + else: + self._pathmap[loc] = MapperEnt(loc, tgt, "Directory", staged) if loc.startswith("_:") or self._follow_dirs: self.visitlisting(obj.get("listing", []), tgt, basedir) elif obj["class"] == "File": @@ -214,7 +243,7 @@ class StagingPathMapper(PathMapper): if "contents" in obj and loc.startswith("_:"): self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile", staged) else: - if copy: + if copy or obj.get("writable"): self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile", staged) else: self._pathmap[loc] = MapperEnt(loc, tgt, "File", staged)