X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/6b0632d7706cbb344331bf5d291cafd541070cbb..d3a82af33386c99c0fd3d6471df7e6696560089c:/sdk/cwl/arvados_cwl/pathmapper.py diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py index e753a185fe..cddb4088b7 100644 --- a/sdk/cwl/arvados_cwl/pathmapper.py +++ b/sdk/cwl/arvados_cwl/pathmapper.py @@ -2,6 +2,7 @@ import re import logging import uuid import os +import urllib import arvados.commands.run import arvados.collection @@ -17,7 +18,7 @@ class ArvPathMapper(PathMapper): """Convert container-local paths to and from Keep collection ids.""" pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$') - pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$') + 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): @@ -30,36 +31,39 @@ class ArvPathMapper(PathMapper): 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", True) - if src not in self._pathmap: + if "#" in src: + src = src[:src.index("#")] + + 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) + + if src not in self._pathmap: + 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") + 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 - else: - raise WorkflowException("File literal '%s' is missing contents" % src) - elif src.startswith("arvwf:"): - self._pathmap[src] = MapperEnt(src, src, "File", True) + self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.unquote(st.fn[5:]), "File", True) 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", True) + 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", []): + self.visit(l, uploadfiles) + with SourceLine(srcobj, "listing", WorkflowException): for l in srcobj.get("listing", []): self.visit(l, uploadfiles) @@ -72,7 +76,7 @@ class ArvPathMapper(PathMapper): for l in obj.get("secondaryFiles", []): self.addentry(l, c, path, subdirs) elif obj["class"] == "Directory": - for l in obj["listing"]: + for l in obj.get("listing", []): self.addentry(l, c, path + "/" + obj["basename"], subdirs) subdirs.append((obj["location"], path + "/" + obj["basename"])) elif obj["location"].startswith("_:") and "contents" in obj: @@ -90,7 +94,7 @@ class ArvPathMapper(PathMapper): 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) @@ -105,17 +109,18 @@ class ArvPathMapper(PathMapper): project=self.arvrunner.project_uuid) for src, ab, st in uploadfiles: - self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File", True) + 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]) for srcobj in referenced_files: + subdirs = [] 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"]: + for l in srcobj.get("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) @@ -123,17 +128,13 @@ class ArvPathMapper(PathMapper): 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", True) - for loc, sub in subdirs: - ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) - self._pathmap[loc] = MapperEnt(ab, ab, "Directory", True) + self._pathmap[srcobj["location"]] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True) 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) @@ -141,13 +142,18 @@ class ArvPathMapper(PathMapper): 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", True) + self._pathmap[srcobj["location"]] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), srcobj["basename"]), + ab, "File", True) if srcobj.get("secondaryFiles"): ab = self.collection_pattern % c.portable_data_hash() - self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory", True) + 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:]) - self._pathmap[loc] = MapperEnt(ab, ab, "Directory", True) + self._pathmap[loc] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), sub[2:]), + ab, "Directory", True) self.keepdir = None