X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/210889a5400ac6ef247078af14c0766f3639b988..f5f5de0ae41e12738a380c422417d5a5e5af7f09:/sdk/cwl/arvados_cwl/pathmapper.py diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py index ffdade0f5e..5e2ee46a87 100644 --- a/sdk/cwl/arvados_cwl/pathmapper.py +++ b/sdk/cwl/arvados_cwl/pathmapper.py @@ -2,20 +2,38 @@ import re import logging import uuid import os +import urllib 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') +def trim_listing(obj): + """Remove 'listing' field from Directory objects that are keep references. + + When Directory objects represent Keep references, it is redundant and + potentially very expensive to pass fully enumerated Directory objects + between instances of cwl-runner (e.g. a submitting a job, or using the + RunInSingleContainer feature), so delete the 'listing' field when it is + safe to do so. + + """ + + if obj.get("location", "").startswith("keep:") and "listing" in obj: + del obj["listing"] + + 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): @@ -24,34 +42,44 @@ class ArvPathMapper(PathMapper): self.collection_pattern = collection_pattern self.file_pattern = file_pattern self.name = name + self.referenced_files = [r["location"] for r in referenced_files] 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: + 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=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) - 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") + 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("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", []): + self.visit(l, uploadfiles) + with SourceLine(srcobj, "listing", WorkflowException): for l in srcobj.get("listing", []): self.visit(l, uploadfiles) @@ -64,20 +92,26 @@ 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: 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 % urllib.unquote(v.resolved[5:]), "File", True) + for srcobj in referenced_files: self.visit(srcobj, uploadfiles) @@ -86,21 +120,23 @@ 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(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) @@ -108,16 +144,13 @@ 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") - for loc, sub in subdirs: - ab = self.file_pattern % (c.portable_data_hash(), sub[2:]) - self._pathmap[loc] = MapperEnt(ab, ab, "Directory") + 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) @@ -125,13 +158,18 @@ 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") + 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") + 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") + self._pathmap[loc] = MapperEnt("keep:%s/%s" % (c.portable_data_hash(), sub[2:]), + ab, "Directory", True) self.keepdir = None @@ -143,36 +181,45 @@ 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): + 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"]) 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", staged) + 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("_:"): - self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile") + if "contents" in obj and loc.startswith("_:"): + self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile", staged) else: if copy: - self._pathmap[loc] = MapperEnt(obj["path"], tgt, "WritableFile") + self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile", staged) else: - self._pathmap[loc] = MapperEnt(obj["path"], tgt, "File") + self._pathmap[loc] = MapperEnt(loc, tgt, "File", staged) 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(): + for path, (ab, tgt, type, staged) in self._pathmap.items(): if type in ("File", "Directory") and ab.startswith("keep:"): - self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type) + self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type, staged) + + +class NoFollowPathMapper(StagingPathMapper): + _follow_dirs = False + def setup(self, referenced_files, basedir): + # type: (List[Any], unicode) -> None + self.visitlisting(referenced_files, self.stagedir, basedir)