1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from past.builtins import basestring
6 from future.utils import viewitems
13 from schema_salad.sourceline import SourceLine, cmap
14 import schema_salad.ref_resolver
16 import arvados.collection
18 from cwltool.pack import pack
19 from cwltool.load_tool import fetch_document, resolve_and_validate_document
20 from cwltool.process import shortname
21 from cwltool.workflow import Workflow, WorkflowException, WorkflowStep
22 from cwltool.utils import adjustFileObjs, adjustDirObjs, visit_class, normalizeFilesDirs
23 from cwltool.context import LoadingContext
25 import ruamel.yaml as yaml
27 from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
28 trim_anonymous_location, remove_redundant_fields, discover_secondary_files,
29 make_builder, arvados_jobs_image)
30 from .pathmapper import ArvPathMapper, trim_listing
31 from .arvtool import ArvadosCommandTool, set_cluster_target
32 from ._version import __version__
34 from .perf import Perf
36 logger = logging.getLogger('arvados.cwl-runner')
37 metrics = logging.getLogger('arvados.cwl-runner.metrics')
39 max_res_pars = ("coresMin", "coresMax", "ramMin", "ramMax", "tmpdirMin", "tmpdirMax")
40 sum_res_pars = ("outdirMin", "outdirMax")
42 def make_wrapper_workflow(arvRunner, main, packed, project_uuid, name):
43 col = arvados.collection.Collection(api_client=arvRunner.api,
44 keep_client=arvRunner.keep_client)
46 with col.open("workflow.json", "wt") as f:
47 json.dump(packed, f, sort_keys=True, indent=4, separators=(',',': '))
49 pdh = col.portable_data_hash()
51 existing = arvRunner.api.collections().list(filters=[["portable_data_hash", "=", pdh], ["owner_uuid", "=", project_uuid]]).execute(num_retries=arvRunner.num_retries)
52 if len(existing["items"]) == 0:
53 col.save_new(name=name, owner_uuid=project_uuid, ensure_unique_name=True)
55 # now construct the wrapper
61 "run": "keep:%s/workflow.json#main" % pdh,
66 for i in main["inputs"]:
68 # Make sure to only copy known fields that are meaningful at
69 # the workflow level. In practice this ensures that if we're
70 # wrapping a CommandLineTool we don't grab inputBinding.
71 # Right now also excludes extension fields, which is fine,
72 # Arvados doesn't currently look for any extension fields on
74 for f in ("type", "label", "secondaryFiles", "streamable",
75 "doc", "id", "format", "loadContents",
76 "loadListing", "default"):
89 for i in main["inputs"]:
91 "id": "#main/step/%s" % shortname(i["id"]),
95 for i in main["outputs"]:
96 step["out"].append({"id": "#main/step/%s" % shortname(i["id"])})
97 wrapper["outputs"].append({"outputSource": "#main/step/%s" % shortname(i["id"]),
101 wrapper["requirements"] = [{"class": "SubworkflowFeatureRequirement"}]
103 if main.get("requirements"):
104 wrapper["requirements"].extend(main["requirements"])
105 if main.get("hints"):
106 wrapper["hints"] = main["hints"]
108 return json.dumps({"cwlVersion": "v1.2", "$graph": [wrapper]}, sort_keys=True, indent=4, separators=(',',': '))
110 def upload_workflow(arvRunner, tool, job_order, project_uuid,
111 runtimeContext, uuid=None,
112 submit_runner_ram=0, name=None, merged_map=None,
113 submit_runner_image=None,
116 packed = packed_workflow(arvRunner, tool, merged_map, runtimeContext, git_info)
118 adjustDirObjs(job_order, trim_listing)
119 adjustFileObjs(job_order, trim_anonymous_location)
120 adjustDirObjs(job_order, trim_anonymous_location)
122 main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
123 for inp in main["inputs"]:
124 sn = shortname(inp["id"])
126 inp["default"] = job_order[sn]
129 name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
131 upload_dependencies(arvRunner, name, tool.doc_loader,
132 packed, tool.tool["id"], False,
135 wf_runner_resources = None
137 hints = main.get("hints", [])
140 if h["class"] == "http://arvados.org/cwl#WorkflowRunnerResources":
141 wf_runner_resources = h
145 wf_runner_resources = {"class": "http://arvados.org/cwl#WorkflowRunnerResources"}
146 hints.append(wf_runner_resources)
148 wf_runner_resources["acrContainerImage"] = arvados_jobs_image(arvRunner,
149 submit_runner_image or "arvados/jobs:"+__version__,
152 if submit_runner_ram:
153 wf_runner_resources["ramMin"] = submit_runner_ram
155 main["hints"] = hints
157 wrapper = make_wrapper_workflow(arvRunner, main, packed, project_uuid, name)
162 "description": tool.tool.get("doc", ""),
163 "definition": wrapper
166 body["workflow"]["owner_uuid"] = project_uuid
169 call = arvRunner.api.workflows().update(uuid=uuid, body=body)
171 call = arvRunner.api.workflows().create(body=body)
172 return call.execute(num_retries=arvRunner.num_retries)["uuid"]
174 def dedup_reqs(reqs):
176 for r in reversed(reqs):
177 if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
178 dedup[r["class"]] = r
179 return [dedup[r] for r in sorted(dedup.keys())]
181 def get_overall_res_req(res_reqs):
182 """Take the overall of a list of ResourceRequirement,
183 i.e., the max of coresMin, coresMax, ramMin, ramMax, tmpdirMin, tmpdirMax
184 and the sum of outdirMin, outdirMax."""
188 for a in max_res_pars + sum_res_pars:
190 for res_req in res_reqs:
192 if isinstance(res_req[a], int): # integer check
193 all_res_req[a].append(res_req[a])
195 msg = SourceLine(res_req, a).makeError(
196 "Non-top-level ResourceRequirement in single container cannot have expressions")
197 exception_msgs.append(msg)
199 raise WorkflowException("\n".join(exception_msgs))
202 for a in all_res_req:
204 if a in max_res_pars:
205 overall_res_req[a] = max(all_res_req[a])
206 elif a in sum_res_pars:
207 overall_res_req[a] = sum(all_res_req[a])
209 overall_res_req["class"] = "ResourceRequirement"
210 return cmap(overall_res_req)
212 class ArvadosWorkflowStep(WorkflowStep):
214 toolpath_object, # type: Dict[Text, Any]
216 loadingContext, # type: LoadingContext
220 ): # type: (...) -> None
222 if arvrunner.fast_submit:
223 self.tool = toolpath_object
224 self.tool["inputs"] = []
225 self.tool["outputs"] = []
227 super(ArvadosWorkflowStep, self).__init__(toolpath_object, pos, loadingContext, *argc, **argv)
228 self.tool["class"] = "WorkflowStep"
229 self.arvrunner = arvrunner
231 def job(self, joborder, output_callback, runtimeContext):
232 runtimeContext = runtimeContext.copy()
233 runtimeContext.toplevel = True # Preserve behavior for #13365
235 builder = make_builder({shortname(k): v for k,v in viewitems(joborder)}, self.hints, self.requirements,
236 runtimeContext, self.metadata)
237 runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
238 return super(ArvadosWorkflowStep, self).job(joborder, output_callback, runtimeContext)
241 class ArvadosWorkflow(Workflow):
242 """Wrap cwltool Workflow to override selected methods."""
244 def __init__(self, arvrunner, toolpath_object, loadingContext):
245 self.arvrunner = arvrunner
247 self.dynamic_resource_req = []
248 self.static_resource_req = []
249 self.wf_reffiles = []
250 self.loadingContext = loadingContext
251 super(ArvadosWorkflow, self).__init__(toolpath_object, loadingContext)
252 self.cluster_target_req, _ = self.get_requirement("http://arvados.org/cwl#ClusterTarget")
254 def job(self, joborder, output_callback, runtimeContext):
256 builder = make_builder(joborder, self.hints, self.requirements, runtimeContext, self.metadata)
257 runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
259 req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
261 return super(ArvadosWorkflow, self).job(joborder, output_callback, runtimeContext)
263 # RunInSingleContainer is true
265 with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
266 if "id" not in self.tool:
267 raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
269 discover_secondary_files(self.arvrunner.fs_access, builder,
270 self.tool["inputs"], joborder)
271 normalizeFilesDirs(joborder)
273 with Perf(metrics, "subworkflow upload_deps"):
274 upload_dependencies(self.arvrunner,
275 os.path.basename(joborder.get("id", "#")),
278 joborder.get("id", "#"),
282 if self.wf_pdh is None:
283 packed = pack(self.loadingContext, self.tool["id"], loader=self.doc_loader)
285 for p in packed["$graph"]:
286 if p["id"] == "#main":
287 p["requirements"] = dedup_reqs(self.requirements)
288 p["hints"] = dedup_reqs(self.hints)
291 if "requirements" in item:
292 item["requirements"] = [i for i in item["requirements"] if i["class"] != "DockerRequirement"]
293 for t in ("hints", "requirements"):
297 if req["class"] == "ResourceRequirement":
299 for k in max_res_pars + sum_res_pars:
301 if isinstance(req[k], basestring):
302 if item["id"] == "#main":
303 # only the top-level requirements/hints may contain expressions
304 self.dynamic_resource_req.append(req)
308 with SourceLine(req, k, WorkflowException):
309 raise WorkflowException("Non-top-level ResourceRequirement in single container cannot have expressions")
311 self.static_resource_req.append(req)
313 visit_class(packed["$graph"], ("Workflow", "CommandLineTool"), visit)
315 if self.static_resource_req:
316 self.static_resource_req = [get_overall_res_req(self.static_resource_req)]
318 upload_dependencies(self.arvrunner,
326 # Discover files/directories referenced by the
327 # workflow (mainly "default" values)
328 visit_class(packed, ("File", "Directory"), self.wf_reffiles.append)
331 if self.dynamic_resource_req:
332 # Evaluate dynamic resource requirements using current builder
333 rs = copy.copy(self.static_resource_req)
334 for dyn_rs in self.dynamic_resource_req:
335 eval_req = {"class": "ResourceRequirement"}
336 for a in max_res_pars + sum_res_pars:
338 eval_req[a] = builder.do_eval(dyn_rs[a])
340 job_res_reqs = [get_overall_res_req(rs)]
342 job_res_reqs = self.static_resource_req
344 with Perf(metrics, "subworkflow adjust"):
345 joborder_resolved = copy.deepcopy(joborder)
346 joborder_keepmount = copy.deepcopy(joborder)
349 visit_class(joborder_keepmount, ("File", "Directory"), reffiles.append)
351 mapper = ArvPathMapper(self.arvrunner, reffiles+self.wf_reffiles, runtimeContext.basedir,
355 # For containers API, we need to make sure any extra
356 # referenced files (ie referenced by the workflow but
357 # not in the inputs) are included in the mounts.
359 runtimeContext = runtimeContext.copy()
360 runtimeContext.extra_reffiles = copy.deepcopy(self.wf_reffiles)
363 remove_redundant_fields(obj)
364 with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
365 if "location" not in obj:
366 raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
367 with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
368 if obj["location"].startswith("keep:"):
369 obj["location"] = mapper.mapper(obj["location"]).target
372 elif obj["location"].startswith("_:"):
375 raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
377 visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
380 if obj["location"].startswith("keep:"):
381 obj["location"] = mapper.mapper(obj["location"]).resolved
383 visit_class(joborder_resolved, ("File", "Directory"), resolved)
385 if self.wf_pdh is None:
386 adjustFileObjs(packed, keepmount)
387 adjustDirObjs(packed, keepmount)
388 self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed, runtimeContext)
390 self.loadingContext = self.loadingContext.copy()
391 self.loadingContext.metadata = self.loadingContext.metadata.copy()
392 self.loadingContext.metadata["http://commonwl.org/cwltool#original_cwlVersion"] = "v1.0"
394 if len(job_res_reqs) == 1:
395 # RAM request needs to be at least 128 MiB or the workflow
396 # runner itself won't run reliably.
397 if job_res_reqs[0].get("ramMin", 1024) < 128:
398 job_res_reqs[0]["ramMin"] = 128
400 arguments = ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl", "cwl.input.yml"]
401 if runtimeContext.debug:
402 arguments.insert(0, '--debug')
405 "class": "CommandLineTool",
406 "baseCommand": "cwltool",
407 "inputs": self.tool["inputs"],
408 "outputs": self.tool["outputs"],
409 "stdout": "cwl.output.json",
410 "requirements": self.requirements+job_res_reqs+[
411 {"class": "InlineJavascriptRequirement"},
413 "class": "InitialWorkDirRequirement",
415 "entryname": "workflow.cwl",
416 "entry": '$({"class": "File", "location": "keep:%s/workflow.cwl"})' % self.wf_pdh
418 "entryname": "cwl.input.yml",
419 "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
423 "arguments": arguments,
426 return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)
428 def make_workflow_step(self,
429 toolpath_object, # type: Dict[Text, Any]
431 loadingContext, # type: LoadingContext
435 # (...) -> WorkflowStep
436 return ArvadosWorkflowStep(toolpath_object, pos, loadingContext, self.arvrunner, *argc, **argv)