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 from cwltool.pack import pack
17 from cwltool.load_tool import fetch_document, resolve_and_validate_document
18 from cwltool.process import shortname
19 from cwltool.workflow import Workflow, WorkflowException, WorkflowStep
20 from cwltool.utils import adjustFileObjs, adjustDirObjs, visit_class, normalizeFilesDirs
21 from cwltool.context import LoadingContext
23 import ruamel.yaml as yaml
25 from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
26 trim_anonymous_location, remove_redundant_fields, discover_secondary_files,
27 make_builder, arvados_jobs_image)
28 from .pathmapper import ArvPathMapper, trim_listing
29 from .arvtool import ArvadosCommandTool, set_cluster_target
30 from ._version import __version__
32 from .perf import Perf
34 logger = logging.getLogger('arvados.cwl-runner')
35 metrics = logging.getLogger('arvados.cwl-runner.metrics')
37 max_res_pars = ("coresMin", "coresMax", "ramMin", "ramMax", "tmpdirMin", "tmpdirMax")
38 sum_res_pars = ("outdirMin", "outdirMax")
40 def upload_workflow(arvRunner, tool, job_order, project_uuid,
41 runtimeContext, uuid=None,
42 submit_runner_ram=0, name=None, merged_map=None,
43 submit_runner_image=None):
45 packed = packed_workflow(arvRunner, tool, merged_map, runtimeContext)
47 adjustDirObjs(job_order, trim_listing)
48 adjustFileObjs(job_order, trim_anonymous_location)
49 adjustDirObjs(job_order, trim_anonymous_location)
51 main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
52 for inp in main["inputs"]:
53 sn = shortname(inp["id"])
55 inp["default"] = job_order[sn]
58 name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
60 upload_dependencies(arvRunner, name, tool.doc_loader,
61 packed, tool.tool["id"], False,
64 wf_runner_resources = None
66 hints = main.get("hints", [])
69 if h["class"] == "http://arvados.org/cwl#WorkflowRunnerResources":
70 wf_runner_resources = h
74 wf_runner_resources = {"class": "http://arvados.org/cwl#WorkflowRunnerResources"}
75 hints.append(wf_runner_resources)
77 wf_runner_resources["acrContainerImage"] = arvados_jobs_image(arvRunner,
78 submit_runner_image or "arvados/jobs:"+__version__,
82 wf_runner_resources["ramMin"] = submit_runner_ram
89 "description": tool.tool.get("doc", ""),
90 "definition":json.dumps(packed, sort_keys=True, indent=4, separators=(',',': '))
93 body["workflow"]["owner_uuid"] = project_uuid
96 call = arvRunner.api.workflows().update(uuid=uuid, body=body)
98 call = arvRunner.api.workflows().create(body=body)
99 return call.execute(num_retries=arvRunner.num_retries)["uuid"]
101 def dedup_reqs(reqs):
103 for r in reversed(reqs):
104 if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
105 dedup[r["class"]] = r
106 return [dedup[r] for r in sorted(dedup.keys())]
108 def get_overall_res_req(res_reqs):
109 """Take the overall of a list of ResourceRequirement,
110 i.e., the max of coresMin, coresMax, ramMin, ramMax, tmpdirMin, tmpdirMax
111 and the sum of outdirMin, outdirMax."""
115 for a in max_res_pars + sum_res_pars:
117 for res_req in res_reqs:
119 if isinstance(res_req[a], int): # integer check
120 all_res_req[a].append(res_req[a])
122 msg = SourceLine(res_req, a).makeError(
123 "Non-top-level ResourceRequirement in single container cannot have expressions")
124 exception_msgs.append(msg)
126 raise WorkflowException("\n".join(exception_msgs))
129 for a in all_res_req:
131 if a in max_res_pars:
132 overall_res_req[a] = max(all_res_req[a])
133 elif a in sum_res_pars:
134 overall_res_req[a] = sum(all_res_req[a])
136 overall_res_req["class"] = "ResourceRequirement"
137 return cmap(overall_res_req)
139 class ArvadosWorkflowStep(WorkflowStep):
141 toolpath_object, # type: Dict[Text, Any]
143 loadingContext, # type: LoadingContext
147 ): # type: (...) -> None
149 super(ArvadosWorkflowStep, self).__init__(toolpath_object, pos, loadingContext, *argc, **argv)
150 self.tool["class"] = "WorkflowStep"
151 self.arvrunner = arvrunner
153 def job(self, joborder, output_callback, runtimeContext):
154 runtimeContext = runtimeContext.copy()
155 runtimeContext.toplevel = True # Preserve behavior for #13365
157 builder = make_builder({shortname(k): v for k,v in viewitems(joborder)}, self.hints, self.requirements,
158 runtimeContext, self.metadata)
159 runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
160 return super(ArvadosWorkflowStep, self).job(joborder, output_callback, runtimeContext)
163 class ArvadosWorkflow(Workflow):
164 """Wrap cwltool Workflow to override selected methods."""
166 def __init__(self, arvrunner, toolpath_object, loadingContext):
167 self.arvrunner = arvrunner
169 self.dynamic_resource_req = []
170 self.static_resource_req = []
171 self.wf_reffiles = []
172 self.loadingContext = loadingContext
173 super(ArvadosWorkflow, self).__init__(toolpath_object, loadingContext)
174 self.cluster_target_req, _ = self.get_requirement("http://arvados.org/cwl#ClusterTarget")
176 def job(self, joborder, output_callback, runtimeContext):
178 builder = make_builder(joborder, self.hints, self.requirements, runtimeContext, self.metadata)
179 runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
181 req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
183 return super(ArvadosWorkflow, self).job(joborder, output_callback, runtimeContext)
185 # RunInSingleContainer is true
187 with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
188 if "id" not in self.tool:
189 raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
191 discover_secondary_files(self.arvrunner.fs_access, builder,
192 self.tool["inputs"], joborder)
193 normalizeFilesDirs(joborder)
195 with Perf(metrics, "subworkflow upload_deps"):
196 upload_dependencies(self.arvrunner,
197 os.path.basename(joborder.get("id", "#")),
200 joborder.get("id", "#"),
204 if self.wf_pdh is None:
205 packed = pack(self.loadingContext, self.tool["id"], loader=self.doc_loader)
207 for p in packed["$graph"]:
208 if p["id"] == "#main":
209 p["requirements"] = dedup_reqs(self.requirements)
210 p["hints"] = dedup_reqs(self.hints)
213 if "requirements" in item:
214 item["requirements"] = [i for i in item["requirements"] if i["class"] != "DockerRequirement"]
215 for t in ("hints", "requirements"):
219 if req["class"] == "ResourceRequirement":
221 for k in max_res_pars + sum_res_pars:
223 if isinstance(req[k], basestring):
224 if item["id"] == "#main":
225 # only the top-level requirements/hints may contain expressions
226 self.dynamic_resource_req.append(req)
230 with SourceLine(req, k, WorkflowException):
231 raise WorkflowException("Non-top-level ResourceRequirement in single container cannot have expressions")
233 self.static_resource_req.append(req)
235 visit_class(packed["$graph"], ("Workflow", "CommandLineTool"), visit)
237 if self.static_resource_req:
238 self.static_resource_req = [get_overall_res_req(self.static_resource_req)]
240 upload_dependencies(self.arvrunner,
248 # Discover files/directories referenced by the
249 # workflow (mainly "default" values)
250 visit_class(packed, ("File", "Directory"), self.wf_reffiles.append)
253 if self.dynamic_resource_req:
254 # Evaluate dynamic resource requirements using current builder
255 rs = copy.copy(self.static_resource_req)
256 for dyn_rs in self.dynamic_resource_req:
257 eval_req = {"class": "ResourceRequirement"}
258 for a in max_res_pars + sum_res_pars:
260 eval_req[a] = builder.do_eval(dyn_rs[a])
262 job_res_reqs = [get_overall_res_req(rs)]
264 job_res_reqs = self.static_resource_req
266 with Perf(metrics, "subworkflow adjust"):
267 joborder_resolved = copy.deepcopy(joborder)
268 joborder_keepmount = copy.deepcopy(joborder)
271 visit_class(joborder_keepmount, ("File", "Directory"), reffiles.append)
273 mapper = ArvPathMapper(self.arvrunner, reffiles+self.wf_reffiles, runtimeContext.basedir,
277 # For containers API, we need to make sure any extra
278 # referenced files (ie referenced by the workflow but
279 # not in the inputs) are included in the mounts.
281 runtimeContext = runtimeContext.copy()
282 runtimeContext.extra_reffiles = copy.deepcopy(self.wf_reffiles)
285 remove_redundant_fields(obj)
286 with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
287 if "location" not in obj:
288 raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
289 with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
290 if obj["location"].startswith("keep:"):
291 obj["location"] = mapper.mapper(obj["location"]).target
294 elif obj["location"].startswith("_:"):
297 raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
299 visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
302 if obj["location"].startswith("keep:"):
303 obj["location"] = mapper.mapper(obj["location"]).resolved
305 visit_class(joborder_resolved, ("File", "Directory"), resolved)
307 if self.wf_pdh is None:
308 adjustFileObjs(packed, keepmount)
309 adjustDirObjs(packed, keepmount)
310 self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed, runtimeContext)
312 self.loadingContext = self.loadingContext.copy()
313 self.loadingContext.metadata = self.loadingContext.metadata.copy()
314 self.loadingContext.metadata["http://commonwl.org/cwltool#original_cwlVersion"] = "v1.0"
316 if len(job_res_reqs) == 1:
317 # RAM request needs to be at least 128 MiB or the workflow
318 # runner itself won't run reliably.
319 if job_res_reqs[0].get("ramMin", 1024) < 128:
320 job_res_reqs[0]["ramMin"] = 128
322 arguments = ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl", "cwl.input.yml"]
323 if runtimeContext.debug:
324 arguments.insert(0, '--debug')
327 "class": "CommandLineTool",
328 "baseCommand": "cwltool",
329 "inputs": self.tool["inputs"],
330 "outputs": self.tool["outputs"],
331 "stdout": "cwl.output.json",
332 "requirements": self.requirements+job_res_reqs+[
333 {"class": "InlineJavascriptRequirement"},
335 "class": "InitialWorkDirRequirement",
337 "entryname": "workflow.cwl",
338 "entry": '$({"class": "File", "location": "keep:%s/workflow.cwl"})' % self.wf_pdh
340 "entryname": "cwl.input.yml",
341 "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
345 "arguments": arguments,
348 return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)
350 def make_workflow_step(self,
351 toolpath_object, # type: Dict[Text, Any]
353 loadingContext, # type: LoadingContext
357 # (...) -> WorkflowStep
358 return ArvadosWorkflowStep(toolpath_object, pos, loadingContext, self.arvrunner, *argc, **argv)