1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
10 from schema_salad.sourceline import SourceLine, cmap
12 from cwltool.pack import pack
13 from cwltool.load_tool import fetch_document
14 from cwltool.process import shortname
15 from cwltool.workflow import Workflow, WorkflowException
16 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class
17 from cwltool.builder import Builder
19 import ruamel.yaml as yaml
21 from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
22 trim_anonymous_location, remove_redundant_fields, discover_secondary_files)
23 from .pathmapper import ArvPathMapper, trim_listing
24 from .arvtool import ArvadosCommandTool
25 from .perf import Perf
27 logger = logging.getLogger('arvados.cwl-runner')
28 metrics = logging.getLogger('arvados.cwl-runner.metrics')
30 max_res_pars = ("coresMin", "coresMax", "ramMin", "ramMax", "tmpdirMin", "tmpdirMax")
31 sum_res_pars = ("outdirMin", "outdirMax")
33 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
34 submit_runner_ram=0, name=None, merged_map=None):
36 packed = packed_workflow(arvRunner, tool, merged_map)
38 adjustDirObjs(job_order, trim_listing)
39 adjustFileObjs(job_order, trim_anonymous_location)
40 adjustDirObjs(job_order, trim_anonymous_location)
42 main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
43 for inp in main["inputs"]:
44 sn = shortname(inp["id"])
46 inp["default"] = job_order[sn]
49 name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
51 upload_dependencies(arvRunner, name, tool.doc_loader,
52 packed, tool.tool["id"], False)
54 # TODO nowhere for submit_runner_ram to go.
59 "description": tool.tool.get("doc", ""),
60 "definition":yaml.round_trip_dump(packed)
63 body["workflow"]["owner_uuid"] = project_uuid
66 call = arvRunner.api.workflows().update(uuid=uuid, body=body)
68 call = arvRunner.api.workflows().create(body=body)
69 return call.execute(num_retries=arvRunner.num_retries)["uuid"]
73 for r in reversed(reqs):
74 if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
76 return [dedup[r] for r in sorted(dedup.keys())]
78 def get_overall_res_req(res_reqs):
79 """Take the overall of a list of ResourceRequirement,
80 i.e., the max of coresMin, coresMax, ramMin, ramMax, tmpdirMin, tmpdirMax
81 and the sum of outdirMin, outdirMax."""
85 for a in max_res_pars + sum_res_pars:
87 for res_req in res_reqs:
89 if isinstance(res_req[a], int): # integer check
90 all_res_req[a].append(res_req[a])
92 msg = SourceLine(res_req).makeError(
93 "Non-top-level ResourceRequirement in single container cannot have expressions")
94 exception_msgs.append(msg)
96 raise WorkflowException("\n".join(exception_msgs))
101 if a in max_res_pars:
102 overall_res_req[a] = max(all_res_req[a])
103 elif a in sum_res_pars:
104 overall_res_req[a] = sum(all_res_req[a])
106 overall_res_req["class"] = "ResourceRequirement"
107 return cmap(overall_res_req)
109 class ArvadosWorkflow(Workflow):
110 """Wrap cwltool Workflow to override selected methods."""
112 def __init__(self, arvrunner, toolpath_object, **kwargs):
113 super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
114 self.arvrunner = arvrunner
115 self.work_api = kwargs["work_api"]
118 def job(self, joborder, output_callback, **kwargs):
119 kwargs["work_api"] = self.work_api
120 req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
122 with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
123 if "id" not in self.tool:
124 raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
125 document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
127 discover_secondary_files(self.tool["inputs"], joborder)
129 with Perf(metrics, "subworkflow upload_deps"):
130 upload_dependencies(self.arvrunner,
131 os.path.basename(joborder.get("id", "#")),
134 joborder.get("id", "#"),
137 if self.wf_pdh is None:
138 workflowobj["requirements"] = dedup_reqs(self.requirements)
139 workflowobj["hints"] = dedup_reqs(self.hints)
141 packed = pack(document_loader, workflowobj, uri, self.metadata)
144 builder.job = joborder
145 builder.requirements = workflowobj["requirements"]
146 builder.hints = workflowobj["hints"]
147 builder.resources = {}
149 res_reqs = {"requirements": [], "hints": []}
150 for t in ("requirements", "hints"):
151 for item in packed["$graph"]:
153 if item["id"] == "#main": # evaluate potential expressions in the top-level requirements/hints
155 if req["class"] == "ResourceRequirement":
156 eval_req = {"class": "ResourceRequirement"}
157 for a in max_res_pars + sum_res_pars:
159 eval_req[a] = builder.do_eval(req[a])
160 res_reqs[t].append(eval_req)
163 if req["class"] == "ResourceRequirement":
164 res_reqs[t].append(req)
165 overall_res_req = {"requirements": get_overall_res_req(res_reqs["requirements"]),
166 "hints": get_overall_res_req(res_reqs["hints"])}
168 new_spec = {"requirements": self.requirements, "hints": self.hints}
169 for t in ("requirements", "hints"):
170 for req in new_spec[t]:
171 if req["class"] == "ResourceRequirement":
172 new_spec[t].remove(req)
173 if overall_res_req[t]:
174 new_spec[t].append(overall_res_req[t])
176 upload_dependencies(self.arvrunner,
177 kwargs.get("name", ""),
183 with Perf(metrics, "subworkflow adjust"):
184 joborder_resolved = copy.deepcopy(joborder)
185 joborder_keepmount = copy.deepcopy(joborder)
188 visit_class(joborder_keepmount, ("File", "Directory"), lambda x: reffiles.append(x))
190 mapper = ArvPathMapper(self.arvrunner, reffiles, kwargs["basedir"],
196 remove_redundant_fields(obj)
197 with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
198 if "location" not in obj:
199 raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
200 with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
201 if obj["location"].startswith("keep:"):
202 obj["location"] = mapper.mapper(obj["location"]).target
205 elif obj["location"].startswith("_:"):
208 raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
210 visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
213 if obj["location"].startswith("keep:"):
214 obj["location"] = mapper.mapper(obj["location"]).resolved
216 visit_class(joborder_resolved, ("File", "Directory"), resolved)
218 if self.wf_pdh is None:
219 adjustFileObjs(packed, keepmount)
220 adjustDirObjs(packed, keepmount)
221 self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
224 "class": "CommandLineTool",
225 "baseCommand": "cwltool",
226 "inputs": self.tool["inputs"],
227 "outputs": self.tool["outputs"],
228 "stdout": "cwl.output.json",
229 "requirements": self.requirements+[
231 "class": "InitialWorkDirRequirement",
233 "entryname": "workflow.cwl",
236 "location": "keep:%s/workflow.cwl" % self.wf_pdh
239 "entryname": "cwl.input.yml",
240 "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
244 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"],
247 kwargs["loader"] = self.doc_loader
248 kwargs["avsc_names"] = self.doc_schema
249 return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder_resolved, output_callback, **kwargs)
251 return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)