sdk/cwl/arvados_cwl/arvworkflow.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import os
   6 import json
   7 import copy
   8 import logging
   9
  10 from schema_salad.sourceline import SourceLine, cmap
  11
  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
  18 from cwltool.context import LoadingContext
  19
  20 import ruamel.yaml as yaml
  21
  22 from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
  23                      trim_anonymous_location, remove_redundant_fields, discover_secondary_files)
  24 from .pathmapper import ArvPathMapper, trim_listing
  25 from .arvtool import ArvadosCommandTool, check_cluster_target
  26 from .perf import Perf
  27
  28 logger = logging.getLogger('arvados.cwl-runner')
  29 metrics = logging.getLogger('arvados.cwl-runner.metrics')
  30
  31 max_res_pars = ("coresMin", "coresMax", "ramMin", "ramMax", "tmpdirMin", "tmpdirMax")
  32 sum_res_pars = ("outdirMin", "outdirMax")
  33
  34 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
  35                     submit_runner_ram=0, name=None, merged_map=None):
  36
  37     packed = packed_workflow(arvRunner, tool, merged_map)
  38
  39     adjustDirObjs(job_order, trim_listing)
  40     adjustFileObjs(job_order, trim_anonymous_location)
  41     adjustDirObjs(job_order, trim_anonymous_location)
  42
  43     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  44     for inp in main["inputs"]:
  45         sn = shortname(inp["id"])
  46         if sn in job_order:
  47             inp["default"] = job_order[sn]
  48
  49     if not name:
  50         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
  51
  52     upload_dependencies(arvRunner, name, tool.doc_loader,
  53                         packed, tool.tool["id"], False)
  54
  55     if submit_runner_ram:
  56         hints = main.get("hints", [])
  57         found = False
  58         for h in hints:
  59             if h["class"] == "http://arvados.org/cwl#WorkflowRunnerResources":
  60                 h["ramMin"] = submit_runner_ram
  61                 found = True
  62                 break
  63         if not found:
  64             hints.append({"class": "http://arvados.org/cwl#WorkflowRunnerResources",
  65                           "ramMin": submit_runner_ram})
  66         main["hints"] = hints
  67
  68     body = {
  69         "workflow": {
  70             "name": name,
  71             "description": tool.tool.get("doc", ""),
  72             "definition":json.dumps(packed, sort_keys=True, indent=4, separators=(',',': '))
  73         }}
  74     if project_uuid:
  75         body["workflow"]["owner_uuid"] = project_uuid
  76
  77     if uuid:
  78         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
  79     else:
  80         call = arvRunner.api.workflows().create(body=body)
  81     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
  82
  83 def dedup_reqs(reqs):
  84     dedup = {}
  85     for r in reversed(reqs):
  86         if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
  87             dedup[r["class"]] = r
  88     return [dedup[r] for r in sorted(dedup.keys())]
  89
  90 def get_overall_res_req(res_reqs):
  91     """Take the overall of a list of ResourceRequirement,
  92     i.e., the max of coresMin, coresMax, ramMin, ramMax, tmpdirMin, tmpdirMax
  93     and the sum of outdirMin, outdirMax."""
  94
  95     all_res_req = {}
  96     exception_msgs = []
  97     for a in max_res_pars + sum_res_pars:
  98         all_res_req[a] = []
  99         for res_req in res_reqs:
 100             if a in res_req:
 101                 if isinstance(res_req[a], int): # integer check
 102                     all_res_req[a].append(res_req[a])
 103                 else:
 104                     msg = SourceLine(res_req, a).makeError(
 105                     "Non-top-level ResourceRequirement in single container cannot have expressions")
 106                     exception_msgs.append(msg)
 107     if exception_msgs:
 108         raise WorkflowException("\n".join(exception_msgs))
 109     else:
 110         overall_res_req = {}
 111         for a in all_res_req:
 112             if all_res_req[a]:
 113                 if a in max_res_pars:
 114                     overall_res_req[a] = max(all_res_req[a])
 115                 elif a in sum_res_pars:
 116                     overall_res_req[a] = sum(all_res_req[a])
 117         if overall_res_req:
 118             overall_res_req["class"] = "ResourceRequirement"
 119         return cmap(overall_res_req)
 120
 121 class ArvadosWorkflow(Workflow):
 122     """Wrap cwltool Workflow to override selected methods."""
 123
 124     def __init__(self, arvrunner, toolpath_object, loadingContext):
 125         super(ArvadosWorkflow, self).__init__(toolpath_object, loadingContext)
 126         self.arvrunner = arvrunner
 127         self.wf_pdh = None
 128         self.dynamic_resource_req = []
 129         self.static_resource_req = []
 130         self.wf_reffiles = []
 131         self.loadingContext = loadingContext
 132
 133     def job(self, joborder, output_callback, runtimeContext):
 134
 135         check_cluster_target(self, self._init_job(joborder, runtimeContext), runtimeContext)
 136
 137         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
 138         if not req:
 139             return super(ArvadosWorkflow, self).job(joborder, output_callback, runtimeContext)
 140
 141         # RunInSingleContainer is true
 142
 143         with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
 144             if "id" not in self.tool:
 145                 raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
 146         document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
 147
 148         discover_secondary_files(self.tool["inputs"], joborder)
 149
 150         with Perf(metrics, "subworkflow upload_deps"):
 151             upload_dependencies(self.arvrunner,
 152                                 os.path.basename(joborder.get("id", "#")),
 153                                 document_loader,
 154                                 joborder,
 155                                 joborder.get("id", "#"),
 156                                 False)
 157
 158             if self.wf_pdh is None:
 159                 workflowobj["requirements"] = dedup_reqs(self.requirements)
 160                 workflowobj["hints"] = dedup_reqs(self.hints)
 161
 162                 packed = pack(document_loader, workflowobj, uri, self.metadata)
 163
 164                 builder = Builder(joborder,
 165                                   requirements=workflowobj["requirements"],
 166                                   hints=workflowobj["hints"],
 167                                   resources={})
 168
 169                 def visit(item):
 170                     for t in ("hints", "requirements"):
 171                         if t not in item:
 172                             continue
 173                         for req in item[t]:
 174                             if req["class"] == "ResourceRequirement":
 175                                 dyn = False
 176                                 for k in max_res_pars + sum_res_pars:
 177                                     if k in req:
 178                                         if isinstance(req[k], basestring):
 179                                             if item["id"] == "#main":
 180                                                 # only the top-level requirements/hints may contain expressions
 181                                                 self.dynamic_resource_req.append(req)
 182                                                 dyn = True
 183                                                 break
 184                                             else:
 185                                                 with SourceLine(req, k, WorkflowException):
 186                                                     raise WorkflowException("Non-top-level ResourceRequirement in single container cannot have expressions")
 187                                 if not dyn:
 188                                     self.static_resource_req.append(req)
 189
 190                 visit_class(packed["$graph"], ("Workflow", "CommandLineTool"), visit)
 191
 192                 if self.static_resource_req:
 193                     self.static_resource_req = [get_overall_res_req(self.static_resource_req)]
 194
 195                 upload_dependencies(self.arvrunner,
 196                                     runtimeContext.name,
 197                                     document_loader,
 198                                     packed,
 199                                     uri,
 200                                     False)
 201
 202                 # Discover files/directories referenced by the
 203                 # workflow (mainly "default" values)
 204                 visit_class(packed, ("File", "Directory"), self.wf_reffiles.append)
 205
 206
 207         if self.dynamic_resource_req:
 208             builder = Builder(joborder,
 209                               requirements=self.requirements,
 210                               hints=self.hints,
 211                               resources={})
 212
 213             # Evaluate dynamic resource requirements using current builder
 214             rs = copy.copy(self.static_resource_req)
 215             for dyn_rs in self.dynamic_resource_req:
 216                 eval_req = {"class": "ResourceRequirement"}
 217                 for a in max_res_pars + sum_res_pars:
 218                     if a in dyn_rs:
 219                         eval_req[a] = builder.do_eval(dyn_rs[a])
 220                 rs.append(eval_req)
 221             job_res_reqs = [get_overall_res_req(rs)]
 222         else:
 223             job_res_reqs = self.static_resource_req
 224
 225         with Perf(metrics, "subworkflow adjust"):
 226             joborder_resolved = copy.deepcopy(joborder)
 227             joborder_keepmount = copy.deepcopy(joborder)
 228
 229             reffiles = []
 230             visit_class(joborder_keepmount, ("File", "Directory"), reffiles.append)
 231
 232             mapper = ArvPathMapper(self.arvrunner, reffiles+self.wf_reffiles, runtimeContext.basedir,
 233                                    "/keep/%s",
 234                                    "/keep/%s/%s")
 235
 236             # For containers API, we need to make sure any extra
 237             # referenced files (ie referenced by the workflow but
 238             # not in the inputs) are included in the mounts.
 239             if self.wf_reffiles:
 240                 runtimeContext = runtimeContext.copy()
 241                 runtimeContext.extra_reffiles = copy.deepcopy(self.wf_reffiles)
 242
 243             def keepmount(obj):
 244                 remove_redundant_fields(obj)
 245                 with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
 246                     if "location" not in obj:
 247                         raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
 248                 with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
 249                     if obj["location"].startswith("keep:"):
 250                         obj["location"] = mapper.mapper(obj["location"]).target
 251                         if "listing" in obj:
 252                             del obj["listing"]
 253                     elif obj["location"].startswith("_:"):
 254                         del obj["location"]
 255                     else:
 256                         raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
 257
 258             visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
 259
 260             def resolved(obj):
 261                 if obj["location"].startswith("keep:"):
 262                     obj["location"] = mapper.mapper(obj["location"]).resolved
 263
 264             visit_class(joborder_resolved, ("File", "Directory"), resolved)
 265
 266             if self.wf_pdh is None:
 267                 adjustFileObjs(packed, keepmount)
 268                 adjustDirObjs(packed, keepmount)
 269                 self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
 270
 271         wf_runner = cmap({
 272             "class": "CommandLineTool",
 273             "baseCommand": "cwltool",
 274             "inputs": self.tool["inputs"],
 275             "outputs": self.tool["outputs"],
 276             "stdout": "cwl.output.json",
 277             "requirements": self.requirements+job_res_reqs+[
 278                 {"class": "InlineJavascriptRequirement"},
 279                 {
 280                 "class": "InitialWorkDirRequirement",
 281                 "listing": [{
 282                         "entryname": "workflow.cwl",
 283                         "entry": '$({"class": "File", "location": "keep:%s/workflow.cwl"})' % self.wf_pdh
 284                     }, {
 285                         "entryname": "cwl.input.yml",
 286                         "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 287                     }]
 288             }],
 289             "hints": self.hints,
 290             "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"],
 291             "id": "#"
 292         })
 293         return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)