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
  17
  18 import ruamel.yaml as yaml
  19
  20 from .runner import upload_dependencies, packed_workflow, upload_workflow_collection, trim_anonymous_location
  21 from .pathmapper import trim_listing
  22 from .arvtool import ArvadosCommandTool
  23 from .perf import Perf
  24
  25 logger = logging.getLogger('arvados.cwl-runner')
  26 metrics = logging.getLogger('arvados.cwl-runner.metrics')
  27
  28 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
  29                     submit_runner_ram=0, name=None):
  30
  31     packed = packed_workflow(arvRunner, tool)
  32
  33     adjustDirObjs(job_order, trim_listing)
  34     adjustFileObjs(job_order, trim_anonymous_location)
  35     adjustDirObjs(job_order, trim_anonymous_location)
  36
  37     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  38     for inp in main["inputs"]:
  39         sn = shortname(inp["id"])
  40         if sn in job_order:
  41             inp["default"] = job_order[sn]
  42
  43     if not name:
  44         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
  45
  46     upload_dependencies(arvRunner, name, tool.doc_loader,
  47                         packed, tool.tool["id"], False)
  48
  49     # TODO nowhere for submit_runner_ram to go.
  50
  51     body = {
  52         "workflow": {
  53             "name": name,
  54             "description": tool.tool.get("doc", ""),
  55             "definition":yaml.round_trip_dump(packed)
  56         }}
  57     if project_uuid:
  58         body["workflow"]["owner_uuid"] = project_uuid
  59
  60     if uuid:
  61         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
  62     else:
  63         call = arvRunner.api.workflows().create(body=body)
  64     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
  65
  66 def dedup_reqs(reqs):
  67     dedup = {}
  68     for r in reversed(reqs):
  69         if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
  70             dedup[r["class"]] = r
  71     return [dedup[r] for r in sorted(dedup.keys())]
  72
  73 class ArvadosWorkflow(Workflow):
  74     """Wrap cwltool Workflow to override selected methods."""
  75
  76     def __init__(self, arvrunner, toolpath_object, **kwargs):
  77         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
  78         self.arvrunner = arvrunner
  79         self.work_api = kwargs["work_api"]
  80         self.wf_pdh = None
  81
  82     def job(self, joborder, output_callback, **kwargs):
  83         kwargs["work_api"] = self.work_api
  84         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
  85         if req:
  86             with SourceLine(self.tool, None, WorkflowException):
  87                 if "id" not in self.tool:
  88                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
  89             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
  90
  91             with Perf(metrics, "subworkflow upload_deps"):
  92                 upload_dependencies(self.arvrunner,
  93                                     os.path.basename(joborder.get("id", "#")),
  94                                     document_loader,
  95                                     joborder,
  96                                     joborder.get("id", "#"),
  97                                     False)
  98
  99                 if self.wf_pdh is None:
 100                     workflowobj["requirements"] = dedup_reqs(self.requirements)
 101                     workflowobj["hints"] = dedup_reqs(self.hints)
 102
 103                     packed = pack(document_loader, workflowobj, uri, self.metadata)
 104
 105                     upload_dependencies(self.arvrunner,
 106                                         kwargs.get("name", ""),
 107                                         document_loader,
 108                                         packed,
 109                                         uri,
 110                                         False)
 111
 112             with Perf(metrics, "subworkflow adjust"):
 113                 joborder_keepmount = copy.deepcopy(joborder)
 114
 115                 def keepmount(obj):
 116                     with SourceLine(obj, None, WorkflowException):
 117                         if "location" not in obj:
 118                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
 119                     with SourceLine(obj, "location", WorkflowException):
 120                         if obj["location"].startswith("keep:"):
 121                             obj["location"] = "/keep/" + obj["location"][5:]
 122                             if "listing" in obj:
 123                                 del obj["listing"]
 124                         elif obj["location"].startswith("_:"):
 125                             del obj["location"]
 126                         else:
 127                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
 128
 129                 adjustFileObjs(joborder_keepmount, keepmount)
 130                 adjustDirObjs(joborder_keepmount, keepmount)
 131
 132                 if self.wf_pdh is None:
 133                     adjustFileObjs(packed, keepmount)
 134                     adjustDirObjs(packed, keepmount)
 135                     self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
 136
 137             wf_runner = cmap({
 138                 "class": "CommandLineTool",
 139                 "baseCommand": "cwltool",
 140                 "inputs": self.tool["inputs"],
 141                 "outputs": self.tool["outputs"],
 142                 "stdout": "cwl.output.json",
 143                 "requirements": workflowobj["requirements"]+[
 144                     {
 145                     "class": "InitialWorkDirRequirement",
 146                     "listing": [{
 147                             "entryname": "workflow.cwl",
 148                             "entry": {
 149                                 "class": "File",
 150                                 "location": "keep:%s/workflow.cwl" % self.wf_pdh
 151                             }
 152                         }, {
 153                             "entryname": "cwl.input.yml",
 154                             "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 155                         }]
 156                 }],
 157                 "hints": workflowobj["hints"],
 158                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
 159             })
 160             kwargs["loader"] = self.doc_loader
 161             kwargs["avsc_names"] = self.doc_schema
 162             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
 163         else:
 164             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)