sdk/cwl/arvados_cwl/arvworkflow.py

   1 import os
   2 import json
   3 import copy
   4 import logging
   5
   6 from schema_salad.sourceline import SourceLine, cmap
   7
   8 from cwltool.pack import pack
   9 from cwltool.load_tool import fetch_document
  10 from cwltool.process import shortname
  11 from cwltool.workflow import Workflow, WorkflowException
  12 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
  13
  14 import ruamel.yaml as yaml
  15
  16 from .runner import upload_dependencies, packed_workflow, upload_workflow_collection, trim_anonymous_location
  17 from .pathmapper import trim_listing
  18 from .arvtool import ArvadosCommandTool
  19 from .perf import Perf
  20
  21 logger = logging.getLogger('arvados.cwl-runner')
  22 metrics = logging.getLogger('arvados.cwl-runner.metrics')
  23
  24 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
  25                     submit_runner_ram=0, name=None):
  26
  27     packed = packed_workflow(arvRunner, tool)
  28
  29     adjustDirObjs(job_order, trim_listing)
  30     adjustFileObjs(job_order, trim_anonymous_location)
  31     adjustDirObjs(job_order, trim_anonymous_location)
  32
  33     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  34     for inp in main["inputs"]:
  35         sn = shortname(inp["id"])
  36         if sn in job_order:
  37             inp["default"] = job_order[sn]
  38
  39     if not name:
  40         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
  41
  42     upload_dependencies(arvRunner, name, tool.doc_loader,
  43                         packed, tool.tool["id"], False)
  44
  45     # TODO nowhere for submit_runner_ram to go.
  46
  47     body = {
  48         "workflow": {
  49             "name": name,
  50             "description": tool.tool.get("doc", ""),
  51             "definition":yaml.round_trip_dump(packed)
  52         }}
  53     if project_uuid:
  54         body["workflow"]["owner_uuid"] = project_uuid
  55
  56     if uuid:
  57         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
  58     else:
  59         call = arvRunner.api.workflows().create(body=body)
  60     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
  61
  62 def dedup_reqs(reqs):
  63     dedup = {}
  64     for r in reversed(reqs):
  65         if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
  66             dedup[r["class"]] = r
  67     return [dedup[r] for r in sorted(dedup.keys())]
  68
  69 class ArvadosWorkflow(Workflow):
  70     """Wrap cwltool Workflow to override selected methods."""
  71
  72     def __init__(self, arvrunner, toolpath_object, **kwargs):
  73         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
  74         self.arvrunner = arvrunner
  75         self.work_api = kwargs["work_api"]
  76         self.wf_pdh = None
  77
  78     def job(self, joborder, output_callback, **kwargs):
  79         kwargs["work_api"] = self.work_api
  80         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
  81         if req:
  82             with SourceLine(self.tool, None, WorkflowException):
  83                 if "id" not in self.tool:
  84                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
  85             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
  86
  87             with Perf(metrics, "subworkflow upload_deps"):
  88                 upload_dependencies(self.arvrunner,
  89                                     os.path.basename(joborder.get("id", "#")),
  90                                     document_loader,
  91                                     joborder,
  92                                     joborder.get("id", "#"),
  93                                     False)
  94
  95                 if self.wf_pdh is None:
  96                     workflowobj["requirements"] = dedup_reqs(self.requirements)
  97                     workflowobj["hints"] = dedup_reqs(self.hints)
  98
  99                     packed = pack(document_loader, workflowobj, uri, self.metadata)
 100
 101                     upload_dependencies(self.arvrunner,
 102                                         kwargs.get("name", ""),
 103                                         document_loader,
 104                                         packed,
 105                                         uri,
 106                                         False)
 107
 108             with Perf(metrics, "subworkflow adjust"):
 109                 joborder_keepmount = copy.deepcopy(joborder)
 110
 111                 def keepmount(obj):
 112                     with SourceLine(obj, None, WorkflowException):
 113                         if "location" not in obj:
 114                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
 115                     with SourceLine(obj, "location", WorkflowException):
 116                         if obj["location"].startswith("keep:"):
 117                             obj["location"] = "/keep/" + obj["location"][5:]
 118                             if "listing" in obj:
 119                                 del obj["listing"]
 120                         elif obj["location"].startswith("_:"):
 121                             del obj["location"]
 122                         else:
 123                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
 124
 125                 adjustFileObjs(joborder_keepmount, keepmount)
 126                 adjustDirObjs(joborder_keepmount, keepmount)
 127
 128                 if self.wf_pdh is None:
 129                     adjustFileObjs(packed, keepmount)
 130                     adjustDirObjs(packed, keepmount)
 131                     self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
 132
 133             wf_runner = cmap({
 134                 "class": "CommandLineTool",
 135                 "baseCommand": "cwltool",
 136                 "inputs": self.tool["inputs"],
 137                 "outputs": self.tool["outputs"],
 138                 "stdout": "cwl.output.json",
 139                 "requirements": workflowobj["requirements"]+[
 140                     {
 141                     "class": "InitialWorkDirRequirement",
 142                     "listing": [{
 143                             "entryname": "workflow.cwl",
 144                             "entry": {
 145                                 "class": "File",
 146                                 "location": "keep:%s/workflow.cwl" % self.wf_pdh
 147                             }
 148                         }, {
 149                             "entryname": "cwl.input.yml",
 150                             "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 151                         }]
 152                 }],
 153                 "hints": workflowobj["hints"],
 154                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
 155             })
 156             kwargs["loader"] = self.doc_loader
 157             kwargs["avsc_names"] = self.doc_schema
 158             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
 159         else:
 160             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)