sdk/cwl/arvados_cwl/arvworkflow.py

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