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