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, del_listing
  15 from .arvtool import ArvadosCommandTool
  16
  17 logger = logging.getLogger('arvados.cwl-runner')
  18
  19 def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
  20     upload_docker(arvRunner, tool)
  21
  22     document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
  23
  24     packed = pack(document_loader, workflowobj, uri, tool.metadata)
  25
  26     adjustDirObjs(job_order, del_listing)
  27
  28     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  29     for inp in main["inputs"]:
  30         sn = shortname(inp["id"])
  31         if sn in job_order:
  32             inp["default"] = job_order[sn]
  33
  34     name = os.path.basename(tool.tool["id"])
  35     upload_dependencies(arvRunner, name, document_loader,
  36                         packed, uri, False)
  37
  38     body = {
  39         "workflow": {
  40             "owner_uuid": project_uuid,
  41             "name": tool.tool.get("label", name),
  42             "description": tool.tool.get("doc", ""),
  43             "definition":yaml.safe_dump(packed)
  44         }}
  45
  46     if update_uuid:
  47         return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
  48     else:
  49         return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
  50
  51 class ArvadosWorkflow(Workflow):
  52     """Wrap cwltool Workflow to override selected methods."""
  53
  54     def __init__(self, arvrunner, toolpath_object, **kwargs):
  55         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
  56         self.arvrunner = arvrunner
  57         self.work_api = kwargs["work_api"]
  58
  59     def job(self, joborder, output_callback, **kwargs):
  60         kwargs["work_api"] = self.work_api
  61         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
  62         if req:
  63             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
  64
  65             workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
  66             workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
  67             packed = pack(document_loader, workflowobj, uri, self.metadata)
  68
  69             upload_dependencies(self.arvrunner,
  70                                 kwargs.get("name", ""),
  71                                 document_loader,
  72                                 packed,
  73                                 uri,
  74                                 False)
  75
  76             upload_dependencies(self.arvrunner,
  77                                 os.path.basename(joborder.get("id", "#")),
  78                                 document_loader,
  79                                 joborder,
  80                                 joborder.get("id", "#"),
  81                                 False)
  82
  83             joborder_keepmount = copy.deepcopy(joborder)
  84
  85             def keepmount(obj):
  86                 if obj["location"].startswith("keep:"):
  87                     obj["location"] = "/keep/" + obj["location"][5:]
  88                 elif obj["location"].startswith("_:"):
  89                     pass
  90                 else:
  91                     raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
  92                 if "listing" in obj:
  93                     del obj["listing"]
  94             adjustFileObjs(joborder_keepmount, keepmount)
  95             adjustDirObjs(joborder_keepmount, keepmount)
  96             adjustFileObjs(packed, keepmount)
  97             adjustDirObjs(packed, keepmount)
  98
  99             wf_runner = {
 100                 "class": "CommandLineTool",
 101                 "baseCommand": "cwltool",
 102                 "inputs": self.tool["inputs"],
 103                 "outputs": self.tool["outputs"],
 104                 "stdout": "cwl.output.json",
 105                 "requirements": workflowobj["requirements"]+[
 106                     {"class": "InlineJavascriptRequirement"},
 107                     {
 108                     "class": "InitialWorkDirRequirement",
 109                     "listing": [{
 110                             "entryname": "workflow.cwl",
 111                             "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 112                         }, {
 113                             "entryname": "cwl.input.yml",
 114                             "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 115                         }]
 116                 }],
 117                 "hints": workflowobj["hints"],
 118                 "arguments": ["--no-container", "--move-outputs", "workflow.cwl#main", "cwl.input.yml"]
 119             }
 120             kwargs["loader"] = self.doc_loader
 121             kwargs["avsc_names"] = self.doc_schema
 122             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
 123         else:
 124             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)