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
  10 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
  11
  12 import ruamel.yaml as yaml
  13
  14 from .runner import upload_docker, upload_dependencies
  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     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  27     for inp in main["inputs"]:
  28         sn = shortname(inp["id"])
  29         if sn in job_order:
  30             inp["default"] = job_order[sn]
  31
  32     name = os.path.basename(tool.tool["id"])
  33     upload_dependencies(arvRunner, name, document_loader,
  34                         packed, uri, False)
  35
  36     body = {
  37         "workflow": {
  38             "owner_uuid": project_uuid,
  39             "name": tool.tool.get("label", name),
  40             "description": tool.tool.get("doc", ""),
  41             "definition":yaml.safe_dump(packed)
  42         }}
  43
  44     if update_uuid:
  45         return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
  46     else:
  47         return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
  48
  49 class ArvadosWorkflow(Workflow):
  50     """Wrap cwltool Workflow to override selected methods."""
  51
  52     def __init__(self, arvrunner, toolpath_object, **kwargs):
  53         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
  54         self.arvrunner = arvrunner
  55         self.work_api = kwargs["work_api"]
  56
  57     def job(self, joborder, output_callback, **kwargs):
  58         kwargs["work_api"] = self.work_api
  59         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
  60         if req:
  61             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
  62
  63             workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
  64             workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
  65             packed = pack(document_loader, workflowobj, uri, self.metadata)
  66
  67             upload_dependencies(self.arvrunner,
  68                                 kwargs.get("name", ""),
  69                                 document_loader,
  70                                 packed,
  71                                 uri,
  72                                 False)
  73
  74             upload_dependencies(self.arvrunner,
  75                                 os.path.basename(joborder.get("id", "#")),
  76                                 document_loader,
  77                                 joborder,
  78                                 joborder.get("id", "#"),
  79                                 False)
  80
  81             joborder_keepmount = copy.deepcopy(joborder)
  82
  83             def keepmount(obj):
  84                 if obj["location"].startswith("keep:"):
  85                     obj["location"] = "/keep/" + obj["location"][5:]
  86                 else:
  87                     raise Exception("Uh oh %s" % obj["location"])
  88                 if "listing" in obj:
  89                     del obj["listing"]
  90             adjustFileObjs(joborder_keepmount, keepmount)
  91             adjustDirObjs(joborder_keepmount, keepmount)
  92             adjustFileObjs(packed, keepmount)
  93             adjustDirObjs(packed, keepmount)
  94
  95             wf_runner = {
  96                 "class": "CommandLineTool",
  97                 "baseCommand": "cwltool",
  98                 "inputs": self.tool["inputs"],
  99                 "outputs": self.tool["outputs"],
 100                 "stdout": "cwl.output.json",
 101                 "requirements": workflowobj["requirements"]+[
 102                     {"class": "InlineJavascriptRequirement"},
 103                     {
 104                     "class": "InitialWorkDirRequirement",
 105                     "listing": [{
 106                             "entryname": "workflow.cwl",
 107                             "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 108                         }, {
 109                             "entryname": "cwl.input.yml",
 110                             "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 111                         }]
 112                 }],
 113                 "hints": workflowobj["hints"],
 114                 "arguments": ["--no-container", "--move-outputs", "workflow.cwl#main", "cwl.input.yml"]
 115             }
 116             kwargs["loader"] = self.doc_loader
 117             kwargs["avsc_names"] = self.doc_schema
 118             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
 119         else:
 120             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)