sdk/cwl/arvados_cwl/arvtool.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 from cwltool.command_line_tool import CommandLineTool, ExpressionTool
   6 from .arvcontainer import ArvadosContainer
   7 from .pathmapper import ArvPathMapper
   8 from .runner import make_builder
   9 from ._version import __version__
  10 from functools import partial
  11 from schema_salad.sourceline import SourceLine
  12 from cwltool.errors import WorkflowException
  13 from arvados.util import portable_data_hash_pattern
  14
  15 def validate_cluster_target(arvrunner, runtimeContext):
  16     if (runtimeContext.submit_runner_cluster and
  17         runtimeContext.submit_runner_cluster not in arvrunner.api._rootDesc["remoteHosts"] and
  18         runtimeContext.submit_runner_cluster != arvrunner.api._rootDesc["uuidPrefix"]):
  19         raise WorkflowException("Unknown or invalid cluster id '%s' known remote clusters are %s" % (runtimeContext.submit_runner_cluster,
  20                                                                                                      ", ".join(list(arvrunner.api._rootDesc["remoteHosts"].keys()))))
  21     if runtimeContext.project_uuid:
  22         cluster_target = runtimeContext.submit_runner_cluster or arvrunner.api._rootDesc["uuidPrefix"]
  23         if not runtimeContext.project_uuid.startswith(cluster_target):
  24             raise WorkflowException("Project uuid '%s' should start with id of target cluster '%s'" % (runtimeContext.project_uuid, cluster_target))
  25
  26         try:
  27             if runtimeContext.project_uuid[5:12] == '-tpzed-':
  28                 arvrunner.api.users().get(uuid=runtimeContext.project_uuid).execute()
  29             else:
  30                 proj = arvrunner.api.groups().get(uuid=runtimeContext.project_uuid).execute()
  31                 if proj["group_class"] != "project":
  32                     raise Exception("not a project, group_class is '%s'" % (proj["group_class"]))
  33         except Exception as e:
  34             raise WorkflowException("Invalid project uuid '%s': %s" % (runtimeContext.project_uuid, e))
  35
  36 def set_cluster_target(tool, arvrunner, builder, runtimeContext):
  37     cluster_target_req = None
  38     for field in ("hints", "requirements"):
  39         if field not in tool:
  40             continue
  41         for item in tool[field]:
  42             if item["class"] == "http://arvados.org/cwl#ClusterTarget":
  43                 cluster_target_req = item
  44
  45     if cluster_target_req is None:
  46         return runtimeContext
  47
  48     with SourceLine(cluster_target_req, None, WorkflowException, runtimeContext.debug):
  49         runtimeContext = runtimeContext.copy()
  50         runtimeContext.submit_runner_cluster = builder.do_eval(cluster_target_req.get("cluster_id")) or runtimeContext.submit_runner_cluster
  51         runtimeContext.project_uuid = builder.do_eval(cluster_target_req.get("project_uuid")) or runtimeContext.project_uuid
  52         validate_cluster_target(arvrunner, runtimeContext)
  53
  54     return runtimeContext
  55
  56
  57 class ArvadosCommandTool(CommandLineTool):
  58     """Wrap cwltool CommandLineTool to override selected methods."""
  59
  60     def __init__(self, arvrunner, toolpath_object, loadingContext):
  61         super(ArvadosCommandTool, self).__init__(toolpath_object, loadingContext)
  62
  63         (docker_req, docker_is_req) = self.get_requirement("DockerRequirement")
  64         if not docker_req:
  65             if portable_data_hash_pattern.match(loadingContext.default_docker_image):
  66                 self.hints.append({"class": "DockerRequirement",
  67                                    "http://arvados.org/cwl#dockerCollectionPDH": loadingContext.default_docker_image})
  68             else:
  69                 self.hints.append({"class": "DockerRequirement",
  70                                    "dockerPull": loadingContext.default_docker_image})
  71
  72         self.arvrunner = arvrunner
  73
  74     def make_job_runner(self, runtimeContext):
  75         if runtimeContext.work_api == "containers":
  76             return partial(ArvadosContainer, self.arvrunner, runtimeContext)
  77         else:
  78             raise Exception("Unsupported work_api %s", runtimeContext.work_api)
  79
  80     def make_path_mapper(self, reffiles, stagedir, runtimeContext, separateDirs):
  81         if runtimeContext.work_api == "containers":
  82             return ArvPathMapper(self.arvrunner, reffiles+runtimeContext.extra_reffiles, runtimeContext.basedir,
  83                                  "/keep/%s",
  84                                  "/keep/%s/%s")
  85
  86     def job(self, joborder, output_callback, runtimeContext):
  87         builder = make_builder(joborder, self.hints, self.requirements, runtimeContext, self.metadata)
  88         runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
  89
  90         if runtimeContext.work_api == "containers":
  91             dockerReq, is_req = self.get_requirement("DockerRequirement")
  92             if dockerReq and dockerReq.get("dockerOutputDirectory"):
  93                 runtimeContext.outdir = dockerReq.get("dockerOutputDirectory")
  94                 runtimeContext.docker_outdir = dockerReq.get("dockerOutputDirectory")
  95             else:
  96                 runtimeContext.outdir = "/var/spool/cwl"
  97                 runtimeContext.docker_outdir = "/var/spool/cwl"
  98         return super(ArvadosCommandTool, self).job(joborder, output_callback, runtimeContext)
  99
 100 class ArvadosExpressionTool(ExpressionTool):
 101     def __init__(self, arvrunner, toolpath_object, loadingContext):
 102         super(ArvadosExpressionTool, self).__init__(toolpath_object, loadingContext)
 103         self.arvrunner = arvrunner
 104
 105     def job(self,
 106             job_order,         # type: Mapping[Text, Text]
 107             output_callback,  # type: Callable[[Any, Any], Any]
 108             runtimeContext     # type: RuntimeContext
 109            ):
 110         return super(ArvadosExpressionTool, self).job(job_order, self.arvrunner.get_wrapped_callback(output_callback), runtimeContext)