sdk/cwl/arvados_cwl/arvworkflow.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import os
   6 import json
   7 import copy
   8 import logging
   9
  10 from schema_salad.sourceline import SourceLine, cmap
  11
  12 from cwltool.pack import pack
  13 from cwltool.load_tool import fetch_document
  14 from cwltool.process import shortname
  15 from cwltool.workflow import Workflow, WorkflowException
  16 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class
  17
  18 import ruamel.yaml as yaml
  19
  20 from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
  21                      trim_anonymous_location, remove_redundant_fields, discover_secondary_files)
  22 from .pathmapper import ArvPathMapper, trim_listing
  23 from .arvtool import ArvadosCommandTool
  24 from .perf import Perf
  25
  26 logger = logging.getLogger('arvados.cwl-runner')
  27 metrics = logging.getLogger('arvados.cwl-runner.metrics')
  28
  29 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
  30                     submit_runner_ram=0, name=None):
  31
  32     packed = packed_workflow(arvRunner, tool)
  33
  34     adjustDirObjs(job_order, trim_listing)
  35     adjustFileObjs(job_order, trim_anonymous_location)
  36     adjustDirObjs(job_order, trim_anonymous_location)
  37
  38     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
  39     for inp in main["inputs"]:
  40         sn = shortname(inp["id"])
  41         if sn in job_order:
  42             inp["default"] = job_order[sn]
  43
  44     if not name:
  45         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
  46
  47     upload_dependencies(arvRunner, name, tool.doc_loader,
  48                         packed, tool.tool["id"], False)
  49
  50     # TODO nowhere for submit_runner_ram to go.
  51
  52     body = {
  53         "workflow": {
  54             "name": name,
  55             "description": tool.tool.get("doc", ""),
  56             "definition":yaml.round_trip_dump(packed)
  57         }}
  58     if project_uuid:
  59         body["workflow"]["owner_uuid"] = project_uuid
  60
  61     if uuid:
  62         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
  63     else:
  64         call = arvRunner.api.workflows().create(body=body)
  65     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
  66
  67 def dedup_reqs(reqs):
  68     dedup = {}
  69     for r in reversed(reqs):
  70         if r["class"] not in dedup and not r["class"].startswith("http://arvados.org/cwl#"):
  71             dedup[r["class"]] = r
  72     return [dedup[r] for r in sorted(dedup.keys())]
  73
  74 class ArvadosWorkflow(Workflow):
  75     """Wrap cwltool Workflow to override selected methods."""
  76
  77     def __init__(self, arvrunner, toolpath_object, **kwargs):
  78         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
  79         self.arvrunner = arvrunner
  80         self.work_api = kwargs["work_api"]
  81         self.wf_pdh = None
  82
  83     def job(self, joborder, output_callback, **kwargs):
  84         kwargs["work_api"] = self.work_api
  85         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
  86         if req:
  87             with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
  88                 if "id" not in self.tool:
  89                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
  90             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
  91
  92             discover_secondary_files(self.tool["inputs"], joborder)
  93
  94             with Perf(metrics, "subworkflow upload_deps"):
  95                 upload_dependencies(self.arvrunner,
  96                                     os.path.basename(joborder.get("id", "#")),
  97                                     document_loader,
  98                                     joborder,
  99                                     joborder.get("id", "#"),
 100                                     False)
 101
 102                 if self.wf_pdh is None:
 103                     workflowobj["requirements"] = dedup_reqs(self.requirements)
 104                     workflowobj["hints"] = dedup_reqs(self.hints)
 105
 106                     packed = pack(document_loader, workflowobj, uri, self.metadata)
 107
 108                     upload_dependencies(self.arvrunner,
 109                                         kwargs.get("name", ""),
 110                                         document_loader,
 111                                         packed,
 112                                         uri,
 113                                         False)
 114
 115             with Perf(metrics, "subworkflow adjust"):
 116                 joborder_resolved = copy.deepcopy(joborder)
 117                 joborder_keepmount = copy.deepcopy(joborder)
 118
 119                 reffiles = []
 120                 visit_class(joborder_keepmount, ("File", "Directory"), lambda x: reffiles.append(x))
 121
 122                 mapper = ArvPathMapper(self.arvrunner, reffiles, kwargs["basedir"],
 123                                  "/keep/%s",
 124                                  "/keep/%s/%s",
 125                                  **kwargs)
 126
 127                 def keepmount(obj):
 128                     remove_redundant_fields(obj)
 129                     with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
 130                         if "location" not in obj:
 131                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
 132                     with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
 133                         if obj["location"].startswith("keep:"):
 134                             obj["location"] = mapper.mapper(obj["location"]).target
 135                             if "listing" in obj:
 136                                 del obj["listing"]
 137                         elif obj["location"].startswith("_:"):
 138                             del obj["location"]
 139                         else:
 140                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
 141
 142                 visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
 143
 144                 def resolved(obj):
 145                     if obj["location"].startswith("keep:"):
 146                         obj["location"] = mapper.mapper(obj["location"]).resolved
 147
 148                 visit_class(joborder_resolved, ("File", "Directory"), resolved)
 149
 150                 if self.wf_pdh is None:
 151                     adjustFileObjs(packed, keepmount)
 152                     adjustDirObjs(packed, keepmount)
 153                     self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
 154
 155             wf_runner = cmap({
 156                 "class": "CommandLineTool",
 157                 "baseCommand": "cwltool",
 158                 "inputs": self.tool["inputs"],
 159                 "outputs": self.tool["outputs"],
 160                 "stdout": "cwl.output.json",
 161                 "requirements": self.requirements+[
 162                     {
 163                     "class": "InitialWorkDirRequirement",
 164                     "listing": [{
 165                             "entryname": "workflow.cwl",
 166                             "entry": {
 167                                 "class": "File",
 168                                 "location": "keep:%s/workflow.cwl" % self.wf_pdh
 169                             }
 170                         }, {
 171                             "entryname": "cwl.input.yml",
 172                             "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
 173                         }]
 174                 }],
 175                 "hints": self.hints,
 176                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
 177             })
 178             kwargs["loader"] = self.doc_loader
 179             kwargs["avsc_names"] = self.doc_schema
 180             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder_resolved, output_callback, **kwargs)
 181         else:
 182             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)