10812: Handle workflow keep references.
[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py
1 import os
2 import json
3 import copy
4 import logging
5
6 from schema_salad.sourceline import SourceLine, cmap
7
8 from cwltool.pack import pack
9 from cwltool.load_tool import fetch_document
10 from cwltool.process import shortname
11 from cwltool.workflow import Workflow, WorkflowException
12 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
13
14 import ruamel.yaml as yaml
15
16 from .runner import upload_dependencies, trim_listing, packed_workflow
17 from .arvtool import ArvadosCommandTool
18 from .perf import Perf
19
20 logger = logging.getLogger('arvados.cwl-runner')
21 metrics = logging.getLogger('arvados.cwl-runner.metrics')
22
23 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
24                     submit_runner_ram=0, name=None):
25
26     packed = packed_workflow(arvRunner, tool)
27
28     adjustDirObjs(job_order, trim_listing)
29
30     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
31     for inp in main["inputs"]:
32         sn = shortname(inp["id"])
33         if sn in job_order:
34             inp["default"] = job_order[sn]
35
36     if not name:
37         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
38
39     upload_dependencies(arvRunner, name, tool.doc_loader,
40                         packed, tool.tool["id"], False)
41
42     # TODO nowhere for submit_runner_ram to go.
43
44     body = {
45         "workflow": {
46             "name": name,
47             "description": tool.tool.get("doc", ""),
48             "definition":yaml.round_trip_dump(packed)
49         }}
50     if project_uuid:
51         body["workflow"]["owner_uuid"] = project_uuid
52
53     if uuid:
54         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
55     else:
56         call = arvRunner.api.workflows().create(body=body)
57     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
58
59 class ArvadosWorkflow(Workflow):
60     """Wrap cwltool Workflow to override selected methods."""
61
62     def __init__(self, arvrunner, toolpath_object, **kwargs):
63         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
64         self.arvrunner = arvrunner
65         self.work_api = kwargs["work_api"]
66
67     def job(self, joborder, output_callback, **kwargs):
68         kwargs["work_api"] = self.work_api
69         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
70         if req:
71             with SourceLine(self.tool, None, WorkflowException):
72                 if "id" not in self.tool:
73                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
74             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
75
76             with Perf(metrics, "subworkflow upload_deps"):
77                 workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
78                 workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
79                 packed = pack(document_loader, workflowobj, uri, self.metadata)
80
81                 upload_dependencies(self.arvrunner,
82                                     kwargs.get("name", ""),
83                                     document_loader,
84                                     packed,
85                                     uri,
86                                     False)
87
88                 upload_dependencies(self.arvrunner,
89                                     os.path.basename(joborder.get("id", "#")),
90                                     document_loader,
91                                     joborder,
92                                     joborder.get("id", "#"),
93                                     False)
94
95             with Perf(metrics, "subworkflow adjust"):
96                 joborder_keepmount = copy.deepcopy(joborder)
97
98                 def keepmount(obj):
99                     with SourceLine(obj, None, WorkflowException):
100                         if "location" not in obj:
101                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
102                     with SourceLine(obj, "location", WorkflowException):
103                         if obj["location"].startswith("keep:"):
104                             obj["location"] = "/keep/" + obj["location"][5:]
105                             if "listing" in obj:
106                                 del obj["listing"]
107                         elif obj["location"].startswith("_:"):
108                             del obj["location"]
109                         else:
110                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
111
112                 adjustFileObjs(joborder_keepmount, keepmount)
113                 adjustDirObjs(joborder_keepmount, keepmount)
114                 adjustFileObjs(packed, keepmount)
115                 adjustDirObjs(packed, keepmount)
116
117             wf_runner = cmap({
118                 "class": "CommandLineTool",
119                 "baseCommand": "cwltool",
120                 "inputs": self.tool["inputs"],
121                 "outputs": self.tool["outputs"],
122                 "stdout": "cwl.output.json",
123                 "requirements": workflowobj["requirements"]+[
124                     {
125                     "class": "InitialWorkDirRequirement",
126                     "listing": [{
127                             "entryname": "workflow.cwl",
128                             "entry": yaml.round_trip_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
129                         }, {
130                             "entryname": "cwl.input.yml",
131                             "entry": yaml.round_trip_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
132                         }]
133                 }],
134                 "hints": workflowobj["hints"],
135                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
136             })
137             kwargs["loader"] = self.doc_loader
138             kwargs["avsc_names"] = self.doc_schema
139             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
140         else:
141             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)